3.

1 Remote Sensing and GIScience in Geomorphology: Introduction

and Overview
MP Bishop, Texas A&M University, College Station, TX, USA
r 2013 Elsevier Inc. All rights reserved.

3.1.1 Introduction 3
3.1.2 Geospatial Technology and Fieldwork 3
3.1.3 Remote Sensing and Geomorphology 5
3.1.3.1 Photography and Videography 5
3.1.3.2 Imaging Spectroscopy 6
3.1.3.2.1 Sensor parameters 6
3.1.3.2.2 Reflectance properties and applications 7
3.1.3.3 Microwave Remote Sensing 10
3.1.3.4 The Atmosphere and Climate Forcing 10
3.1.3.5 Land-Cover Assessment and Mapping 12
3.1.3.6 Near-Surface Geophysics 12
3.1.4 GIS and Geomorphology 12
3.1.4.1 Digital Terrain Modeling (DTM) 13
3.1.4.2 Terrain Analysis 13
3.1.4.3 Landform Mapping 14
3.1.4.4 Spatial Hydrology 15
3.1.4.5 Erosion Modeling 17
3.1.4.6 Natural Hazards 18
3.1.4.7 Visualization 19
3.1.5 Conclusions 20
References 21

Glossary philosophical, cognitive, and scientific treatments of spatio-

Digital elevation model A digital elevation model (DEM)
is generally a land-surface model that attempts to accurately
portray the altitude field of the topography. In
geomorphology, it commonly takes the form of a
raster data layer representing a field of square
tessellations. The grid cell resolution is based upon
the source data and the desired scale for representing the
topography.
Digital terrain modeling Digital terrain modeling (DTM)
refers to a workflow process of acquiring data that samples
the altitude field, preprocessing the data to generate a digital
elevation model, and error and uncertainty analysis to
identify and remove systematic and random errors.
Geographic information science Geographic
information science (GIScience) is an emerging
multidisciplinary field that attempts to understand the
nature of spatio-temporal information, with a focus on
geographic representation, spatial analysis and modeling,
and addressing scientific problems. It represents a body of
geographical and technical knowledge concerned with
temporal theory and concepts. GIScience is often used as an
umbrella term to refer to developments in a variety of fields
that have contributed theory, concepts, and new
information technology. It also refers to those disciplines
involved in the investigation of the utility of information
technology for applied information production and
problem solving.
Geographic information systems A geographic
information system (GIS) is a software system that can be
used to store, manage, manipulate, analyze, and display
spatially referenced data. It can also be thought of as a
decision support system, as spatial data are analyzed to
produce information that is used to support management
and planning decisions.
Geomorphological mapping Geomorphological
mapping is a general term that refers to mapping various
aspects of the geomorphological system. This can include
numerous themes such as land systems, hydrology, surface
material and structure, morphometry, sediment transfer,
surface-process regimes, and chronology.

Bishop, M.P., 2013. Remote sensing and GIScience in geomorphology:

introduction and overview. In: Shroder, J. (Editor in Chief), Bishop, M.P.
(Ed.), Treatise on Geomorphology. Academic Press, San Diego, CA, vol. 3,
Remote Sensing and GIScience in Geomorphology, pp. 1–24.

Treatise on Geomorphology, Volume 3 http://dx.doi.org/10.1016/B978-0-12-374739-6.00040-3 1

2 Remote Sensing and GIScience in Geomorphology: Introduction and Overview

Geomorphometry The discipline that is concerned with
the science of quantitative land-surface characterization.
Imaging spectroscopy Imaging spectroscopy refers to the
collection of surface spectra from a spectrally contiguous
region of the electromagnetic spectrum that includes the
visible, near-infrared, and short-wave infrared regions of the
electromagnetic spectrum. The spectra are collected using a
very narrow spectral interval (8–12 nm) and hundreds of
images are usually acquired. This allows detailed
biophysical assessment and mapping of the Earth surface.
Imaging spectroscopy is also commonly called imaging
spectrometry or hyperspectral remote sensing, as the
dimensionality of spectral data (images) is very high. The
accuracy of the surface spectra for each pixel is dependent
on the spatial, spectral, and radiometric characteristics of
the sensor.
Land-surface objects Land-surface objects are spatial
entities that represent a meaningful segmentation of the
Earth’s surface. They are generated from land-surface
parameters using a variety of algorithms and methods, and
generally relate to the morphology of the topography in
terms of landform elements, features, or functional units.
Terrain segmentation can also relate to topographic
position and structure, surface material, or process
domains, if properly defined and delineated.
Land-surface parameters Land-surface parameters are
also called geomorphometric parameters, and they attempt
to quantitatively characterize various aspects of the
topography. They can be defined and classified based on
geometry, scale, and by surface-process characterization.
They are used to generate land-surface objects and
characterize process mechanics in surface-process modeling.
A variety of parameters such as slope, slope azimuth,
curvature, surface roughness, and relief are used for
studying geomorphological systems and for
geomorphological mapping.
Light detection and ranging Light detection and ranging
(LiDAR) is a form of active remote sensing that is based on
laser-light technology to obtain information about the
Earth surface. LiDAR data can be used to produce high-
resolution digital elevation models that attempt to
characterize a surface or a bare-earth representation of the
topography.
Near infrared The near-infrared (NIR) represents a region
of the electromagnetic spectrum that ranges from 0.7 to
1.3 mm. Optical sensors record NIR light reflected from the
surface. NIR imagery can be used to assess and map
vegetation, water, and other environmental characteristics.
Object-oriented analysis Object-oriented analysis refers
to the analysis of surface objects to determine their inherent
properties. Terrain or image segmentation is first required to
group tessellations into meaningful spatial entities that
represent land-cover or terrain features. Spatial analysis is
required to isolate surface objects at a variety of scales.
Object-oriented analysis is then used to characterize object
properties such as size, shape, and topographic conditions.
Spatial analysis of objects can then provide detailed
information about the context and topological
relationships among surface objects. This analysis has been
found to be superior to pixel-based methods for mapping a
variety of Earth-surface features.
Short-wave infrared The short-wave infrared (SWIR)
represents a region of the electromagnetic spectrum that
ranges from 1.3 to 3.0 mm. Optical sensors record SWIR
light reflected from the surface. SWIR imagery is routinely
used for snow and ice, mineralogical, and lithological
mapping.
Spatial hydrology Spatial hydrology refers to the linking
of remote sensing and GIS with hydrological modeling
efforts. Numerous levels of coupling of GIScience and
hydrology exist, although a major focus has been on using
various forms of data and methods to estimate key
hydrological parameters that drive hydrological models.
Spectral absorption feature A spectral absorption feature
is a wavelength-dependent feature of a spectral reflectance
curve. It is characterized by a localized decrease in reflectance
due to the absorption of energy by surface matter, given its
composition and chemistry. The prominence of the
absorption feature is determined by its depth, width, and
shape. A spectral curve can exhibit numerous spectral
absorption features depending on sensor-system
characteristics and surface compositional variations. Many
surface materials such as biochemicals or minerals have
diagnostic absorption features that can be used to assess
surface biophysical characteristics. Imaging spectroscopy
makes use of spectral absorption features to assess and map
biophysical conditions on the Earth’s surface.
Thermal infrared The thermal infrared (TIR) represents a
region of the electromagnetic spectrum that effectively ranges
from 3.0 to 15 mm. This region of the spectrum is used to
assess the thermal properties of the landscape and emission is
related to surface temperature. Thermal imagery is used in a
variety of disciplines including geomorphology, volcanology,
glaciology, and for mineralogical/lithological mapping.

Abstract

Geospatial technologies are having a profound effect on geomorphology. Remote sensing and geographic information
system studies are now commonplace in Earth science investigations. Significant advances have occurred in sensors,
geodesy, photogrammetry, geophysics, computer science, statistics, and pattern recognition. Consequently, it is now pos-
sible to quantify landscape morphology, investigate climate forcing, link process with patterns and form, and enhance our
understanding of scale dependence and the polygenetic nature of landscape evolution. This chapter introduces current
capabilities and new developments that are relevant to geomorphological investigations. The emphasis is on using data and
new analysis approaches to better understand geomorphological systems and landforms.
 Remote Sensing and GIScience in Geomorphology: Introduction and Overview
3.1.1 Introduction
Earth science investigations are increasingly utilizing geospatial
technologies (Bishop and Shroder, 2004b; Hengl and Reuter,
2009; Bishop et al., 2012). The rapid proliferation of geospatial
technologies is related to advances in geodesy, photogram-
metry, geophysics, computer science, statistics, remote sensing,
geographic information technology (GIT), and numerical
modeling, which have collectively revolutionized the field of
geomorphology (Bishop et al., 2001; Shroder and Bishop,
2003; Bishop and Shroder, 2004a;). Scientists are routinely
utilizing new spatio-temporal data, geocomputational algo-
rithms, and processing approaches and models, which now
allow assessments far beyond traditional geomorphological
mapping. It is now possible to quantify landscape morphology
(Pike, 2000; Hengl and Reuter, 2009), assess surface biophysical
conditions (Florinsky, 1998; Liang, 2007; Smith and Pain,
2009; Tarolli et al., 2009), assess near-surface conditions, link
process with form and patterns (Allen and Walsh, 1993;
Montgomery et al., 2004), and improve our understanding of
scale dependence and the polygenetic nature of landscape
evolution (Walsh et al., 1997; Tate and Wood, 2001; Bishop
et al., 2003). Nevertheless, such rapid utilization of information
technologies must be carefully examined, given the empirical
nature of utilizing geographic information systems (GIS), the
need for formalization to address numerous issues, and re-
quirements of accuracy and repeatability (Bishop et al., 2012).
Numerous conceptual/theoretical and methodological
issues are at the heart of effectively utilizing spatio-temporal
data and GIT to study geomorphological systems (Bishop
et al., 2012). Therefore, Earth scientists need to be fully aware
of current capabilities as well as the issues and challenges re-
lated to geomorphology and geographic information science
(GIScience) (Bishop and Shroder, 2004b; Bishop et al., 2012).
Geospatial technologies can be used to address various con-
ceptual and practical issues such as heterogeneous surface
composition with fuzzy-classification membership (Warner
and Shank, 1997), indeterminant boundaries and features
(Burrough, 1989; Usery, 1996; Burrough et al., 2000; Smith
et al., 2000; Deng and Wilson, 2008), hierarchical organiza-
tion and spatial analysis using object-oriented technology
(Ralston, 1994; Brändli, 1996; Schmidt and Dikau, 1999),
scale dependence of properties and patterns using geostatistics
(Tate and Wood, 2001), and objective mapping using different
analytical approaches (e.g., descriptive statistics, inferential
statistics, artificial intelligence, and various analytical rea-
soning technologies).
Nevertheless, numerous limitations are associated with the
use of existing cartographic representations of environmental
and geomorphological information, as parameterization
schemes that uniquely characterize the formal structure of
natural systems and information associations (i.e., attributes,
space, and time), and the formal linkage of multiple processes
and forms in space–time is not readily available. Con-
sequently, cartographic representational schemes are still ef-
fectively being used, although more complex representational
schemes typically occur outside of GIS environments. Ad-
dressing numerous science issues will most likely require
multidisciplinary collaboration between Earth and infor-
mation scientists (Bishop and Shroder, 2004b).
3
From a more methodological perspective, advances in
the development and deployment of new sensors and mobile
platforms will provide for new forms of data across a multi-
tude of scales. This has enabled researchers to address nu-
merous issues at the requisite scale. Active research areas
include the assessment of surface processes and their impact
on erosion and landscape evolution (e.g., Finlayson et al.,
2002; Finlayson and Montgomery, 2003) and the evaluation
of data and methods for geomorphological mapping (e.g.,
Saadat et al., 2008; Schneevoigt et al., 2008). The role of re-
mote sensing in generating high-quality digital elevation
model (DEMs) is essential (Wilson and Gallant, 2000), as the
topography inherently incorporates a multitude of morphol-
ogies that are generated due to the interaction of climate,
tectonic, and surface processes. The development and evalu-
ation of new techniques and analytical approaches for infor-
mation extraction from multispectral, multitemporal, and
DEMs is an active research area (Bishop and Shroder, 2004b;
Bishop et al., 2012). Investigators have focused on the tech-
nical aspects of developing GIS databases (Gustavsson et al.,
2008), developing geomorphometric mapping software
(Klingseisen et al., 2008), mapping specific landform features,
and developing new ways to visualize geomorphological
information (Vitek et al., 2008). A plethora of quantitative
metrics and approaches exist. The advantages and limitations,
however, of numerous algorithms and multistage processing
approaches have not been rigorously evaluated and compared
for specific applications. Furthermore, standardized for-
malizations for specific applications are urgently needed.
Consequently, numerous issues related to taxonomy, scale,
process mechanics, feedback mechanisms, system dynamics
and states, representational schemes, algorithms and process-
ing protocols, visualization of complex information, and
effective information distribution need to be accounted for.
The objective of this chapter is to place the current volume
into perspective and provide an overview of remote sensing
and GIScience contributions to geomorphology and the Earth
sciences in general. It is important that significant capabilities
be highlighted so that scientific inquiry may be facilitated by
the use of spatio-temporal data and GIT. It is also necessary to
highlight the challenges and issues associated with the use and
evolution of GIT, so that rapidly evolving capabilities effect-
ively address scientific and practical issues. Given the ex-
ceptionally diverse fields of study that encompass Earth
science and GIScience, it is not possible to cover all develop-
ments in a comprehensive fashion. Rather, the topics that have
the most direct significance for understanding geomorpho-
logical systems and landforms have been selected.
3.1.2 Geospatial Technology and Fieldwork
Fieldwork has been and will continue to be an important as-
pect of scientific inquiry and mapping. The nature of field
equipment has radically changed from analog to digital de-
vices, and automated sensors and systems ensure specific
levels of accuracy (e.g., location, spectra-based mineral iden-
tification, microclimate). Field sensors can be used for re-
cording many different landscape parameters related to
microclimate, ecological, hydrological, sedimentological, and
4 Remote Sensing and GIScience in Geomorphology: Introduction and Overview
lithological systems. Telecommunications and computers
allow the rapid collection and analysis, whereas mobile sensor
platforms allow expanded coverage.
Relatively new quantitative spatio-temporal data are related
to advances in geodesy. A significant milestone was the de-
velopment of the global positioning system (GPS). A con-
stellation of GPS satellites communicate their position with
each other and ground receivers, such that attribute infor-
mation collected in the field is tagged with precise spatio-
temporal coordinates. Spatial and temporal information can
be used to compute distances, directions, and rates of change.
Consequently, GPS technology is routinely used in fieldwork,
and GPS data have significantly contributed to the under-
standing of environmental change, glaciation and glacieriza-
tion, tectonics, and facilitate soils and landform mapping
(e.g., Bilham et al., 1997; Gao and Liu, 2001; Banerjee and
Bürgmann, 2002; Sella et al., 2007; Bishop et al., 2010; Flowers
et al., 2011).
Similarly, advances in laser, computer, and communication
technology have resulted in new surveying equipment.
Field devices include laser-based total stations, laser range-
finders, hand-held spectral radiometers, radiation sensors, and
many other portable sensors that transfer data to loggers or
satellites for subsequent downloading. Interactive data loggers
can be used to visually examine digital aerial photography or
satellite imagery in the field, and GIS data layers can also be
viewed to facilitate fieldwork. This allows improved data col-
lection for geomorphological mapping, sampling, and tar-
geted investigations.
Terrestrial laser scanners can now be used to study surface
processes in great detail. These laser-scanning devices can
be mounted on tripods, jeeps, or terrain rovers to collect
high-resolution 3-D point clouds (Figure 1). High-resolution
DEMs can be generated over relatively short time intervals.
Rates of erosion, deposition, and ablation can be estimated by
comparing altitude values over time. Furthermore, process
mechanics can be studied, as the relationships between pro-
cess, form and topographic evolution can be quantitatively
characterized.
New developments in field-based radiation sensors also
facilitate field and remote sensing studies of geomorpho-
logical systems. Investigators can use field-based spectral
radiometers and spectra-based mineralogical identification
systems to collect spectra in the field and identify the miner-
alogical composition of rocks, sediments, and soils. Specific
biophysical characteristics of surface materials include primary
silicate minerals, secondary silicate minerals, water content,
and organic matter. Furthermore, samples of material col-
lected in the field can be brought back to the laboratory for
subsequent spectral analysis.
Other radiation sensors are used to measure the magnitude
of incoming and outgoing radiation that governs many surface
processes and the surface-energy budget. Consequently,
pyranometers, pyrheliometers, albedometers, and shortwave
and longwave net radiometers are used to measure various
components of the radiation-transfer cascade. Microclimatic
conditions influence weathering, ablation and melt water
production, glacier erosion, avalanching, and other mass
movement processes. Such field-based investigations are crit-
ical for evaluating the information content in multispectral
Figure 1 Terrestrial laser scanning at the Chalk Cliffs Natural Debris
Flow Laboratory (Summer 2010). A Leica HDS C10 laser scanner was
used for repeat surveying. Pictured in the image are Thad Wasklewicz
(yellow helmet) and Dennis Staley (red helmet). Photograph courtesy
of Thad Wasklewicz, East Carolina University.
satellite imagery and validating the accuracy of thematic in-
formation generated from satellite-derived data and numerical
models. Such validations (e.g., surface temperature and albedo
estimates from imagery) provide new opportunities for geo-
morphological assessment via remote sensing.
Given the increasing miniaturization of sensors and
near-field communication capabilities, investigators are
recognizing the potential of establishing field-site sensor net-
works to characterize and monitor changes on the landscape.
The parameter list for such networks is ever increasing
and potentially includes atmospheric variables, photography,
and surface parameters. Depending on the spatial density of
station sensors, data can be transmitted to satellites and
downloaded to produce spatio-temporal information layers
via spatial interpolation. Future developments in sensor net-
works will most likely include robotic rovers, where infor-
mation collected from the network governs the location of
sensors, such that the network adapts to environmental con-
ditions in an attempt to better characterize the spatio-tem-
poral variability of phenomena being collected by the sensor
network.
Communication technology now allows ground control of
unmanned ground and aerial vehicles, which provides new
opportunities for collecting data in the field. The University of
Arizona, in a project led by Jeffrey Kargel, and assisted by the
Kathmandu-based International Centre for Integrated Moun-
tain Development (ICMOD), has conducted feasibility studies
for the use of remotely controlled boats equipped with side
scan sonar and other instruments to study glacial lakes
(Figure 2). Featured here is a boat deployment on Imja Lake
(Imja Glacier in the background), near Mount Everest, Nepal.
These studies have investigated lake bathymetry and sus-
pended sediments, and density stratification of the water col-
umn. Routine deployments in hazardous or dangerous lakes
are the goal of these studies. The potential advantage of boat
deployments, as opposed to conventional human-piloted boat
deployments, is that the small watercrafts are more readily
transported to remote locations and their deployment leaves
the scientific crew safely onshore.
 Remote Sensing and GIScience in Geomorphology: Introduction and Overview 5

Figure 2 Remote-controlled boat on Imja Lake (Imja Glacier in the background), near Mount Everest, Nepal. The boat has a sensor payload that
allows the acquisition of bathymetric information, video, and water temperature. Such mobile platforms and sensors are useful for studying
supraglacial lake development and monitoring flood hazards associated with proglacial lakes and landslide-valley impoundments that cause
catastrophic flooding. Photography courtesy of Jeffrey Kargel, University of Arizona.

Deployment of terrain rovers with sensors is also being
tested at the University of Arizona by Wolfgang Fink. Research
is aimed toward the development and deployment of truly
autonomous observing systems. These platforms and sensors
involve semi-autonomous navigation, where input from
scientists would be greatly reduced. Such technology and
equipment involves advancements in semi-automated data
analysis and autonomous scientific interpretation and de-
cision making using various forms of artificial intelligence.
Other possibilities also include aircraft and gliders with vari-
ous sensor arrays.
Supplementary material related to this article can be found
online at doi:10.1016/B978-0-08-088523-0.00040-X.
A host of different geoscience fieldwork applications are
also becoming available for field-portable laptop computers,
wherein checklists for data collection, section measuring, and
other typical field data can be collected electronically. Given
many new possibilities, fieldwork is becoming more complex
and automated. Special attention must be paid to the choice
and suitability of sensors and technologies, sensor-system
deployment and sampling, key data parameters, and storage
issues. Collectively, geospatial technologies have greatly en-
hanced fieldwork and data collection.
3.1.3 Remote Sensing and Geomorphology
Advances in optical, microwave, and light detection and ran-
ging (LiDAR) remote sensing now allow a multitude of
landscape information to be acquired. Data and information
characteristics are closely tied to sensor characteristics that
must be carefully considered. Improvements in spatial reso-
lution enable better recognition of small-scale objects. A re-
cent trend is high-resolution imagery which is now available
via commercial systems, although cost is an issue for many.
Similarly, the evolution of sensors has brought about an in-
crease in spectral resolution that includes measuring more
regions of the electromagnetic spectrum with improved spec-
tral bandwidths. Hyperspectral sensors, or imaging spec-
trometers, record hundreds of spectral images using 8–12 nm
bandwidth ranges. Better radiometric resolution (10–12 bits)
also allows sensors to be more sensitive to reflectance and
emission variations. Microwave remote sensing has also seen
significant advancements, and LiDAR remote sensing provides
numerous possibilities for the study of the atmosphere and
landscape. Consequently, a plethora of spatio-temporal data
can be used by Earth scientists.
3.1.3.1 Photography and Videography
Ground photography has historically played a significant role
in conducting fieldwork. Pictures of landforms, topography,
surface cover, and environmental conditions have been used
to document landscape features, and provide the basis for
qualitative interpretations of landforms and surface processes.
Photography, based on the chemical processing of film, was
mainstream for a considerable amount of time. Technological
advances have recently resulted in the use of ‘digital photog-
raphy’ and fieldwork now involves the use of digital cameras
that generate megapixel images (Figure 3). Nevertheless, the
ability to acquire repeat ground ‘photography’ is important in
documenting the Earth’s surface and for change detection.
Classic examples involve monitoring lake water levels, glacier
advancement and retreat, soil erosion, vegetation succession,
mass movements, and in general, landscape evolution (given
sufficient time).
Digital camera technology is rapidly advancing and it will
soon be possible to obtain giga-pixel digital images. Research
is underway to produce a digital camera that can take pan-
oramic images. A single-shot image of a landscape would
allow a detailed examination of the entire landscape, as such
high-resolution systems generate detailed data for subareas
within the panoramic scene. This technology, coupled with
photogrammetry and GPS technology, will markedly improve
6 Remote Sensing and GIScience in Geomorphology: Introduction and Overview
Figure 3 Ground ‘photograph’ of the north-facing Braldu Valley in
the central Karakoram Himalaya of Pakistan. Glaciation has had a
profound influence on the topographic evolution of the Karakoram,
and such data document glacially polished valley walls, high-altitude
erosion surfaces, and a variety of depositional features associated
with mass movement and catastrophic flooding. Fieldwork and
ground photography allow geomorphological mapping and
reconstruction of significant erosion events.
our ability to comprehensively document the landscape as
never before. Critical issues involving data acquisition include
data volume and storage, computer memory, and processor
speeds.
Such technology will inevitably be placed on mobile
platforms, thereby providing detailed spatial coverage. One
may also envision giga-pixel multispectral images. Con-
sequently, these systems would represent state-of-the-art sen-
sors for aerial and space ‘photography,’ and such systems
would provide for a tremendous volume of multitemporal
data for change-detection studies.
It should be kept in mind, however, that a tremendous
amount of historical aerial and space photography can be
used to study geomorphological systems. Classic examples
include the use of multitemporal aerial photography for
studying coastline changes and sediment transport, aerial and
space photography for monitoring glacier fluctuations, space
photography and imagery for studying regional geomorpho-
logical conditions, and ground and aerial photography for
inventory and assessment of mass movements. It is critical to
understand that multitemporal ‘photography’ serves as base-
line information on which existing data for quantifying en-
vironmental change can be compared.
Developments in videography can also considerably im-
prove our understanding of surface processes and landforms.
Ground and aerial video of catastrophic events document such
events and provide insights into the nature of process mech-
anics. Examples include debris flows and landslide events,
earthquakes, and associated landscape changes, tsunamies,
and flooding events. Thermal videography is also being in-
creasingly utilized in volcanology (e.g., Vaughan et al., 2005),
and in many surface applications to study energy budget
and depositional processes (e.g., Hardgrove et al., 2009).
Many examples are presented in Chapter 3.2 by Shroder, as he
specifically addresses the use of photography and videography
in geomorphology.
3.1.3.2 Imaging Spectroscopy
Optical imaging sensors record the magnitude of reflected
and emitted radiation from planetary surfaces. Sensors can be
mounted on aircraft and satellite platforms to obtain
multispectral and multitemporal information about surface
characteristics. The data are qualitatively and quantitatively
analyzed to generate thematic and quantitative biophysical
information. The nature and accuracy of remotely derived
information is highly dependent on sensor characteristics re-
garding spatial, spectral, and radiometric sensitivity. Con-
sequently, not all sensors can provide similar information, as
they have been specifically designed to address issues related
to sensor-system evolution and application objectives.
With the advent of resource satellites, starting in the 1970s
and 1980s, Earth scientists were able to assess and map the
regional geomorphological conditions on the Earth (e.g.,
Short and Blair, 1986). Numerous governments subsequently
launched their own satellites with different sensor payloads,
such that currently, a multitude of information products and
satellite data are routinely used in Earth science applications.
The evolution of imaging sensors has been one of improving
spatial, spectral, and radiometric resolution and signal-to-
noise ratio. This evolution has already had a profound influ-
ence on Earth science, and high spatial-resolution sensors
(1–5 m) allow improved thematic mapping capabilities.
Nevertheless, many sensors do not have a comparable spectral
resolution, exhibiting a limited number of spectral bands that
may or may not cover key regions of the electromagnetic
spectrum.
The primary sensor characteristic that governs image in-
formation content is spectral resolution. Therefore, more
spectral bands theoretically facilitate obtaining unique infor-
mation. In practice, this is not the case, as the amount of
generalization associated with the spectral bandwidth deter-
mines the degree of multicollinearity in the data. Imaging
spectroscopy represents the collection of many narrow, spec-
trally contiguous bands, such that each pixel contains a
spectrum. In the literature, imaging spectroscopy is also
commonly referred to as imaging spectrometry or hyperspec-
tral remote sensing. Imaging spectroscopy has existed
for the past three decades, and has witnessed exponential
growth recently, as new sensors are being developed and
evaluated, and hyperspectral data are being increasingly util-
ized in a large variety of domains including ecology, hydrol-
ogy, soil science, geology, and geomorphology (Schaepman
et al., 2009).
3.1.3.2.1 Sensor parameters
Electro-optical sensors consist of detectors that record the in-
cident at-sensor radiance as an electronic signal. The signal
varies over time, space, wavelength, and amplitude depending
on sensor responsivity, which in general is called resolution.
The sensor characteristics can significantly alter the signal
and ultimately determine the level of generalization and the
nature of the information represented in the imagery. Fur-
thermore, knowledge of sensor characteristics is required for
interpretation and analysis, given sensor influence on spectral
variability.
 Remote Sensing and GIScience in Geomorphology: Introduction and Overview
The spatial resolution of a sensor is usually referred to as
the ground instantaneous field-of-view (GIFOV). It is the
geometric projection of the detector width, w, and is defined as
 
IFOV H data are not useful for biophysical applications, given the SRF
GIFOV ¼ 2:0H tan ¼w ; ½1
2:0 f (Figure 4). Other sensors provide more detailed spectral in-
where IFOV is the instantaneous field-of-view, H is the height
of the sensor above the Earth surface, and f is the focal length
of the sensor. The GIFOV determines whether an object will be
resolved by the system, as GIFOV and object dimensions de-
termine the sensor’s object-discrimination capabilities. The
sensor-system averages the recorded signal over the spatial
extent of a pixel. Specifically, the spatial response of a sensor is
characterized by the optical point-spread function (PSF). It is a
weighting function that spatially distorts the inherent spatial
geometry at the surface of the Earth. A common model for the
PSF is the 2-D Gaussian function such that
1:0
PSFðx,yÞ ¼ expðx2 =2a2 Þexpðy2 =2b2 Þ; ½2
2:0pab
where a and b determine the width of the PSF in the cross- and
in-track directions. For well-designed optics, a ¼ b. Con-
sequently, moderate- to coarse-resolution sensors produce
spectral data that are not representative of a single type of
matter or biophysical property, and the system produces
composite spectra.
The spectral resolution of the sensor is also an important
consideration, as sensors exhibit spectral bands that measure
energy in different regions of the spectrum. In theory, different
regions of the spectrum can be used to obtain different infor-
mation, as matter/energy interactions can be fundamentally
different. Consequently, it is advantageous to sample the visible
(0.4–0.7 mm), near-infrared (0.7–1.3 mm), shortwave infrared
(SWIR) (1.3–3.0 mm), and thermal (3.0–100 mm) regions of the
spectrum. The total amount of energy measured in each spectral
band must be averaged over a spectral bandwidth, and each
detector has a spectral-response function (SRF) that character-
izes the wavelength weighting. An ideal SRF is modeled as a
Gaussian function such that
" #
ðl  lc Þ2
WðlÞ ¼ exp ; ½3
2s2
where lc is the wavelength corresponding to the peak
weighting and s ¼ FWHM(8 log(2))0.5. The full-width at half-
maximum (FWHM) effectively represents the spectral band-
width, although technically this represents the width where W
is 0.5. A wider spectral bandwidth results in more spectral
generalization. The central wavelength peak and bandwidth
determine whether spectral absorption features will be char-
acterized in recorded spectra. The spectral averaging can be
represented as
Pl2
l1 WðlÞLðlÞ
LðlÞ ¼ Pl2 ; ½4
l1 WðlÞ
where L represents the average radiance value and L represents
the radiance recorded by the sensor.
7
In practice, sensor selection is an important aspect of re-
mote sensing, as resolution characteristics determine applic-
ability for problem solving. For example, some panchromatic
sensors may have a higher spatial resolution, although the
formation with less spectral generalization (Figure 5). Finally,
imaging spectroscopy depends on a very fine spectral response,
such that there are a large number of spectral bands and
narrow spectral bandwidths (Figure 6).
The electronic signal must also be amplified to provide for
a sufficient signal for quantification. The electronic gain and
offset values are set based on an expectation of the range in the
magnitude of incident radiance from different environments.
The amplified signal is then sampled and quantified into
digital number (DN) values. Therefore, a DN value for each
pixel is represented as
DNðlÞ ¼ int½gðlÞes þ oðlÞ; ½5
where es is the electronic signal, g is the electronic gain, and o
is the electronic offset. The number of discrete DN values that
represent the magnitude of radiance defines the radiometric
resolution of the sensor. This represents 2n bits over the dy-
namic range in radiance. This characterizes the sensitivity of
the sensor to record variations in the magnitude of energy
reflected from the landscape. Most modern sensor have an
8–12-bit radiometric resolution.
3.1.3.2.2 Reflectance properties and applications
The reflectance properties of surface features are based on their
composition and molecular processes. Incident radiation is
preferentially absorbed by molecules. The total energy of a
molecule is the sum of electronic, vibrational, and rotational
energy. Changes in the energy state are governed by electronic
and vibrational processes. The absorption of energy results in
absorption features in reflectance spectra. Emission of photons
results from a transition to a lower energy state. Consequently,
reflectance spectra can be used to diagnostically detect the
presence of a variety of materials, as many exhibit absorption
features and unique spectral reflectance patterns.
Imaging spectroscopy can be used for a variety of geo-
logical and soil-science applications. Mineral detection is
critical for many applications. Imaging spectroscopy has long
been used to explore for mineral deposits and for lithological
mapping (Goetz and Rowan, 1981; Kruse et al., 1993). This is
possible as electronic transition and charge transfer processes
associated with transition metal ions determine the position
of diagnostic absorption features in the spectra of minerals
(Burns, 1970). Vibrational processes in H2O and OH also
produce overtone absorptions. Consequently, the position,
shape, depth, width, and asymmetry of absorption features are
controlled by the crystal and chemical structure of the mineral
(van der Meer and de Jong, 2006). Absorption bands can be
found in the visible, NIR, and SWIR regions of the spectrum.
For a more detailed characterization of mineral reflectance
spectra, see Grove et al. (1992).
The reflectance spectra of minerals are well known, and
numerous spectral libraries can be used to facilitate remote-
sensing studies. Specifically, the United States Geological
8 Remote Sensing and GIScience in Geomorphology: Introduction and Overview

1.0

0.9

0.8

Spectral response weight (w)

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0.0
0.45 0.50 0.55 0.60 0.65 0.70 0.75 0.80 085 0.90 0.95 1.00
Wavelength (µm)
Figure 4 Landsat-7 Panchromatic sensor spectral response function. The sensor is more sensitive to the near-infrared region of the spectrum,
and this reduces the atmospheric effects in the imagery. The wide spectral bandwidth does not allow the imagery to be utilized to assess specific
biophysical conditions because extensive spectral averaging does not accurately characterize narrow absorption features.

0.80

0.75

0.70

0.65

0.60
Reflectance

0.55

0.50
ASTER illite spectral curve
0.45 Landsat 7 TM Illite spectral curve
0.40

0.35

0.30

0.25
0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4
Wavelength (µm)
Figure 5 Simulated ASTER and Landsat-7 Thematic Mapper spectral-reflectance curves for the secondary-silicate mineral illite. Sensor spectral-
response functions determine the degree of generalization associated with the spectral curve. This example demonstrates that ASTER data depict
the illite absorption feature in the SWIR, whereas the TM spectral curve does not. Consequently, spectral resolution must be carefully considered
for different applications.

Survey Spectral Library and the ASTER spectral library can be
used to simulate composite spectra using spectral mixing
models. In addition, spectral matching algorithms can be used
to assess the degree of similarity between pixel spectra and
laboratory spectra. These types of analyses allow assessment of
the potential presence of specific minerals at the surface. Rock
spectra can be simulated based on standard or assumed
mineralogical composition. Linear and nonlinear spectral
mixing models and model inversion can be used for
lithological and soils mapping and finding mineral end-
member distributions, respectively.
Imaging spectroscopy provides many new capabilities for
studying soil properties (Ben-Dor et al., 2009). Soil degradation
due to salinity, and caused by rising water tables or irrigation,
can be assessed via hyperspectral remote sensing (Taylor et al.,
1994; Metternicht and Zinck, 2003). Specifically, Taylor et al.
(1994) showed that soil salinity could be mapped using air-
borne spectrometer data. More specifically, Taylor (2004)
 Remote Sensing and GIScience in Geomorphology: Introduction and Overview
12.5
12.0
11.5
AVIRIS FWHM (nM) 11.0
10.5
10.0
9.5
9.0
0.4 0.6 0.8
Figure 6 Airborne visible/infrared imaging spectrometer (AVIRIS) spectral response-function characteristics. The sensor exhibits 224 spectral
bands with very narrow spectral bandwidths. Notice that the wavelength interval and FWHM values vary across the spectrum. The very high
spectral resolution allows improved characterization of measured spectra.
demonstrated that the depth and width of the hydroxyl ab-
sorption feature at 220 nm changes with increased soil salinity.
Others have found that surface gypsum is highly correlated with
the NaCl content in some soils and can be spectrally identified.
Other soil processes can be examined and include soil
erosion and deposition (Ben-Dor et al., 2009). Using airborne
visible/infrared imaging spectrometer (AVIRIS) data, Hill et al.
(1995) used a spectral mixing model with end-member
spectra to describe the status of soil erosion. They estimated
the relative abundance of parent material and soil particles on
the surface. This represents the mapping of erosion state, as it
is related to the mixing ratio between developed substrates
and components of the parent material.
Another approach to assessing soil erosion involves
quantitative assessment of soil chemical properties (Ben-Dor
et al., 2009). For example, Hill and Schütt (2000) suggest that
organic carbon can be a tracer substance for identifying ac-
cumulation areas and relatively stable soil conditions.
Stable conditions correspond to higher infiltration and water
retention capacity; therefore, erosion and depositional areas
can be mapped based on detecting organic carbon. The
curvature of the spectral continuum from 0.4 to 1.6 m was
used to derive the organic carbon content of soils (Hill and
Schütt, 2000). Research has indicated that spectral differences
are associated with variations in weathering and natural versus
eroded soils (Demattê, 2002).
Other capabilities are also feasible and include soil mapping
and classification, extraction of information regarding soil gen-
esis and formation, and assessment of soil contamination and
swelling (Ben-Dor et al., 2009). These capabilities are related to
assessment of specific soil properties including: (1) presence and
absence of primary and secondary silicate minerals; (2) quan-
tification of iron oxides; (3) content, composition, and maturity
of organic matter; (4) quantitative estimation of heavy metals;
and (5) assessment of soil moisture content.
9
1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4
AVIRIS wavelength center (µm)
Imaging spectroscopy is also used in vegetation science and
ecology. Its utility is based on assessment of specific physical
and chemical characteristics of vegetation. In the visible region
of the spectrum, plant pigments absorb radiation and ab-
sorption features can be used to assess a variety of plant pig-
ments. In the NIR region, plants exhibit relatively high
reflectance and transmission. The distribution of air spaces
and the size, shape, and arrangement of cells determine the
passage of light through plant leaves and the amount of scatter
in the mesophyll layer of leaves. In the SWIR region, leaf
moisture and foliar biochemicals influence reflectance, and
this region contains strong water absorption features and
minor biochemical absorption features.
Variations in environmental conditions can cause variations
in chlorophyll production, leaf cellular structure, and leaf
moisture conditions. Consequently, variations in these and
other canopy characteristics (e.g., leaf area index) can be de-
tected via imaging spectrometer data, and provide an insight
into surface/subsurface conditions, as plants respond to soil
geochemical, lithological, and structural conditions. Numerous
studies have evaluated the use of spectra and the shifting of the
red edge to detect plant stress and their response to mineral
deposits, lithological changes, and other environmental char-
acteristics (e.g., Collins et al., 1983; Boochs et al., 1990).
Finally, imaging spectroscopy can be used for assessing
water in various phases within the Earth’s natural systems, and
this supports new strategies for hydrological research and as-
sessment of quality and distribution (Green et al., 2006). Re-
mote sensing has commonly been used for the assessment
and mapping of inland and coastal water quality conditions.
With the advent of spectrometers, atmospheric water-vapor
conditions and ice-crystal size variations in snow can be as-
sessed quantitatively (Dozier et al., 2009). This allows dis-
tributed water-budget assessment in high-altitude basins that
facilitates water management and planning activities.
10 Remote Sensing and GIScience in Geomorphology: Introduction and Overview
Each phase of water exhibits absorption features between
400 and 2500 nm. Water vapor exhibits an extremely fine
spectral absorption structure, and liquid water and ice exhibit
broad molecular absorption bands (Green et al., 2006). This
allows spectroscopic separation of the three phases. Con-
sequently, imaging spectroscopy will play an ever-increasing
role in hydrological research, as detailed information re-
garding surface-energy conditions (i.e., albedo) can be used
for energy-budget modeling, the update and validation of
distributed snowmelt and runoff models, and in assessing
water-quality condition and bathymetry in relatively shallow
inland and coastal areas. Unfortunately, a chapter on Imaging
Spectroscopy was not completed in time for inclusion in this
Treatise volume.
3.1.3.3 Microwave Remote Sensing
Unlike optical-based sensors that depend on reflected and
emitted radiation from objects and surfaces, microwave sen-
sors collect information from the microwave region of the
electromagnetic spectrum with frequencies ranging from 0.3
to 40 GHz. Passive and active sensors exist; however, active
sensors play a dominant role.
Numerous sensor types provide a wealth of information
obtained at planetary, region, and local scales. Consequently,
it is essential to be familiar with the advantages and dis-
advantages associated with the use of specific microwave
sensors. These include radio detection and ranging (radar)
altimeters, synthetic aperture radar (SAR), polarimetric
SAR, stereo SAR, and interferometric SAR. The choice of a
particular sensor and analysis for information extraction is
slightly more complicated compared with optical-based sen-
sors, as the backscatter of microwave radiation is dependent
on frequency selection, sensor-system imaging geometry, po-
larization, surface composition and roughness, near-surface
structure, and the electrical composition of the surface that
determines the amount of the energy reflected, absorbed, and
transmitted.
Radar imagery can be qualitatively evaluated using human
interpretation techniques. Quantitative analysis is used to
produce topographic information (i.e., DEMs) that is rou-
tinely utilized by Earth scientists. In addition, the ability to
assess deformation patterns resulting from natural (e.g.,
earthquakes) and anthropogenic events (e.g., oil and ground
water extraction) allows new insights into the nature of nu-
merous processes related to landscape evolution and natural
hazards. Numerous applications involving geodesy, land
cover, ecology, hydrology, geology, geomorphology, and gla-
ciology are possible.
Hensley and Farr, in Chapter 3.3, provide an authoritative
treatment of microwave remote sensing in geomorphology.
They specifically address the different types of active micro-
wave sensors and relate scale to various geomorphological
applications. Microwave remote-sensing principles are also
addressed so that users take into consideration the multitude
of factors that influence backscatter variations in imagery such
as frequency, resolution, polarization, scattering, and pene-
tration. For each sensor, the theory of operation is examined
and specific geomorphological applications are presented.
This treatment of the topic should provide readers with an
insight into the complex nature of matter–energy interactions
in the microwave region of the spectrum, while highlighting
the wealth of landscape information that can be obtained and
utilized by the Earth science community.
3.1.3.4 The Atmosphere and Climate Forcing
The linkages between climate and surface processes have long
been established by the geomorphological community. At-
mospheric conditions regulate surface energy, temperature,
and precipitation. Climate forcing is an active research area
and is known to govern process domains, rates of erosion, and
the presence/absence of specific landforms. A relatively recent
development has been the recognition of the complex inter-
relationships between climate, surface processes, and tectonics
(Molnar and England, 1990; Shroder and Bishop, 2000). This
has prompted vigorous debate regarding the magnitude
of erosion, role of surface processes, and issues of climate
versus tectonic forcing in orogenesis. Investigations into
climate–geomorphology linkages require quantitative esti-
mates of atmospheric variables and maps depicting the spatial
variability of atmospheric conditions. Such detailed infor-
mation has not been historically available to geomorph-
ologists. Recent advances now allow a detailed examination of
the atmospheric conditions that facilitate climate forcing
studies, and remote sensing provides new avenues for geo-
morphological research.
The moderate resolution imaging spectro-radiometer
(MODIS) on the Terra and Aqua satellites can be used to
collect a variety of atmospheric parameters. It has a swath
width of 2330 km and can nearly provide global coverage of
atmospheric conditions using 36 spectral bands ranging from
the visible to the thermal region of the electromagnetic spec-
trum. Specifically, it allows global monitoring of atmospheric
profiles, precipitable water-vapor amount, aerosol particles,
and cloud characteristics. An atmospheric profile algorithm is
used to estimate the atmospheric temperature and moisture
conditions. Consequently, the spatial pattern of the atmos-
pheric temperature and moisture variations at a particular
level in the atmosphere can be examined. See King et al.
(2003) for technical details and a more complete description
of data products generated from the sensor.
Other imaging spectrometers can also be used to assess
atmospheric conditions. For example, Green et al. (2006) used
AVIRIS over Mount Rainer in Washington to assess water vapor
content. At the water vapor absorption bands (near 940, 1150,
1380, and 1900 nm), the upwelling radiance at the sensor
varies in strength as a function of the column water vapor
content. Numerous investigators have found a strong rela-
tionship between water vapor and elevation (Green et al.,
2006). Consequently, AVIRIS data can be used to generate
high-resolution atmospheric moisture maps.
The magnitude of precipitation is another key atmospheric
parameter. Satellite remote sensing is an indispensable means
of measuring and monitoring precipitation on a global scale
(Masunaga et al., 2002). The tropical rainfall measuring mis-
sion (TRMM) has made significant contributions to the Earth
sciences, as scientists have access to a variety of precipitation
 Remote Sensing and GIScience in Geomorphology: Introduction and Overview 11

information products. This information is critical for water-
budget studies involving ground water recharge and ex-
traction, surface runoff and human impact on basin hydrol-
ogy, water storage and resources in mountain environments,
and in the assessment of flooding and flood inundation.
Furthermore, this parameter is critical for studying the influ-
ence of climate forcing on mountain topographic evolution,
studying specific climate systems (e.g., Houze et al., 2007),
and validating climate simulations.
Numerous researchers have examined spatial patterns in
precipitation in an attempt to quantitatively relate climate to
topography, erosion, and specific process domains (e.g., Bur-
bank et al., 2003; Wulf et al., 2010; Bishop et al., 2010). Re-
search indicates a coupling between precipitation and
topography (Nesbitt and Anders, 2009) and magnitude vari-
ation can characteristically reveal the presence of orographic
precipitation and erosion zones that may be related to uplift
patterns. For example, Wulf et al. (2010) examined seasonal
precipitation gradients using TRMM data to determine their
impact on fluvial sediment flux in the Himalaya. They found
that the Indian summer monsoon is the main driving force for
erosional processes, despite more precipitation falling in the
winter season. Similarly, Bishop et al. (2010) used TRMM data
to find that the highest amounts of precipitation in
the Karakoram occur during the spring by the westerlies
(Figure 7). They also noted a spatial coincidence between a
precipitation anomaly in the Hunza region and increased
landscape dissection and steep slopes, suggesting the presence
of a high-magnitude erosion zone.
Remotely derived atmospheric information has not been
thoroughly utilized in geomorphology and numerous research
opportunities exist. For example, an important part of land-
scape evolution is related to physical and chemical weathering
and regolith production. Research has demonstrated links
between weathering and surface and atmospheric conditions
(Curtis, 1976; Trudgill, 1976). Spatial distribution patterns of
air and surface temperature, air and surface moisture, surface
mineralogy, and vegetation characteristics should provide in-
sights into the magnitude and distribution of weathering and
regolith production. Perhaps modelers might be able to de-
velop new parameterization schemes that incorporate satellite-
derived atmospheric and surface parameters. This would sig-
nificantly improve the ability to account for the spatial vari-
ability in weathering and sediment transport. Nevertheless,
there would be other difficult issues to address that include
accounting for biological processes, acid production, and the
integrated influence of climate change that extends beyond the
temporal availability of the data.
Atmospheric information is also critical for understanding
and predicting the surface-energy budget that governs many
surface processes including weathering, moisture availability,
erosion and sediment transport. Numerous field studies typi-
cally use point station data to obtain measurements of at-
mospheric conditions, not knowing the degree of variability
associated with the shortwave and longwave net radiation flux.
Atmospheric conditions, surface albedo and temperature, and
topographic information via remote sensing allow more so-
phisticated modeling that can be used for hydrological and

12

11

10

8 0
20
7 40
Month

60
6 80
50
60
5
70
80
4 90

1
1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008
Year
Figure 7 Tropical rainfall mapping mission (TRMM) precipitation data (3B43V6) for the Baltoro-Mustagh region in the Karakoram Himalaya,
Pakistan. The temporal variation in cumulative precipitation (mm per month) depicts the influence of the westerlies during the spring and the
combined influence of the southwestern monsoon and westerlies during the summer months. TRMM data can also be used to examine spatial
patterns in precipitation. Such satellite data allow a spatio-temporal evaluation of precipitation conditions.
12 Remote Sensing and GIScience in Geomorphology: Introduction and Overview

glaciological investigations. Consequently, satellite-derived image elements (i.e., tone, texture, size, shape, pattern, site)
atmospheric information can greatly improve the under- and the linkage of spatial objects across multiple scales.
standing of many aspects of geomorphological systems. Un- This approach is also valuable for mapping specific
fortunately, a chapter on this topic was not completed in time landforms based on the segmentation of terrain units using
for inclusion in this Treatise volume. geomorphometry.
Allen and others in Chapter 3.4 specifically address the
issues of land-cover and land-use assessment via optical re-
mote sensing. They highlight the data sources that are valuable
3.1.3.5 Land-Cover Assessment and Mapping
for land-cover mapping and review some of the methodo-
Aerial photography and satellite imagery have been historic- logical approaches that are routinely utilized. Finally, they
ally utilized to generate land-cover and land-use information. provide classic application examples of how land-cover in-
A multitude of new high-resolution and multispectral sensors formation is used in geomorphology and provide a treatment
acquire data that can be used to characterize a wide range on land-cover change detection.
of biophysical landscape properties. Collectively, this infor-
mation can be used for thematic mapping of land-cover
characteristics. Research has focused on evaluating the infor- 3.1.3.6 Near-Surface Geophysics
mation content within multispectral datasets, evaluation and
Remote sensing of the subsurface is commonly required to
comparison of pattern-recognition techniques for improved
assess geomorphological systems and to accurately map the
classification of thematic content, and the development
three-dimensional extension of landforms and structural fea-
of new methodological approaches for data fusion, spectral-
tures. Subsurface compositional variations and the occurrence
feature extraction, spatial-feature extraction, and multi-
of subsurface structures can have a significant influence on
temporal analysis. Although challenges remain with respect
surface processes and landscape evolution.
to addressing increased data volumes and increased spectral
Information on subsurface materials and characteristics
and temporal variability, existing software systems allow
can be obtained by passive gamma-ray spectrometry and
routine mapping of fundamental land-cover classes. Thematic
geophysical techniques such as seismic, gravity, aeromagnetics,
mapping and study of land-cover dynamics represent an
electromagnetics, and ground penetrating radar (GPR).
important component of geomorphological assessment and
Gamma-ray spectrometry may indicate the composition of
mapping. Furthermore, land-cover and land-use patterns
materials in the upper 50 cm of the surface (Smith and Pain,
provide an insight into the magnitude of anthropogenic
2009), whereas gravity, aeromagnetics, electromagnetics, and
forcing.
GPR can be used to assess density, subsurface features, con-
New capabilities include developments in a variety of sub-
ductivity variations, and depths, respectively (Lane, 2002;
disciplines. For example, data fusion is an approach to mapping
Wilford, 2002). Consequently, subsurface lithological vari-
and analysis that exploits the power of multiple representations
ation can be compared with surface morphometry and other
of the landscape. This involves integrating data with different
biophysical properties to characterize the 3-D nature of
spatial, spectral, and radiometric resolutions. A classic example
landforms. The cost and availability of such subsurface in-
is merging multispectral satellite data with higher-resolution
formation is currently a serious limitation, as expensive air-
panchromatic data. In a GIS, multiresolution airborne and
borne or field surveys are required.
satellite data can be fused with a DEM, terrestrial photography,
Nevertheless, the use of geophysical techniques in geo-
maps, and graphics. Digital mapping can be accomplished by
morphology crosses many subdisciplines, as revealed by Kruse
utilizing various feature sets that represent multiple landscape
in Chapter 3.5. She summarizes the relative significance of
dimensions and perspectives.
different geophysical methods in various subdisciplines of
Object-oriented land-cover mapping also represents a
geomorphology. She also provides practical advice for pro-
relatively new development. It first requires meaningful seg-
spective users of near-surface geophysics and highlight the
mentation based on specific criteria to generate spatial entities
importance of reference data collected in the field, in addition
called objects. Initial segmentation is typically based on in-
to geophysical data. There is a need for the comparison of
formation in imagery and DEMs. Numerous approaches to
methods and caution in the interpretation based on forward
segmentation can be used including homogeneity and shape
or inverse modeling. Kruse notes that the uses of near-surface
analysis, region growing, pattern recognition, and rule-based
geophysics in geomorphology are not fundamentally different
segmentation. Segmentation results are then analyzed via
from the uses in other geoscience disciplines, and that geo-
spatial clumping to identify individual homogeneous spatial
morphologists would be well served by examining methods
entities. These objects then serve as a spatial constraint for
and results from tectonic, hydrogeophysical, applied geo-
subsequent analysis. Object-oriented analysis involves com-
physical, and engineering studies.
puting the attributes of individual objects such as object lo-
cation, size, shape, and its topological relationships with other
objects on the landscape.
Mapping can be facilitated by spatial aggregation and 3.1.4 GIS and Geomorphology
spatial intersection of objects and by identifying unique pat-
terns of object attributes in an n-dimensional feature space. The rapid proliferation of GIT allows improved data man-
This approach is widely recognized as superior to purely pixel- agement, manipulation, analysis, modeling, and visualization
based classification procedures, as it allows the integration of capabilities. Various forms of spatio-temporal data can be
 Remote Sensing and GIScience in Geomorphology: Introduction and Overview
stored in GIS databases, and a plethora of software tools
allows scientists to effectively study spatio-temporal patterns
and relationships.
Such new capabilities represent a substantial evolution in
geomorphological assessment and mapping compared with
traditional approaches (Bishop et al., 2012). Yet, the tradi-
tional approaches of information integration via analytical
reasoning, which is the pillar of qualitative interpretation, are
poorly represented by statistical metrics and mathematical
operators that are commonly used in GIS analysis. Further-
more, the results of GIS-based quantitative analysis and nu-
merical modeling are dependent on numerous factors and
simplifying assumptions, and may not be representative of
objective measurements obtained in the field. Consequently,
conceptual and practical issues need to be recognized,
and Bishop et al. (2012) have identified a number of issues
that should be considered. These include representation,
the predominantly empirical nature of using modern-based
GISs, scale, and mapping perspectives. In general, advances
in geomorphology have resulted from addressing these
issues and utilizing geospatial technologies to address specific
problems.
3.1.4.1 Digital Terrain Modeling (DTM)
Quantitative land-surface information is required in geo-
morphology. A major contribution of remote sensing has been
the development and use of passive and active sensors to
generate DEMs. A variety of techniques can be utilized for
DTM including image photogrammetry, radar or laser alti-
metry, and interferometric SAR. Photogrammetric applications
utilizing Satellite Pour l’Observation de la Terre (SPOT) and
Advanced Spaceborne Thermal Emission and Reflection
Radiometer (ASTER) data are commonly used by scientists. In
the case of SPOT imagery, alternate view perspectives from
multiple satellite passes enable stereoscopic representations,
whereas the ASTER system relies on forward- and back-looking
telescopes to characterize topography through a merged
characterization. Similarly, radar imagery and specifically
shuttle radar topographic mapping mission (SRTM) data are
widely used. The SRTM and ASTER mission objectives were
specifically designed to produce a global DEM data product to
facilitate Earth science applications. These DEMs have resulted
in many new developments to characterize surface morph-
ology and better assess and map the landscape.
More recently, airborne high-resolution LiDAR systems and
terrestrial-laser-scanning systems now generate millions of 3-D
point measurements. These ‘point clouds’ must be analyzed
and manipulated to ensure accurate interpolation to generate
a bare-Earth altitude field. LiDAR high-resolution DEMs allow
detailed geomorphometric characterization of the surface and
greater mapping accuracy (Figure 8). Such data allow devel-
opments in geomorphometry to be exploited, whereas the
same techniques may not be as useful, given a coarser DEM
measurement scale. For example, DEM differencing is an im-
portant aspect of change detection suitable for examining
spatial patterns of surface dynamics and volumetric analysis,
but the availability of high-resolution, geo-referenced ele-
vation grids is critical. Numerous studies have revealed the
13
Figure 8 One meter digital elevation model generated from LiDAR
data with an average point spacing of approximately 5 m. ESRI
terrain dataset generated from original bare-earth LiDAR points from
the North Carolina Floodplain Mapping Program. The 1 m DEM was
created to enhance hydrographic modeling operations within the
watershed. DEM courtesy of Jeffrey Colby, Appalachian State
University.
significance of using an optimum resolution to appropriately
characterize geomorphometric parameters for erosion and
mapping investigations (Zhang and Montgomery, 1994;
Napieralski and Nalepa, 2010).
DTM involves many issues related to the nature of data
acquisition and sampling, preprocessing, spatial interpolation,
quantitative characterization of error and uncertainty, and
postprocessing. Evaluation of DTM for geomorphological
applications is an active research theme. Wasklewicz and
others in Chapter 3.6 discuss the generation and utility of
DEMs in geomorphology. They address the technical advances
in measuring the topography including laser-scanning, shut-
tle-based radar, and terrestrial photogrammetry techniques.
They also address numerous technical issues that must be
accounted for via preprocessing and postprocessing. They
correctly acknowledge the need to understand a variety of
DTM issues that are critical in analysis and communication of
information generated from a DEM.
3.1.4.2 Terrain Analysis
The quantification and analysis of the land surface is
called gemorphometry (Pike, 1995, 2000). Geomorphometry
plays a central role in studying surface processes and
for geomorphological mapping. Consequently, it has been
characterized as general and specific geomorphometry,
respectively.
Geomorphometry addresses issues of: (1) sampling attri-
butes of land surfaces; (2) geodesy and DTM; (3) DEM pre-
processing and error assessment; (4) generation of land-
surface parameters, indices, and objects; and (5) geomorphic
information production and problem-solving using par-
ameters and objects. Each aspect of geomorphometry repre-
sents a research subdiscipline and contributes significantly
toward the development of software tools and geospatial
technology. Its significance in geomorphology is expected to
increase, as it can be used for assessing and mapping geology
and tectonics, landform elements and landform, functional
units related to water resources and hydrology, process do-
mains, erosion patterns, as well as climate and meteorological
conditions (Bishop et al., 2012).
14 Remote Sensing and GIScience in Geomorphology: Introduction and Overview

Progress has focused on: 3.1.4.3 Landform Mapping

• The development and use of geomorphometric algorithms. Historically, landform mapping was primarily a field-based
New and modified forms of parameters and indices are
activity, although it has rapidly evolved, given the advent of
being developed and evaluated for assessment and mapping.
geospatial technologies and the ever-increasing availability of
Spatial analysis involves neighborhood operations, subgrid
spatio-temporal data (Bishop et al., 2012). New imagery and
operations, and multiscale analysis. The primary math-
DEMs, along with new algorithms and processing approaches,
ematical approach has been statistical analysis and prob-
now allow assessment of surface materials, assessment of near-
ability theory; however, geostatistics, artificial intelligence,
surface conditions using geophysical techniques, and assess-
and fuzzy-set theory are increasingly being utilized.
ment of terrain morphology. These new mapping capabilities,
• New software tools and systems for geomorphometric an- however, bring about new challenges involving theory, con-
alysis and mapping. A number of programs are specifically
cepts, and technology that need to be carefully addressed. This
designed to compute numerous geomorphometric par-
will require new developments in data, representation, an-
ameters, although many GISs are limited to basic parameters.
alysis, and visualization.
Consequently, geomorphometry and mapping can be carried
A fundamental issue in mapping is that of spatial com-
out using ESRI software (Reuter and Nelson, 2009), SAGA
plexity and taxonomy, as robust and reliable classification is
(Olaya and Conrad, 2009), ILWIS (Maathuis and Wang,
required for landform mapping. Bishop (2009) addressed this
2009), LandSerf (Wood, 2009), MicroDEM (Guth, 2009), TAS
issue and discussed the need for taxonomic criteria that fa-
GIS (Lindsay, 2009), GRASS GIS (Hofierka et al., 2009), and
cilitate the classification of volcanic landforms. Spatial com-
River-Tools (Peckham, 2009), just to name a few.
plexity was addressed using a generic approach that described
• Existing and new applications. Numerous algorithms and the complexity of landform regardless of scale, climate regime,
approaches for characterizing spatial variation, scale, land-
or geological and tectonic setting. This concept is critical for
scape position, fuzzy boundaries, and complexity exist, and
GIS-based landform mapping as formalized taxonomies seem
many landforms and features such as drainage basins and
warranted to address the indeterminant nature of landform
networks, ridges, and peaks can be mapped to various de-
terminology.
grees. Nevertheless, researchers have a daunting task of de-
Similarly, the indeterminant nature of landform bound-
termining which metrics and approaches are best for specific
aries is an issue, as many landforms do not exhibit a discrete
objectives. Geomorphometry has significantly contributed
demarcation in material and form (Burrough, 1996; Lagach-
to geological, soil, vegetation, landform, ecological, hydro-
erie et al., 1996; Bishop et al., 2012). For example, Deng and
logical, mass movements, hazards, meteorological, and
Wilson (2008) used fuzzy theory and fuzzy spatial entities to
agricultural mapping applications, and new applications are
map mountain peaks. Peaks were mapped as multiscale en-
likely to evolve (Gessler et al., 2009).
tities with modifiable boundaries. Their approach allowed
• Mapping other aspects of the geomorphic system related to vague peak entities to be assessed. The approach addresses
climate and tectonic forcing, process domains, and erosion,
landform delineation, spatial continuity, and multicriteria
however, is more complex and may require very different
definition, all important concepts in landform mapping.
morphometric approaches. For example, quantifying the
Assessment of the spatial uncertainty associated with pro-
extent to which geomorphic parameters or landforms and
cess domains and landform mapping is another critical area of
landform elements can be used to assess and characterize
investigation. Increasingly, landform mapping is being ac-
tectonic signals, or the influence of tectonics on the land-
complished by the development and assessment of complex
scape, remains a key challenge in the Earth Sciences (Boulton
spatial models that inherently incorporate the error that is
and Whittaker, 2009; Whipple, 2009). A typical approach
associated with data, algorithm selection, processing pro-
includes the analysis of drainage basins and patterns, and an
cedures, and model assumptions. There is a need to quanti-
evaluation of the longitudinal profiles of bedrock rivers.
tatively characterize the degree of uncertainty in such mapping
Asymmetric drainage patterns, elongated drainage basins,
models, and selected techniques including artificial intelli-
and convexities and the presence of knick points are thought
gence (i.e., neural networks and fuzzy uncertainties) and fuzzy
to reflect the system response to ongoing tectonic uplift
theory can be used to facilitate such investigations. For ex-
(Jamieson et al., 2004; Boulton and Whittaker, 2009). Other
ample, Luoto et al. (2010) produced spatial uncertainty maps
applications, such as the sampling and estimation of surface
based on the agreement between different modeling techni-
cosmogenic nuclides, allow estimates of catchment erosion
ques. The uncertainty maps reflected the reliability of assessing
rates using GIS. This requires knowledge of the production
periglacial landforms.
rate of various isotopes related to the incoming cosmic-ray
A multitude of studies have focused on improving map-
flux, which is governed by latitude, altitude, slope, azimuth,
ping capabilities for assessing fluvial and hydrological systems
and topographic shielding (Figure 9).
(e.g., Schaper et al., 1999; Marcus et al., 2003), glacial and
It is essential that geomorphologists be familiar with periglacial systems (e.g., Schneevoigt et al., 2008; Kääb and
geomorphometric parameters and know how to use them Vollmer, 2000; Frankl et al., 2010), and various aspects of the
for various applications. Wilson and Bishop in Chapter 3.7 sediment transfer cascade including mass movements (e.g.,
provide a treatment of geomorphometry that highlights vari- Wichmann et al., 2009). Napieralski and others in Chapter 3.8
ous parameters and how they can be used for study- address the ways in which remote sensing and GIS can be used
ing geomorphological systems and for geomorphological for mapping a variety of landforms. Specifically, they highlight
mapping. the numerous forms of data and methods that can be used.
 Remote Sensing and GIScience in Geomorphology: Introduction and Overview 15

Figure 9 Skyview-factor image over the Mount Everest region in Nepal. The skyview factor represents the degree to which the atmosphere can
be viewed, given hemispherical topographic variation. Flat terrain exhibits a skyview factor of 1 (white) and lower values (darker greytones)
depict valley bottoms exhibiting more relief. The algorithm examines the relief characteristics around each pixel using an azimuth direction
interval out to a specified distance. The magnitude of the parameter is directly related to topographic shielding of incoming radiation.
Consequently, this parameter is valuable in depicting variation in the cosmic-ray flux required for cosmogenic-based erosion modeling and in
assessing surface irradiance conditions for surface energy-budget modeling, as it governs the diffuse-skylight irradiance. Its evaluation provides
new research opportunities in geomorphology, hydrology, and glaciology.

They also cover landslide mapping and relate this to hazard
studies. This treatment represents many of the traditional and
new forms of analysis that have been used in geomorpho-
logical mapping.
3.1.4.4 Spatial Hydrology
Understanding the spatio-temporal complexities of the
hydrological system is essential for characterizing geo-
morphological systems and for the management of water re-
sources. Detailed information is required about the spatial
variability of functional units and processes that govern the
collection, flow, and storage of water (Beven and Moore,
1993). Consequently, information regarding climate, geology,
topography, land cover, soils, and regolith is needed to char-
acterize various components of the system.
Meso-scale information regarding atmospheric moisture
and precipitation rates can be obtained with passive and active
sensors (e.g., MODIS and TRMM). Spatio-temporal variations
in surface-water conditions are assessed by mapping water,
snow, and ice/glacier distributions (Bishop et al., 2004; Green
et al., 2006; Dozier et al., 2009). Active sonar systems can
produce detailed bathymetric data to facilitate volume esti-
mates for rivers and lakes. Remotely sensed data and analysis
can also be used to estimate the grain size of the snow, the
fraction of each pixel covered by snow, and the amount of
radiative forcing caused by absorbing impurities (Dozier et al.,
2009).
Satellite gravimetry allows assessment of water mass vari-
ations in the cryosphere and subsurface (Rodell et al., 2009;
Matsuo and Heki, 2010). Data from the gravity recovery and
climate experiment (GRACE) satellite mission can provide
monthly estimates of the Earth’s gravitational field (Tapley
et al., 2004). Time variations in the gravitational field can be
used to determine changes in the Earth’s mass distribution.
Water mass variations represent a dominate signal that can
influence the gravitational field (Wahr et al., 1998). Con-
sequently, if the data are preprocessed appropriately, and mass
variations due to the atmospheric and tectonic conditions are
accounted for, GRACE data can be used to evaluate water mass
variations in the cryosphere (e.g., Chen et al., 2007; Matsuo
and Heki, 2010) and detect large regional anomalies in ground
water fluctuations including the Amazon and ground water
depletion in India (Syed et al., 2005; Rodell et al., 2009).
Consequently, remote-sensing investigations have already
provided new insights into hydrological systems at local, re-
gional, and global scales.
Remote sensing and GIS have also contributed significantly
to hydrological studies, given the advent of DEMs and terrain
analysis capabilities (Beven and Moore, 1993; Wilson and
Gallant, 2000). The production of high- and moderate-reso-
lution DEMs using active sensor systems has markedly
16 Remote Sensing and GIScience in Geomorphology: Introduction and Overview
influenced hydrological modeling. Numerous topographic
parameters directly and indirectly control hydrological and
geomorphological processes. In general, these parameters have
been classified as primary and secondary (compound) par-
ameters that play a significant role in the spatial complexity of
the hydrological system (Beven and Moore, 1993; Wilson and
Gallant, 2000). Topographic information is now routinely
used for assessing and delineating drainage basins and
drainage networks, and DEMs are required for assessing water
flow direction and regimes.
Topographic parameters govern a variety of climate and
lithological processes including precipitation, surface-energy
balance, erosion, deposition, and rock stress fields and
strength. These in turn influence specific hydrologic processes
and storage including recharge, evaporation, infiltration, soil
moisture content, and surface saturation zones. One of the
most commonly used hydrologically based topographic par-
ameters or indexes is the wetness index, which is used to de-
termine the effects of the topography on the location and size
of saturated source areas (Wilson and Gallant, 2000). Land-
cover conditions also govern the magnitude of erosion and
infiltration capacity, and remote sensing provides detailed
information required for watershed analysis and the predic-
tion of discharge. Detailed mapping of impermeable surfaces
and GIS-based modeling of urban expansion represent other
aspects of watershed modeling and planning.
Terrain analysis is also critical for assessing precipitation
patterns, and topographic parameters have been found to in-
fluence precipitation rates. The orographic precipitation
mechanism is well understood and results in spatial anomal-
ies and gradients in precipitation magnitude in mountainous
terrain. However, valley structure and orientation can also
influence precipitation, as atmospheric flow can be topo-
graphically directed into basins or regions. This is the case in
Asia, where monsoon conditions move further inland due to
large glacial valleys that do not restrict air flow inland. Simi-
larly, slope angles and relief also govern precipitation, as these
parameters regulate the rate at which the air will be forced
upward. In orogens such as the Himalaya, precipitation can
vary significantly over relatively short distances. Even a
microtopographical variation has been shown to be related to
precipitation variations (Sharon et al., 1988).
The surface-energy budget is a critical component of the
hydrological system. It regulates ablation, snow melt, evap-
oration, and transpiration. Energy-budget modeling represents
the most rigorous approach for assessing surface-water runoff
and basin discharge. This type of modeling is critical for the
assessment of water resource potential and can be used to
address issues of sustainability. Currently, governments
around the world are monitoring drought conditions, snow-
pack variations, and glacier fluctuations to assess future water
supplies. Information related to the key energy-balance par-
ameters can be estimated via remote sensing and numerical
modeling including surface irradiance, albedo, and surface
temperature. Specifically, the net shortwave radiation com-
ponent can be assessed as atmospheric, topographic, and
surface albedo information is required to predict the direct,
diffuse-skylight, and adjacent-terrain irradiance. GIS-based ir-
radiance modeling can be used, and it is feasible to account
for multiscale topographic effects. Spectral and surface albedo
can be estimated using satellite imagery and bidirectional re-
flectance distribution models. Similarly, the net longwave ra-
diation component can also be evaluated.
The aforementioned discussion summarizes the many ways
in which remote sensing and GIS play a fundamental role in
hydrological modeling. Data analysis and spatial modeling
provide spatio-temporal information for characterizing the
spatial structure of hydrological systems and key parameters.
Hydrological modeling, however, is mostly concerned with the
flow of water over the surface and in the subsurface (Maidment,
1993). Historically, hydrological modeling has focused on the
temporal evolution of systems, characteristically assuming
uniform spatial properties for various system components.
Spatial hydrology represents the linkage of remote sensing and
GIS with hydrological models to more accurately account for
the spatio-temporal complexities in the hydrological system.
Maidment (1993) reviewed hydrological modeling in-
dependent of geospatial technologies and characterized the
nature of such models. They typically focus on surface water
hydrology, surface water quality, groundwater flow, and
groundwater transport. The spatial components associated
with the GIS-hydrological model linkage include watersheds,
pipes and stream channels, aquifers, lakes, and estuaries. A
treatment of the process mechanics of the fundamental flow
systems is beyond the scope of this chapter, although there are
several levels of GIS-based coupling that are itemized below.
• Hydrological assessment involves the mapping of hydro-
logical factors that relate to a situation. A classic example is
the use of the DRASTIC model for mapping groundwater-
contamination potential, which characterizes the likelihood
that the groundwater will be contaminated based on point
and areal sources of contamination, and topographic and
subsurface conditions. It represents the utility of GIS-based
spatial modeling and does not utilize any explicit physical
laws. Such spatial modeling is highly empirical and the re-
sults are also a function of ranking and weighting of infor-
mation layers.
• Hydrologic parameter determination is a very active research
area in hydrology. The objective is to accurately estimate the
parameters that go into hydrological models, based on at-
mospheric, terrain, and land-cover analysis. Many examples
have been previously presented.
• GIS-based hydrologic modeling represents a more detailed
level of coupling. This is possible, provided that the mod-
eling does not require significant temporal simulations.
Maidment (1993) indicates that one- and two-dimensional
steady-flow computations may be carried out using GIS-
based modeling, although often times, numerous assump-
tions are used to eliminate or reduce temporal variability.
Hydrologic modeling of flood inundation is an example of a
2-D GIS-based hydrological model where the groundwater
conditions are not taken into account.
• Hydrological modeling that links GIS-based modeling with
hydrological subsystem models accounts for the connection
between the surface and subsurface systems to examine
piezometric head surface and contamination plumes.
The contributions of remote sensing and GIS to hydrology
have the potential to open up new fields of study, as new
 Remote Sensing and GIScience in Geomorphology: Introduction and Overview
information and methodologies allow for the exploration of
new ways to characterize hydrological systems and processes.
The limiting factor is not the ability to characterize hydro-
logical processes mathematically or to solve the resulting
equations, but to accurately characterize the model par-
ameters, given their scale dependencies (Maidment, 1993).
Consequently, recent developments in geostatistics, object-
oriented analysis, geomorphometry, and spatial analysis and
modeling have considerably strengthened hydrology. New
developments in space–time representation and temporal
analysis are also required to facilitate GIS-based hydrological
modeling. Unfortunately, a chapter on this topic was not
completed in time for inclusion in this Treatise volume.
3.1.4.5 Erosion Modeling
Landscape evolution theories indicate that the topography in-
herently records the interaction between climatic, tectonic, and
surface processes. Consequently, geological and topographic
information can be used to assess erosion and topographic
evolution. Rather than focusing on empirical relationships,
landscape-evolution modeling attempts to formalize the
understanding of process mechanics and systems coupling.
Clearly, a better understanding of process–form relation-
ships is needed for developing improved erosion models.
Depending on the climatic and tectonic setting, variations in
weathering, fluvial, mass movement, and glacial processes
make it difficult to accurately assess the magnitude and spatial
distribution of erosion and denudation.
Erosion modeling facilitates the understanding of such
complications, as models account for the conservation of mass
and energy, and are based on a series of mass continuity
equations to address the erosion and deposition of rock and
sediment. Continuity assumes that the rate of change of alti-
tude is proportional to the volumetric sediment flux.
Uplift should account for isostatic and tectonic forcing
components. Tectonic forcing includes the advection of rock
mass, given structural controls, and the alteration of rock
strength, given topographic stress fields and far-field velocities.
The tectonics component requires the integration of mech-
anical models, as feedbacks exist between the topographic
stress field, rock strength, and erosion and uplift (Koons,
1995; Koons et al., 2002). The magnitude of denudation also
influences the isostatic compensation, which is a function of
the flexural rigidity of the crust and the wavelength of the
topography (Gilchrist et al., 1994).
Most models use a flexible parameterization scheme that
accounts for the depth of regolith production from weathering
(Tucker and Hancock, 2010). Parameterization schemes
should account for variations in lithology, precipitation, and
surface temperature and moisture conditions. Remote sensing
and terrain analysis of surface and atmospheric conditions can
be used to generate maps that may be associated with wea-
thering patterns. Key variables include surface irradiance,
temperature variation, atmospheric water vapor content, and
precipitation patterns. In mountain environments, tempera-
ture and precipitation variations may be considerable, given
highly variable topography and forcing factors (Barros et al.,
2006).
17
Hillslope sediment flux can be accounted for using a linear
or a nonlinear relation with the hillslope gradient. Diffusivity
coefficients are commonly used, with different values for dif-
ferent environments.
Bedrock river incision is important in many regions and
investigators have utilized the stream-power bedrock river in-
cision law to account for fluvial erosion. In general, the change
in elevation is modeled as:
qz
¼ KAm Sn ½6
qt
where K is bedrock erodability and A is the upstream catch-
ment area that is used as a proxy for discharge. The exponents
m and n are constants used to differentiate between the stream
power and shear stress-based rules.
Simulations of glacier erosion have also been conducted
where erosion is based on basal-sliding velocity and ice
thickness (MacGregor et al., 2000; Tomkin and Braun, 2002;
Pelletier et al., 2010). An abrasion model (Hallet, 1979) can be
used such that the rate of erosion is
qz
¼  aubs ½7
qt
where us is the basal sliding speed, and a and b are empirical
coefficients usually set to 1 or 2. Basal sliding is primarily
dependent on the basal shear stress, tb, and a bed-friction
parameter. Simulations of glacier erosion demonstrate the
complexity associated with relating process to form as glacier
erosion can enhance or reduce relief and controls valley spa-
cing and slope variability (Harbor, 1992; Bishop et al., 2003;
Tomkin and Braun, 2002; Pelletier et al., 2010).
Although such landscape-evolution modeling could be
implemented in a GIS (numerous issues do exist), most
models for orogen evolution simulations exist outside of GIS
environments. Common GIS-based erosion models focus on
soil erosion modeling, given relatively mild to moderate
topographic conditions. Soil erosion and sediment transport
studies are important in agriculture, water quality, and sedi-
ment budget modeling, and in determining the magnitude
of anthropogenic forcing. Geospatial technologies are ideally
suited for soil erosion modeling and numerous models
exist that use different sediment transport equations. Mitasova
and others in Chapter 3.9 describe different types of erosion
models and provide a physical and mathematical foundation
for understanding their ability to predict the magnitude and
spatial distribution of soil erosion over the landscape. Special
emphasis is on using existing GIS-based erosion models and
the visualization of erosion patterns.
Ultimately, remote sensing and GIS can provide new
information that can be used to develop more rigorous
parameterization schemes for erosion modeling. Geomor-
phometric characterization is required at each time interval to
drive process mechanics and process domain states, such that
modeling allows parameters, processes, and system charac-
teristics to be mapped. Further advances in GIS space–time
representations and formal process–form linkages are urgently
needed to facilitate improved GIS-based erosion modeling.
18 Remote Sensing and GIScience in Geomorphology: Introduction and Overview
3.1.4.6 Natural Hazards
Advances in remote sensing, GIT, and numerical modeling
have greatly improved the ability to assess a variety of natural
hazards. To date, remote sensing and GIS are routinely used
for assessing volcanic, earthquake, flood, slope stability, me-
teorological, and other environmental hazards. Given popu-
lation growth and rapid environmental change, extreme
hydro-meteorologic events are expected to increase, thereby
causing a higher frequency of hazards.
The literature is replete with research investigating the use
of remote sensing and GIS technology for landslide inventory,
slope-failure susceptibility mapping, and landslide hazard
assessment. Researchers are actively involved with developing,
testing, and validating new GIS-based spatial models to pre-
dict slope failures in order to keep landslide hazards from
becoming disasters in highly populated areas. The methodo-
logical approach is typically based on empirical relationships
or heuristics, statistical analysis, and the use of deterministic
physical-based models. Carrara and Pike (2008) indicated that
despite the variety of approaches, our abilities to spatially
predict slope failure and hazards are based on unsuitable data,
lack of numerical modeling, or improper characterization of
processes in slope-failure modeling.
A classic GIS example involves the common approach of
utilizing a criterion-weighted scheme based on controlling/
triggering factors that include topographic parameters, lith-
ology, land cover, and other environmental variables. These
factors are assumed to be causative and are subjectively ranked
based on knowledge of the location and existing inventory
information. Spatial overlay and weighting of GIS layers are
used to produce a composite index that is assumed to be re-
lated to slope-failure susceptibility. Such GIS-based empirical
approaches have significant limitations as key causative fac-
tors/variables involving earthquake activity, rock strength and
structure, topographic stress fields, soil depth, cohesion and
moisture, and climate forcing factors (precipitation and wind
direction) are not usually accounted for. It is important to
realize that many of these factors are dynamic and not ac-
curately characterized by static cartographic representations
using an ordinal measurement scale. The weighting schemes
also do not necessarily portray the relative importance of se-
lected processes and the dominance of site-specific feedback
mechanisms.
Other more sophisticated spatial analysis and pattern rec-
ognition approaches involving geostatistics and artificial in-
telligence techniques have also been evaluated, although it is
essential to recognize the difference between characterizing
spatial and nonlinear patterns and process-based modeling
based on fundamental physics and rock geomechanical
properties. Given the numerous approaches, both qualitative
and quantitative, remote sensing and GIS technology have
promoted more quantitative slope-failure studies. Research
indicates that there is a high degree of uncertainty with respect
to a dominant GIS-based approach to landslide-hazard pre-
diction (Carrara and Pike, 2008), as research tends to focus
more on investigating the method or technology rather than
focusing on inventories and causative factors. GIS data
manipulation cannot be expected to accurately characterize
key parameters and processes. GIS spatial analysis may
potentially produce new information that can help provide an
insight into better understanding mass-movement processes
and generate improved predictive spatial patterns. This most
likely will require an emphasis on the generation of new
causal factors and their integration into GIS-based numerical
models.
Another key example of the role of remote sensing and GIS
in hazards assessment is in characterizing flood magnitude
and inundation. Numerous types of flooding around the
world pose various risks to populations and infrastructure.
Flash, management-induced, and catastrophic flooding
(caused by landslide break-out floods and glacier hazards) can
all drastically alter the landscape and create flood disasters in
more populated areas. For example, in the spring of 2010 in
the Hunza region of Pakistan, a large landslide blocked the
Hunza River. Given the relatively large river discharge and the
increasing meltwater contributions by glaciers in the region,
the water impoundment grew at a significant rate. Multi-
temporal satellite imagery was used to monitor the progress of
the rapidly growing Lake Gojal (Figure 10). When combined
with topographic data and discharge information, the water
volume and spill-over time can be accurately predicted. It is
still slightly unclear as to the reasons why a catastrophic
breakout flood did not occur, although landslide experts
speculate that it is probably due to remobilized lake clays
imbedded with boulders greater than 10–20 m, making an
ideal lake dam.
Given the complexity of assessing various types of natural
hazards, it is reasonable to assume that an integrated approach
involving the use of numerical modeling, field data and
mapping, and remote sensing may be required. Pelletier et al.
(2005) used this approach for flood-hazard assessment on
alluvial fans. Raster-based hydraulic modeling, satellite-image
change detection, field mapping of recent flood inundation,
and surficial geological mapping were used to characterize
specific spatial details that are lacking in standard GIS ap-
proaches. Model predictions of flood inundation and flow
depths were tested against field and satellite-based flood maps
for two extreme events. They were able to predict spatially
complex flood hazards that strongly reflect small-scale topo-
graphic and geologic conditions.
Geomorphologists typically address the issue of using re-
mote sensing and GIS from a specific hazard susceptibility
perspective, although this is only the first phase or component
in a series of steps and analyses that are required for hazards
assessment and disaster risk management. Chapter 3.10 by van
Westin provides a comprehensive treatment of this topic,
highlighting the significance and difficulty of assessing specific
natural hazards, but also incorporating multihazard assess-
ment as part of a system to understand and manage complex
cascading hazard influences. Specifically, van Westin describes
the importance of the integration of approaches for hazards
assessment and the need to conduct hazard and risk assess-
ments at different scales, from global to community levels. van
Westin reveals that each scale or level of analysis has its own
objectives and spatial data requirements for hazard inven-
tories, causal factors, and elements at-risk mapping and
database development. van Westin also addresses vulnerability
assessment approaches. Collectively, his treatment reveals that
very little research has been conducted on establishing
 Remote Sensing and GIScience in Geomorphology: Introduction and Overview 19

Lake Gojal Growth

12 Jan, 2010: SPOT
16 Mar, 2010: ALI
02 May, 2010: ASTER
25 May, 2010: ASTER
07 July, 2010: ALI

(a) (b) (c)

Figure 10 Development of Lake Gojal. (a) Prerockslide ASTER FCC image mosaic of the Hunza Valley. Red represents vegetation, mainly
agricultural fields associated with villages. (b) ASTER false-color image 4 day before spillover. Note extensive late spring snowfields and glaciers
feeding Lake Gojal. (c) Advanced Land Imager (ALI) near-true-color base image, 7 July 2010, showing the growth of Lake Gojal based on SPOT,
ALI, and ASTER. Reproduced from Kargel, J.S., Leonard, G.J., Crippen, R.E., Delaney, K.B., Evans, S.G., Schneider, J.S., 2010. Satellite
monitoring of Pakistans rockslide-dammed Lake Gojal. Eos 91(43), 394–395, with permission from AGU.

complete GIS-based multihazard assessments, and that geo-
morphologists typically focus on evaluating individual hazard
types. Although geospatial technologies serve as a basis for
hazard and risk assessment, the scientific advances in hazards
and risk assessment have remained in the scientific com-
munity and have not effectively reached end-users such as
planner, managers, and policy-decision makers.
3.1.4.7 Visualization
Visualization is an emerging science that represents research
developments in cartography, computer graphics, computa-
tional geometry, cognitive science, and psychology (Gahegan,
2000). It is used in a variety of ways to view imagery, integrate
information, view spatial and temporal patterns, and inter-
actively examine abstract-data-spaces and numerical simu-
lations. In the Earth sciences, many have recognized the
significance of visualizing and perceiving patterns of structure,
processes, features, and relationships, in an attempt to validate
analysis and modeling approaches and understand many
concepts in geomorphology regarding processes, landforms,
and system dynamics. With the advent of rapidly improving
computer and visualization technologies, a trend has emerged
involving more human–computer interaction that exploits the
significance of the human visualization system.
In general, visualization is the interaction between com-
puters and humans based on a graphical-user-interface en-
vironment that makes use of a variety of visualization
techniques. In the geosciences, visualization is used in a var-
iety of ways, and Earth scientists need to be familiar with
visualization techniques suitable for working with specific
data sets and applications. The major functional uses can be
categorized as follows, although these are not necessarily
mutually exclusive.
• Exploratory visual analysis (EVA). The goal is to explore the
nature of the data such that spatial and temporal relation-
ships and structure become apparent. This is analogous to
data mining or knowledge discovery and is closely related to
finding statistical anomalies in the data. Classic examples
include visualizations of scatterplots and spectral-feature
spaces in remote sensing. Such visualizations allow the ex-
ploration of spectral end-members that can be used in linear
and nonlinear spectral mixing to facilitate accurate mapping
of surface composition and assessing the composition mix-
ture of materials that influence surface processes (e.g., sur-
face energy budget, ablation).
A critical EVA example in geomorphology is visualizing
data validity and uncertainty. DTM and the accuracy of
DEMs is a critical first step toward meaningful terrain an-
alysis. Three-dimensional perspective viewing of the land-
scape using anaglyphs or fly-by simulations can be
effectively utilized to detect high-frequency errors in DEMs
and low-frequency patterns caused by spatial interpolation
algorithms. Similarly, the use of pattern-recognition algo-
rithms such as neural networks and fuzzy classifiers can be
used to generate fuzzy uncertainties or likelihoods of
20 Remote Sensing and GIScience in Geomorphology: Introduction and Overview
accurate land-cover or surface matter classifications. These
approaches and quantitative measures communicate the
concepts of uncertainty, error, validity, or probability.
A geomorphometry example of EVA involves the explor-
ation of scale dependence and prediction of spatial patterns
in understanding complex landscapes. Surface parameter
magnitudes such as relief, slope, and other important mul-
tiscale topographic parameters are dependent on the com-
putational scale of analysis (Bishop et al., 2012).
Consequently, visualization of magnitude-scale relations
and distribution must be explored to determine the appro-
priate scale. Spatial scale dependence involves analysis and
display to determine the scale of spatial autocorrelation and
the directional fabric of the topography. Semivariograms can
be used to characterize scale dependence and the anisotropic
nature of the topography, although visualization is required
to appropriately assign semivariogram models to experi-
mental variograms in order to view simulated spatial pat-
terns (Webster and Oliver, 2007). Furthermore, visualization
is critical in an attempt to understand the anisotropic nature
of topography caused by various landscape evolution com-
ponents, as erosion, deformation, rock strength, and faulting
govern the directional dependence. To date, this aspect of
EVA in geomorphometry is yet to be fully exploited, al-
though the potential for new knowledge generation and new
theory exists.
• Visual analysis and modeling. A strong empirical basis is
associated with the use of geospatial technologies for an-
alysis and modeling (Bishop and Shroder, 2004b; Deng,
2007; Bishop et al., 2012). Whereas this provides for flexi-
bility in developing metrics, software tools, and new algo-
rithms and analysis approaches, it also raises important
questions concerning the validity of analysis and the use of
geomorphological information in integrative science.
Spatial analysis commonly relies on an index approach to
characterize various types of landscape information. These
indices or metrics are generally based on a association
with spatial position or a topographic parameter, although
they do not adequately characterize process mechanics,
scale dependencies, or temporal dynamics. Furthermore,
manipulations of these indices are based on concepts of
ranking, weighting, membership, prototypicality, scaling,
thresholding, heuristic rules, and additional empirical co-
efficients that can be used to alter the spatial extent of an-
alysis results. Whereas these metrics have value in terms of
flexibility to address issues associated with semantic mean-
ings, spatial uncertainty, subjective interpretations, data in-
tegration, classification, and variable definitions, the results
can be highly variable. Visualization is required to evaluate
the magnitude of index/metric values and determine whe-
ther or not such patterns actually represent ‘reality’ (i.e.,
morphology, physical properties, genetics, dynamics, land-
forms). Although new spatio-temporal information can be
potentially generated in this way, spatial patterns and the
delineation of boundaries and zones must be carefully
examined to determine the sensitivity of combinations of
parameters. Consequently, visualization is required for ad-
equate and accurate analysis and information production.
Finally, numerical modeling of climate, surface processes,
and tectonics can provide valuable insights into erosion,
relief production, feedback mechanisms, and the role of
surface processes in landscape evolution. Animations of
simulations can provide three-dimensional perspectives on
the nature of polygenetic evolution and the genesis of
landforms. Visualizations of the temporal dynamics of sys-
tem-critical parameters such as precipitation, surface-sedi-
ment flux, influx of mass due to uplift, and topographic
constraints can be very useful for evaluating existing repre-
sentations of process mechanics and dominant forcing fac-
tors. Consequently, visualization will play an ever-increasing
role in fundamental analysis and modeling efforts.
• Visual Synthesis. Visualization approaches that allow an
evaluation of various types of thematic information can fa-
cilitate geomorphological studies. Classic examples include
examination of false-color composite images to facilitate
geological and land-cover mapping. Spatial overlay and the
use of symbols, color, transparency, and material properties
can be used to depict a variety of landscape conditions. In
essence, geomorphological mapping represents a visual syn-
thesis of landscape conditions, as traditional and modern-day
geomorphological maps are commonly generated based on
the integration of information reflecting climatic, geological,
morphometric, pedogenic, land cover, and geochronological
variation. Bishop et al. (2012) provided a treatment of the use
of geospatial technologies for geomorphological mapping,
and indicate the role of visualizing and integrating a multi-
tude of data types for assessing geomorphological conditions.
Given the complexity of polygenetic evolution and the gen-
eral lack of standard protocols for information integration
and mapping, new techniques and cartographic approaches
seem warranted, as there are different integration require-
ments based on a variety of mapping perspectives.
• Presentation. Numerous techniques and approaches can be
used for presenting information. These techniques and
products include traditional graphic products such as cross
sections, profiles, images, maps, animations, virtual globes,
and virtual realities. Effective communication and dis-
semination of geomorphological information to planners,
managers, and policy makers is a central theme. This aspect
of visualization is essential to promote and facilitate the use
of scientific information in the decision-making process.
Geomorphological research has been greatly facilitated by
rapid advances in geospatial technologies and augmented
realities. Chapter 3.11 by Smith and others provides a more
detailed examination of some fundamental aspects of visual-
ization in the geosciences. Specifically, they provide examples
of techniques and approaches that allow various aspects of
geomorphology to be effectively studied. Important examples
involving remote sensing and terrain analysis are included.
The treatment clearly demonstrates that visualization techni-
ques and approaches are valuable in Earth science
investigations.
3.1.5 Conclusions
The rapid advancement of geospatial technologies has had a
profound effect on the discipline of geomorphology. Remote
 Remote Sensing and GIScience in Geomorphology: Introduction and Overview
sensing and GIS studies are now commonplace in geo-
morphological investigations, as a better understanding of
surface processes and landscape evolution is sought. Earth
scientists are increasingly using new spatio-temporal datasets
and GIS technology for analyzing and modeling various as-
pects of geomorphological systems and for addressing con-
ceptual and practical issues (i.e., scale dependency, process-
form and process-pattern relationships, digital geomorpho-
logical mapping, landscape evolution modeling, natural haz-
ard modeling).
The 41st Binghamton symposium highlighted many of the
aforementioned developments in geomorphology related to
geospatial technologies. The delegates were amazed at the
rapid evolution of geospatial technologies and were interested
in the many new ways to study geomorphological systems and
facilitate practical-problem solving. Nevertheless, the delegates
realized the difficulty of effectively utilizing new data and
analysis/modeling approaches, because the effective use of
remote and GIS in geomorphology requires multidisciplinary
domain knowledge including radiation transfer, matter/energy
interactions, sensor-system characteristics, mathematical under-
pinnings of algorithms, analysis approaches, models, and
Earth science. It is not possible to push a button and obtain
the desired results, as expertise and time is required to effect-
ively produce accurate information. It is also essential to rec-
ognize the inherently empirical nature associated with using
GIS in geomorphology (Bishop et al., 2012). Consequently, it
is now important for geomorphologists to take remote sensing
and GIS coursework.
Clearly, the incorporation of remote sensing and GIScience
investigations into geomorphology has greatly contributed to
the quantitative evolution in geomorphology, compared with
its more qualitative beginnings. Geospatial technologies have
revolutionized the way in which scientists study the Earth, and
quantitative analysis and modeling of the landscape offer
many new research opportunities. It should be kept in mind,
however, that qualitative information derived from the field
and via human interpretation of data can contribute greatly to
the interpretation of quantitative results. The inherent digital
and quantitative nature of modern-day analysis should be
carefully examined with respect to GIS-based empiricism and
the ability to produce repeatable results. Only through quan-
titative formalization of geomorphological concepts and the-
ories can such repeatable results be expected, which will go a
long way toward establishing geomorphology as a required
component in integrated science. Advances in remote sensing
and GIScience contribute to this important goal.
References
Allen, T.R., Walsh, S.J., 1993. Characterizing multitemporal alpine snowmelt
patterns for ecological inferences. Photogrammetric Engineering and Remote
Sensing 59(10), 1521–1529.
Banerjee, P., Bürgmann, R., 2002. Convergence across the northwest Himalaya from
GPS measurements. Geophysical Research Letters 29(13), 1–4.
Barros, A.P., Chiao, S., Lang, T.J., Burbank, D., Putkonen, J., 2006. From weather
to climate-seasonal and interannual variability of storms and implications for
erosion processes in the Himalaya. In: Willett, S.D., Hovius, N., Brandon, M.T.,
Fisher, D.M. (Eds.), Tectonics, Climate, and Landscape Evolution. No. 398 in
Special paper. Geological Society of America, Boulder, Colorado, pp. 17–38.
21
Ben-Dor, E., Chabrillat, S., Demattê, J.A.M., Taylor, G.R., Hill, J., Whiting, M.L.,
Sommer, S., 2009. Using imaging sepctroscopy to study soil properties. Remote
Sensing of Environment 113, S38–S55.
Beven, K.J., Moore, I.D. (Eds.), 1993. Terrain Analysis and Distributed Modelling in
Hydrology. Advances in Hydrological Processes. John Wiley and Sons,
Chichester.
Bilham, R., Larson, K., Freymueller, J.Project Idylhim Members, 1997. GPS
measurements of present-day convergence across the Nepal Himalaya. Nature
386, 61–64.
Bishop, M.A., 2009. A generic classification for the morphological and
spatial complexity of volcanic (and other) landforms. Geomorphology 111,
104–109.
Bishop, M.P., Barry, R.G., Bush, A.B.G., et al., 2004. Global land-ice measurements
from space (GLIMS): remote sensing and GIS investigations of the Earth’s
cryosphere. Geocarto International 19(2), 57–84.
Bishop, M.P., Bonk, R., Kamp, Jr. U., Shroder, Jr. J.F., 2001. Terrain analysis
and data modeling for alpine glacier mapping. Polar Geography 24(4),
257–276.
Bishop, M.P., Bush, A.B.G., Copland, L., Kamp, U., Owen, L.A., Seong, Y.B.,
Shroder, J.F., 2010. Climate change and mountain topographic evolution in the
Central Karakoram, Pakistan. Annals of the Association of American Geographers
100(4), 772–793.
Bishop, M.P., James, L.A., Shroder, J.F., Walsh, S.J., 2012. Geospatial technologies
and digital geomorphological mapping: concepts, issues and research.
Geomorphology 136, 5–26.
Bishop, M.P., Shroder, Jr. J.F. (Eds.), 2004a. Geographic Information Science and
Mountain Geomorphology. Springer-Praxis, Chichester.
Bishop, M.P., Shroder, Jr. J.F. (Eds.), 2004b. GIScience and mountain
geomorphology: overview, feedbacks, and research directions. Springer-Praxis,
Chichester, pp. 1–31.
Bishop, M.P., Shroder, Jr. J.F., Colby, J.D., 2003. Remote sensing and
geomorphometry for studying relief production in high mountains.
Geomorphology 55, 345–361.
Boochs, F., Kupfer, G., Dockter, K., Kuhbauch, W., 1990. Shape of the red edge as
vitality indicator for plants. International Journal of Remote Sensing 11,
1741–1753.
Boulton, S.J., Whittaker, A.C., 2009. Quantifying the slip rates, spatial distribution
and evolution of active normal faults from geomorphic analysis: field examples
from oblique-extensional graben, southern Turkey. Geomorphology 104,
299–316.
Brändli, M., 1996. Hierarchical models for the definition and extraction of terrain
features. In: Burrough, P., Frank, A.U. (Eds.), Geographic Objects with
Indeterminate Boundaries. Taylor and Francis, London, pp. 257–270.
Burbank, D.W., Blythe, A.E., Putkonen, J., et al., 2003. Decoupling of erosion and
precipitation in the Himalayas. Nature 426, 652–655.
Burns, R.G., 1970. Mineralogical Application to Crystal Field Theory. Cambridge
University Press, Cambridge, UK.
Burrough, P., 1989. Fuzzy mathematical methods for soil survey and land
evaluation. Journal of Soil Science 40, 477–492.
Burrough, P., 1996. Natural objects with indeterminate boundaries. In: Burrough, P.,
Frank, A.U. (Eds.), Geographic Objects with Indeterminate Boundaries. Taylor
and Francis, London, pp. 3–28.
Burrough, P., van Gaans, P., MacMillan, R., 2000. High-resolution landform
classification using fuzzy k-means. Fuzzy Sets and Systems 113(1), 37–52.
Carrara, A., Pike, R.J., 2008. GIS technology and models for assessing landslide
hazard and risk. Geomorphology 94, 257–260.
Chen, J.L., Wilson, C.R., Tapley, B.D., Blankenship, D.D., Ivins, E.R., 2007.
Pategonia Icefield melting observed by Gravity Recovery and Climate Experiment
(GRACE). Geophysical Research Letters 34, L22501.
Collins, W., Chang, S.H., Raines, G., Canney, F., Ashley, R., 1983. Airborne
biogeochemical mapping of hidden mineral deposits. Economic Geology 78,
737–743.
Curtis, C.D., 1976. Chemistry of rock weathering: fundamental reactions and
controls. In: Derbyshire, E. (Ed.), Geomorphology and Climate. John Wiley and
Sons, New York, Chapter 2, pp. 25–57.
Dematt̂e, J.A.M., 2002. Characterization and discrimination of soils by their reflected
electromagnetic energy. Brazilian Journal of Agricultural Research 37,
1445–1458.
Deng, Y., 2007. New trends in digital terrain analysis: landform definition,
representation, and classification. Progress in Physical Geography 31, 405–419.
Deng, Y., Wilson, J.P., 2008. Multi-scale and multi-criteria mapping of mountain
peaks as fuzzy entities. International Journal of Geographical Information Science
22(2), 205–218.
22 Remote Sensing and GIScience in Geomorphology: Introduction and Overview
Dozier, J., Green, R.O., Noline, A.W., Painter, T.H., 2009. Interpretation of snow
properties from imaging spectrometry. Remote Sensing of Environment 113,
S25–S37.
Finlayson, D.P., Montgomery, D.R., 2003. Modeling large-scale fluvial erosion in
GIS. Geomorphology 53, 147–164.
Finlayson, D.R., Montgomery, D.R., Hallet, B., 2002. Spatial coincidence of rapid
inferred erosion with young metamorphic massifs in the Himalayas. Geology 30,
219–222.
Florinsky, I.V., 1998. Combined analysis of digital terrain models and remotely
sensed data in landscape investigations. Progress in Physical Geography 22(1),
33–60.
Flowers, G.E., Roux, N., Pimentel, S., Schoof, C.G., 2011. Present dynamics and
future prognosis of a slowly surging glacier. The Cryosphere 5, 299–313.
Frankl, A., Nyssen, J., Calvet, M., Heyse, I., 2010. Use of Digital Elevation Models
to understand and map glacial landforms-The case of the Canigou Massif
(Eastern Pyrenees, France). Geomorphology 115, 78–89.
Gahegan, M., 2000. Visualization as a tool for geocomputation. In: Openshaw, S.,
Abrahart, R.J. (Eds.), GeoComputation. Taylor and Francis, London, pp.
253–274.
Gao, J., Liu, Y., 2001. Applications of remote sensing, GIS and GPS in glaciology:
a review. Progress in Physical Geography 25(4), 520–540.
Gessler, P., Pike, R., MacMillan, R.A., Hengl, T., Reuter, H.I., 2009. The future of
geomorphometry. In: Hengl, T., Reuter, H.I. (Eds.), Geomorphometry: Concepts,
Software, and Applications. Developments in Soil Science. Elsevier, Amsterdam,
The Netherlands, pp. 637–651.
Gilchrist, A.R., Summerfield, M.A., Cockburn, H.A.P., 1994. Landscape dissection,
isostatic uplift, and the morphologic development of orogens. Geology 22,
963–966.
Goetz, A.F.H., Rowan, L.C., 1981. Geologic remote sensing. Science 211,
781–791.
Green, R.O., Painter, T.H., Roberts, D.A., Dozier, J., 2006. Measuring the expressed
abundance of the three phases of water with an imaging spectrometer over
melting snow. Water Resources Research 42, W10402.
Grove, C.I., Hook, S.L., Paylor, E.D., 1992. Laboratory reflectance spectra of 160
minerals, 0.4 to 2.5 micrometers. Technical Report 92-2, NASA-Jet Propulsion
Laboratory, Pasadena.
Gustavsson, M., Seijmonsbergen, A.C., Kolstrup, E., 2008. Structure and contents
of a new geomorphological GIS database linked to a geomorphological map –
with an example from Liden, central Sweden. Geomorphology 95, 335–349.
Guth, P., 2009. Geomorphometry in MicroDEM. In: Hengl, T., Reuter, H.I. (Eds.),
Geomorphometry: Concepts, Software, and Applications. Developments in Soil
Science. Elsevier, Amsterdam, The Netherlands, pp. 351–366.
Hallet, B., 1979. A theoretical model of glacial abrasion. Journal of Glaciology 23,
39–50.
Harbor, J., 1992. Numerical modeling of the development of U-shaped valleys by
glacial erosion. Geological Society of America Bulletin 104, 1364–1375.
Hardgrove, C., Moersch, J., Whisner, S., 2009. Thermal imaging of alluvial fans: a
new technique for remote classification of sedimentary features. Earth and
Planetary Science Letters 285, 124–130.
Hengl, T., Reuter, H.I. (Eds.), 2009. Geomorphometry: Concepts, Software, and
Applications. Developments in Soil Science. Elsevier, Amsterdam, The
Netherlands, 33, 765 pp.
Hill, J., Mégier, J., Mehl, W., 1995. Land degradation, soil erosion and
desertification monitoring in Mediterranean ecosystems. Remote Sensing
Reviews 12, 107–130.
Hill, J., Schütt, B., 2000. Mapping complex patterns of erosion and stability in dry
Mediterranean ecosystems. Remote Sensing of Environment 74, 557–569.
Hofierka, J., Mitasova, H., Neteler, M., 2009. Geomorphometry in GRASS GIS.
In: Hengl, T., Reuter, H.I. (Eds.), Geomorphometry: Concepts, Software, and
Applications. Developments in Soil Science. Elsevier, Amsterdam, The
Netherlands, pp. 387–410.
Houze, R.A., Wilton, D.C., Smull, B.F., 2007. Monsoon convection in the Himalayan
region a seen by the TRMM Precipitation Radar. Quarterly Journal of Royal
Meteorological Society 133, 1389–1411.
Jamieson, S.S.R., Sinclair, H.D., Kirstein, L.A., Purves, R.S., 2004. Tectonic forcing
of longitudinal valleys in the Himalaya: morphological analysis of the Ladakh
Batholith, North India. Geomorphology 58, 49–65.
Kääb, A., Vollmer, M., 2000. Surface geometry, thickness changes and flow fields
on creeping mountain permafrost: automatic extraction by digital image analysis.
Permafrost and Periglacial Processes 11, 315–326.
Kargel, J.S., Leonard, G.J., Crippen, R.E., Delaney, K.B., Evans, S.G., Schneider,
J.S., 2010. Satellite monitoring of Pakistans rockslide-dammed Lake Gojal. Eos
91(43), 394–395.
King, M.D., Menzel, P., Kaufman, Y.J., et al., 2003. Cloud and aerosol properties,
precipitable water, and profiles of temperature and water vapor from MODIS.
IEEE Transactions on Geoscience and Remote Sensing 41(2), 442–458.
Klingseisen, B., Metternicht, G., Paulus, G., 2008. Geomorphometric landscape
analysis using a semi-automated GIS-approach. Geomorphology 23, 109–121.
Koons, P.O., 1995. Modeling the topographic evolution of collisional belts. Annual
Review Earth and Planetary Science 23, 375–408.
Koons, P.O., Zeitler, P.K., Chamberlain, C.P., Craw, D., Meltzer, A.S., 2002.
Mechanical links between erosion and metamorphism in Nanga Parbat, Pakistan
Himalaya. American Journal of Science 302, 749–773.
Kruse, F.A., Lefkoff, A.B., Dietz, J.B., 1993. Expert system-based mineral mapping in
Northern Death Valley, California/Nevada, using the Airborne Visible/Infrared
Imaging Spectrometer (AVIRIS). Remote Sensing of Environment 44, 309–336.
Lagacherie, P., Andrieux, P., Bouzigues, R., 1996. Fuzziness and uncertainty of soil
boundaries: from reality to coding in GIS. In: Burrough, P., Frank, A.U. (Eds.),
Geographic Objects with Indeterminate Boundaries. Taylor and Francis, London,
pp. 275–286.
Lane, R., 2002. Ground and airborne electromagnetic methods. In: Papp E. (Ed.).
Geophysical and Remote Sensing Methods for Regolith Exploration. CRC LEME.
Open File Report, http://crcleme.org.au/Pubs/OPEN20FILE20REPORTS/
OFR20144/OFR144.html.
Liang, S., 2007. Recent developments in estimating land surface biophysical
variables from optical remote sensing. Progress in Physical Geography 31(5),
501–516.
Lindsay, J.B., 2009. Geomorphometry in TAS GIS. In: Hengl, T., Reuter, H.I. (Eds.),
Geomorphometry: Concepts, Software, and Applications. Developments in Soil
Science. Elsevier, Amsterdam, The Netherlands, pp. 367–386.
Luoto, M., Marmion, M., Hjort, J., 2010. Assessing spatial uncertainty in predictive
geomorphological mapping: a multi-modeling approach. Computers and
Geosciences 36, 355–361.
Maathuis, T.H.B.H.P., Wang, L., 2009. Geomorphometry in ILWIS. In: Hengl, T.,
Reuter, H.I. (Eds.), Geomorphometry: Concepts, Software, and Applications.
Developments in Soil Science. Elsevier, Amsterdam, The Netherlands, pp.
309–331.
MacGregor, K.R., Anderson, R.S., Anderson, S.P., Waddington, E.D., 2000.
Numerical simulations of glacial-valley longitudinal profile evolution. Geology
28, 1031–1034.
Maidment, D.R., 1993. GIS and hydrological modeling. In: Goodchild, M.F., Parks,
B.O., Steyaert, L.T. (Eds.), Environmental Modeling with GIS. Oxford University
Press, New York, pp. 147–167.
Marcus, W.A., Legleiter, C.J., Aspinall, R.J., Boardman, J.W., Crabtree, R.L., 2003.
High spatial resolution, hyperspectral (HSRH) mapping of instream habitats,
depths, and woody debris in mountain streams. Geomorphology 55, 363–380.
Masunaga, H., Iguchi, T., Oki, R., Kachi, M., 2002. Comparison of rainfall products
derived from TRMM Microwave Imager and Precipitation Radar. Journal of
Applied Meteorology 41, 849–862.
Matsuo, K., Heki, K., 2010. Time-variable ice loss in Asia high mountains from
satellite gravimetry. Earth and Planetary Science Letters 290, 30–36.
Metternicht, G.I., Zinck, J.A., 2003. Remote sensing of soil salinity: potentials and
constraints. Remote Sensing of Environment 85, 1–20.
Molnar, P., England, P., 1990. Late Cenozoic uplift of mountain ranges and global
climate change: chicken or egg? Nature 346, 29–34.
Montgomery, D.R., Hallet, B., Yuping, L., Finnegan, N., Anders, A., Gillespie, A.,
Greenberg, H.M., 2004. Evidence for Holocene megafloods down the Tsangpo
River gorge, Southeastern Tibet. Quaternary Research 62, 201–207.
Napieralski, J., Nalepa, N., 2010. The application of control charts to determine the
effect of grid cell size on landform morphometry. Computers and Geosciences
36, 222–230.
Nesbitt, S.W., Anders, A.M., 2009. Very high resolution precipitation climotologies
from the Tropical Rainfall Measureing Mission precipitation radar. Geophysical
Research Letters 36, L15815.
Olaya, V., Conrad, O., 2009. Geomorphometry in SAGA. In: Hengl, T., Reuter, H.I.
(Eds.), Geomorphometry: Concepts, Software, and Applications. Developments in
Soil Science. Elsevier, Amsterdam, The Netherlands, pp. 293–308.
Peckham, S.D., 2009. Geomorphometry in RiverTools. In: Hengl, T., Reuter, H.I.
(Eds.), Geomorphometry: Concepts, Software, and Applications. Developments in
Soil Science. Elsevier, Amsterdam, The Netherlands, pp. 411–430.
Pelletier, J.D., Comeau, D., Kargel, J., 2010. Controls of glacial valley spacing on
Earth and Mars. Geomorphology 116, 189–201.
Pelletier, J.D., Mayer, L., Pearthree, P.A., House, P.K., Demsey, K.A., Klawon, J.E.,
Vincent, K.R., 2005. An integrated approach to flood hazard assessment on
alluvial fans using numerical modeling, field mapping, and remote sensing.
Geological Society of America Bulletin 117(9/10), 1167–1180.
 Remote Sensing and GIScience in Geomorphology: Introduction and Overview
Pike, R.J., 1995. Geomorphometry – progress, practise and prospect. Zeitschrift für
Geomorphologie 101(7), 221–238.
Pike, R.J., 2000. Geomorphometry – Diversity in quantitative surface analysis.
Progress in Physical Geography 24(1), 1–20.
Ralston, B.A., 1994. Object oriented spatial analysis. In: Fotheringham, S.,
Rogerson, P. (Eds.), Spatial Analysis and GIS. Taylor and Francis, London, pp.
165–185.
Reuter, H.I., Nelson, A., 2009. Geomorphometry in ESRI packages. In: Hengl, T.,
Reuter, H.I. (Eds.), Geomorphometry: Concepts, Software, and Applications.
Developments in Soil Science. Elsevier Amsterdam, The Netherlands, pp.
269–291.
Rodell, M., Velicogna, I., Famiglietti, J.S., 2009. Satellite-based estimates of
groundwater depletion in India. Nature 460, 999–1002.
Saadat, H., Bonnell, R., Sharifi, F., Mehuys, G., Namdar, M., Ale-Ebrahim, S., 2008.
Landform classification from a digital elevation model and satellite imagery.
Geomorphology 100, 453–464.
Schaepman, M.E., Ustin, S.L., Plaza, A.J., Painter, T.H., 2009. Earth system science
related imaging spectroscopy- An assessment. Remote Sensing of Environment
113, S123–S137.
Schaper, J., Martinec, J., Seidel, K., 1999. Distributed mapping of snow and
glaciers for improved runoff modelling. Hydrological Processes 13(12-13),
2023–2031.
Schmidt, J., Dikau, R., 1999. Extracting geomorphic attributes and objects from
digital elevation models - Semantics, methods, future needs. In: Dikau, R.,
Saurer, H. (Eds.), GIS for Earth Surface Systems: Analysis and Modelling of the
Natural Environment. Gebrder Borntraeger, Berlin, Stuttgart, pp. 153–173.
Schneevoigt, N.J., van der Linden, S., Thamm, H., Schrott, L., 2008. Detecting
alpine landforms from remotely sensed imagery. A pilot study in the Bavarian
Alps. Geomorphology 93, 104–119.
Sella, G.F., Stein, S., Dixon, T.H., Craymer, M., James, T.S., Mazzotti, S., Dokka,
R.K., 2007. Observation of glacial isostatic adjustment in ‘‘stable’’ North America
with GPS. Geophysical Research Letters 34, L02306.
Sharon, D., Morin, J., Moshe, Y., 1988. Micro-topographical variations of rainfall
incident on ridges of a cultivated field. Transactions of the American Society of
Agricultural Engineering 31, 1715–1722.
Short, N.M., Blair, R.W. (Eds.), 1986. Geomorphology From Space: A Global
Overview of Regional Landforms. NASA, Washington, DC.
Shroder, Jr. J.F., Bishop, M.P., 2000. Unroofing of the Nanga Parbat Himalaya.
In: Khan, M.A., Treloar, P.J., Searle, M.P., Jan, M.Q. (Eds.), Tectonics of the
Nanga Parbat Syntaxis and the Western Himalaya. No. 170 in Special
Publication. Geological Society, London, pp. 163–179.
Shroder, Jr. J.F., Bishop, M.P., 2003. A perspective on computer modeling and
fieldwork. Geomorphology 53, 1–9.
Smith, G.R., Woodard, J.C., Heywood, D.I., Gibbard, P.L., 2000. Interpreting
Pleistocene glacial features from SPOT HRV data using fuzzy techniques.
Computers and Geosciences 26, 479–490.
Smith, M.J., Pain, C.F., 2009. Applications of remote sensing in geomorphology.
Progress in Physical Geography 33(4), 568–582.
Syed, T.H., Famiglietti, J.S., Chen, J., Rodell, M., Seneviratne, S.I., Viterbo, P.,
Wilson, C.R., 2005. Total basin discharge for the Amazon and Mississippi river
basins from GRACE and a land-atmosphere water balance. Geophysical Research
Letters 32, L24404.
Tapley, B., Bettapur, S., Ries, J., Thompson, P.F., Watkins, M., 2004. GRACE
measurements of mass variability in the Earth system. Science 305, 503–505.
Tarolli, P., Arrowsmith, J.R., Vivoni, E.R., 2009. Understanding Earth surface
processes from remotely sensed digital terrain models. Geomorphology 113,
1–3.
Tate, J.N., Wood, J., 2001. Fractals and scale dependencies in topography. In: Tate,
J.N., Atkinson, P.M. (Eds.), Modelling Scale in Geographical Information
Science. John Wiley and Sons, Chichester, pp. 35–52.
23
Taylor, G.R., 2004. Field and image spectrometry for soil mapping. Proceedings
of the 12th Australian Remote Sensing Conference. Fremantle, WA,
Australia.
Taylor, G.R., Bennett, B.A., Mah, A.H., Hewson, R.D., 1994. Spectral properties of
salinised land and implications for interpretation of 24 channel imaging
spectrometry. Proceedings of the First International Remote Sensing Conference
ane Exhibition. Vol. 3. Strasbourg, France.
Tomkin, J.H., Braun, J., 2002. The effect glaciation sas on the relief of a fast
growing orogen: a numerical modelling study. American Journal of Science 302,
169–190.
Trudgill, S.T., 1976. Rock weathing and climate: quantitative and experimental
aspects. In: Derbyshire, E. (Ed.), Geomorphology and Climate. John Wiley and
Sons, New York, Chapter. 3, pp. 59–99.
Tucker, G.E., Hancock, G.R., 2010. Modeling landscape evolution. Earth Surface
Processes and Landforms 35, 28–50.
Usery, E.L., 1996. A conceptual framework and fuzzy set implementation for
geographic features. In: Burrough, P., Frank, A.U. (Eds.), Geographic Objects
with Indeterminate Boundaries. Taylor and Francis, London, pp. 71–85.
van der Meer, F.D., de Jong, S.M. (Eds.), 2006. Imaging Spectrometry: Basic
Principles and Prospective Applications. Remote Sensing and Digital Image
Processing. Springer, Dordrecht.
Vaughan, R.G., Hook, S.J., Ramsey, M.S., Realmuto, V.J., Schneider, D.J., 2005.
Monitoring eruptive activity at Mount St. Helens with TIR image data.
Geophysical Research Letters 32, L19305.
Vitek, J.D., Giardino, J.R., Fitzgerald, J.W., 2008. Mapping geomorphology: a
journey from paper maps through computer mapping to GIS and virtual reality.
Geomorphology 16, 233–249.
Wahr, J., Molenaar, M., Bryan, F., 1998. Time variability in the Earth’s gravity field:
Hydrological and oceanic effects and their possible detection using GRACE.
Journal of Geophysical Research 103, 30,205–30,229.
Walsh, S.J., Moody, A., Allen, T.R., Brown, D.G., 1997. Scale dependence of NDVI
and its relationship to mountainous terrain. In: Quattrochi, D.A., Goodchild, M.F.
(Eds.), Scale in Remote Sensing and GIS. CRC/Lewis Publishers, Boca Raton,
pp. 27–55.
Warner, T., Shank, M., 1997. An evaluation of the potential for fuzzy classification
of multispectral data using artificial neural networks. Photogrammetric
Engineering and Remote Sensing 63(11), 1285–1294.
Webster, R., Oliver, M.A., 2007. Geostatistics for Environmental Scientists. Statistics
in Practice. John Wiley and Sons Ltd., Chichester.
Whipple, K.X., 2009. The influence of climate on the tectonic evolution of mountain
belts. Nature Geoscience 2, 97–104.
Wichmann, V., Heckmann, T., Haas, F., Becht, M., 2009. A new modeling approach
to delineate the spatial extent of alpine sediment cascades. Geomorphology 111,
70–78.
Wilford, J., 2002. Airborne gammay-ray spectrometry. In: Papp, E. (Ed.),
Geophysical and Remote Sensing Methods for Regolith Exploration. CRC LEME.
Open File Report, http://crcleme.org.au/Pubs/OPEN20FILE20REPORTS/
OFR20144/OFR144.html.
Wilson, J.P., Gallant, J.C. (Eds.), 2000. Terrain Analysis: Principles and
Applications. John Wiley and Sons, New York.
Wood, J., 2009. Geomorphometry in LandSerf. In: Hengl, T., Reuter, H.I. (Eds.),
Geomorphometry: Concepts, Software, and Applications. Developments in Soil
Science. Elsevier, Amsterdam, The Netherlands, pp. 333–349.
Wulf, H., Bookhagen, B., Scherler, D., 2010. Seasonal precipitation gradients and
their impact on fluvial sediment flux in the Northwest Himalaya. Geomorphology
118, 13–21.
Zhang, W., Montgomery, D.R., 1994. Digital elevation model grid size, landscape
representation, and hydrological simulations. Water Resources Research 30,
1019–1028.
24 Remote Sensing and GIScience in Geomorphology: Introduction and Overview

Biographical Sketch
Dr Michael P Bishop is a professor and Haynes Chair in Geosciences in the Department of Geography at Texas
A&M University. He received his PhD from the Indiana State University (1987) in physical geography with a focus
on geographic information science (GIScience). His areas of expertise are in remote sensing, geographic infor-
mation systems (GIS), geomorphometry, numerical modeling, and mountain geomorphology. He has published
more than 30 articles in scientific journals, three books, and numerous book chapters on topics including
radiation transfer, image and terrain spatial analysis, surface processes and landforms, climate and glacier change,
and landscape evolution modeling. Furthermore, he has presented more than 200 national and international
professional papers on various remote sensing, GIScience, and mountain geomorphology-related topics. Financial
support for his research has been obtained through the National Geographic Society, National Science Foun-
dation, NASA, DOE, USGS, and numerous foundations. His current research is focused on the use of satellite
imagery and terrain analysis for characterizing surface processes and features in complex mountain environments.
 3.2 Ground, Aerial, and Satellite Photography for Geomorphology and
Geomorphic Change
JF Shroder, University of Nebraska at Omaha, Omaha, NE, USA
r 2013 Elsevier Inc. All rights reserved.

3.2.1 Introduction 26
3.2.2 Data Acquisition 26
3.2.2.1 Photographic Scale 29
3.2.2.2 Temporal Coverage 30
3.2.2.3 Digital Cameras and Videography 31
3.2.2.3.1 Gigapan technology 32
3.2.2.4 Thermal Imaging Technology and Geomorphology 33
3.2.3 Image Interpretation 33
3.2.3.1 Change Detection 34
3.2.4 Conclusions 38
References 40

Glossary Photogrammetry Practice of determining geometric

Change detection The use of repeat photography or
images acquired at different time periods to detect and map
changes on the landscape.
Colorimetry Science and technology used to describe
physically, the human perception of color.
Film-return satellite An early generation of orbital
satellites with onboard cameras whose film was ejected
periodically by parachute return vehicle to be collected and
developed.
Forward-looking, infrared radiometer (FLIR) A
portable device to obtain thermal or temperature-based
images of phenomena.
Gigapan technology Gigapixel-panorama technology
using digital cameras for an aggregated single image that
can be browsed and zoomed at multiple scales from
macroscale to microscale.
Megageomorphology Geomorphology of large regions
(regional geomorphology) enabled by space photography
or satellite images that cover broad areas.
properties of objects from photographic images, in some
cases with dimensionally accurate and precisely rectified
aerial photography.
Spectrophotometry Quantitative measurement of the
spectral reflectance or transmittance of an object as a
function of wavelength.
Stereoscopy A technique for creating an illusion of depth
by presenting two offset photographs or images to the left
and right eye of the viewer.
Thermal imaging (thermography) Thermal imaging
cameras detect radiation in the thermal infrared region of
the electromagnetic spectrum and produce images of the
emitted radiation as thermograms.
Time-lapse photography Individual picture frames of
slowly moving or changing objects taken repeatedly from a
fixed position over some intermittent time period, and then
sequentially viewed to depict environmental change.
Videography The process of capturing sequential images
that depict object or landscape change.

Abstract

Historically, repeat coverage enabled change detection and study of landscape evolution. Today, digital cameras and
videography permit new capabilities in terms of producing ground, aerial, and space ‘photography,’ as geomorphologists
still rely on spatial data collected from camera sensors that enable them to collect field data and document geomorphic
events and environmental change. Active geomorphological research now involves the acquisition of digital photography/
imagery, videography, and the use of historical photography, as historical data must serve as baseline data that is compared
to recently collected data for detecting and measuring geomorphic change. Digital camera technology has rapidly evolved,

Shroder, J.F., 2013. Ground, aerial, and satellite photography for

geomorphology and geomorphic change. In: Shroder, J. (Editor in Chief),
Bishop, M.P. (Ed.), Treatise on Geomorphology. Academic Press, San Diego,
CA, vol. 3, Remote Sensing and GIScience in Geomorphology, pp. 25–42.

Treatise on Geomorphology, Volume 3 http://dx.doi.org/10.1016/B978-0-12-374739-6.00041-5 25

26 Ground, Aerial, and Satellite Photography for Geomorphology and Geomorphic Change
and scientists now use these sensors to record flooding, volcanic, mass movement, and other events. Movies and videos of
geomorphic phenomena, especially transient processes such as landslides, debris flows, or time-lapse glacial activities are an
essential research and teaching element. Finally, new gigapixel-panorama (gigapan) technology will provide new cap-
abilities to assess multiscale geomorphic phenomena.
3.2.1 Introduction
Understanding surface processes and process–form relation-
ships, and conveying that understanding requires conceptual,
qualitative description, quantitative characterization, and
graphic depiction. Several centuries ago, graphic depiction
consisted solely of scientific illustrations, primarily artistic
field sketches of greater or lesser capability and complexity
(Merriam, 2009). Line drawings, sketches, or watercolors were
commonly used in the field, and a well-trained young scien-
tist, particularly in Europe, normally had some formal edu-
cation in drawing to assist their field work. This useful manner
of touristic, military, or scientific illustration was largely sup-
planted by widespread use of cameras by the early twentieth
century, and the ability to make skillful field sketches and
hand-drawn illustrations has been replaced with newer tech-
nologies, involving new platforms and sensor technologies for
acquisition of various forms of photography.
The idea of using a camera for field documentation,
however, is almost as old as the first cameras themselves. For
example, in 1838–42, F. Catherwood used camera-lucida
projections to make accurate drawings of Mayan ruins, while
also documenting their weathered condition with daguerre-
otypes only shortly after the invention of that camera-film type
(von Hagen, 1947). Malde (1973) noted that the camera ‘sees’
all of the intricate qualities of the terrain that might otherwise
escape notice and never make it into the field notebook. Such
things include weathering stains on outcrops, soil cracks and
small rills on bare ground, plant species and growth patterns,
and countless other features and environmental characteristics
that are impossible to specifically map and describe by any
other reasonable means.
The earliest known use of what is referred to by the military
as an ‘overhead platform’ to observe the ground during the
French Revolution in the late eighteenth century was when
aerostiers, or balloonists took to the skies. In the nineteenth
century, military photographs from balloons in the American
Civil War were used to document enemy positions in land-
scapes, as well as to obtain high-angle views of the terrain. This
capability from the air has undergone various stages of devel-
opment over time that have proven quite useful to geomorph-
ologists. For example over a century ago, G. Lawrence devised
cameras weighing 4453 kg (1000 lb) that took panoramic
pictures as large as 1.4 m  2.4 m (4.5 ft  8 ft) from heights of
4600 m from balloons and kites. Of relevance to geomorph-
ology were his panoramas aloft of the results of the great San
Francisco earthquake (Rosenberg et al., 1966). In the early
twentieth century, shortly after the invention of the airplane,
aerial cameras were beginning to be used for mapping, and
between World War I (WWI) and World War II (WWII),
aerial-survey techniques were extensively developed. Cameras in
airplanes were flown for reconnaissance before WWI, but geo-
morphologic studies with them were not done at that time. This
is thought in part to be because early conceptions of idealized
and hypothetical terrain conditions and preconceived notions
of geomorphic genesis through time were presented primarily
as artistic block diagrams (Lobeck, 1958; Hayden, 1986).
Between the two world wars of the twentieth century, an
enhanced development of aerial photography was undertaken
and the photogrammetric principles of aerial photography for
mapping were elucidated (Bagley, 1922; Lee, 1922; Reeves,
1927; Ashworth, 1937). Then with the advent of new aerial
photography for WWI, extensive development of the use of
aerial photographs occurred. Where direct overflights were
possible, vertical aerial photographs prevailed, in which
overlapping photographs along flight paths allowed stereog-
raphy, wherein pseudo-three-dimensional (3-D) relief was
possible to visualize. Commonly however, low-oblique aerial
photographs (without the horizon), or high-oblique photos
(with the horizon) had to be taken from a distance, and then
projected into the vertical using the principles of photo-
grammetry in order to make a map. After the war, the sys-
tematic use of such postwar photography permitted detailed
topographic mapping (Smith, 1943), thereby allowing this
geomorphic research medium to achieve its apogee of tech-
nical formulation.
In the latter half of the twentieth century, the Cold-War
drive to ‘command the high ground’ led to the race into space
by the USA and the Soviet Union. This endeavor resulted in
the development of a host of new technologies, including
photographic and imaging sensors aboard a multitude of
satellites.
Numerous sensing devices have moved from experimental
to effectively operational, and from aerial to space-based, or
from space-based to ground-based, such that new sensors for
producing digital photography and imagery are common-
place. Many of these technologies overlap considerably in
terms of data acquisition, manipulation, analysis, and geo-
morphic application. Therefore, the objective of this chapter is
provide examples of the use of historical photography, as well
as some newer technologies that are not be covered elsewhere
in the Treatise on Geomorphology.
3.2.2 Data Acquisition
The first known photographs were daguerreotypes produced
in 1839, and shortly thereafter the science of photogrammetry
started, although the term ‘photogrammetry’ did not come
into common usage until the mid-twentieth century (Whit-
more and Thompson, 1966). In the mid-nineteenth century,
photography began to be used to make topographic maps by
combining surveying theodolites with cameras. In the latter
half of the nineteenth century many early photographers of
the new (to them) lands of the western USA, such as W.H.
Jackson, A.J. Russell, and others, were led to document the
exciting new landscapes that were being discovered. These
professional photographs, and any number of more amateur
 Ground, Aerial, and Satellite Photography for Geomorphology and Geomorphic Change
attempts up to the present day, constitute a collection of
photographs that are valuable for landform analysis as well as
constituting historical documents. Consequently, much effort
is being focused nowadays on electronic storage and computer
analysis to generate digital presentations and animations of
geomorphic change.
Cameras, of course, can be used in a variety of ways to
permit accurate delineation and measurement of geomor-
phologic phenomena. It is not the purpose of this chapter,
however, to relate the geometric or algebraic details of
photogrammetry to enable quantitative geomorphic meas-
urements because other sources do that in detail (e.g.,
Thompson, 1966; Malde, 1973; Graf, 1985; Lane et al., 1993).
Nonetheless, it should be noted that major objectives of
ground-based geomorphologic research are characterizing,
measuring, and explaining surficial processes as they vary
spatially and temporally across landscapes. Measuring di-
mensions of landforms is a standard method to make such
observations, and the methods of photogrammetry using
ground-based and aerial cameras enable many useful meas-
ures to be obtained.
In an interesting new usage of photography in documen-
tation for later geomorphologic interpretation, J.P. MacCalpin
has described what he terms as objective, 2-D photomosaic,
and 3-D photogrammetric logging of exposed trench faces (see
Chapter 14.12). These techniques (Coe et al., 1991; Fairer
et al., 1989) are designed to capture all the essential details of
sediment clast size, grain orientation, color, structure, and all
other fine information to document exactly what a subsurface
trench wall looks like so as to be best able to interpret past
events in the stratigraphy relating to the formation of the
landforms. Processes of ground rupture and sedimentation
attendant to past earthquakes, slope failures, or karst subsid-
ence may be much better understood and explained in this
fashion in the investigation of the subsurface of fault scarps,
landslides, or sinkholes.
Beginning in the late 1940s, the desire to monitor devel-
opments inside the Soviet Union and its satellite countries to
prevent another ‘Pearl Harbor’ surprise attack led to a plethora
of classified airborne and spaceborne platforms for photog-
raphy of the Earth’s surface. The reconnaissance versions of
high-altitude bombers, especially the RB-26, RB-29, RB-50,
RB-47, RB-57, RB-58, and the later purpose-built reconnais-
sance aircraft – the U-2 and the SR-71 – allowed photography
during the Cold War, but little useful geomorphology is
known to have ever been obtained from them.
Before the declassification in 1995 of the historic photo-
graphs taken by the generations of spy satellites, the few
available photographs of the Earth’s terrain from space in the
early days were those taken by orbiting US astronauts. At first
in the manned space programs in the 1960s, handheld camera
photographs by the astronauts through porthole windows in
space capsules were only barely possible, and their use in
geomorphology has been rather limited. The four Mercury
orbital missions primarily used the Swedish-made precision
Hassleblad cameras, but it was only on the last two missions,
MA-8 and MA-9, that systematic terrain photography was a
formal experiment (Lowman, 1996). Astronauts W.M. Schirra
and L.G. Cooper took photographs that received widespread
publicity.
27
The follow-after Gemini program (Underwood, 1967;
Lowman and Tiedemann, 1971) had a Synoptic Terrain Pho-
tography Experiment (S005) that produced some 1300 usable
70 mm color pictures of Earth’s terrain (Figure 1). Some re-
sulting Gemini photographs were published in Life, National
Geographic, and many newspapers to engage the public’s
interest. It was only later that the progressive diminutions of
Lake Chad in Africa and the Aral Sea in Central Asia were
documented through repeat photography that the values of
such data were realized for change detection. Photography
obtained from the Gemini astronaut program was the stimu-
lus to what ultimately became the Landsat series of fairly
standardized acquisition of satellite multispectral imagery. In
1966, the US Geological Survey had proposed an Earth Re-
sources Observation Satellite (EROS), whose name was later
changed to Earth Resources Technology Satellite (ERTS), and
finally, Landsat. Early satellite imagery was low resolution that
showed only broad areas at small scale, but even so this en-
abled views of megageomorphology (wherein huge areas of
the Earth’s crust could be viewed collectively to gain under-
standings of regional geomorphology).
Well before astronaut photographs entered the public do-
main, several series of satellites were acquiring greater than
860 000 detailed photographs of many parts of the Earth’s
surface, but the photos were unavailable for general scientific or
geomorphologic use. After declassification by the Clinton–Gore
executive branch of government in 1995, geomorphologists
were able to access the treasure trove of available espionage-
satellite data for the first time (Figure 2). In spite of the un-
availability of some critical photogrammetric documentation
and deficient metadata that impedes some usages, the historical
data from widely dispersed parts of the world are an unusual
source of information that can be used in any number of
change detection or other studies in geomorphology.
Figure 1 Astronaut photograph from the Gemini 4 mission of 1966
showing the delta of the Nile River in the foreground and center, with
the Suez Canal area in the middle ground and the Red Sea, Gulf of
Aqaba, and Dead Sea in the background. Courtesy of US Government.
28 Ground, Aerial, and Satellite Photography for Geomorphology and Geomorphic Change

1,000 500 0 1.000 Meters 400 200 0 400 Meters

(a) (b)

150 75 0 150 Meters 200 100 0 200 Meters

(c) (d)

500 250 0 500 Meters 200 100 0 200 Meters

(e) (f)

Figure 2 Periglacial geomorphological features on the Bykovsky Penninsula of the northeast Siberian coast identified from Corona photographs
of July 1969. Courtesy of US Government.
 Ground, Aerial, and Satellite Photography for Geomorphology and Geomorphic Change
Table 1 Early generations of espionage satellites with film return by
ejected canister in the Keyhole program whose large number of
declassified photographs can be accessed from the US Geologic
Survey’s EROS Data Center for change detection and other studies in
geomorphology. Each satellite number may represent multiple satellite
systems as 144 satellites alone were launched in the Corona program
Satellite number Satellite name Resolution (m) System operation
KH-1 Corona B7.5 m 1960–72
KH-3 Corona B7.5 m
KH-4 Corona B7.5 m
KH-4A Corona B2.75 m
KH-4B Corona B1.8 m
KH-7 Gambit B48 cm 1963–67
KH-8 Gambit B10 cm 1966–84
KH-9 Octagon B15 cm 1971–86
The US National Reconnaissance Office (NRO) has oper-
ated satellites for the national intelligence community for
some time. Beginning in June of 1959 the NRO launched and
operated a number of Earth-observation satellites. Only the
early film-return satellites are discussed here (Table 1); imagery
from later satellites has not been declassified. In general the
film-return CORONA, ARGON, and LANYARD satellites were
in elliptical orbits ranging from a low of B280 km to a high of
B1006 km above the Earth’s surface. Each satellite passed over
its assigned observation target on the ground approximately
twice a day. At appropriate times, canisters of exposed films
were ejected over oceans, made reentry and deployed para-
chutes at lower altitudes, where most were successfully caught
by US aircraft, although some floated for a time until the US
Navy could intercept them. Others that were not recovered
quickly were designed to sink to deny them to other espionage
agencies. A large number (144) of satellites were launched
under the CORONA program and 102 returned usable data.
Many of the now-declassified scenes were acquired in ste-
reographic mode from forward and aft-pointed cameras
(relative to the direction of motion of the satellites). Some
stereo viewing of the ground surface is possible, although
spatial distortions caused by forward camera motion, scan
time, and imaging of a ‘bow–tie’ shaped area of the ground
that was compressed into a rectangular image frame in the
Corona system makes such viewing rather difficult (Casana
and Cothren, 2008). Distortions in the Corona images can be
corrected using a rigorous model (e.g., Schenk et al., 2003;
Sohn et al., 2004), or a simple photogrammetric frame model
(Altmaier and Kany, 2002). The distortions can be further re-
moved using ground-control points for geometrically cor-
recting the data.
In practice, one might have expected such declassified data
to have been used much more in geomorphology. For ex-
ample, in the Global Land Ice Measurements from Space
(GLIMS) project begun in the late 1990s that was designed to
monitor the world ice masses, researchers have used de-
classified satellite data to identify the terminus positions of
glaciers in the Hindu Kush and the Himalaya (Bishop et al.,
2004; Kargel et al., 2005), and stereoscopic mapping of
rockslide denudation there as well (Shroder et al., 2010). Rao
(2009) used declassified satellite data to map beach ridges on
29
the Godavari delta in India, and Grosse et al. (2005) were able
to map a wide variety of periglacial geomorphology (ther-
mokarst depressions, lakes, and lagoons; thermoerosional
cirques and valleys; pingos) in northeast Siberia with the im-
agery (Figure 2).
The US Geological Survey maintains these images at its
EROS Data Center, in Sioux Falls, SD, and beginning in 2005
began distributing images scanned directly from the original
film strips at resolutions of up to 7 mm, which produces much
crisper images than the previous contact prints provided
(Casana and Cothren, 2008). Inasmuch as over 800 000 of
these images were acquired and the collection of Corona
material alone included 2.1 million ft of film in 39 000 cans,
the terrain information contained in these sources is enor-
mous and can be used to facilitate geomorphological research.
Other official NASA campaigns of terrain photography
were also undertaken (Robinson et al., 2002). These include
data acquisition during the Earth-orbiting Apollo missions
(Colwell, 1971), the Apollo–Soyuz mission (El-Baz, 1977;
El-Baz and Warner, 1979), Skylab (NASA, 1974; Wilmarth
et al., 1977), some Shuttle missions (Jones et al., 1996), and
the Shuttle–Mir missions (Evans et al., 2000). As of 30
September 1999, some 378 461 photo frames had been
included in the database (Office of Earth Sciences, 2000), but
approximately 50% have been deemed not useful for remote
sensing applications, which leaves some 190 911 possibly
suitable for Earth science related studies (Robinson et al.,
2002).
With the large number available, astronaut photographs
can be an excellent source of data for different studies, and the
best case resolutions are similar to Landsat or SPOT imagery
with pixel resolutions down to o10 m (Robinson et al., 2002).
As public domain information, costs are minimal to nothing,
and these images can be quite useful in filling in time-series
gaps where other imagery are not available. Access to the
complete and available database of astronaut photographs,
including low-spatial-resolution browse images, is available via
the web (Office of Earth Sciences, 2000). Such images posted
on the web are of low quality, but high-quality images can be
obtained on request for no cost (Robinson et al., 2002).
In recognition of the great advances made in visualizing
the geomorphology of the Earth’s surface from space, NASA
produced a large volume with copious pictures that represent
an atlas of regional landforms (Short and Blair, 1986). This
atlas has as its core, 237 color or black and white plates, each
of which consists of a space image with accompanying com-
mentaries, explanations, together with three or four informa-
tive and enhancing aerial and/or ground photographs
(Lattman, 1987). Astronaut photographs, Landsat images,
radar, and thermal images are included as well. Twelve ex-
tensively illustrated chapters on regional, tectonic, and global
geomorphology are included, as well as detailed information
regarding process types and geomorphological mapping.
3.2.2.1 Photographic Scale
The mere existence of historical photography does not mean
that the data are suitable for specific geomorphological stud-
ies. In addition to the typical problems of cloud cover,
30 Ground, Aerial, and Satellite Photography for Geomorphology and Geomorphic Change
vegetation cover, and differential illumination causing
extensive shadowing, the issue of photographic scale must be
accounted for. This is especially the case for process-based
versus mapping-based studies, as photographic scale in re-
lation to the spatial complexity of the landscape, and the
phenomena of interest, dictates photographic-scale require-
ments. In general, large-scale photography can be used for
detailed geomorphological mapping, although aerial coverage
is limited and requires the mosaicking of photographs. Me-
dium scale photography addresses this issue as long as the
landforms or features can be detected and differentiated from
other objects. Regional geomorphological mapping can be
facilitated by medium- and small-scale photography although
the level of detail progressively worsens. These issues have
been extensively addressed in the past and readers are directed
to manuals and textbooks, especially those published by the
American Society for Photogrammetry and Remote Sensing.
3.2.2.2 Temporal Coverage
Two types of repeat photography or repeat imagery can be
recognized: (1) obtaining historical ground or aerial photo-
graphs for modern replication and (2) establishment of
photographic or videographic monitoring stations that are
meant to be replicated in the future for various purposes.
Sources of historical photography are varied and commonly
obscure enough that Graf (1985) recognized a convergence of
interest between the geomorphologist, historian, and even the
detective to ferret out locations of known historical photo-
graphs, as well as to find out whether photographs exist at all
for certain important areas. Government agencies are obvious
sources of a plethora of photographs, and local historical so-
cieties are another major source. Postcard collections in an-
tique shops and the popular parlor stereopticon pictures of the
nineteenth century provide useful historical photographs as
well. A major problem in all searches for useful photographs to
use in rephotography studies is that indexes are generally not at
all relevant, and the search for photographs to use
in repeat photography forces the researcher to use visual
inspection of historical photographs, one photo at a time. In
some cases somewhat fortuitous photographs have given geo-
morphologists the ability to ‘see’ long-term processes in action.
For example, Stephenson et al. (Chapter 10.11 in this volume)
and Shepard and Kuhn (1983) used historical photographs to
document the evolutionary development of coastal sea caves
on peninsulas into arches by wave erosion.
Historical photographs of any area in the world may or
may not be difficult to acquire, although by far the most dif-
ficult job once the photograph has been obtained, is to rees-
tablish the original standpoint in the target photograph. In
some cases, only a reasonable approximation is possible,
whether because the original standpoint was on a boat on a
body of water, or was inundated subsequently, or is covered
with new or growing vegetation, or might be too hard to find
or to get to (A. Byers, oral communication, 2007; B. Molnia,
oral communication, 2009). In order to facilitate finding
difficult locations, Hanks (2006) for example, has engaged
in ‘virtual repeat photography’ wherein digital aerial photo-
graphic data are draped over a digital elevation model (DEM)
to locate past camera stations that would have been difficult or
impossible to find otherwise.
In terms of temporal coverage where historical photo-
graphs were not available, R. McInnes of the Isle of Wight
consultancy, Coastal and Geotechnical Services in the UK,
compiled an array of more than 1000 sketches, prints, water-
colors, and oil paintings of the Isle of Wight off the southern
coast of England and the adjacent Hampshire coast that
record the changes and landform evolution there since 1770
(Johnson, 2009a). He and his colleagues developed a ranking
system of the art based on up to five factors for accuracy de-
terminations that enabled useful comparison for delineation
of change over time. In general, the watercolors of the nine-
teenth century were the most accurate because they were de-
signed to record the actual landscape, especially as tourists or
military artists wanted to record information to explain their
travels and observations (Figure 3).
The capture of photographs of geomorphic processes in
action have been obtained over the years by people with
movie or video cameras at the right place and time in the
field, or in wind tunnel or flume experiments, and other such
laboratory manipulations (Table 2). For example, videos of
landslides have been taken, such as the famous quick-clay
failure in Rissa, Norway, in April 1978, as well as a wide
variety of other slope failures, or rapid wet debris flows
(Video 1) in various parts of the world (California, Japan,
China, Afghanistan, and Pakistan). Many of these videos are
available online through You Tube, Flickr, and other media
outlets. Other processes that have been captured using
videography include saltating sand grains, river meandering,
flow and flooding (Gough, 2007), action of waves in eroding
landforms and transporting coastal sediment, as well as a
wide variety of volcanic processes including catastrophic
eruptions, lava flows, formation of pillow lavas underwater,
and so forth.
Figure 3 Wash sketch of rocks displaced by mass movement and
an erosional residual pillar in Sinjao Nullah (gulley) a few km west of
Herat Afghanistan, painted in 1885 by Edward Law Durand of the
Afghan Frontier Commission during the border delineations of the
country. Such sketches could be quite accurate delineations of
landscapes in the 18th and 19th centuries when cameras were
unavailable or quite rare. Print purchased from India Office Library,
1979; identification India Office, British Library, Prints and Drawings,
shelfmark WD427. Copyright r The British Library Board.
 Ground, Aerial, and Satellite Photography for Geomorphology and Geomorphic Change 31

Table 2 List of geomorphology videos that are available online at http://serc.carleton.edu/NAGTWorkshops/geomorph/visualizations.html

(accessed 10 February 2011). Each of these topics listed in the original website opens onto a plethora of animations, images, and videos that
show geomorphic processes at work in real field situations, or contrived laboratory ones

• River Geomorphology Videos – Videos of geomorphic process at work in rivers and river beds.
• Chemical Weathering – Animations and images dealing with chemical weathering.
• Coastal Wave Mechanics (in ocean systems) – Animations and movies depicting how waves and water molecules act along the shore and in deep
water settings.
• Cryosphere – Visualizations dealing with the cryosphere as a whole.
• Deltas and Plumes – Images and animations that illustrate both modern and ancient delta systems. Photos and 3-D animations depict the
development of deltas at various spatial and temporal scales.
• Dunes: Process and Form – Animations describing conditions necessary for producing form variations in sand dunes.
• Examples of Deglaciation – Animations of the deglaciation of North America and the removal of ice from the Bering Strait.
• Glacial Landforms Resulting from Erosion and Deposition – Animations presenting how depositional landforms like moraines and outwash
plains form.
• Glacier Physics – Animations, images, and movies revealing how a glacier forms, moves, retreats, and in the case of tidewater glaciers, calving.
• Longshore Drift and Depositional Landforms – Animations and images showing a variety of depositional landforms resulting from longshore drift
and the impact of river and coastal engineering projects on sediment transport processes.
• Mass Wasting/Landslides – Animations showing different types of landslides like slumps, slides, and falls in a variety of environments.
• Mountain Uplift and Erosion (in structural geology) – Visualizations illustrating the physical processes interacting to create mountain uplift and
erosion.
• Physical Weathering – Animations and images dealing with physical weathering.
• Processes of River Erosion, Transport and Deposition – Animations showing processes of river erosion, transport and deposition.
• River Systems: Process and Form (in sedimentary geology) – Visualizations and supporting material that can be used effectively to teach students
about physical processes acting in rivers and their floodplains.
• Rocky Coastlines and Erosional Landforms – Animations and images related to erosional landforms such as cliffs, arches, sea stacks and other
morphological features.
• Soil Erosion – Animations of soil erosion, images of soils, and audio files about the Dust Bowl.
• Soil Horizons – Animations illustrating the development of soil horizons and their characteristics.
• Soil Orders – Animations showing distinct soil orders.
• Soils Physical Properties – Animations about the physical properties of soil.
• Waterfall Formation/Nick Point Migration – Animations showing how waterfalls are created; examples of spectacular waterfalls from around the
world, and how falling water is used in hydroelectric power.

Figure 4 Extreme panoramic Gigpan picture of the north wall of Yosemite National Park. On the website, 14 snapshots can be selected and
then zoomed in and out from one to another, showing fine details, such as climbers on walls, details of the rock and water, individual people,
and other features. Interactive version available on http://www.xrez.com/, with permission from xRez Studio.

The online version of this chapter contains a Video with an
Animation 1. The online version can be found at doi:10.1016/
B978-0-12-374739-6.00041-5
Some largely imperceptible processes such as the slow ice
flow of glaciers, or terminus retreat and calving are best cap-
tured by time-lapse photography in which individual picture
frames are taken automatically once an hour or day, or some
other time period. These data can then be used together to
depict the movement of the phenomena that provides insights
into process and mechanism. Where this is done by film
production companies, they produce a product that provides
for a relatively seamless transition, and maintain lighting
conditions. In the real world of science given logistics, such
conditions are not usually obtained. Instead, cameras are set
to run automatically with batteries or solar power regardless of
lighting conditions or processes velocities, which generally
produces a jumpy, light-flickering, time-lapse series of images.
Even so, and depending on the results captured, this can
produce quite dramatic footage. For instance, the Extreme Ice
Survey (EIS) is an example of recent (B2008) expedition work
that Balog (2009) has done for glacier advance, retreat, and
extreme calving for glaciers in Alaska, Greenland, Iceland, the
Alps, and Andes, as part of the attempt to capture major
cryospheric-change visuals for public presentation.
3.2.2.3 Digital Cameras and Videography
The transition from film to digital photography is a collision
of interests and debate that goes deeply into arguments of
feature resolution with random silver halide crystals of various
sizes in film, versus pixel resolution and radiometric sensitivity
(Galer and Horvat, 2005; Rand et al., 2005). In addition,
32 Ground, Aerial, and Satellite Photography for Geomorphology and Geomorphic Change

digital sensor geometry is arranged in a rectangular grid pat-

tern, which makes images susceptible to moiré interference-
pattern artifacts, whereas the random orientation of film
grains precludes this problem. In general, to approach 35 mm
film quality with digital cameras requires at least 6 million
pixels, but such an oversimplified statement needs to be
amended very much on what one wishes to do subsequently
with the acquired image (enlargement, publication, etc.). Part
of the issue is also the dynamic range one is interested in
capturing in the image, which is the amount of detail that can
be seen in the shadows before the bright areas start to lose
texture. In general one needs uncompressed file sizes 430 MB
to come close to film when making enlargements. At the
outset, many digital cameras were not as capable as film for
capturing the level of detail that geomorphologists generally
needed, although that is no longer the case. A number of
digital cameras now exhibit a huge dynamic range, compared
to either print or slide film, and film production companies
are gradually ceasing to produce film for use anymore, as it
becomes progressively more obsolete.
In addition, the line between ‘still’ and video digital cam-
eras has become blurred, if not quite nonexistent, and digital
camera technology is one of explosive growth and sudden
collapse. For example, the once quite popular Flip video
cameras, introduced in the spring of 2006, ended production
in spring of 2011, as ‘still’ cameras increased onboard storage
to the point that significant lengths of video imagery could
be recorded. Cameras suitable for ground-based acquisition
of instructional or simple photographs are in many smart
phones and tablet computers. A profusion of more complex
‘viewfinder’ or digital single-lens reflex cameras suitable for
very detailed studies are available for less than $1500, and
prices are continuing to fall. Photogrammetric quality cam- (a)
eras, being more complex and less popular, are rarer and more
expensive. Because of the variety of new cameras available in
this growing field, expert advice should be sought before
investing in a new digital camera for collecting field data.

3.2.2.3.1 Gigapan technology

Gigapan or gigapixel-panorama technology is a new system
using digital cameras for acquiring composite pictures com-
posed of billions of picture elements (pixels). The system
produces extremely high-resolution panoramic images that are
capable of resolutions from macropanoramas from a con-
siderable distance away, zooming down without leaving the
scene to microresolutions of rock thin sections (Johnson,
2009b). The goal of gigapan technology is to facilitate acqui-
sition and presentation of data as a single image that can be
stored efficiently on the web, and that can be examined and
displayed at multiple scales. The system is established as single
or multiple robotic cameras set to full zoom with automated (b)
image acquisition of multiple individual pictures arranged
Figure 5 Photographs of the following: (a) Russia’s Klyuchevskoy
across a grid. The motorized camera system then automatically
volcano showing a thermal image (upper), and a photograph of the
moves, taking hundreds to thousands of slightly overlapping
volcano (lower image. Reproduced from http://
scenes until an entire scene or viewscape is captured. Such news.nationalgeographic.com/news/bigphotos/3626065.html). (b)
images are then downloaded to a personal computer and Italy’s Stomboli Island volcano showing a thermal image of openings
software ‘stitches’ the individual pictures into one large of a new fracture vent in the middle of a flowfield along the Sciara
explorable, gigapixel-sized, super-image, or gigapan. del Fuoco, which is the hillside from which the lava flows to the sea.
Numerous examples of gigapan use in geomorphology are A curtain of smoke and gas emissions obscured the view in
available, from full-gigapixel anaglyph (red/blue 3-D images) visible-light but not this infrared.
 Ground, Aerial, and Satellite Photography for Geomorphology and Geomorphic Change 33

Table 3 List of geomorphology picture libraries that are available online with numerous photographs available of landforms and geomorphic
processes. Many of these photographs are free (F) but some require payment ($)

Lisa Wells Geomorphology Images – http://geoimages.berkeley.edu/GeoImages/Wells/wells.html (F)

Geomorphology from Space (NASA) – http://disc.sci.gsfc.nasa.gov/geomorphology (F)
Gallery of Landform Images – http://www.geomorph.org/gal/mslattery/world.html (F)
Ediafocus Stock Photos (Geomorphology) – http://www.mediafocus.com/image-search/geomorphology-stock-photos.html ($)
About.Com: Geology (Landform Picture Gallery) – http://geology.about.com/library/bl/images/bllandformindex.htm (F)
Overview of Geomorphology Sites – http://www.falw.vu/Bbalr/geomorphology.htm (F)

of sea stacks and uplifted wave-cut terraces in Sonoma County,
CA (Johnson, 2009b) that can be viewed stereoscopically, to
the Extreme Resolution Panoramic Imaging Project (ERPIP) in
Yosemite National Park, where comprehensive assessment and
mapping of rockfall hazards was required (Figure 4). Because
the Yosemite Valley undergoes numerous large rock falls every
year, with 4600 recorded since 1850, evaluating, mapping,
and quantifying the geomorphic hazard was deemed a major
goal of the ERPIP activities. Thus high-resolution imagery of
the precipitous valley walls was needed to establish a baseline
datum for before-and-after comparisons. With establishment
of a collaboration between the National Park Service and xRez
Studios of Los Angeles, 410 000 concurrent images of the
425 km of granite walls in Yosemite were obtained at the
same time on 9 May 2010 to ensure the same lighting and
atmospheric conditions.
The basic technology included ground-based, gigapixel
panoramic photography, light detecting and ranging (LiDAR)
data acquisition from the air (plane and hang glider), and 3-D
computer rendering. Twenty photo-shooting teams totaling 70
photographers between them ascended a total of B11 000
vertical meters of trail. At each vantage point 4500 overlap-
ping shots were obtained, which when merged and printed at
magazine-quality, 300 dpi resolution, the photos extended
uninterrupted for 412 m (Madrigal, 2008). All 20 gigapixel
panoramas were projected onto a 1-m resolution DEM using
Maya 3-D animation software, unifying the 25 km of Yosemite
walls into two single vertical orthographic (undistorted) views,
which yielded a unique, nonperspective elevational view of
the valley walls, which is a first in landscape photography.
In other future studies, obtaining such high-resolution
imagery can be used to establish baseline conditions for be-
fore-and-after comparisons of geomorphic phenomena.
Evaluation, mapping, and quantification of process rates,
geomorphic hazard, and other useful measures of geomorphic
change could be obtained in this way.
3.2.2.4 Thermal Imaging Technology and Geomorphology
A thermal imaging camera (TIC) records emitted thermal ra-
diation and can be used to assess thermal properties and
surface temperature, which is a useful for monitoring volcanic
activity, especially in remote areas from the air, or where
dangerous eruptions threaten and some distance needs to be
maintained. Low resolution (1 km pixel 1) satellites such as
the Advanced Spaceborne Thermal Emission and reflection
Radiometer (ASTER) on board the Terra satellite produce
thermal imagery that can be compared to portable thermal
camera systems. An example of a portable system is the for-
ward-looking infrared radiometer (FLIR) that can then be used
for generating high-resolution thermal images, such as from a
helicopter hovering near a volcano (Figure 5), or a fixed
camera position to record thermal change through time as
lavas are emplaced (Video 2).
The online version of this chapter contains a Video with an
Animation 2. The online version can be found at doi:10.1016/
B978-0-12-374739-6.00041-5
Such instruments enable scientists to assess volcanic haz-
ards more accurately because they can not only be used to
assess surface temperatures, but they can also measure the
presence of phenomena in the sky above the ground, such as
ash or steam. From such data, calculations of emitted volumes
of gas, rock, ice, water, ash, sulfur dioxide, and lava are pos-
sible. For example, in this sort of work on Erebus volcano in
Antarctica in 2004, radiative heat outputs calculated for the
Ray (B1400 m2) and Werner (B1000–1200 m2) lava lakes
were 30–35 and 20 MW, respectively (Calkins et al., 2008).
The estimated magma flux necessary to sustain the combined
heat loss was B250–710 kg s 1, the minimum volume of the
magma reservoir underground was B2 km3, and the radius of
the conduit feeding the Ray Lava Lake was B2 m.
3.2.3 Image Interpretation
Human interpretation of photography and imagery is a
standard approach for information extraction that has been
used for teaching and research for almost as long as the dis-
cipline of geomorphology has been practiced. Photo inter-
pretation is based on the fundamental photographic or image
information elements that include the following: (1) tone, or
the relative brightness and color of landforms, soils and other
surface features that make up the landscape; (2) texture, the
local spatial variability in tone caused by variations in the
surface structure or morphology of the landscape or feature;
(3) size, both absolute and relative, that may be important
with respect to classification; (4) shape, which is related to
morphology and may be distinctive for specific landform
features; (5) shadow, that reflects the shape and the nature of
the object and topographic relief; (6) pattern, in which the
spatial arrangements of landforms may be significant with
respect to repetition, layout, and ordering; and (7) site, the
topographic position of landforms on the landscape (Lille-
sand and Kiefer, 1994).
Collectively these elements constitute the basic infor-
mation that can be extracted and synthesized via human
34 Ground, Aerial, and Satellite Photography for Geomorphology and Geomorphic Change

visualization and analytical reasoning. When combined with

a-priori experience, geographic-, and discipline-domain
knowledge, effective analysis and information can be pro-
duced. Computer-assisted analysis that attempts to mimic the
human interpretation process and information production
can be utilized with digital ‘photography,’ although quantita-
tive formalization of human interpretation still remains an
active research area.
Human interpretation of the landscape and landforms is
relatively more accurate under complex landscape scenarios,
although great progress is being made in quantitative land-
form analysis and mapping from digital data (Bishop et al.,
2011).
A number of web sites provide pictures of landscape and
landforms that can be used most effectively in education
(Table 3). Commonly ignored as an important means of
(a)
doing research or making effective presentations, the ability to
obtain field photography and images, and interpreting field
data is an important skill that needs to be thought about
carefully involving a fair degree of competence. This skill is not
generally taught in graduate school or field camps, although it
is an essential part of any project in geomorphology. Practical
exercises in such landform analysis occur in some older texts,
such as Miller (1961), Wanless (1986), and Way (1973), and
the aforementioned manuals from the American Society of
Photogrammetry and Remote Sensing. The Remote Sensing
Tutorial, by Dr. Nicholas Short, contains several sections of
interest to geomorphologists desiring further instruction in
image interpretation, and is available at multiple web sites.

3.2.3.1 Change Detection

(b)
The elementary, although significant aspect of acquiring repeat
photographs and images from marked locations seems to have
been first used in Europe in the 1880s to monitor glacier
changes, and then in 1896 in the USA by Israel Russell (1898),
who used the technique to monitor and measure the terminus
of the Nisqually Glacier on Mount Rainier. After a hiatus of
approximately five decades, the technique of change detection
using photography was finally taken up again with more at-
tention and utilized by a number of people looking for other
geologic change (Shepard and Grant, 1947; Phillips, 1963;
Veatch, 1969; Baars and Molenaar, 1971). This flurry of
interest by a number of people led to more general recognition
of the potential of repeat photography to provide important
new information, with the result that the United States Geo-
logical Survey (USGS) decided to utilize its vast historical
collections of photographs to accomplish major efforts in
(c)
change detection by trying to reoccupy many remote lo-
cations. The early expeditions by the USGS into the western Figure 6 Tufa knobs, Pyramid Lake, NV: (a) Photograph by T.
USA in the 1860s and 1870s had resulted in thousands of O’Sullivan for the USGS, 1867. (b) Photograph by M. Klett for the
photographs by such renowned photographers as W.H. Jack- Rephotographic Survey Project, 1979. (c) Photograph by M. Klett and B.
Wolfe for the Third View Project, 2000. Such tufa deposits are a
son, T. O’Sullivan, J.K. Hillers, and A.J. Russell that were en-
characteristic of many playa lakes that were active in the late Pleistocene
visioned in the USGS as ripe for rephotography. In the late and some of the Holocene in the western USA but the progressive
1970s then, the USGS began a second phase of taking new abstraction of ever more water from these sites for irrigation and other
photographs at the same date and time of day in the same human uses indicates their progressive degradation thereafter. Such
places as an experiment in repeat photography. A third phase photographic documentation enables measurement of former water
of this repeat photography began in 1997 (Klett, 2004) depths and potential erosional and depositional effects above and below
(Figures 6 and 7). the former water line through time. Courtesy of US Government.
 Ground, Aerial, and Satellite Photography for Geomorphology and Geomorphic Change
(a) (b)
Figure 7 Eroded sandstones in Monument Park area of Colorado Springs, CO: (a) Photograph by W.H. Jackson for the USGS, 1873. (b) Photograph
by J. Verburg for the Rephotographic Survey Project, 1977. (c) Photograph by M. Klett, B. Wolfe, T. Ueshina, K. Bajakian for the Third View Project,
1997. The prominent changes due to rock erosion between scenes allow some calculation of episodic erosion rate. Courtesy of US Government.
In other cases, commercial interests in developing the
American West also eventually provided opportunity for re-
peat photography, as when in 1889–90 R.B. Stanton took
photographs every few km along the bottom of the Grand
Canyon in connection with a railroad that was planned close
to the water level but that was never constructed. A century
later, Webb (1996), a USGS hydrologist investigating debris
flows in the Canyon, was able to replicate all 445 of Stanton’s
photographs. As with so many rephotography projects, the
changes in vegetation were most obvious, but also apparent
were changes exerted by the newer Glen Canyon Dam
upstream so that eroded sand bars, aggraded debris fans, and
new marshes have resulted from this anthropogenic–
geomorphic interference with the natural river flows. The ex-
perience gained in this project enabled Webb et al. (2010) to
write a text on the history, techniques, and applications of
repeat photography that other geomorphologists may find
quite useful.
Similarly, P. Bierman set about to replicate as many old
photographs in the state of Vermont as he could find, with the
result that he was able to document major landscape change
in the state, especially as the clear-cutting of old growth forest
in the nineteenth century led to extensive soil erosion, mass
movement, debris-fan growth, and flooding exacerbated by
denuded watersheds. Subsequently in the twentieth century,
as farm fields were abandoned wholesale back to second-
growth forests, dirt roads were paved, and major new interstate
routes were constructed, the anthropogenic forcings of geo-
morphic change resulted in major changes in sedimentation
and slope and river-valley changes. White and Hart (2007) did
a similar rephotography project for the Canadian Rockies,
which was dominantly about vegetation change, but also in-
cluded change in some glaciers and other landform features.
Some landscapes in the Himalaya, such as those docu-
mented by R. Finsterwalder in the 1930s on Nanga Parbat,
35
(c)
were so well photographed as major parts of their large-scale
photogrammetry-based mapping, that it became desirable to
see what changes have occurred there in the intervening dec-
ades. With the exception of the later German (Schmidt and
Nüsser, 2009) and other rephotographic work on Nanga
Parbat (Figure 8) in the western Himalaya, however, repho-
tography has been predominantly confined to the eastern
Himalaya where so many iconic mountains, valleys, and gla-
ciers occur. For example in the 1950s, several Austrian and
Swiss scientists traversed the terrain in Nepal thoroughly and
made extensive photographic records that A. Byers rephoto-
graphed in 2007 (Figure 9). In the western Himalaya, how-
ever, many of the superb high-altitude panoramas of the
Karakoram Himalaya taken by Sella (De Filippi, 1912) on the
expedition with the Duke of Abruzzi in 1909, remain to be
rephotographed, although Diolaiuti et al. (2003) have
reproduced some.
In other areas multitemporal photographs of glaciers have
proven useful in geomorphologic change detection. This in-
cludes the glaciers of Alaska by B. Molnia after many other
original photographers (Figure 10), or in a more restricted area
such as Glacier National Park – those by D. Fagre after many
other photographers (Figure 11). Many other areas of interest
make ground-based, repeat photography very useful to under-
stand geomorphic (and ecologic) change through time, espe-
cially with the rapidly changing and labile ice masses of the
world. Although commonly hard to do, ancillary quantitative
measurements from the photographs are possible as well.
Repeat aerial photography has also allowed a number of
special geomorphologic applications. For instance, geomorphic
change can be captured and better understood by limited
extrapolation between successive scenes, as for example,
where a sequence of migrating river meanders can reasonably
be constructed between scenes and linked into an
animation – Video 3 (Gough, 2007). In another example, an
36 Ground, Aerial, and Satellite Photography for Geomorphology and Geomorphic Change

(a) 1934 (b) 1985

(c) 1991 (d) 1993

(e) 1994 (f) 1995

(g) 1997 (h) 2006

Figure 8 Pictures of the terminus of Raikot (Rakhiot) Glacier on the north face of Nanga Parbat, western Himalaya, Pakistan that were taken
from the same standpoint at the northeast end of Fairy Meadows. In part after Schmidt, S., Nüsser, M., 2009. Fluctuations of Raikot Glacier
during the last 70 years – a case study from the Nanga Parbat massif, northern Pakistan. Journal of Glaciology 55(194), 949–959. (a)
Photograph by R. Finsterwalder, 1934. (b) Photograph by J. Gardner, 1985. (c) Photogaph by J. Shroder in 1992. (d) Photograph by M. Nüsser
in 1994. (e) Photogaph by J. Shroder in 1996. (f) Photograph by M. Nüsser in 2006. Such variation allows determination of the highly labile
nature of glacier geomorphology.
 Ground, Aerial, and Satellite Photography for Geomorphology and Geomorphic Change 37

(a) Late 1950s

(b) 2007
Figure 9 Taboche (6367 m) in center, Jobo Lapstan, the peak leaning to the right, and Khumbu Valley from the east near Mount Everest
(Sagamatha) in the eastern Himalaya of Nepal: (a) ca. 1955 photograph by Erwin Schneider, courtesy of the Association for Comparative Alpine
Research, Munich. Archives of Alton C. Byers, The Mountain Institute. (b) Photograph by A. Byers, The Mountain Institute, 2007. The clean,
debris-free ice below the summit of Taboche has been reduced considerably, as have the size of many small glaciers. Tsholo Tso is the moraine-
dammed lake at the base of Jobo Lapstan and the moraine dam can be seen to have been more active half a century ago, whereas the lake was
higher in 2007.

intermittently active landslide in the Slovakian Carpathian
mountains was photographed at large scale from the air four
times at Bdecadal intervals and used to construct a detailed
DEM that enabled calculation of the net vertical and hori-
zontal, mass-flux volumes through the system – Figure 12
(Prokešová et al., 2010). Similarly Chandler (1989) and
Chandler and Brunsden (2006) applied analytical photo-
grammetry to archival photographs of five epochs of move-
ment between 1946 and 1988 of the Black Venn landslide in
Dorset, England (Figure 13). Digital terrain models were pro-
duced to assess slope changes through time.
The online version of this chapter contains a Video with an
Animation 3. The online version can be found at doi:10.1016/
B978-0-12-374739-6.00041-5
In an unusual application of repeat photography to geo-
morphic process, albeit of an anthropogenic nature, the
Limestone Project in Oxford, England, was designed to assess
the catastrophic decay of building limestone with a wide
variety of tools (Meneely et al., 2008). Chief among these
techniques is rephotography (Thornbush and Viles, 2005,
2007; Thornbush, 2010) wherein Adobe PhotoshopTM was
used as image-processing software to obtain histogram-based
measurements of soiling and weathering of the building stone.
Results showed that photographs could be used to measure
and quantify change using integrated digital photos and
image-processing techniques calibrated by photographic in-
corporation of a grayscale and spectrophotometry. In addition
to the digital photography used in these analyses, com-
prehensive studies were also done using colorimetry, per-
meability, ground penetrating radar, thermography, X-ray
fluorescence, and stone-condition surveys. Such detailed ana-
lyses devoted to buildings can also be applied to landforms in
the natural rather than anthropogenic world.
Photomonitoring of sites is thus increasingly recognized as
scientifically advantageous, especially in vegetation studies,
range management, or ecological analysis (Rasmusson and
Voth, 2001), and assessment of geomorphological conditions
could be done in more locations. Equipment required to get
started is minimal, and includes a camera, film or camera
computer chips, a photo-information board, a reference-scale
pole, evaluation forms, and a notebook or field computer in
which to record the relevant information. The reference pole,
which is particularly important in the background of the
photograph for range managers to assess long-term health of
vegetation, would be less essential for many geomorphologic
situations, but might be relevant in gauging such things as
ripple and dune size and character, badland topography,
erosional scarps, depositional moraines, and such.
Most essential in setting up the picture for long-term
monitoring is to make absolutely sure that a distinctive and
38 Ground, Aerial, and Satellite Photography for Geomorphology and Geomorphic Change
(a) 1941
(b) 1950
(c) 2004
Figure 10 Muir and Riggs glaciers, Glacier Bay National Monument,
AL: (a) Photograph in August 1941 after nearly two centuries of
retreat. (b) Photograph in August 1950. (c) Photograph in August
2004 by B. Molnia. Muir Glacier has retreated out of the field of view
and is now nearly 8 km to the northwest. Riggs Glacier is still in view
but has retreated as much as 610 m and thinned or downwasted as
much as 250 m.
permanent landmark occurs in the background or on the
skyline. One must make sure that the camera frame includes
the skyline, and that there are distinctive rock outcrops,
mountain slopes, or other geologic or geomorphic features
that will remain over long periods of human history so that
the site can be found again and again. A secondary, but also
important feature is the photoboard set up in the foreground
that has the data and location of the monitoring site. Both the
photoboard and the skyline or landmarks need to be quite
(a) 1913
(b) 2005
Figure 11 Shepard Glacier in Glacier National Park, MT: (a)
Photograph in 1913 by W. Alden. (b) Photograph by B. Reardon. The
highly labile nature of glacier ice is obvious.
visible in the photograph. Finally, a plan for preservation and
long-term storage of the photographic images and written
records is critical if they are to ever be of real scientific use to
anyone in the future. This can be a problem if a geomorph-
ologist has not had time to put things in order before retire-
ment, or who has left inadequate documentation at death.
Special efforts should be exerted by older geomorphologists
who have had well published careers to see that their field
photographs and mapped materials are not lost to science
after their retirement. Film slides and photographs should
be digitized at high resolution and stored in well indexed
electronic media.
3.2.4 Conclusions
The uses of photography as a primary data-collection and
information-recording tool, as well as a device for obtaining
some quantitative information about landforms and pro-
cesses, has been established in geomorphology for some time.
These qualitative as well as quantitative procedures can be
used to aid geomorphologists in studying landscape dynamics
and for geomorphological mapping. Most geomorphology
papers use photographs of surface processes or landforms as a
means to illustrate important aspects of the landscape or vital
concepts.
 Ground, Aerial, and Satellite Photography for Geomorphology and Geomorphic Change 39

1977–1969
−7.0 −5.0 −3.0 −1.7 +1.7 +3.0 +5.0 +7.0 m

[m a.s.l]
650

445
0
200

400

600

800

N
1000
Landslide boundary
Obscured areas
1200

1400
[m]
Figure 12 Differential image of the Ľubietová landslide in the western Carpathian Mountains of Slovakia that is draped over a 1977 DEM that
shows the removal (blue) and accumulation of transported slope debris (red) measured from four sets of stereographic aerial photographs
during the 1969–77 period. Reproduced from Prokešová, R., Kardoš, M., Medved’ova, A., 2010. Landslide dynamics from high-resolution aerial
photographs: a case study from the Western Carpathians, Slovakia. Geomorphology 115, 90–101.

25

20 1988
1976
1969
1958
15
Percentage

1946

10

0
0 10 20 30 40 50 60 70
Interval 5 degrees
Figure 13 Percentages of changing slope angles of the Black Venn landslide measured from aerial photographs acquired in 1946, 1958, 1969,
1976, and 1988. Reproduced from Chandler, J.H., 1989. The acquisition of spatial data from archival photographs and their application to
geomorphology. Ph.D. thesis, The City University, London, England, 300 pp., unpublished, and Lane, S.N., Richards, K.S., Chandler, J.H., 1993.
Developments in photogrammetry; the geomorphological potential. Progress in Physical Geography 17(3), 306–328, with permission from Sage.
40 Ground, Aerial, and Satellite Photography for Geomorphology and Geomorphic Change
The great diversity of general photographs of the surface of
Earth is a vast collection scattered in various places that has
only been accessed and used in a limited way. Thus the dis-
cipline of geomorphology over the past century has produced
large collections of still photographs of landforms, as well as
hosts of movies and videos that show geomorphic processes
in action. On top of the plethora of ground photographs, the
huge amounts of aerial and space-based photos taken
worldwide in the past 40 years are also tremendous sources
that could be processed and stored in databases so that they
could be used better than they have been. Stereographic aerial
photographs are a long-term staple of geomorphologic re-
search, and with the addition of photos from astronauts and
film-return satellites, the potential data sources have grown
ever larger over time. A problem is that with the millions of
potentially rewarding and available historical photographic
materials, a certain amount of time and energy is necessary to
access, review, and store the scattered sources. This effort can
put off busy people so that they do not make effective use of
the materials of which they are unaware, or for which they
lack the requisite skills to access effectively. The addition of
new LiDAR and gigapan technologies and thermal sensor
technologies add to the photographic disciplinary diversity
as well.
New research using photographic techniques could cer-
tainly include greater use of the enormous store of Corona
film returned from space, as well as the many historical
photographs preserved unused in the many historical societies
in towns and cities all across the world where such photo data
are kept. Any number of quantitative or qualitative change-
detection studies remain to be done with permafrost, glaciers,
mass movement, fluvial, aeolian or coastal environments. The
common availability of video cameras of various types, even
in some cases only low resolution cell phones, sometimes
have captured rare geomorphic phenomena in action, which
can be useful in analysis. Different velocities, viscosities, and
grain sizes of such unusual complex debris-flow continua are
being caught increasingly. Eventually, such features as the high
speed and enigmatic mass movements, such as long runout
zone landslides or slow debris flows may be captured in mo-
tion photographically and assessed for better understanding of
mechanics of their motion. Many other somewhat enigmatic
processes may be similarly assessed in the future with these
techniques.
In sum, the use of photographs is an essential and
central part of almost all geomorphologic research, and the
ubiquity and availability of so many old and new cameras
and photographic techniques enhances geomorphological
investigations. Attention to the details of making or obtaining
good digital photographic-image coverage of any particular
geomorphologic problem is viewed as an important, even
essential part of the modern discipline that should be taught
in some fashion in the more advanced graduate programs.
References
Altmaier, A., Kany, C., 2002. Digital surface model generation from CORONA
satellite images. ISPRS Journal of Photogrammetry and Remote Sensing 56(4),
221–235.
Ashworth, F.L., 1937. Aerial mapping. Leicia Photography 6(9), 7 and 16.
Baars, D.L., Molenaar, C.M., 1971. Geology of canyonlands and cataract canyon.
Four Corners Geological Society, Sixth Field Conference, Cataract Canyon River
Expedition, 1971, 99 pp.
Bagley, J.W., 1922. Concerning aerial photographic mapping: a review.
Geographical Review 12, 628–635.
Balog, J., 2009. Extreme Ice Now: Vanishing Glaciers and Changing Climate; A
Progress Report. National Geographic Society, Washington, DC, 118 pp.
Bishop, M.P., James, L.A., Shroder, Jr. J.F., Walsh, S.J., 2012. Geospatial
technologies and digital geomorphological mapping: concepts, issues and
research. Geomorphology, Binghamton Symposium Special Issue 137(1), 5–26.
http://dx.doi.org/10.1016/j.geomorph.2011.06.027.
Bishop, M.P., Olsenholler, J.A., Shroder, J.F., et al., 2004. Global land ice
measurements from space (GLIMS): remote sensing and GIS investigations of
the Earth’s cryosphere. Geocarto International 19(2), 57–84.
Calkins, J., Oppenheimer, C., Kyle, P.R., 2008. Ground-based thermal imaging of
lava lakes at Erebus volcano, Antarctica. Journal of Volcanology and Geothermal
Research 177(3), 695–704.
Casana, J., Cothren, J., 2008. Stereo analysis, DEM extraction and orthorectification
of CORONA satellite imagery: archaeological applications from the Near East.
Antiquity 82(317), 732–749.
Chandler, J.H., 1989. The acquisition of spatial data from archival photographs and
their application to geomorphology. Ph.D. thesis, The City University, London,
England, 300 pp., unpublished.
Chandler, J.H., Brunsden, D., 2006. Steady state behaviour of the Black Ven
mudslide: the application of archival analytical photogrammetry to studies
of landform change. Earth Surface Processes and Landforms 20(3),
255–275.
Coe, J.A., Taylor, E.M., Schilling, S.P., 1991. Close-range geophotogrammetric
mapping of trench walls using multi-model stereo restitution software. ACSM-
ASPRS annual convention. Primary Data Acquisition 5, 30–43.
Colwell, R.N., 1971. Monitoring Earth Resources from Aircraft and Spacecraft,
NASA SP-275. National Aeronautics and Space Administration, Washington, DC.
De Filippi, F., 1912. Karakoram and western Himalaya 1909: an account of the
expedition of H.R.H. Prince Luigi Amedeo of Savoy Duke of the Abruzzi. E.P.
Dutton and Company, NY, 496 pp., 18 panoramas.
Diolaiuti, G., Pecci, M., Smiraglia, C., 2003. Liligo Glacier, Karakoram, Pakistan: a
reconstruction of the recent history of a surge-type glacier. Annals of Glaciology
36, 168–172.
El-Baz, F., 1977. Astronaut Observations from the Apollo–Soyuz Mission.
Smithsonian Studies in Air and Space. Smithsonian Institution Press,
Washington, DC, vol. 1, 400 pp.
El-Baz, F., Warner, D.M., 1979. Apollo–Soyuz Test Project, vol. II, Earth
Observations and Photography, NASA SP-412; Houston, Texas; National
Aeronautics and Space Administration, Lyndon B. Johnson Space Center,
692 pp.
Evans, C.A., Lulla, K.P., Dessinov, L.V., Glazovskiy, N.F., Kasimov, N.S.,
Knizhnikov, Yu.F., 2000. Shuttle–Mir Earth science investigations: studying
dynamic Earth environments from the Mir space station. In: Lulla, K.P.,
Dessinov, L.V. (Eds.), Dynamic Earth Environments: Remote Sensing
Observations from Shuttle–Mir Missions. John Wiley & Sons, New York, pp.
1–14.
Fairer, G.M., Whitney, J.W., Coe, J.A., 1989. A close-range photogrammetric
technique for mapping neotectonic features in trenches. Bulletin of the
Association of Engineering Geologists 26, 521–530.
Galer, M., Horvat, L., 2005. Digital Imaging. Elsevier, Amsterdam.
Gough, S., 2007. River Geomorphology Videos. DVD. Little River Research &
Design, Carbondale, IL.www.emriver.com
Graf, W.L., 1985. Geomorphic measurements from ground-based photographs.
In: Pitty, A. (Ed.), Geomorphology Themes and Trends, pp. 211–225.
Grosse, G., Schirrmeister, L., Kunitsky, V.V., Hubberten, H.-W., 2005. The use
of CORONA images in remote sensing of periglacial geomorphology: an
illustration from the NE Siberian coast. Permafrost and Periglacial Processes 16,
163–172.
von Hagen, V.W., 1947. Maya explorer, John Lloyd Stephens and the Last Cities
of Central America and Yucatán. University of Oklahoma Press, Norman, OK,
324 pp.
Hanks, T.C., 2006. Repeat photography, virtual repeat photography, and Earth-
surface change in the photographic era. Geological Society of America 38(7),
451(Philadelphia Annual Meeting, Abstracts with Programs).
Hayden, R.S., 1986. Geomorphological mapping. In: Short, N.M., Blair, Jr. R.W.
(Eds.), Geomorphology from Space: A Global Overview of Regional Landforms,
pp. 637–656.
 Ground, Aerial, and Satellite Photography for Geomorphology and Geomorphic Change
Johnson, B.F., 2009a. Art: old paintings give new insight into coastal change. Earth
54(8), 66–67.
Johnson, B.F., 2009b. GigaPan imaging. Earth 54(11), 66–67.
Jones, T.D., Godwin, L.M., Wisoff, P.J., Amsbury, D.L., Evans, C.A., 1996. Astronaut
Earth observations during the Space Radar Laboratory missions. Proceedings of
the IEEE Aerospace Applications Conference, Snowmass at Aspen, Colorado,
Institute of Electrical and Electronics Engineers, vol. 2, 3–10 February 1996,
pp. 29–46.
Kargel, J.S., Abrahams, M.J., Bishop, M.P., et al., 2005. Multispectral imaging
contributions to global land ice measurements from space. Remote Sensing of
Environment 99, 187–219.
Klett, M. (Ed.), 2004. Third Views, Second Sights: A Rephotographic Survey of the
American West. Museum of New Mexico Press, Albuquerque, 256 pp.
Lane, S.N., Richards, K.S., Chandler, J.H., 1993. Developments in photogrammetry;
the geomorphological potential. Progress in Physical Geography 17(3), 306–328.
Lattman, L.H., 1987. Landform studies from space. Science 238, 1298.
Lee, W.T., 1922. The Face of the Earth as Seen from the Air: A Study in the
Application of Airplane Photography to Geography. American Geographical
Society Special Publication No. 4, New York.
Lillesand, T.M., Kiefer, R.W., 1994. Remote Sensing and Image Interpretation. John
Wiley and Sons, NY, 750 pp.
Lobeck, A.K., 1958. Block Diagrams: And Other Graphic Methods used in Geology
and Geography. Emerson-Trussell Book Co, Amherst, MA, 212 pp.
Lowman Jr. P.D., 1996. The human remote sensor in space: astronaut photography. 6
pp. http://fas.org/irp/imint/docs/rst/Sect12/Sect12_1.html (accessed 9 June 2010).
Lowman Jr. P.D., Tiedemann, H.A., 1971. Terrain Photography from Gemini
Spacecraft. Final Geologic Report, Report X-644-71-15. National Aeronautics
and Space Administration, Goddard Space Flight Center, Greenbelt, MD, 15 pp.
Madrigal, A., 2008. Extreme resolution photos and lasers create virtual Yosemite.
Wired Science 31. October 2008.
Malde, H.E., 1973. Geologic bench marks by terrestrial photography. US Geological
Survey Journal of Research 1, 193–206.
Meneely, J.D., Smith, B.J., Viles, H.A., Gomez-Heras, M., 2008. In-situ monitoring
of limestone buildings in Oxford. In: Tiano, P., Pardini, C. (Eds.), In-Situ
Monitoring of Monumental Surfaces. CNR, Florence, pp. 327–334.
Merriam, D.F., 2009. A lost art: geological illustrations. Special Commentary, GSA
Today 19(11), 30–34.
Miller, V.C., 1961. Photogeology. McGraw-Hill, New York, NY.
National Aeronautics and Space Administration, 1974. Skylab Earth Resources Data
Catalog. Lyndon B. Johnson Space Center, Houston, TX, JSC-09016.
Office of Earth Sciences, NASA-Johnson Space Center, 14 March 2000. The
Gateway to Astronaut Photography of Earth. http://eol.jsc.nasa.gov/sseop/
default.htm.
Phillips, W.S., 1963. Photographic documentation, vegetational changes in northern
Great Plains. University of Arizona, Agricultural Experiment Station Report 214,
185 pp.
Prokešová, R., Kardoš, M., Medved’ova, A., 2010. Landslide dynamics from high-
resolution aerial photographs: a case study from the Western Carpathians,
Slovakia. Geomorphology 115, 90–101.
Rand, G., Litschel, D., Davis, R., 2005. Digital Photographic Capture. Elsevier,
Amsterdam.
Rao, K.N., 2009. CORONA satellite photographs – A new (Old) tool for earth
scientists. Journal of the Geological Society of India 73(3), 425–429
http://dx.doi.org/10.1007/s12594-009-0022-z.
Rasmusson, G.A., Voth, K., 2001. Repeat photography monitoring made easy. Utah
State University Cooperative Extension, USEPA NR504, 11 pp. (obtained in-line
6 July 2001).
Reeves, D.M., 1927. Aerial Photographs, Characteristics and Military Applications.
The Ronald Press, NY, 307 pp.
Robinson, J.A., Amsbury, D.L., Liddle, D.A., Evans, C.A., 2002. Astronaut orbital
photographs as digital data for remote sensing: spatial resolution. International
Journal of Remote Sensing 23(20), 4403–4438.
Rosenberg, P., Katz, A.H., Rosenfield, G.H., 1966. Photogrammetry in the space age.
In: Thompson, M.M. (Ed.), Manual of Photogrammetry, vol. II, pp. 1049–1089.
Schenk, T., Csatho, B., Shin, S.W., 2003. Rigorous panoramic camera model for
DISP imagery. Proceedings of the Joint ISPRS/EARSeL Workshop: High
41
Resolution Mapping from Space 2003, Hannover, Germany, October 6–8,
www.ipi.uni-hannover.de/fileadmin/institut/pdf/schenk.pdf.
Schmidt, S., Nüsser, M., 2009. Fluctuations of Raikot Glacier during the last 70
years – a case study from the Nanga Parbat massif, northern Pakistan. Journal
of Glaciology 55(194), 949–959.
Shepard, F.P., Grant, IV U.S., 1947. Wave Erosion along the Southern California
Coast. Bulletin of the Geological Society of America 58, 919–926.
Shepard, F.P., Kuhn, G.G., 1983. History of sea arches and remnant stacks of La
Jolla, California, and their bearing on similar features elsewhere. Marine
Geology 51, 139–161.
Short, Sr. N.M., Blair, Jr. R.W. (Eds.), 1986. Geomorphology from Space: A Global
Overview of Regional Landforms. National Aeronautics and Space Administration
(NASA), Washington, DC, 717 pp.
Shroder, Jr. J.F., Owen, L.A., Seong, Y.B., et al., 2010. The role of mass movements
on landscape evolution in the Central Karakoram: discussion and speculation.
Quaternary International, http://dx.doi.org/10.1016/j.quaint.2010.05.024.
Smith, H.T.U., 1943. Aerial Photographs and Their Applications. The Century Earth
Science Series. Appleton-Century Crofts, Inc., New York, 372 pp.
Sohn, H.G., Kim, G.H., Yom, J.H., 2004. Mathematical modeling of historical
reconnaissance Corona KH-4B imagery. The Photogrammetric Record 19(105),
51–66.
Thompson, M.M. (Editor-in-Chief), 1966. Manual of Photogrammetry. Third ed.,
vols. 1 and 2. American Society of Photogrammetry, Falls Church, VA, 1199 pp.
Thornbush, M.J., 2010. Measurements of soiling and colour change using outdoor
rephotography and image processing in Adobe Photoshop along the southern
fac- ade of the Ashmolean Museum, Oxford. Geological Society, London, Special
Publications 331, 231–236.
Thornbush, M.J., Viles, H.A., 2005. The changing fac- ade of Magdalen College,
Oxford: reconstructing long-term soiling patterns from archival photographs and
traffic records. Journal of Architectural Conservation 11(2), 40–57.
Thornbush, M.J., Viles, H.A., 2007. Photo-based decay mapping of replaced stone
blocks on the boundary wall of Worcester College, Oxford. In: Prikryl, R., Smith,
B.J. (Eds.), Building Stone Decay: From Diagnosis to Conservation. Geological
Society Special Publication, vol. 271, pp. 69–75.
Underwood, R.W., 1967. Space photography. Proceedings Gemini Summary
Conference; NASA-SP-138, 231–290.
Veatch, F.M., 1969. Analysis of a 24-Year Photographic Record of Nisqually Glacier,
Mount Rainier National Park, Washington. U.S. Geological Survey Professional
Paper 631.
Wanless, Harold, 1986. Aerial Stereo Photographs. Hubbard Scientific, Fort Collins, CO.
Way, D.S., 1973. Terrain Analysis. Dowden, Hutchinson, and Ross, Stroudsburg, PA.
Webb, R.H., 1996. Grand Canyon: A Century of Change. Rephotography of the
1889–1890 Stanton Expedition. University of Arizona Press, Tucson, AZ, 290 pp.
Webb, R.H., Boyer, D.E., Turner, R.M., 2010. Repeat Photography: Methods and
Applications in the Natural Sciences. Island Press, Washington, DC, 530 pp.
White, C., Hart, E.J., 2007. The Lens of Time: A Repeat Photography of Landscape
Change in the Canadian Rockies. University of Calgary Press, Calgary, Alberta,
312 pp.
Wilmarth, V.R., Keltenbach, J.L., Lenoir, W.B., 1977. Skylab Explores the Earth.
NASASP-380. National Aeronautics and Space Administration, Washington, DC,
517 pp.
Relevant Websites
www.ExtremeIceSurvey.com
Extreme Ice Survey.
http://www.fas.org/irp/imint/docs/rst/Front/tofc.html
Federation of American Scientists.
www.lib.utulsa.edu/Speccoll/collections/rephotosurvey/index.htm
McFarlin Library, University of Tulsa.
http://rst.gsfc.nasa.gov/
NASA Remote Sensing Tutorial.
www.nro.gov/corona/facts.html
National Reconnaissance Office.
42 Ground, Aerial, and Satellite Photography for Geomorphology and Geomorphic Change

Biographical Sketch

Jack Shroder has had a strong interest in geomorphology ever since working in the Mojave Desert as a field
assistant on the US Air Force-funded Playa Project in 1963 for his professors, HTU Smith and Ward Motts. As a
result of this early experience in geomorphology, rather than becoming an oil company geologist, that he was
expected to do, he instead elected to study the landslides of Utah for his dissertation at the University of Utah, UT,
USA, in 1967. His eventual pursuit of landslides in the Hindu Kush and Himalaya (HKH) ultimately led him to
glacier studies in the region, and especially the use of repeat satellite imagery and photography to discern long-
term changes in the highly labile geomorphology of glaciers. Dr. Shroder is planning on maintaining his active-
funded research in the HKH for the foreseeable future.
Dr. Shroder has published more than 150 scientific papers in journals and books, 30 edited books, and several
hundred published abstracts, opinion editorials, and other writings. He is a Trustee of the Foundation of the
Geological Society of America (GSA), as well as a GSA Scientific Fellow, and also an elected Fellow of the American
Association for the Advancement of Science. Dr. Shroder won Distinguished Career awards from both the Geo-
morphology Specialty Group of the Association of American Geographers in 2010, as well as their Mountain
Specialty Group in 2001.
 3.3 Microwave Remote Sensing and Surface Characterization
S Hensley and T Farr, Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA, USA
r 2013 Elsevier Inc. All rights reserved.

3.3.1 Types of Microwave Sensors 45

3.3.2 Microwave Remote-Sensing Principles 47
3.3.2.1 Frequency or Wavelength Considerations 47
3.3.2.2 Resolution 48
3.3.2.2.1 Beam-limited resolution 48
3.3.2.2.2 Pulse-limited resolution 48
3.3.2.3 Polarization 49
3.3.2.4 Scattering, Surface Compositions, and Surface Roughness 50
3.3.3 Altimeters 51
3.3.3.1 Theory of Operation 51
3.3.3.2 Example Echo Profile 53
3.3.3.3 Geomorphological Applications 56
3.3.4 Synthetic-Aperture Radars 57
3.3.4.1 Theory of Operation 58
3.3.4.2 Geomorphological Applications 62
3.3.5 Stereo SAR 65
3.3.5.1 Theory of Operation 65
3.3.5.2 Geomorphological Applications 68
3.3.6 Interferometric SAR 69
3.3.6.1 Theory of Operation 69
3.3.6.2 Geomorphological Applications 74
3.3.7 Summary 75
References 77

Glossary Azimuth resolution Resolution in the along-track

Active sensor An active sensor is a sensor that generates its
own illumination such as a radar or light detection and
ranging (LiDAR).
Airborne Synthetic Aperture Radar (AIRSAR) A NASA/
JPL airborne synthetic aperture radar that operated from
1986 to 2004. It was a multi-frequency (C-, L-, and
P-bands) fully polarimetric radar that also had both along-
track and cross-track interferometric observing capabilities
and C- and L-bands.
Altimeter A radar altimeter is a device for determining the
height or elevation above a surface by recording the round-
trip transit time for a pulse to propagate to the surface and
return.
Antenna A device used to radiate and receive
electromagnetic radiation.
Azimuth compression The signal processing needed to
combine data from multiple pulses to achieve fine resolution
in the along-track direction for microwave sensors.
direction (direction of travel of the sensor).
Backscatter The amount of electromagnetic signal
reflected back toward the direction of incidence.
Bandwidth The difference between the highest and lowest
frequencies in a contiguous set of frequencies in a radar
waveform. Bandwidth is related to the achievable resolution
of a radar sensor.
Beam-limited Sensor whose ground resolution is
determined from the size of the antenna footprint on the
ground.
Beamwidth The half-power angular width of the
mainlobe of an antenna.
Chirp A type of radar waveform that varies the frequency
linearly over the transmitted pulse.
Correlation See interferometric correlation.
Decorrelation The loss of signal coherence between the
pair of signals in a radar interferometer. Mathematically, it is
one minus the correlation.

Hensley, S., Farr, T., 2013. Microwave remote sensing and surface
characterization. In: Shroder, J. (Editor in Chief) Bishop, M.P. (Ed.), Treatise
on Geomorphology. Academic Press, San Diego, CA, vol. 3, Remote Sensing
and GIScience in Geomorphology, pp. 43–79.

Treatise on Geomorphology, Volume 3 http://dx.doi.org/10.1016/B978-0-12-374739-6.00044-0 43

44 Microwave Remote Sensing and Surface Characterization
Dielectric constant A physical property of a material that
determines how much electric energy is reflected and
absorbed.
Digital elevation model (DEM) A digital representation
of the topography of a surface.
Echo profile The sequence of received signals for
multiple-range samples within the antenna footprint of a
radar altimeter.
Ellipticity One of two parameters (the other being tilt)
necessary to fully describe the polarization state of an
electromagnetic wave. Ellipticity measures the elongation of
the ellipse formed in the plane perpendicular to
the direction of propagation for an elliptically
polarized wave.
Foreshortening A form of geometric distortion occurring
in radar imagery of sloped surfaces that results in the
apparent contraction or expansion of these features relative
to their true dimensions.
Gain The relative increase in radiation at the maximum
point compared to an idealized isotropic radiator.
Geometric decorrelation The loss of correlation in an
interferometer resulting from spatial separation of the
observing vantages.
H Electromagnetic waves whose electric field orientation
is fixed and parallel to the ground surface are referred to as
being horizontally polarized.
Image parallax The apparent shift or displacement of an
object between images taken at different vantages whose
elevation is above or below a reference surface.
Incidence angle The angle between the direction of
incident radiation and a vector normal or perpendicular to
the surface.
Interferometric correlation A measure of the similarity of
the pair of signals received by a radar interferometer.
Correlation equals zero if there is no similarity and equals
one if they are identical.
Interferometry The science and techniques used to study
the interference of superimposed electromagnetic waves
transmitted from spatially and/or temporally separated
vantages.
Jet Propulsion Laboratory (JPL) One of the major NASA
centers primarily responsible for robotic exploration of the
solar system.
Layover This occurs when the radar beam reaches the top
of a tall feature before it reaches the base. In this case, the
top of the object appears displaced closer to the sensor from
its base and hence appears to lay over its base.
Mainlobe The angular region or lobe containing the
maximum power of an antenna.
Microwave sensor Any sensor transmitting or receiving
microwaves.
Microwaves A form of electromagnetic radiation with
wavelengths from 1 mm to 1 m. Microwaves are about
100 000 times longer than visible light waves and are
employed in a variety of Earth science applications.
National Aeronautics and Space Administration
(NASA) The United States space agency.
Noise decorrelation The loss in interferometric
correlation due to thermal noise.
Passive sensor One that senses the emitted or reflected
radiation from object.
Polarimetry The science and techniques used to study the
polarimetric properties of electromagnetic waves and the
scattering properties of these waves.
Polarization signature A plot of the backscattered power
received from specific a scatterer as a function of the
polarizations of the incident and scattered
electromagnetic waves.
Pulse repetition frequency (PRF) The frequency at which
a radar emits pulses and is measured in units of hertz (s1).
Pulse repetition interval (PRI) The time between pulses
emitted by a radar.
Pulse length The duration of a transmitted radar pulse.
Pulse-limited Sensor whose ground resolution is
determined from the range resolution of the transmitted
waveform.
Radar Any sensing device that transmits and then receives
reflected microwaves to measure the properties of the object
or scene under study.
Range bandwidth See bandwidth.
Range compression The signal processing needed to
achieve fine resolution in the range or line-of-sight direction
for microwave sensors.
Range resolution Resolution in the line-of-sight or
direction of propagation.
Resolution A measure of the ability to spatially separate
signals obtained from a sensor. For radar sensors, the
resolution is a function of the bandwidth.
Root mean square (RMS) A statistical measure of the
magnitude of a varying quantity.
Shadow Those regions in a radar image that are occluded
from view resulting either from obstruction by another
object or a portion of the terrain.
Sidelobes The angular region outside the antenna
mainlobe.
Signal-to-noise ratio (SNR) The ratio of the signal
strength to the background thermal noise.
Small perturbation model (SPM) An electromagnetic
scattering model that predicts the amount of radar
backscatter from a surface subject to conditions on the
surface geometry.
Speckle Speckle noise is a granular noise that inherently
exists because of the coherent form of the radiation that
degrades the quality of the active radar and SAR images.
Shuttle Radar Topography Mission (SRTM) A single pass
radar interferometer that flew aboard the space shuttle in
February 2000 and generated fine-resolution topographic
maps of the Earth’s surface between 7601 latitude.
Synthetic aperture radar (SAR) A type of radar that uses
data from multiple pulses to synthesize a long antenna,
thereby achieving fine resolution in the along-track or
azimuth direction.
Temporal decorrelation The loss of interferometric
correlation as a result of changes within a resolution
element between observations.
Tilt One of two parameters (the other being ellipticity)
necessary to fully describe the polarization state of an
electromagnetic wave. Tilt measures the orientation of the
 Microwave Remote Sensing and Surface Characterization 45

ellipse formed in the plane perpendicular to the direction of
propagation for an elliptically polarized wave.
Uninhabited Aerial Vehicle Synthetic Aperture Radar
(UAVSAR) UAVSAR is a NASA/JPL L-band actively
scanned synthetic aperture radar operational since
2008.
V Electromagnetic waves whose electric-field orientation is
fixed and parallel to the local vertical are referred to as being
vorizontally polarized.
Volumetric decorrelation The loss of interferometric
correlation as a result of a vertical distribution of scatterers
within a resolution element.

Nomenclature c^ Unit look vector (  )

k Radar wavelength (m) fd Beam-limited ground resolution (m)
C Speed of light (m s1) ~f Pulse-limited ground resolution (m)
d
F Frequency (Hz) NL Number of looks (  )
L Antenna length (m) / Interferometric phase (radians)
Ds Along-track spacing (m) c Interferometric correlation (  )
vp Platform velocity (m s1) dR Ground area element (m2)
h3dB Antenna half-power beamwidth (radians) rh Elevation precision (m)
G Gain (dB) r/ Phase precision (radians)
A Antenna area (m2) Dqext Range extent of antenna footprint on
Db Beam-limited spatial resolution (m) ground (m)
q Range (m) ht Terrain elevation (m)
tp Propagation time (s) Laz Azimuth antenna length (m)
s Pulse length (s) !
b Interferometric baseline vector (m)
B Bandwidth (Hz) Lsyn Length of synthetic aperture (m)
Dg Ground spacing (m) Dp, Dpr Optical and radar parallax (m)
Dq Range resolution or spacing (m) Dssyn Azimuth resolution from synthetic
yc Look angle (radians) aperture (m)
hi Incidence angle (radians) b Baseline length (m)
sq Cross-track slope (radians) a Baseline orientation angle (radians)
PRF Pulse repetition frequency (Hz) !
d Surface deformation vector (m)
ro Radar backscatter (m m1)
e Relative complex dielectric constant (  )

Abstract

The scientific study of landforms and the processes that shape them have benefited tremendously from the rapid progress in
remote sensing and computer technologies. Microwave sensors have the ability to inform geomorphological studies at
length scales that range from planetary scales of thousands of kilometers to micro-topography of surfaces at the centimeter
scale. This tremendous range of applicability of microwave sensors is a result of the variety of sensor types and the large
frequency range over which these sensors operate. Microwave sensors, operating from both airborne and space-borne
platforms, have had a profound influence on our understanding of landforms and the processes that shaped them. In
addition to their standalone intrinsic value to geomorphological studies, these sensors can be combined synergistically with
other sensor data (e.g., light detection and ranging (LiDAR), optical, or hyperspectral data). This chapter focuses on the
basic theory of sensor operation and selected applications in geomorphology.

3.3.1 Types of Microwave Sensors where c is the speed of light. The intensity is a measure of
the strength or amplitude of the electromagnetic radiation and
Microwaves are a form of electromagnetic radiation. Electro- is related to the amount of energy carried by the wave.
magnetic waves are comprised of oscillating electric The electric and magnetic fields comprising an electro-
and magnetic fields that propagate at the speed of light or magnetic wave are vector quantities, which means they have
299 792 458 m s1. They are characterized by several key both a magnitude and direction. The direction of the electric
properties that include their frequency or wavelength, their field (and the magnetic field) lies in the plane perpendicular
intensity, or strength and polarization. Frequency, f, is the rate to the direction of propagation and defines the polarization
at which the amplitude of an electromagnetic-wave changes of the wave as shown in Figure 1. The electric field can point
and is related to wavelength via the relationship in a fixed direction as the wave propagates, called ‘linear po-
c larization’, or it can trace out a circular (circular polarization)
l¼ ½1
f or more generally an elliptical path in the plane perpendicular
46 Microwave Remote Sensing and Surface Characterization

to the direction of propagation. The vector nature of electro-
magnetic radiation gives rise to sensitivity to the orientation of
an object from which it scatters, and, hence, provides add-
itional information about surface morphology.
Sensors that employ electromagnetic waves with frequen-
cies in the range of 0.3–40 GHz (wavelengths from 1 mm to
1 m) are referred to as ‘microwave sensors’. These sensors can
operate either passively, that is, they sense natural microwave
radiation emitted by a surface, or they may be active sensors,
that is, sensors that emit microwaves and then sense reflected
microwaves from the object or scene under observation.
Microwave sensor systems are currently operating over a wide
range of frequencies depending on their intended applications.
Operating frequencies vary from as low as 300 MHz to as high
as 40 GHz. The choice of frequency is dictated by a number of
factors including intended application, platform and power
constraints, and availability of the spectrum at the desired
frequency range. Table 1 shows the correspondence between
frequency, wavelength, and the band designation letter code
(assigned in World War II for security reasons) that are often
used to specify the operating frequency of microwave sensors.
Microwave sensors capable of making measurements
suitable for geomorphological applications can be grouped
into five basic types (Table 2). All are active sensors (that is,
they emit and then record reflected microwaves, to make
measurements of a surface. (Passive microwave sensors, those
that receive microwaves naturally emitted from a surface, can
also provide useful information about a surface; e.g., passive
microwave sensors are used to measure soil moisture and sea-
surface salinity. This chapter restricts its coverage to active
sensors.) Sensor types have distinctive measurement paradigms
and provide information at length scales from thousands of
kilometers down to centimeters. They include the following:

→ • Altimeters. These determine the height of a surface by

E measuring the round-trip time it takes for a microwave to
reflect from the surface and converting it to a distance that is
subtracted from the platform altitude.
→ • Synthetic aperture radar (SAR). These radars generate fine-
B resolution imagery in the along-track or azimuth direction
using the motion of the platform to synthesize a long an-
Direc tenna. The magnitude of the reflected energy from the sur-
t
prop ion of
agat face, called the ‘backscatter’, in each pixel is related to both
Pola ion
rizat the surface composition, through the dielectric constant,
plan ion
e and the surface roughness at the wavelength scale. SARs
Figure 1 The polarization plane is defined by the electric field and operate during both day and night and can operate during
the direction of propagation vectors. inclement weather as well (e.g., clouds or rain, thus pro-
viding remote-sensing observations in regimes where other
sensors cannot operate).
Table 1 Frequency and wavelength relationship table • Polarimetric SAR. Polarimetric SARs are SAR systems that
generate imagery from multiple polarizations. Polarimetric
Frequency band Wavelength range Band identification
information helps resolve surface roughness information
(MHz) (cm)
from surface composition effects on the backscatter. Sensi-
26 500–40 000 1.13–0.75 Ka tivity to orientation and increased information on surface
18 000–26 500 1.66–1.13 K scattering permits more detailed surface classification and
12 500–18 000 2.4–1.66 Ku more accurate quantitative parameter estimation.
8000–12 500 3.75–2.4 X • Stereo SAR. Stereo uses SAR imagery collected from different
4000–8000 7.5–3.75 C vantages to determine topographic information. Like optical
2000–4000 15–7.5 S image stereo pairs, objects at different elevations in SAR
1000–2000 30–15 L
images give rise to parallax, or image displacement, that is
300–900 100–33 P or UHF
proportional to the height above a reference surface.

Table 2 Types of microwave sensors

Sensor type Length scale Geomorphological applications

Altimeter 10–1000 s km Geodetic studies of earth shape, ice profiling.

SAR 1 cm–100 s km Surface roughness, fluvial and aeolian processes, land-cover types.
Polarimetric SAR 1 cm–100 s km Surface roughness, fluvial and aeolian processes, land-cover types,
paleohydrology.
Stereo SAR 10 m–100 s km Topographic mapping, hydrological processes, fault mapping, hypsometry,
glaciology.
Interferometric SAR 1 m–00 s km Topographic mapping, subcanopy topographic mapping, hydrological
processes, fault mapping, hypsometry, glaciology.

• Interferometric SAR. Like stereo SAR, interferometric SARs use
data collected from different vantages to determine topo-
graphic or surface displacement information. In interfero-
metric systems, the parallax is typically much less than a
pixel; so, the topographic information is obtained from a
phase measurement that makes extremely accurate parallax,
or range difference, measurements possible. These phase
measurements are then converted into elevation measure-
ments or surface displacements in the case of differential
radar interferometry. Differential radar interferometry can
measure surface displacements at the millimeter level at fine
resolution and over wide spatial scales, thus providing de-
tailed surface morphological change information for both
natural (e.g., earthquakes and volcanoes) and anthropogenic
causes (e.g., pumping of oil and groundwater).
3.3.2 Microwave Remote-Sensing Principles
Although microwaves are a form of electromagnetic radiation,
they interact differently with surfaces compared to shorter-
wavelength electromagnetic waves (e.g., visible light). More-
over, key sensor parameters of interest (e.g., resolution) are
often specified in a manner not usual for optical sensors. This
section provides a brief overview of key sensor parameters of
interest and a very elementary introduction to the scattering of
microwaves from surfaces (Raney, 1999).
3.3.2.1 Frequency or Wavelength Considerations
Frequency and wavelength are related as shown by eqn [1].
Choice of frequency or wavelength is governed by the in-
tended application and platform constraints. Microwaves tend
to interact most strongly with objects that are 1/20th of a
wavelength or larger, and penetrate more deeply into soil,
vegetation, and ice for longer wavelengths. Thus, depending
on the desired application, the wavelength is chosen to
maximize the information content contained in the radiation
0
−10
Intensity or gain (dB)
−20
−30
−40 Sidelobes
−50
−60
−60 −50 −40 −30
Figure 2 Intensity pattern for a typical antenna is divided into two primary regions. The mainlobe, shaded in light green, is the primary region
where a microwave sensor ideally would supply and receive illumination. The sidelobe region, indicated by the red arrows, is a region of reduced
gain where undesired energy is both emitted and received. Techniques exist for reducing sidelobes in particular applications; however, the laws
of physics preclude their complete elimination.
Microwave Remote Sensing and Surface Characterization 47
reflected from the surface related to the quantity or quantities
of interest. For example, C-band or 5-cm wavelength systems
are often desired for ice imaging applications, whereas L-band
(23-cm wavelength) or P-band (80-cm wavelength) is used for
vegetation and land-cover applications due to their increased
penetration capabilities.
Antennas are used to both transmit and receive microwave
radiation. The size of an antenna very much depends on the
wavelength. To see why this is the case we consider how the
beamwidth and gain depend on wavelength and antenna size.
Beamwidth is an angular measure of how much the microwave
radiation spreads as it propagates from an antenna. Gain relates
to the intensity of the radiation in a given direction relative to an
idealized antenna that radiates equally in all directions. The in-
tensity or gain pattern of a typical antenna can be separated into
two regions. First, is the mainlobe where most of the energy is
directed and the second is the sidelobe region which is the area
outside of the mainlobe. The boresight direction is the direction
of peak gain of the antenna. Quantitatively, the beamwidth is
measuring the angular width of the mainlobe where the inten-
sity drops by a factor of 2. (Intensity or gain is often measured in
decibels. The gain in dB, GdB, is equal to 10 log10(G), where G is
intensity. In decibels a factor of 2 reduction in intensity is a loss
of 3 dB since 10 log(0.5)E  3.0103.) Figure 2 shows the in-
tensity pattern of a typical antenna with the mainlobe and
sidelobe regions identified.
The relationship between beamwidth, y3dB, and antenna
size, L, is given by
l
y3dB ¼ k ½2
L
where l is the wavelength and k is constant that depends on
the design of the antenna. Typical values of k are between 0.3
and 1.0. (For simplicity k will be assumed equal to 1
throughout this chapter.) Notice, that as the antenna size be-
comes larger, the beamwidth becomes smaller, and, hence, the
microwaves are directed into a narrower angular region. As the
antenna dimensions can be different in different directions,
the beamwidth will also vary accordingly.
Beamwidth
Mainlobe Sidelobes
−20 −10 0 10 20 30 40 50 60
Angle off-boresight (deg)
48 Microwave Remote Sensing and Surface Characterization

The peak intensity or gain, G, of an antenna at the center of
the mainlobe is given by
4pA
G¼
l2
where A is the effective area of the antenna. In order to obtain
useful microwave signals reflected from a surface, a microwave
sensor must employ an antenna of appropriate size. Practical
limits exist for the size of an antenna that can be accommodated
on either space-borne or air-borne platforms, and, hence, places
constraints on frequency selection for microwave sensors.
3.3.2.2 Resolution
One of the most important parameters in selecting a sensor for
any remote-sensing application is the resolution, or the ability
to spatially discriminate signals from that sensor. Microwave
sensor resolution, depending on the type and operation of the
sensor, can be limited either by the antenna beamwidth, called
beam-limited resolution, or by the form of the transmitted
signal called the waveform, called pulse-limited resolution.
3.3.2.2.1 Beam-limited resolution
Beam-limited systems are ones for which the range variation
within a beam footprint is less than the range resolution of the
system. The size of the antenna footprint on the ground is
approximately the range, or distance between the antenna and
ground, times the beamwidth measured in radians. Figure 3
shows the region on the ground illuminated by an antenna at
a range, r.
Beam-limited spatial resolution, Db, derived from eqn [2]
is given by
l resolution (i.e., ability to discriminate objects in the direction
Db ¼ r
L of propagation of the microwaves) is limited by the width of
Laz
Lel
where r is the range. The important fact to note about beam-
limited resolution is that it depends on the distance, with
greater distances resulting in coarser resolution.
½3
3.3.2.2.2 Pulse-limited resolution
Before proceeding directly to the definition and quantification
of pulse-limited resolution, it is useful to understand a little
about the operational aspects of microwave sensors. As men-
tioned earlier, active sensors supply their own illumination by
transmitting microwaves. Microwaves may be transmitted
continuously, called ‘continuous wave (CW) systems’, or be
transmitted in segments called pulses of limited duration that
are interrupted by periods when the sensor is not transmitting,
called ‘pulsed operation’. As nearly all the systems used for
geomorphological applications are of a pulsed nature, we will
assume pulsed operation of microwave sensors for the re-
mainder of this chapter. For pulsed systems, the time duration
between pulses is called the ‘pulse repetition interval’ (PRI)
and the associated pulse frequency is called the ‘pulse repe-
tition frequency’ (PRF ¼ 1/PRI). The time duration of the
pulse is called the ‘pulse length’.
Locating where the microwaves are reflected from a surface
depends on the sensor type; however, all the microwave sensor
types under consideration use range or the distance from the
antenna to an object for location determination. Range is
measured by recording the time it takes from the generation of
a pulse till it propagates to an object and then returns to the
radar. The one-way distance or range to an object, r, is then
given by
ctp
r¼ ½5
2
where tp is the round-trip propagation time. Nominally, range
½4
the transmitted pulse, since energy returned from any point of
a pulse cannot be distinguished from another point within the
same pulse. In equation form, the range resolution, Dr, is
given as
ct
Dr ¼ ½6
2

where t is the pulse length. Many operational systems find it

 impractical to transmit pulses narrow enough to achieve their
desired resolution due to peak power limitations or other

 hardware considerations. In order to reduce the peak power in
Laz a transmitted pulse yet maintain the same average power, it is
desirable to have longer pulses without somehow sacrificing
 range resolution. This is achieved by encoding the transmitted
 pulse in such a way as to be able to distinguish where within a
Lel
pulse the returned energy originated. Several schemes exist for
Figure 3 An antenna of dimensions Laz in the along-track or azimuth achieving this discrimination; however, the method used by
direction and Lel in the cross-track or range direction illuminate the most microwave sensors is chirp encoding, where the frequency
ground. The shaded region shows the area on the ground illuminated is linearly changed across the pulse as illustrated in Figure 4.
by the mainlobe of an antenna that has approximate dimensions given The amount of frequency variation across a pulse is called
by the range, r, multiplied by the beamwidth of the antenna. The red
the ‘range bandwidth’. Range resolution is inversely pro-
arrow shows the range or elevation antenna footprint dimension that is
portional to the bandwidth, B, and is given by
proportional to the elevation beamwidth, and the blue arrow shows the
along-track antenna footprint dimension that is proportional to the c
azimuth beamwidth. Dr ¼ ½7
2B
 Microwave Remote Sensing and Surface Characterization
PRI
Frequency
Bandwidth

0 Time
Figure 4 A typical microwave sensor emits a sequence of pulses
separated in time. The time duration between pulses is called the
pulse repetition interval (PRI) and the associated pulse frequency is
called the pulse repetition frequency (PRF ¼ 1/PRI). The pulse
duration, t, is called the pulse length. For many microwave sensors,
each pulse is frequency encoded with a linear-frequency ramp across
the pulse known as a chirp.
Table 3 Bandwidth to resolution table
Bandwidth (MHz) Resolution (m)
400 0.37
300 0.50
160 0.95
80 1.90
40 3.70
20 7.50
10 15.0
From eqns [6] and [7], the bandwidth associated with a
nonchirped pulse of duration, t, is simply 1/t. For coded
pulses, the desired range resolution is achieved only after a
signal-processing step called ‘range compression’. Note that in
this case range resolution is independent of range! In fact, 4-m
resolution imagery of the moon has been obtained from the
Earth at a distance of 373 000 km using this technique
(Hensley et al., 2009). Table 3 shows the conversion between
range resolution and bandwidth for a wide range of currently
operational systems.
Resolution in a direction different from the range direction
is needed to generate two- (or three-)dimensional image
products. Direction along the platform flight path is called the
along-track or azimuth direction and is generally the second
spatial coordinate that is not aligned with the range direction
for Earth (or planetary) observing sensors. How resolution is
obtained is a function of the type of sensor and the sensor
imaging geometry. It is possible to express azimuth resolution
in terms of bandwidth, but this will not be pursued here.
Specific details concerning azimuth resolution are discussed in
the sections covering the sensor types.
Range resolution, which is measured along the line of
sight, can be converted into resolution along the ground, Dg,
and is given by
Dr Dr
Dg ¼ ¼ ½8
sinyi sinðyl  tr Þ
where yi is the incidence angle, which is the angle between the
line-of-sight vector and the local normal vector to the surface,
^, yl is the look angle, which is the angle between the line of
n and V polarization on every pulse. Radars can also operate in
49
θl
n̂
θi
Δg
θi
Δ

Figure 5 The ground resolution, or resolution along the surface
being imaged, is a function of the intrinsic range resolution of the
sensor and the imaging geometry. Variations in the local incidence
angle, yi, which is a function of the look angle, yl, and the cross-
track slope, tr, results in changes to the ground resolution, Dg. The
sensor resolution, Dr, is shown as bounded by the two green
dashed lines and the ground resolution, Dg, by the thick brown line.
Larger incidence angles result in finer resolution, whereas small
incidence angles result in coarser resolution.
sight and nadir, and tr is the terrain slope in the range dir-
ection as illustrated in Figure 5.
When the ground-projected range resolution is less than
beam-limited resolution, the system is said to be pulse-
limited.
3.3.2.3 Polarization
A microwave signal impinging on a resolution element (area
of the surface contained within a single range and azimuth
resolution cell) will in general scatter energy in all directions.
The signal reflected back toward the microwave sensor or radar
is referred to as the ‘backscatter’. The amount of backscattered
energy is a function of a number of parameters including the
polarization that was introduced in Section 3.3.1. Polarization
diversity can be a powerful tool for characterizing the type of
surface and for separating the various components contrib-
uting to the backscatter magnitude. Although even a cursory
treatment of radar polarization theory is beyond the scope of
this chapter, a few salient points and terminology will prove
useful to subsequent sections. See Mott (2007), Lee and
Pottier (2009), and Cloude (2010) for more details.
Microwave polarization is characterized by the shape of the
path that the electric-field vector traces out in the plane per-
pendicular to the direction of propagation. The most general
path is an ellipse that can be characterized by two parameters.
Tilt measures the orientation of the ellipse with respect to
horizontal direction and ellipticity is a measure of ellipse
elongation, as illustrated in Figure 6.
Now although there are an infinite number of polarization
states, it turns out that by transmitting and receiving data on
two suitably chosen polarizations (a sufficient requirement is
that they be orthogonal), it is possible to synthesize all other
polarizations from a combination of these representative po-
larizations. The most common representative set used by most
radars are horizontal and vertical polarization denoted by H
and V, respectively. Polarimetric radars operate by transmitting
H and V polarization on alternate pulses and receiving both H
50 Microwave Remote Sensing and Surface Characterization
∧ |E |y0
y surface and depends weakly on the radar frequency. The di-
|E |x0
 fects the dielectric and will affect the level of backscatter.
∧ measure of how much the surface morphology varies in a
x
Figure 6 Microwaves are a form of electromagnetic radiation
whose polarization state affects the returned signal from a
microwave sensor. Polarization is characterized by the shape the
electric-field vector traces out in the plane perpendicular to the
direction of propagation. The most general shape traced out by the
electric-field vector is an ellipse that is typically specified by two
numbers. Illustrated above are the tilt, f, a measure of the
orientation of the ellipse and the ellipticity, a measure of the
oblateness of the ellipse.
restricted polarization combinations, that is, transmitting only
H and receiving H and V, or transmitting H and receiving H
and transmitting V and receiving V. When the sensor is oper-
ated in a restricted polarization combination, it is not possible
to synthesize all possible polarizations any longer.
Various polarization combinations or quantities that
can be computed from the various polarizations (e.g., polar-
ization signatures, eigenvalue and eigenvector decompositions,
and entropy) are useful both for land classification and for
determining quantitative information about the surface. For
example, the pedestal height in the polarization signature can
be related to surface roughness. Polarization diversity can
also be used to identify dominant scattering mechanisms
within a resolution element, as well as determine quantitative
information about the orientation of objects. A nice example
of the sensitivity to surface orientation is the estimation of
surface slope in the along-track direction from shifts in the peak
of the polarization signature. See Lee et al. (2000) for more
details.
3.3.2.4 Scattering, Surface Compositions, and Surface
Roughness
In order to understand how wavelength-scale roughness, that
is, commonly much finer than the resolution of the radar, can
be inferred from microwave-sensor observations, it is neces-
sary to understand a little about how microwaves interact with
surfaces. As this subject is very rich and can be quite technical,
we will only touch upon this topic to illustrate how one might
go about obtaining centimeter roughness data from micro-
wave measurements.
Backscatter strength is a function of the composition of the
surface, its structure, polarization, and imaging geometry.
Electrical composition of a surface is characterized by its
dielectric constant. The dielectric constant of a material
determines how much energy is absorbed or reflected from the
electric constant is a complex quantity, that it is has a real and
imaginary part, with the real part determining the amount of
energy reflected and the imaginary part determining the
amount of penetration and absorption. Surface moisture af-
Surface structure is generally characterized by roughness, a
resolution element. Roughness is measured in terms of the
incident radiation’s wavelength, so surfaces that are smooth at
one wavelength may appear rough at another wavelength. As a
general rule of thumb, the rougher the surface, the greater the
backscatter. For example, a road that is relatively flat and free
of potholes or other major imperfections may appear very
smooth when imaged using an L-band (23-cm wavelength)
radar, since a typical road’s micro-topography may be less than
5 mm. The same road imaged with a Ka-band (7.5-mm
wavelength) radar, however, may seem quite rough and appear
relatively bright in a radar image. Figure 7 shows the same
area imaged at three different wavelengths.
Imaging geometry is generally characterized in terms of the
incidence angle, which is the angle between the incident wave
and the local surface normal, as illustrated in Figure 5. In
general, the backscatter decreases as the incidence angle be-
comes larger. Thus, the largest backscatter is achieved when
the incoming wave is perpendicular to the surface (i.e., zero
incidence angle) and the smallest backscatter occurs for in-
coming waves parallel to the surface (i.e., grazing incidence or
901 incidence angle).
One of the simplest mircrowave-backscatter models is the
small-perturbation model (SPM) that we will use to illustrate
the several ways surface-roughness information can be obtained
from radar data. This model is valid for surfaces that are not too
rough at the wavelength scale. The horizontal and vertical po-
larized backscatter, sxx, where (xx ¼ hh or vv) is given by
sxx ¼ 4pk4 h2 cos4 yi 9axx 9Wð2sinyi Þ ½9
where k ¼ 2p/l, h is the root-mean-square surface roughness, yi
is the incidence angle, W is the two-dimensional normalized
surface-roughness spectrum, and axx are given by
e1
ahh ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2
ðcosyi þ e  sin2 yi Þ
ðe  1Þ½ðe  1Þsin2 yi þ e
avv ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ½10
ðecosyi þ e  sin2 yi Þ2
where e is the complex dielectric constant of the surface. Note
from eqns [9] and [10] that the backscatter is a function of the
surface roughness and the dielectric constant, as well as the
imaging geometry. Surface roughness can be separated from
these other factors in several ways, provided enough meas-
urements are available.
One method of separating surface-roughness information
uses polarization diversity. In cases where the SPM model is
valid, this can be seen in principle by noting that the ratio
of shh/svv ¼ ahh/avv depends only on the dielectric constant
and incidence angle. Using this relationship to solve for the
 Microwave Remote Sensing and Surface Characterization 51

Figure 7 GeoSAR X-band (left), AISRAR L-band (middle), and GeoSAR P-band (right) orthorectified SAR images. Notice how the vegetated
areas in the center portion of the images have much greater contrast at L-band (23-cm wavelength) and P-band (85 cm wavelength) than X-band
(3-cm wavelength). This contrast differential results from open areas appearing smoother at P-band than X-band, whereas the vegetated areas
appear rough at both wavelengths.

dielectric constant, eqn [8] can then be inverted for the surface
roughness. More complicated models, both empirical and
theoretical, of the polarization dependence on surface
roughness are used in cases where the simple SPM model does
not apply.
Another means of trying to separate out surface roughness
from the other scattering factors uses incidence-angle diversity.
Measurements of the backscatter over a range of incidence
angles, preferably covering as large an angle range as possible
and with measurements close to normal incidence where the
backscatter variation is typically the largest, can be inverted to
give surface-roughness information under appropriate con-
ditions. In some cases, quantitative surface-roughness infor-
mation does not have the desired precision but may allow
classification of the surface into smooth, moderately rough,
and very-rough classes. Again, we note that surface roughness
means surface roughness at the wavelength scale. Figure 8 Altimeters are nadir-pointing microwave sensors that
determine terrain height by measuring the time delay for a pulse to
transit to and from the surface. Measured elevation is a measure
3.3.3 Altimeters ‘average’ elevation in the antenna footprint shaded in light green.
Elevation profiles are obtained by taking measurements from multiple
pulses while the sensor moves in the along-track direction (i.e., in
We begin the discussion of microwave sensors with altimeters
the direction of the velocity vector shown in black above). Additional
which have the simplest implementation of the sensors we elevation measurement footprints are shown shaded in yellow.
will describe and use it to introduce several other key concepts
of microwave remote sensing. Although the basic concept of altimeter operation is very
simple, in practice understanding the measurement is complex
due to the fact that the terrain elevation is not constant within
3.3.3.1 Theory of Operation
the footprint of the antenna beam on the ground and the
Microwave or radar altimeters are downward- or nadir-pointing manner in which microwaves backscatter from the terrain.
sensors that measure terrain elevation. Altimeters measure ter- Altimeters are profiling instruments that measure terrain
rain elevation by taking the difference of the platform height elevation beneath the aircraft or spacecraft by emitting a se-
and the range to the surface as shown in Figure 8. quence of pulses and measuring the distance to the surface for
52 Microwave Remote Sensing and Surface Characterization

each pulse. Spatial resolution in the cross-track or range dir- Beam-limited Pulse-limited
ection depends on whether the altimeter is a beam-limited or
pulsed-limited sensor. Beam-limited systems are those for
which the beam size on the ground, fd, given by
l
fd ¼ r ½11
L
is less than the size of a ground-projected range resolution
element at nadir, ~fd , given by

Δ


+Δ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffipffiffiffiffiffiffiffiffiffi  

+
~f ¼ 2 ðr þ DrÞ2  r2 E2 2 rDr ½12
d

where Dr is the range resolution of the sensor and r is the range

to the nadir point (point directly beneath the sensor). Figure 9
shows the pulse-limited and beam-limited dimensions. Figure 9 In beam-limited altimeters (left) the antenna beam
From eqns [11] and [12], an altimeter is pulsed-limited if footprint on the ground is less than the distance to the point where a
 2 resolution element centered at nadir intersects the terrain (illustrated
l Dr by the dotted line). For pulse-limited altimeters (right), however, the
8 ½13
L r antenna footprint (green-shaded region) on the ground is wider than
the dimension of a range-resolution cell centered at nadir (area on
A pulse-limited altimeter can have multiple range-resolution ground inside the dotted line).
elements within an antenna footprint and thus provide finer-
resolution topography measurements within the antenna
footprint. Since the range varies for range-resolution elements
within the antenna footprint as a result of topographic vari-
ation within the footprint, it means there are multiple possible
elevation values that can be assigned to a given pulse de-
pending on the algorithm chosen. Possible algorithms for
choosing the reported elevation are the highest, lowest, or
average’’ elevation within the antenna footprint as illustrated in
Figure 10. Understanding of what a particular altimeter is
measuring is critical to proper interpretation of the data. Using o
the echo profile, which is the echo intensity as a function of
range, can provide information on the topographic variability Δ
and surface composition at the scale of the footprint size.
What can be estimated and to what level of accuracy depends
on the applicability of the particular scattering model used to
invert the echo-profile data for the geophysical parameters of
interest. Figure 10 A pulse-limited altimeter will make multiple range
measurements within an antenna footprint that is related to the
Spatial resolution in the along-track direction depends on
amount of topographic variation. Depending on the algorithm used to
several factors, including how the data is processed. Without
process the data, the reported elevation may correspond to the
sophisticated processing to sharpen the along-track resolution highest (blue line), lowest (red line), or some average (yellow line)
similar to synthetic aperture radars (discussed in the next elevation within the echo profile.
section), the along-track resolution is the greater of the beam
footprint size in the along-track direction and the spacing or
distance between pulses. If the distance between pulses in the footprint have differential Doppler shifts, sophisticated signal-
along-track direction, Ds, given by processing algorithms can be used to improve resolution.
The Doppler shift is the change in frequency of the radar
vp
Ds ¼ vp PRI ¼ ½14 signal when an object is moving relative to the radar antenna.
PRF (For the radar systems of interest in this chapter, it is the
where vp is the platform velocity and PRI is the inter-pulse platform that is moving and generating the relative motion
period, is greater than the antenna footprint size, then the between the ground and platform). Objects moving toward
along-track resolution equals Ds. However, if the along-track the radar have compressed or smaller wavelengths, hence a
spacing between pulses in less than the footprint size, then the higher frequency, whereas objects moving away from the radar
along-track resolution will be less than or equal to the foot- have longer wavelengths, hence lower frequency, as illustrated
print size. It is possible to refine the resolution in the along- in Figure 11.
track direction to better than the antenna footprint size if the The magnitude of the frequency shift is proportional to
PRF can be commanded high enough, such that the pulse the projection of the velocity vector of the radar onto the line-
spacing is less than half the antenna length in the along-track of-sight vector. Thus, objects located perpendicular to the
direction. By exploiting the fact that points within the antenna velocity-vector have zero Doppler shift. Explicitly, the Doppler
 Microwave Remote Sensing and Surface Characterization 53

Table 4 Cassini radar-altimeter parameters

→

Parameter Value Symbol

Radar parameters
Bandwidth 4.24886 MHz B
Wavelength 2.1 cm l
PRF 5000 Hz PRF
Range resolution 15.0 m Dr
Antenna beamwidth 0.351 y3dB
Number of pulses 8 Np
Platform parameters
Figure 11 Objects moving relative to an antenna experience a
Platform altitude 5000 km hp
frequency shift called the Doppler shift. Objects moving toward the
Along-track velocity 2.0 km/s1 v
radar have higher frequencies, since the reflected waves are
Radial velocity 4.5 km/ss1 r_
compressed relative to the transmitted wave (reflected wave shown
in blue), whereas objects moving away from the radar have lower Backscatter function parameters
frequencies, since the received wavelengths are elongated compared Backscatter function Gaussian –
to the transmitted waves (reflected wave shown in red). RMS surface slope 101 k
Dielectric constant 3 e
shift, f, is given by

2 2 average over that element. This means if a resolution element

f ¼ /!
v p , c^ S ¼ vp cosysq ½15
l l
where ! v p is the velocity vector, vp is the magnitude of
the velocity vector, c^ is the unit line-of-sight vector, /!x ,!
yS
! !
represents the dot product between two vectors x and y , and
ysq is the angle between the velocity vector and line-of-sight
vector called the ‘squint angle’. Points in the beam forward of
the radar have positive Doppler shifts since the radar is
moving toward them, whereas points in the beam aft of the
radar have negative Doppler shifts, as the radar is moving
away from them. By combining data from multiple pulses and
using signal-processing techniques similar to those employed
in SAR, it is possible to refine the along-track resolution by
separating points by their Doppler shifts. This process is called
‘Doppler-beam sharpening’. The increase in along-track
resolution depends on how many pulses are combined, as is
discussed in greater detail in Section 3.3.4.
The precision to which an altimeter can measure elevation
is roughly equal to the range resolution of the sensor. Some-
what better performance is possible in cases where the scat-
tering from the surface can be well modeled (e.g., ocean
surfaces), where more sophisticated algorithms can be em-
ployed to resolve the range to a fraction of range-resolution
element or range bin. Other factors contributing to the ele-
vation measurement accuracy are the signal-to-noise ratio
(SNR, i.e., how strong the radar signal is in comparison to
background thermal noise and the accuracy of the platform
position). With the advent of global position system (GPS)-
aided navigation systems, the position accuracy is normally
not a dominating factor in the overall altimeter accuracy for
Earth-based systems, but represents a larger component for
planetary altimeters.
3.3.3.2 Example Echo Profile
To better understand the complexity in an altimeter-echo
profile, it is useful to see more explicitly how various terms
combine to form the returned signal. The signal returned from
each resolution element represents the backscatter-weighted
on the ground is broken up into small subareas, the subareas
with the largest backscatter will contribute most to an echo-
profile measurement from that range. Since the returned
power level is a function of the antenna gain, backscatter
function, incidence angle, and distance to the antenna, all
these terms factor into the returned signal. Here, we exhibit a
couple of simulated echo profiles from an altimeter with
characteristics similar to the Cassini-radar altimeter used to
map the surface of Titan to illustrate these points. Figure 9
shows the notional imaging geometry and Table 4 provides a
list of the radar-altimeter parameters.
Using eqn [13] and values for the altitude and antenna
beamwidth (recall it is equal to l/L) in Table 4, it follows that
the Cassini-radar altimeter is a pulse-limited system. There-
fore, it is possible to generate an echo profile with multiple
range samples within an antenna footprint. To generate
simulated altimeter returns, the area within the antenna
footprint is divided into many small subareas that are much
smaller than either the pulse-limited dimension or the along-
track resolution of a resolution cell. Subareas within a reso-
lution cell will be called a ‘scattering element’, or ‘scatterer’ for
short. The amount of backscattered power from each scatterer
is computed from the radar equation below.
The range from the platform to a scatterer denoted by, r, is
simply the distance from scatterer to the platform. The in-
tensity or power, P, contributing to the returned signal for each
scatterer is computed using a simplified version of the radar
equation given by
so ðyi ÞG2 ðZÞdS
P ¼ ks ½16
r4
where so is the surface backscatter function, yi is the incidence
angle, G is the antenna gain, Z is the angle off boresight, dS is
the area on surface within the scattering element, and ks is an
overall scaling factor that is function of the transmitted power
and various gains and losses in the system.
Each of the terms in the radar equation has a simple
physical basis. To understand the r4 term in the denominator,
54 Microwave Remote Sensing and Surface Characterization
consider a microwave that is radiating isotropically. At a dis-
tance r from the point of origin of the wave, the energy is
distributed evenly on a sphere of that radius. As the radiated
energy is evenly distributed over a sphere of radius equal to the
range, and since the area of a sphere of radius R is 4pR2,
the power drops off, as 1/r2. Since this effect occurs for both
the transmitted and reflected waves, the resulting overall re-
duction in power is 1/r4. In Section 3.3.2, we learned that
antennas angularly direct the energy that results in non-
isotropic radiation. The gain factor, G, accounts for the direc-
tivity of the antenna (i.e., how much the radiation is focused in
a particular direction), and it is squared in the radar equation
to account for two-way transmission to and from the scatterer.
The final two terms determine the strength of the reflection
from the surface. Somewhat obviously, the larger the amount
of surface area reflecting the energy, the larger the return
should be. The surface area over the scattering element appears
as a simple multiplicative factor because of the way the back-
scatter function, so, is defined. The backscatter function is a
measure of how much energy is reflected from the surface back
toward the radar. It is a function primarily of the surface re-
flectivity, which is in turn a function of the complex dielectric
constant, and the local surface roughness or slope distribution
at wavelength scales. Backscatter from a surface depends on the
angle between the incident wave and the surface that is par-
ametrized in terms of the incidence angle. Backscatter func-
tions vary quite dramatically depending on the surface
composition, and many models, both theoretical and empir-
ical, have been developed for various types of surfaces.
Finally, we specify a simple backscatter function to illus-
trate how it affects the reflected signal. We chose a standard
Gaussian backscatter model where the backscatter, as a func-
tion of incidence angle, is given by
2
egtan ðyi Þ
so ðyi Þ ¼ gm ½17
cos4 ðyi Þ
where
pffiffi 2
1 e1
g¼ , m¼ pffiffi ½18
tan2 ðkÞ eþ1
where k is the RMS surface slope within a scattering element
and e is the dielectric constant. Values for the RMS surface
slope and dielectric constant used in the echo-profile simu-
lations are given in Table 4, and Figure 12 shows a plot of the
backscatter function as a function of incidence angle. Note the
reduction in backscatter as the incidence angle increases (this
is typical of the backscatter of many surfaces).
For the first example, we consider the echo profile for a
perfectly flat surface. The left side of Figure 13 shows an image
of the amplitude, including antenna gain, backscatter, and
range effects, for each scattering element within the antenna
footprint and extending into the sidelobe region. On the right
is a plot of the range difference for the scattering elements
within the footprint. From the graphic, we can see there is
approximately 300 m of range variation within the antenna
footprint. With the 15-m range-resolution size, this corres-
ponds to approximately 20 range samples (300/15) in the
echo profile. Eight pulses were simulated including both
4
0
Backscatter (dB)
−4
−8
−12
−16
0 4 8 12 16 20
Incidence angle (deg)
Figure 12 Plot of the Gaussian backscatter function as a function of
incidence angle. Note the reduction in backscatter as the incidence
angle increases.
thermal noise added to each pulse. The lower part of Figure 13
shows the echo power for the eight pulses and the average
profile obtained from these pulses.
Several features of the echo profile should be noted. First,
the leading edge of the echo profile (i.e., location in the profile
where intensity rises out of the background noise) corres-
ponds to the range closest to the radar or the highest elevation
within the antenna footprint. Using the leading edge to de-
termine the resulting elevation value gives the unambiguous
value expected for this example. Determining the location of
the leading edge is simple for this terrain; however, finding the
leading edge can be quite complicated for more complex ter-
rain. Moreover, it is not always clear that this the best method
for assigning an altitude, as the next example illustrates. The
width of the average echo profile is fairly narrow in this ex-
ample and roughly corresponds to the range extent of the
beamwidth of the antenna, Drext, which is given by
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 
hp y3dB 2 hp y23dB
Drext ¼ h2p   hp E ½19
2 8
This equates to 23 m using the parameters in Table 4. Thus, a
wider beam or increased altitude results in a wider echo profile
and increased difficulty in extracting the location of the
leading edge.
For more complex topography, we consider terrain con-
sisting of sinusoidally rolling hills or dunes in the along-track
direction using a height of 200 m and a spatial wavelength
of 2 km. This terrain has a maximal along-track slope of ap-
proximately 321. Figure 14 shows the intensity and range
variation images for the sinusoidal terrain. Observe in the
intensity image the horizontally stripped banding of the
scattering element intensities. This banding results from
changes in the backscatter function, as the incidence angle
varies on the terrain. The points of maximal return occur when
the tangent plane to the terrain is perpendicular to the nadir
direction (i.e., points where the incidence angle is zero, which
occurs at the peaks and troughs of the terrain). Thus, the
highest and lowest points in the terrain contribute the most to
echo intensity in this example. The delta-range image also
exhibits the same horizontal banding corresponding to the
200-m variation in height of the terrain. Echo profiles for eight
 Microwave Remote Sensing and Surface Characterization 55

Intensity image Delta range image

Δ range (m)
600
500

400
300
200
100
0

Compressed echo profile

Cassini altimeter simulation

flat terrian
1.0
P1
P2
P3
P4
0.8 P5
P6
P7
Normalization power

P3
AVG
0.6

0.4

0.2

0
4999.5 5000 5000.5
Range (km)
Figure 13 Echo profiles and the average profile for eight simulated pulses over flat terrain. The top-left image in this figure shows the amount
of signal reflected back toward the radar for scattering elements within the antenna footprint. Lighter-colored areas correspond to stronger
returns. Since the range is smallest and the antenna gain is largest, the greatest reflected signal occurs at the center of the beam. The top-left
image shows the variation in range within the antenna footprint where distance is color-coded via the scale at right. The bottom graph shows the
echo profile, signal strength as a function of range, for the simulated flat terrain. Note, the echo power is centered and sharply at the distance or
range of the platform to the surface. In this case, we get a nice unambiguous measure of surface elevation.

pulses and their average are shown at the bottom of Figure 14.
Observe that the width of the profile is now approximately
300 m, a factor of 10 larger than in the previous example.
Here, the leading edge of the echo profile is fairly well defined
since it originates from the high-intensity portion of the echo
coming from the peaks of the hills. The profile is double
peaked with the second peak corresponding to the troughs,
and is considerably more complex than the flat terrain ex-
ample. Obviously, more complex terrain will result in more
complex profiles.
Generically, given an echo profile, it is not possible to fully
reconstruct the terrain elevations for all points within the
antenna footprint. Note that the same underlying terrain can
yield many different echo profiles simply by spatially varying
the backscatter function. Different backscatter functions arise
naturally from different terrain-cover types (e.g., vegetated, wet
vs. dry, and lava or sand). Generally, one must be content with
providing an elevation value and some metric(s) of the terrain
variability with the antenna footprint. Deciding which ele-
vation value to pick and what metrics to use depends on the
type of surface being mapped with the altimeter.
In cases where the terrain being mapped is fairly homo-
geneous and a well-defined backscatter function can be iden-
tified, it is possible to generate a library of template-echo
profiles that can then be compared with actual ones to de-
termine the best match in a least-square or maximal-likeli-
hood sense. Terrain characteristics are then assigned based on
the parameters used to simulate the echo profile that best
matches the actual echo profile. An example of this paradigm
was Magellan, a National Aeronautics and Space Adminis-
tration (NASA) mission that used radar to map the surface of
Venus. In that case, the echo profiles were generated based on
RMS surface slope and radar reflectivity using a Hagfor’s
backscatter model. Echo profiles were inverted to provide an
56 Microwave Remote Sensing and Surface Characterization

Intensity image Delta range image

Δ range (m)
600
500

400
300
200
100
0

Compressed echo profile

Cassini altimeter simulation

Dunes: 2000 m spacing - 200 m height
1.0
P1
P2
P3
P4
0.8 P5
P6
P7
Normalizd power

P8
0.6 AVG

0.4

0.2

0
4999.5 5000 5000.5
Range (km)
Figure 14 Simulated altimeter–echo profile for sinusoidal hills of 200 m in height. Echo power shown in the upper right shows that in addition
to the range and antenna-gain variation we had in the first example, the backscatter variation due to changes in local slope leads to a horizontal
banding of the returned power. Similar banding is seen in the range-variation image shown in the top left. Echo profiles for eight pulses and their
average are plotted in the bottom figure. This echo profile is more complex than the flat terrain example. The shape of the profile has a double-
peaked appearance, where the peaks correspond to the higher-powered returns from the peaks and valleys in the terrain. More complex terrain
variation exhibits more complex echo profiles, requiring more sophisticated signal-processing algorithms to assign an elevation value.

elevation measurement with 20-km resolution as well as ter-
rain reflectivity and RMS surface slope at that scale (Ford and
Pettengill, 1992).
3.3.3.3 Geomorphological Applications
Radar altimeters have been used to map the Earth as well as
several planetary bodies including Venus and Titan. Most
radar altimeters used by remote-sensing scientists are flown on
orbiting platforms. Due to the nadir-pointing geometry and
limited swath and orbital constraints, radar-altimeter tracks
rarely consist of contiguous tracks providing complete cover-
age of the planet. A notable exception is the Magellan radar
altimeter that generated topographic data for nearly the entire
surface of Venus at 20–40-km resolution. Nevertheless, radar-
altimeter data can be used in many geomorphological studies.
Earth-orbiting radar altimeters, for example, Topex/Poseidon
and Jason I altimeters have been used primarily to measure
sea-surface heights (e.g., Nerem et al., 2006) and in deter-
mining Earth geoid parameters (e.g., Nerem et al., 1994).
As an example of applying radar-altimeter data to geo-
morphology, we consider the Cassini-radar altimeter. Cassini-
radar data have been applied to understand the surface
morphology of Titan from planetary length scales down to
wavelength scales. This shows the versatility of microwave re-
mote sensing to obtain multi-scale topographic information
that characterizes the morphology of the terrain over very large
length scales.
Large-scale planetary shape parameters can be computed
from radar-altimeter data. Given altimeter tracks with suf-
ficient geographic distribution on a planetary body, it is

Altimeter
90
45
North latitude (deg)
−45
−90
360 270
West longitude (deg)
Figure 15 Altimeter and radar monopulse-derived topographic measurements of Titan (elevations relative to a reference sphere of 2575 km)
used to estimate the ellipsoidal and higher-order shape parameters of Titan’s surface. Reproduced from Zebker, H.A., Stiles, B., Hensley, S.,
Lorenz, R., Kirk, R.L., Lunine J., 2009. Size and shape of Saturn’s moon Titan. Science 324(5929), 921–923, with permission from AAAS.
possible to solve for low-order shape parameters such as tri-
axial ellipsoid axes and their orientation relative to the spin
axis (Zebker et al., 2008, 2009). Using a combination of radar
altimetry tracks and surface topography derived from mono-
pulse measurements (Stiles et al., 2009) as shown in Fig-
ure 15, spherical, bi-axial, and tri-axial ellipsoidal fits to
Titan’s surface were made. The best-fit tri-axial ellipsoid to
Titan has axial parameters of 2575.15 and 2574.78 km in the
equatorial plane, and 2574.47 km in the direction of the spin
axis. These parameters describe the shape of Titan at the
planetary scale.
Depending on the density of tracks, higher-order fits are
possible (e.g., Legrendre polynomials), providing greater de-
tails on the global-shape distribution. Shape parameters can in
turn be used to estimate geophysical quantities relating to the
interior density distribution. Indeed Legendre polynomial fits
to seventh order were generated (Zebker et al., 2009) and used
to characterize higher-order facets to Titan’s shape. These
global shape parameters were compared with a planetary
model for Titan’s shape, assuming it were in hydrostatic
equilibrium for a body at its present orbital location and ro-
tation state. Its observed larger oblateness suggested that the
icy crust has appreciable rigidity.
Long tracks of altimeter data are ideally suited to charac-
terizing the power- spectral-density (PSD) function of the
topography. From the PSD, other quantities such as structure
functions and fractal dimension can be computed. PSDs can
be computed for spatial scales ranging from the length of an
altimeter track, hundreds-to-thousands of kilometers down to
the resolution of the altimeter, and several-to-tenths of
Microwave Remote Sensing and Surface Characterization 57
800
Elevation (m)
400
0
−400
Monopulse −800
180 90 0
kilometers depending on the altitude. Along-track slope in-
formation, also at these scales, can be estimated providing
constraints on surface composition and geophysical processes.
Comparisons of simulated echo profiles to actual terrain can
be used to further constrain morphology at the subresolution
length scales, albeit these inferences may in general impose
broad constraints.
Information at the wavelength and sub-wavelength scales
can be derived from altimeter-backscatter data. These estimates
can be obtained using altimeter data alone via backscatter
models or in conjunction with backscatter and radiometer data
at other incidence angles to obtain better estimates. Titan’s
surface parameters, dieletric constant, and surface roughness
were estimated using a combination altimeter, scatterometer
(backscatter measured at a variety of incidence angles), and
radiometer data (Zebker et al., 2009). The estimated surface
dielectric constants varied from 1.75 to 2.5, consistent with a
surface composed of solid hydrocarbons or CO2 with con-
siderable small-scale structure. Inferred RMS surface slopes at
the 10–100-wavelength scale (Cassini has a 2.1-cm wavelength)
were in the range of 10–161. Additional modeling of the
scattering further indicated volume scattering from millimeter-
sized bubbles embedded in the ice or structures similar to ice
pipes and lenses occurring in terrestrial ice sheets.
3.3.4 Synthetic-Aperture Radars
In the early 1950s a technique that enabled the generation of
fine-resolution imagery using radar was developed called SAR.
58 Microwave Remote Sensing and Surface Characterization

n
d irectio
track
Along
Ra
ng
e
di
re
ct
io
n

Figure 16 A typical SAR-imaging geometry has a platform containing a radar instrument moving in the along-track direction and imaging the
terrain to one side of the flight path. The SAR transmits a series of pulses at regular intervals along track that simultaneously illuminates an area
in the along-track direction, much greater than the desired azimuth resolution. By recording the returned echo from each pulse, and using signal-
processing techniques to synthesize a larger antenna, fine resolution in azimuth is achieved. The blue square in the center of beam shows the
size of a resolution element compared with the illuminated area from a single pulse indicated in green.

Since then many applications both civilian and military have 

Lsyn = 2 ⎯⎯
been developed that make use of SAR imagery. Laz

3.3.4.1 Theory of Operation

SAR can be used to produce fine-resolution imagery from ei-
ther air-borne or space-borne platforms (Raney, 1999; Kovaly,
1976; Curlander and McDonough, 1991). Unlike optical
sensors operating at wavelengths between 3 nm and 30 mm,
such as photogrammetric or hyperspectral systems that form
images from reflected solar radiation, SAR systems transmit
their own radiation and record the signals reflected from the
terrain. With optical systems, images are generally formed
instantaneously, whereas for SAR, data collected from multiple
pulses along the flight path are required in order to achieve
useful resolution in the along-track, or azimuth, direction.
(Exceptions in optical systems include scanning optical sys-
tems such as Landsat where the optics are scanned and the
image is generated one line (or pixel) at a time. This is not a
fundamental sensor constraint, that is, if adequate lenses and
optical-recording technology are available, then a full two-
dimensional image could be made instantaneously.) Rather
sophisticated image processing is required to form recogniz-
able images from the raw data. The resolution and quality of
the imagery depend on a number of system parameters, as
well as how the data are collected and processed.
SAR takes advantage of the motion of the platform to
synthesize a large antenna that may be many hundreds of
meters in length to achieve fine along-track resolution. Fig-
ure 16 shows the typical SAR-imaging geometry with the SAR-
platform moving along in flight. The radar antenna points in a
direction perpendicular to the flight path called the range or
cross-track direction imaging the terrain below. At approxi-
mately regular intervals along the flight path, the radar trans-
mits a signal called a pulse, and then records the returned echo.


Lfp =  ⎯⎯
Laz
Figure 17 An object (yellow triangle) is being illuminated by an
antenna of azimuth dimension Laz at multiple points along the flight
trajectory. The synthetic-aperture length is the distance over which
the object is illuminated and is equal to twice the azimuth-footprint
size on the ground.
Azimuth resolution is achieved by synthesizing a large
antenna from the echoes received from a sequence of pulses
illuminating a target. Without signal processing, the intrinsic
azimuth resolution from a single transmitted pulse would be
the azimuth-angular width of the antenna beam times the
range. By combining the echoes using appropriate signal-
processing algorithms from all the pulses imaging a point, the
azimuth resolution is dramatically improved. Azimuth reso-
lution after processing is determined by the size of the syn-
thetic aperture (or antenna), which is the length of flight track
over which a fixed point stays within the azimuth-antenna
beamwidth. Figure 17 shows an object (yellow triangle) at
 Microwave Remote Sensing and Surface Characterization 59

range, r, being illuminated by an antenna of length, Laz, from
multiple points along its trajectory.
The distance that the object is illuminated, called the
synthetic-aperture length, Lsyn, is twice the azimuth antenna-
footprint size and is given by
l
Lsyn ¼ 2Lf p ¼ 2r
Laz
The corresponding azimuth resolution, Dssyn, that is achieved
from a synthesized antenna of length, Lsyn, is given by
data-rate reduction are achieved at the expense of azimuth
resolution and a reduction in SNR.
• Spotlight mode: In spotlight mode, the radar transmits a
sequence of pulses while directing the antenna to a specified
point on the ground. This increases the amount of time an
object is within the antenna beam and consequently the
½20 length of the synthetic aperture, and thereby the azimuth
resolution. A consequence of this mode of operation is that
areal coverage is limited compared with the long-continuous
imagery obtained in strip mode.

l l Laz Figure 18 illustrates the basic modes of SAR operation.

Dssyn ¼ r ¼r ¼ ½21 During the time a target is in the beam, the range to the
Lsyn 2rðl=Laz Þ 2
target is changing from pulse to pulse. After generating an SAR
image, we identify a target’s location in the image by its azi-
Note that the resolution is simply half the azimuth-antenna
muth or along-track coordinate and range position as shown
length, independent of range! This remarkable feature of SAR
in Figure 19. To select a unique range from the family of
systems enables them to make very fine resolution imagery at
ranges that are changing from pulse to pulse during the syn-
large distances with moderately sized antennas. If data are not
thetic aperture, the angle from the velocity vector to the target
combined over the full synthetic-aperture length, then there is
(or equivalently the Doppler frequency – see eqn [14]) is
a corresponding reduction in the along-track or azimuth
specified for processing, as shown in Figure 20.
resolution. The process of forming the synthetic aperture to
The bold-dashed line from pulse N-2 to the target indicates
achieve the increased azimuth resolution is called ‘azimuth
the desired Doppler (or equivalently angle) at which the target
compression’.
will be imaged. Observe that the selection of the Doppler
A natural question that arises is if there is a way to achieve
frequency affects not only the range at which a target is
an even finer azimuth resolution than half the azimuth-an-
tenna length. An alternate question is if there is some resource
savings that the system could incur if it did not need the full
azimuth resolution. The answer is yes to both questions, and
Range
depends on the operational mode of the SAR that may be one
of the following: Δ

• Strip mode: In strip mode, the radar transmits a continuous

sequence of pulses over the area to be imaged. Achievable
Azimuth

azimuth resolution is half the along-track antenna length.

• ScanSAR or burst mode: In ScanSAR or burst-mode systems,
the radar transmits a sequence of pulses at the pulse-repe-
tition frequency (PRF) followed by a period of no trans-
mission. This allows the radar to either scan to another
elevation angle during the no-transmission period, thereby o
 = o + j Δ
increasing swath width (ScanSAR mode), or remain silent,
thereby reducing the data rate. The increased swath-width or Figure 19 Shown above is a target imaged in the j th range bin.

Strip mode Burst mode Spotlight mode

Figure 18 Figure illustrating three basic modes of SAR operation which are strip, burst, and spotlight modes. In strip mode (left image), the
radar emits a continuous stream of pulses indicated by the red rectangles that allow fine resolution to be achieved in long-imaging strips. Burst
mode (middle images) transmits a short sequence of pulses (typically tens to several hundred) called a burst, followed by a suspension of
transmission usually of similar duration. This enables wider swaths by scanning to another location during the interburst period or a reduction in
the amount of recorded data at the expense of azimuth resolution. The finest resolutions are achieved by spotlight-mode systems that direct the
antenna at a point on the ground to keep in the beam for a longer period of time, and, hence, improve azimuth resolution.
60 Microwave Remote Sensing and Surface Characterization

N+3
N+2
N+1
N
N-1
N-2
N-3

Target

Figure 20 This figure shows the space shuttle imaging a fixed point on the ground from a number of pulses in a synthetic aperture. The range
at which a target appears in a synthetic-aperture image depends on the processing parameters and algorithm used to generate the image. For
standard range/Doppler processing, the range is fixed by choosing the pulse that has a user-defined fixed angle between the velocity vector and
the line-of-sight vector to the target. This is equivalent to selecting the Doppler frequency.

imaged, but also the corresponding position of the platform
(and hence azimuth location) when the target is imaged.
It is useful for our subsequent discussions to distill the
above information on SAR-image coordinates to the simple
geometry of the intersection of two surfaces. As discussed
earlier, range information is obtained by measuring the time it
takes a radar pulse to propagate from the antenna to a target
and return. Azimuth location is determined from the Doppler-
frequency shift that is related to the angle from the velocity
vector when a target is imaged. Viewing SAR-target location
geometrically, the range/azimuth location locus is the inter-
section of a sphere centered at the antenna, with radius equal
to the radar range and a cone with generating axis along the
velocity vector, with cone-angle proportional to the Doppler
frequency, as shown in Figure 21. A target in the radar image
could be located anywhere on the intersection locus, which is
a circle in the plane formed by the radar line of sight to the
target and vector pointing from the aircraft to nadir. Since the
intersection is a curve in three-dimensional space, further in-
formation is required in order to locate a target uniquely.
Since the range direction is not parallel to ground co-
ordinates as shown in Figure 22, SAR images are distorted
relative to a planimetric view. In many applications, this dis-
tortion can adversely affect data interpretation, particularly
when one is not well acquainted with SAR imagery (Leberl,
1990). This distortion can be removed as part of the geocod-
ing process if a digital elevation model (DEM) is available, or
from three-dimensional position data generated directly from
SAR-stereo or interferometric systems, at points where these
techniques produce useful data. These corrected images are
often easier both to interpret and to register with other data
layers that might be required for analysis. Three common
features observed in SAR imagery that warrants particular
mention include foreshortening, layover, and shadow.
Foreshortening in radar imagery results from the fact that
relief displacement is toward the direction of the radar. Be-
cause the range increases more slowly than ground coordin-
ates on slopes facing toward the radar (higher elevations
contend with increasing ground-distance slowing the range
increase), they tend to appear bunched relative to a plani-
metric view. The opposite occurs on slopes facing away from
the radar (lower elevations coupled with increasing ground-
distance speeds the range increase) where they tend to expand
out when compared to a planimetric view. Both situations are
illustrated in Figure 22. Equation [8] can be used to determine
the amount of foreshortening from the imaging geometry and
terrain slope in the range direction. Note that foreshortening
in radar images is opposite to that of optical imagery where
relief displacement is away from the direction of the camera.
Layover is a limiting case of foreshortening where points
arranged with increasing ground coordinates appear reversed
in the radar imagery. Layover occurs because the range to
objects with larger ground coordinates is less than the range to
other objects with smaller ground coordinates. Geometrically
this happens when the slope of the terrain is greater than the
angle the incident radiation makes with respect to vertical.
Also, we note that for interferometric or stereometric radar
systems, layover causes a loss of useful signal and therefore
precludes the determination of elevation in layover regions.
Shadow occurs when the radar beam cannot reach a por-
tion of the terrain being imaged, because it is occulted by
other parts of the terrain or other objects in the scene. Where
the terrain is shadowed, the radar image will appear dark and
the signal in these range cells is only due to thermal noise.
 Microwave Remote Sensing and Surface Characterization 61

Range sphere

Doppler cone

Aircraft
position

Velocity
vector

Assumed reference
plane
Scatter is assumed at the intersection of range
sphere, doppler cone and reference plane
Figure 21 Geometric perspective of target location in SAR imagery is illustrated above. Range information locates a target on a sphere centered
about the SAR platform and the Doppler frequency locates a target on a cone centered about the velocity vector. The intersection locus of these
two surfaces is a circle, and thus the three-dimensional location of a target cannot be uniquely determined from a single image.

Ground coordinates
Figure 22 The three-dimensional world is collapsed to two dimensions in conventional SAR imaging. After image formation, the radar return is
resolved into an image in range–azimuth coordinates. This figure shows a profile of the terrain at constant azimuth, with the radar-flight track
into the page. Notional radar-backscatter values are shown in gray scale above the radar pixels that are delineated by dotted lines. Radar
foreshortening on slopes is indicated by the green and blue arrows above. On the slope facing toward the radar it is imaged in three radar pixels,
whereas the slope facing away from the radar is imaged in five radar pixels. Notional layover and shadow regions are indicated by the
semitransparent green and blue ellipses, respectively. Note the bright return indicated for the layover region and the dark return for the
shadowed region. Finally, the Leaning Tower of Pisa illustrates a classic layovered object where the top of the image appears nearer in range
than the base.
62 Microwave Remote Sensing and Surface Characterization
Minor PRF
2 1 2 1
Major PRF
Tx LH1 Tx LV1
Rx LH1 Rx LH1
Rx LV1 Rx LV1
Channel 1 data LH1LH1 LV1LH1
Channel 2 data LH1LV1 LV1LV1
Figure 23 Polarization diversity is achieved by transmitting H
polarization on alternate pulses (designated by a 1 above) and V
polarization on the adjacent pulses (designated by a 2 above). Four
images are formed from the data, an HH (H transmit, H receive) HV,
VH, and VV. From these four basic images, it is possible to
synthesize any other polarization combination.
As with layover regions, shadowed regions have no useful
signal and consequently no elevation values or other geo-
physical parameters for the surface can be estimated.
Because radar primarily interacts with structures that have
lengths comparable to the wavelength or larger, longer wave-
length (lower frequency) radars tend to penetrate deeper into
the vegetation canopy or ground surface. The amount of
penetration in a vegetation canopy depends on the structure
and density of the vegetation. Radar wavelengths less than
roughly 10 cm mostly scatter off the upper portions of can-
opies, whereas wavelengths longer than 20 cm penetrate
deeper into a canopy. This differential penetration effect for
lower-frequency radars has led to the development of radar
systems designed to exploit this phenomenon. Ground-surface
penetration depends on the type and composition of the
ground layers, ground cover, and soil moisture. Longer-wave-
length radars have been known to penetrate several meters or
more in dry sandy soil, and even deeper into certain types
of ice.
The vector nature of electromagnetic radiation can be ex-
ploited to gain further information about a surface. SAR sys-
tems that transmit and/or receive multiple polarizations of an
electromagnetic wave are called polarimetric SAR systems.
Systems that transmit and receive on two orthogonal polar-
izations (usually H and V) are said to be fully or quad-
polarimetric. Most polarimetric-airborne systems are fully
polarimetric, whereas a number of polarimetric-space-borne
systems operate primarily using a restricted set of polarization
combinations (e.g., HH and HV or HH and VV) due to per-
formance constraints. Polarization diversity is achieved by
transmitting H and V polarized waves on alternate pulses and
then receiving H and V data for every pulse as illustrated in
Figure 23.
Imagery from multiple polarizations is often combined to
form a false-color composite image (Figure 24). For example,
from the NASA/JPL L-band fully polarimetric radar over Death
Valley, CA, the color-composite image shown in Figure 24 was
formed with the red layer from an HH image, the blue layer
from the VV image, and the green layer from the HV image.
Color-coding of the image components was chosen to be
Figure 24 NASA/JPL uninhabited aerial vehicle synthetic aperture
radar (UAVSAR) color composite (red ¼ HH, blue ¼ VV, green ¼ HV)
image of polarimetric data collected over Death Valley, CA. Color
scheme is chosen to highlight various scattering mechanisms useful
in the delineation and classification of various surface types.
suggestive of particular scattering mechanisms that aid in
classification of the surface types, such as red for double
bounce, green for volume scattering, and blue for surface
scattering. In this image, we see that green represents the al-
luvium and the vegetated regions where volume scattering
dominates. Blue regions correspond to surfaces where the
radar is not penetrating and surface scattering dominates.
3.3.4.2 Geomorphological Applications
Radar images can be used for a variety of geomorphic appli-
cations that make use of both qualitative, photointerpretation
techniques and quantitative inversions of the data. Photo-
interpretation relies on the basic image elements including
tone and texture, which in radar images, are dominated both
by the imaging geometry and by the surface roughness and
dielectric constant. Because radar images are obtained by im-
aging to the side of the flight line, geometry is more important
in radar image interpretation than for interpreting optical-
sensor data such as satellite imagery or aerial photography. As
described above, radar-image layover and shadowing occur for
extreme combinations of radar look angle and topography.
Small radar look angles combined with steep topography can
lead to layover, and large look angles combined with high
relief can cause shadows. Both cases lead to loss of data.
Outside of those extremes, radar images at small look angles
emphasize low relief, much like airphotos at low sun angle.
This is at the cost of distorting higher relief through fore-
shortening or layover. By contrast, radar images obtained at
larger look angles are more map like, and tone in these images
is more directly related to surface characteristics such as
roughness near the scale of the radar wavelength and dielectric
constant.
We illustrate image interpretation aspects of using SAR data
for geomorphological and geological application using data
collected by the NASA/JPL AIRSAR in Australia (Tapley and
Craig, 1995; Tapley, 1996, 1998, 2000). Here, polarimetric
multi-frequency, C-, L-, and P-bands (wavelengths 5.6, 24, and
70 cm, respectively) are combined in color-composite imagery
 Microwave Remote Sensing and Surface Characterization 63

and filtered to highlight various geologic units. Figure 25
shows a three-frequency false-color composite image of the
Ophthalmia Range region of the Hamersley Basin in Western
Australia. Interpretation of the radar imagery relies on the
ability to distinguish the interplay between surface morph-
ology and composition of landform units. The labels 1, 2, and
3 indicate several morphological elements that are discernable
in this imagery. Outcrops of the Ophthalmia Range have been
and continue to be a supply of an erosional material forming
both colluvial (location 1) and alluvial fans (location 2), as
well as sheetwash plains (location 3).
Figure 25 AISRAR three frequency false-color composite image
consisting of red from C-VV, blue from L-VV, and green from P-VV in
the vicinity of the Ophthalmia Range, Hamersley Basin in Western
Australia. Image dimensions are approximately 16  8 km.
Another interesting application of the sensitivity of radar
images to slope angle occurs for sand dunes. As the angle of
repose for loose sand is about 331, it has been observed that
radar images of sand dunes at look angles smaller than that
value contain many bright spots caused by direct reflection
from the dune slip faces and other lower-angle surfaces (Blom
and Elachi, 1981). At larger look angles, the smooth sand
yields little scattering of the radar waves, as described in the
following.
Surface roughness, although not a quantity that most Earth
scientists consider routinely, is related to a number of geologic
characteristics. Roughness of lava flows, particularly the dif-
ferences between pahoehoe and aa, reflects differences in
eruption and composition. Weathering of different lithologies
typically yields varying sizes of debris that can be detected
easily in radar images. Other surficial processes such as
mantling by aeolian dust, desert pavement formation, and salt
weathering, all serve to modify surface roughness (Farr, 1992;
Farr and Chadwick, 1996).
Radar imagery collected over the Big Island of Hawaii by
the NASA/JPL uninhabited aerial vehicle synthetic aperture
radar (UAVSAR) fully polarimetric L-band (23-cm wavelength)
radar is used to illustrate how both qualitative and quantita-
tive aspects of surface roughness are reflected in polarimetric
imagery. Figure 26 shows HH, HV, and VV individual polar-
ization images and a color-composite image from a 23-km-
wide strip of data imaging the Kilauea volcano. Lava
flows, vegetation, volcanic craters, fault zones, and ocean

HH HV VV Composite
Figure 26 Individual polarization (HH, HV, and VV) and false-color composite imagery from the NASA/JPL L-band polarimetric radar collected
over the Kilauea volcano on the Big Island of Hawaii. To the right are expanded views of the three orange-boxed areas. Lava flows, vegetation,
volcanic craters, fault zones, and ocean are all easily differentiated using typical photo-interpretation methodologies.
64 Microwave Remote Sensing and Surface Characterization

Co-polarization signature Co-polarization signature

1 1
Normalized power

Normalized power
0.8 0.8
0.6 0.6
0.4 0.4
0.2 0.2
0 0
150 −40 150 −40
−20 −20
100 0 100 0
 50 20  50 20
0 40  0 40 

Co-polarization signature Co-polarization signature

1 1

Normalized power
Normalized power

0.8 0.8
0.6 0.6
0.4 0.4
0.2 0.2
0 0
150 −40 150 −40
−20 −20
100 0 100 0
 50 20  50 20
0 40  0 40 

Figure 27 Co-polarized radar signatures generated for four surface types using the NASA/JPL UAVSAR L-band polarimetric image data shown
in Figure 26. The top-left signature is of the lava near the Puu Oo cone; top right is the signature of a forested area; bottom left is a signature of
ocean data; and, finally, the bottom right is a signature of a building. Note how the shape and pedestal height of the signature vary as a function
of the surface type. Pedestal height can often be linked to the amount of surface roughness, and the shape and number of peaks provide insight
into the scattering mechanism involved.

are all easily discerned using standard photo-interpretation
strategies. On the left are expanded views of the regions en-
closed by the orange boxes. Lava flows (reddish-purple) and
volcanic craters and forested regions (green) are clearly visible
in the top two images. Wind-roughed ocean surfaces show up
as magenta-like color in the bottom image. Fault zones and
other geologic units are also clearly distinguishable in the
imagery.
More quantitative interpretations using polarization
signatures or polarimetric decompositions can also be used
to distinguish surface types and morphologies. Figure 27
shows the polarization signatures for lava near the Puu Oo
cone, a forested region, ocean, and building in the Kilauea
radar image shown in Figure 26. Polarization signatures
depict the normalized power return from the surface, as the
polarization is varied (i.e., as the tilt and ellipticity are
varied). Pedestal height of the polarimetric signatures can
generally be related to surface roughness and the shape of
the polarization signature can provide information con-
cerning the dominant scattering mechanism (e.g., surface or
volume scattering), providing additional information about
the surface.
Vegetation appears rough to radar sensors and radar images
have been used extensively for vegetation type and density
mapping. In fact, quantitative relationships between biomass
and radar-image brightness have been developed and used for
mapping large regions (Saatchi et al., 2007).
Smooth surfaces (at the scale of the wavelength) show up
as dark surfaces in radar images because they scatter the inci-
dent waves away from the receiving antenna. Common ex-
amples of these types of surfaces include calm-water bodies
and smooth-sand sheets. An interesting effect occurs, however
for flooded-vegetated areas. Even if the flooded surface is not
visible due to vegetation cover, enough of the radar signal
penetrates the vegetation to reflect off of the smooth water
surface only to then scatter among the vegetation. This has
been observed to enhance the radar brightness of a flooded
forest (Hess et al., 1990).
Generally, radar waves do not penetrate significantly into
rock, soil, or ice surfaces. If the soil is very dry and uniform
(no large rocks or horizons) or the ice very cold, however,
radar images may be obtained of the subsurface. The amount
of penetration is also proportional to the wavelength and has
been documented most for L-band (25 cm) (Schaber et al.,
 Microwave Remote Sensing and Surface Characterization
1997) and C-band (5 cm) (Robinson, 2002). Nonimaging
ground-penetrating radars, which operate at the surface at
wavelengths ranging from 25 cm to several meters, have been
used in the field for some time to characterize geologic and ice
volumes (Bristow and Jol, 2003). Radar-wave penetration has
been found to be important at Saturn’s moon Titan, where the
composition and temperature of the surface promotes pene-
tration of the short (2.2 cm)-wavelength Cassini SAR up to
several meters.
Another, secondary effect on radar-image brightness is the
surface dielectric constant. On Earth, most geologic materials
have a range of dielectric constants from about 3 to 10. How-
ever, water has a dielectric constant of about 80 and so dom-
inates dielectric constant differences. Thus, wet soils can
commonly be differentiated from drier soils and quantitative
models of radar scattering can be used to estimate soil moisture.
Scattering models allow more quantitative interpretations of
radar images in terms of surface roughness, structure, and
moisture content. Generally, additional radar-image parameters
are utilized in modeling including multiple wavelengths and
polarizations. Multiple wavelengths provide information about
multiple scales of roughness and the polarization data allow
inferences about surface or vegetation structure.
3.3.5 Stereo SAR
We have seen that surface morphology can be obtained dir-
ectly by examination of SAR imagery, or, more quantitatively
in some cases, from the backscatter measurements. In these
cases, however, a direct measurement of topography was not
obtained. This naturally leads to the question of whether
topographic information can be obtained directly from SAR
data. The answer is affirmative, provided imagery is acquired
from two spatially separated vantages having the appropriate
geometry.
3.3.5.1 Theory of Operation
Measurement of topography using stereo techniques uses
relative displacement in imagery collected from two distinct
vantages to extract elevation information. This technique in
the SAR area is a natural extension of the classical stereo
techniques employed very successfully by the photogrammetry
community. Sensor geometry and differences in the nature of
optical versus SAR imagery result in some substantial differ-
ences in the processing required to generate topographic
maps. We start with a brief overview of optical-stereo geom-
etry. Figure 28 shows a typical optical-stereo imaging geom-
etry. Two cameras are positioned to view an area to be mapped
from with off-nadir angles of y1 and y2 (angles to left of nadir
in the figure are negative and angle to the right are positive)
typically in the along-track direction. An object of height, h,
above a reference image plane is displaced by amount Dxi
(i ¼ 1,2) as shown in Figure 28. Image parallax, defined as the
relative displacement of object as seen in the two images, Dp,
is equal to
Dp ¼ Dx2  Dx1 ½22
65
b
1 2
− +
H
h
Δx1
Δx2
Δp
Figure 28 A typical optical-stereo viewing geometry is illustrated.
Two optical cameras view an object of height, h, from two distinct
vantages from which relative image displacements in the resulting
imagery are converted to elevation measurements.
b
1
2
H
h
Δpr Δx1
Δx2
Figure 29 SAR stereo-imaging geometry where a scene is imaged
from two vantages with incidence angles y1 and y2.
that is related to height by
Dp
h¼ ½23
tany2  tany1
Note that since y1 and y2 have opposite signs, eqn [23] is
well defined and reduces to Dp=ð2tany1 Þ when the two vant-
ages have viewing angles that are of equal magnitude.
Although a couple of distinct imaging geometries are
possible for SAR-stereo topography generation, we will con-
centrate on the cross-track stereo geometry that is most com-
monly employed in this section. Figure 29 shows the SAR
cross-track stereo imaging geometry.
Unlike optical images, where objects appear in imagery
based on the viewing angle, objects in SAR images are pos-
itioned based on range. An object of unknown height is lo-
cated in an SAR image where the sphere of that range
intersects the image reference plane. Figure 29 shows an ob-
ject of height, h, being imaged from two vantages. The range
sphere can be well approximated by a line perpendicular to
the line of sight, and the location of an object in an SAR image
is the intersection of that line with the reference surface as
shown in Figure 29. The radar parallax, Dpr, is the relative
position of an object in the two images which is given by
Dpr ¼ Dx2  Dx1 ¼ hðcoty2  coty1 Þ ½24
66 Microwave Remote Sensing and Surface Characterization

that can be inverted for height to give generated via SAR stereo. When no a priori elevation data are
available, geocoding to a flat reference surface is done.)
Dpr
h¼ ½25 Figure 31 shows data collected over Mt. Everest in Nepal by
coty2  coty1 shuttle radar SIR-C in 1994 at two incidence angles of 541 and
581. By collecting data with relatively large incidence angles,
Given eqn [25], several important observations about SAR layover problems were reduced at the expense of greater
stereo can be made. The sensitivity of height to the parallax shadowing in the imagery.
measurement is given by After geocoding, automated scene matching is used to
qh 1 identify corresponding points in the two images. Careful
¼ ½26 filtering of the match data is required to eliminate bad mat-
q Dpr coty2  coty1
ches obtained from automated-matching algorithms. Often
pyramid- or hierarchical-matching schemes are employed that
Thus, the larger the difference in incidence angles, the greater
start at low resolution and then go to progressively higher
is the height sensitivity. This relationship would seem to imply
levels of resolution until the desired resolution level is
that to maximize stereo-mapping accuracy, large incidence-
reached. Filtering algorithms can then take advantage of the
angle differences are desired. However, another aspect of ste-
various matching levels to eliminate bad matches. From the
reo processing dominates radar-stereo-derived elevation
matched points, the parallax can be computed and converted
measurement accuracy that constrains the incidence angle
into a height measurement. In practice, height determination
difference selection. If the difference in incidence angles is not
is done in a more rigorous fashion than simply using eqn [30].
too large (which is usually the case – see discussion below on
An SAR sensor-imaging model that incorporates all the
automated scene matching), then eqn [26] can be approxi-
measurements is used and solves for the three-dimensional
mated as
location of a point, which is discussed later in this section.
qh sin2 y If a rasterized-elevation product is desired, the irregularly
¼ ½27
q Dpr Dy

where y is the average of y1 and y2, Dy ¼ y2  y1 and, therefore,

the elevation error, sh, in terms of the matching accuracy,
sDpr , is

sDpr sin2 y
sh ¼ ½28
Dy

with typical automated stereo-matching algorithms having an

accuracy of 0.3 a pixel or larger and y between 301 and 401,
and Dy about 101, the height accuracy is about ½ the pixel
size. This simple rule of thumb is often heard with regard to
SAR-stereo height accuracy, but it is predicated on several as-
sumptions about the stereo geometry and matching accuracy.
Figure 30 shows the generic processing flow for SAR-stereo
topographic map generation.
Radar-stereo processing consists of four main steps. The Incidence angles
process begins by processing SAR imagery for each of the two 56°−60°
channels and then geocoding to some reference surface. The
reference surface may be a simple plane or some higher- Incidence angles
fidelity representation of the surface (e.g., a low-resolution 52°−57°
digital elevation model. (Usually, the reference surface has Figure 31 Stereo pair of Mt. Everest collected by the shuttle radar
much lower resolution than the product that is being SIR-C in 1994.

Ephemeris
Raw SAR image data
Geocode
data formation
image 1 Low Scene Match Height
resolution
dem matching filtering determination
Raw SAR image
Geocode
data formation Rasterize to
image 2
desired grid

Figure 30 The four main steps in the generic processing flow used to generate SAR-stereo topographic maps.
 Microwave Remote Sensing and Surface Characterization 67

spaced points obtained in the height-determination
process must be resampled or interpolated onto a chosen
map grid.
Scene matching of SAR imagery is the limiting factor af-
fecting stereo elevation-estimation accuracy. There are two
primary mechanisms that limit the matching accuracy. First,
the two vantages with different incidence angles result in
image distortions related to SAR image formation. For ex-
ample, foreshortened and shadow regions have different di-
mensions due to the difference in incidence angles. Moreover,
the location of the brightest point on a slope facing toward the
radar may come from different physical locations that can be
misidentified in the matching process, as illustrated in
Figure 32.
The larger the incidence-angle difference, the greater the
relative distortion between the images and the larger the
matching error. However, reducing the incidence-angle dif-
ference results in decreased height acuity and, hence, the op-
timal incidence-angle difference is a balance between better
matching accuracy and height acuity. Studies have shown in-
cidence-angles differences of about 101 result in the best SAR
topographic maps.
The second factor limiting matching accuracy is SAR
speckle. Speckle is point-to-point random fluctuations in
the image magnitude that is a result of using a mono-
chromatic imaging source. This type of noise is not present in
optical imagery and results in reduced matching accuracy of
SAR data relative to optical data. It can be shown that the
magnitude of single-look SAR imagery is exponentially dis-
tributed, meaning that the mean and standard deviation are
equal. The random fluctuations are uncorrelated between
images when the incidence-angle difference exceeds a critical
threshold that is typically the case for SAR stereo-image pairs.
Speckle noise is reduced via spatial averaging of pixels that
results in reduced spatial resolution. The amount of spatial
averaging to reduce speckle noise needs to be balanced against
having higher-resolution data for the matching process.
Nominally, the minimum amount of spatial averaging com-
mensurate with a significant reduction in speckle noise
is done.
Low-contrast, shadowed, and layover regions generally are
not mapped by SAR stereo. Selecting the incidence angles used
in mapping can minimize the area affected by shadow and
layover. SAR stereo collections in mountainous areas from
opposite sides can also reduce dropout area from shadow and
layover. Low-contrast featureless terrain precludes good stereo
matches and these areas are better mapped via other techni-
ques such as SAR interferometry. Figure 33 shows the output
of an automated hierarchical matching algorithm for the SIR-
C Mt. Everest data. The matching algorithm also outputs a
match quality metric (estimated covariance error in pixels)
that is used in the filtering algorithm and to generate a height-
error map for stereo-derived heights. (The height error (more
accurately height precision) map is the estimated statistical

2

1

Images from 1 Images from 2

Mismatched
object

Figure 32 Illustration of the matching ambiguity that can occur with the radar stereo-imaging geometry. In this example, a hill is illuminated
from two different incidence angles with the brightest returns in their associated images corresponding to the points on the hill near normal
incidence. An automated matcher can misidentify the bright streaks as the same points and thereby lead to error in the SAR-stereo elevation.
68 Microwave Remote Sensing and Surface Characterization
75 pixel contour wrap
Radar stereo disparity Matching error
Figure 33 Disparity or parallax maps generated from the output of
an automated matching and filtering algorithm applied to the SIR-C
Mt. Everest data.
25
6000 m contour wrap
Elevation error
0 be useful when deciding the appropriateness of radar stereo-
Elevation Elevation error
Figure 34 Elevations and elevation precision (height-error map)
derived from SIR-C radar-stereo data over Mt. Everest. Color
contours of elevations with a 6000-m color wrap and height errors
with a 25-m color wrap are overlain on radar-backscatter data.
height accuracy of a point, and not the actual height error
relative to the reference map. This layer is very valuable when
using elevation data for quantitative analysis, or filtering data
that contain points not meeting a desired map accuracy for a
particular application.)
Equation [30] relates the height of a point in terms of its
observed parallax; however, for precision mapping appli-
cations, a set of equations based on direct SAR imaging
measurements is used. SAR imaging from two vantages gives
rise to a set of four nonlinear equations for the target-position
vector that will be referred to as the nonlinear stereo equa-
tions. These equations geometrically represent the intersection
of two range spheres centered at the platform positions when
Mt Everest
N
6000 m per color cycle
Figure 35 UTM projection of the SIR-C stereo-derived DEM of Mt.
Everest.
the target was imaged, and the corresponding two Doppler
cones. Exact closed-form solutions are available to these
equations; however, the exact solution requires the use of only
three of the four equations (Hensley, 2008). Use of all the
measurement data from the four stereo equations is possible
using least-square techniques. There are a couple advantages
to using least squares to obtain the stereo-solution vector.
First, by using all the measurement data and by providing
input-error covariance information on the matching accuracy,
platform ephemeris, and range data, a formal position-error
estimate can be obtained for every point in the map. This can
derived elevation measurements to particular geomorpho-
logical applications. A second benefit is that corrections to the
range for atmospheric delay and bending can be done as part
of the least-square iterative procedure. Figure 34 shows ele-
vations and estimated elevation precisions derived from the
SIR-C stereo data collected over Mt. Everest. Elevations in the
scene varied from 3500 m to 8800 m. Formal height-precision
estimates that varied from 15 to 50 m were verified using
higher-accuracy radar interferometric data. Average estimated
elevation precision was 25 m with a standard deviation 5.3 m.
Finally, Figure 35 shows the radar-derived DEM in a
standard map projection (Universal Transverse Mercator
(UTM)).
3.3.5.2 Geomorphological Applications
Radar-stereo data have not been used extensively for Earth-
based geomorphological investigations primarily due to the
ready availability of more accurate topographic data obtained
 Microwave Remote Sensing and Surface Characterization 69

from optical or radar-interferometric techniques. However,

planetary investigations, particularly Venus and Titan, have
benefited from radar-stereo data, which to this point is the
only available means of obtaining high-resolution topo-
graphic data. Because the applications and interpretation
techniques for stereo SAR are similar to optical stereo, the
extensive literature on the use of optical-stereo images will not
be repeated here. Main differences between optical- and radar-
stereo-derived topographic data are related to the geometric
differences between side-looking radar images and nadir-
looking optical images. The presence of radar image speckle
and geometric distortions induced by the incidence-angle H
difference discussed earlier are a major factor limiting stereo
topographic mapping accuracy. This contributes to high spa- Swath width
tial frequency noise in the resulting topographic maps that can
cause difficulty interpreting small features and can yield Figure 36 Interferometric SAR for topographic mapping uses two
anomalously large values for small-scale slopes. apertures separated by a baseline to image the surface. The phase
Radar stereo has been used to constrain impact crater difference between the apertures for each image point, along with the
range and knowledge of the baseline, can be used to infer the precise
morphology on Venus (Cochrane and Ghail, 2006). The
shape of the imaging triangle (in red) to determine the topographic
Magellan radar systematically mapped Venus in the early
height of an object.
1990s owing to its optically opaque atmosphere, precluding
other means of obtaining high-resolution imagery of its sur-
face. Mapping was conducted in three cycles, each in one
Venus day or 243 earth days in duration. On the first and third
imaging cycles, left-looking radar-stereo data were obtained
D
for approximately 30% of the planet’s surface before recorder
2 4
problems forced the suspension of further mapping oper-
ations. Stereo data from approximately 421 N to 441 S latitude
and 511E to 901 latitude containing the Aphrodite region were
used to study the morphology of impact craters. By comparing
depth-to-diameter trends, the authors found that the data were
not in accordance with that predicted from theory. They found
that craters smaller than 18 km in diameter were unexpectedly
shallow, leading them to conclude that midsize bolides 1 2
(200–1000 m in diameter) fragment in the dense Venusian
atmosphere and then disperse over an area 10–20 times the D
Phase = 2
diameter of the bolide diameter. 
Radians per wavelength Number of Wavelengths
Figure 37 Distance in a radar system is converted to a phase
3.3.6 Interferometric SAR measurement by dividing the distance by the radar wavelength to get
the distance in wavelengths and then multiplying by 2p, the number
of radians per wavelength to get to phase. Measured-phase values in
3.3.6.1 Theory of Operation a radar image do not contain the total number of wavelengths but
By augmenting a conventional SAR system with another spatially only retain the fractional portion of a wavelength. Typically, phase in
separated receiving antenna, as illustrated in Figure 36, it is pos- a radar interferometric image is color coded as shown above, and
one wavelength’s worth of phase change is called a fringe.
sible to extract topographic information. More details on inter-
ferometric synthetic-aperture radar (IFSAR) systems and
processing can be found in Bamler and Hartl (1998), Fran- obtaining a basic understanding of the measurement con-
ceschetti and Lanari (1999), Madsen and Zebker (1999), Rosen cept.) Phase information in a single SAR image encodes sub-
et al. (2000), and Hensley et al. (2001a). By receiving data on wavelength range information that normally is not accessible.
both antennas, it is possible to extract fine-resolution topographic Extraction of this sub-wavelength range information is only
information from either space-borne or air-borne platforms. possible by using image pairs satisfying geometric imaging
Up to this point, we have considered SAR images to consist constraints. There are two primary components to the phase
of only a brightness or magnitude value at each image pixel. In measurement consisting of a systematic and a random part.
actuality, SAR-pixel values are complex numbers consisting of The systematic part is the range to the resolution element that
both a magnitude and phase. (Although glossed over when is converted to a phase value by dividing by the wavelength
polarimetric-imaging radars were introduced, phase infor- (gives distance in unit of wavelengths) and then multiplying
mation is also critical to the understanding and use of po- by 2p, which is the number of radians in one wavelength as
larimetric data. We suppressed this complication until the illustrated in Figure 37. Even though there may be many
introduction of interferometry where it is essential to millions of wavelengths (hence many multiples of 2p) from
70 Microwave Remote Sensing and Surface Characterization

Range from antenna Coherent sum of elemental

Phase =
to center of cell
Systematic component desired
by interferometric measurement
+ +
scatters arranged randomly in cell
Random component that if look direction is nearly the same
and scatterers within cell do not move relative to each other
this component cancels in the interferogram formation process.
Thermal noise
Random component that does not cancel
and results in interferometric phase noise

Figure 38 Phase is composed of a systematic component that encodes the range difference between the two antennas and point, and a
random part that consists of both random thermal noise and random-arranged elemental scatterers within a resolution element.

the antenna to a resolution element, only the principal value Baseline

(a number between  p and p) can be extracted from a
complex-valued resolution element. D2
The random component may also be thought of as con-
sisting of two components summarized in Figure 38. The first D1
component, thermal noise, corrupts all electronic measure-
ments and is present in every pixel value. Its impact on the
phase depends on the relative magnitude of the reflected signal
magnitude to the noise level. The second component is geo- L4
L3
metric in nature and depends on the arrangement of so-called
elemental scatterers, points within a resolution element that L2
L1
dominate the signal value, and their contribution depends only
on the viewing geometry and the composition of the scatterers.
In general, because the distribution of elemental scatterers
within a resolution element changes from element to element
and the range converted to phase modulo 2p (subtracting off
the integer multiples of 2p) to a resolution element is randomly Differential ranges
distributed, phase values in SAR images are also randomly
distributed. It is important for the interferometric SAR dis-
cussion to follow to note that if the viewing geometry is nearly D1 = L2 − L1 D2 = L4 – L3
unchanged and the elemental scatterers within a resolution cell
are undisturbed, then this portion of the random phase re- Figure 39 Topographic information can be extracted from the
differential-distance encoded in the interferometric phase obtained
mains the same. Thus, the thermal noise is random in time and
from two antenna separated in distance in the cross-track direction
the elemental scatterer noise is random in space.
called the baseline. D1 and D2 are the differential distances measured
As mentioned previously by using two images sufficiently by the interferometric phase for two points at different elevations.
similar in imaging geometry and combining the radar data Since the elevation affects the differential distance and hence the
coherently, the elemental-scatterer portion of the interfero- phase, it contains topographic information.
metric phase cancels out, leaving only the difference of the
systematic portions of the phase. Thus, interferometry recovers
the sub-wavelength range information encoded in the phase
from two images to extract differential-range information with A2
an accuracy of a fraction of a wavelength. The interferometric- B
phase difference is essentially related to the geometric path-
length difference to the image point, which depends on the A1
topography as is shown in Figure 39. With knowledge of the Δ
interferometer geometry, the phase difference can be con-
 2
verted into an elevation measurement for each image point.
Differential-radar interferometry uses observations separated 1
in time to measure sub-wavelength changes in the surface h
topography. Observations may be separated by time intervals
ranging from fractions of a second to many years.
Previously, it was shown that knowing the SAR coordinates
of a point constrained its location to be on the intersection
locus of a sphere and cone that from Figure 21 was seen to be hτ
a circle. Parametrizing the location on this circle by an angle,
Figure 40 Simplified broadside looking (zero Doppler) radar-
referred to as the elevation angle, y, reduces the three-di- interferometry geometry. The difference in range from the two
mensional location problem to determining this angle. For observing antennas to the target is approximately equal to the
this, we need the interferometric measurement. Given two projection of the baseline vector onto the line-of-sight vector shown
identical antennas, A1 and A2, are receiving radar-echo signals in blue. This range difference can be related to a phase measurement
from a single point as shown in Figure 40. and forms the primary interferometric observable.
 Microwave Remote Sensing and Surface Characterization 71

The path-length difference, Dr, of the signals received by where l is the radar wavelength, and p equals 1, when the
the two antennas is approximately given by path-length difference is associated with the one-way path
difference, or 2 for the two-way path difference, as is the case
Dr ¼ r2  r1 Ebsinðy  aÞ ½29 for Ping-Pong or repeat-pass systems described below. Geo-
metrically, the phase measurement represents a second cone
where ri indicates the range from antenna i to the target, b is
with cone axis aligned with the interferometric baseline.
the length of the baseline vector which is the vector pointing
Intersecting the phase cone with the range sphere and Doppler
from antenna 1 to antenna 2, y is the desired elevation (or
cone determines the elevation angle to the target, and there-
look) angle, and the baseline-orientation angle, a, is the angle
fore the target’s full three-dimensional location, as shown in
the baseline vector makes with respect to the horizontal.
Figure 41.
Observe that the range difference to a good approximation for
Using the simplified geometry of Figure 39, the height of a
most systems is simply the length of the projection of the
target, ht, is given by
baseline vector onto the line of sight. The range difference, Dr,
may be obtained by measuring, f, the phase between the two ht ¼ h  r cosðyÞ, ½31
interferometer signals, using the relation
2ppDr where h is the altitude of the radar antenna and r is the slant
f¼ , p ¼ 1,2 ½30
l range from the antenna to the target. Since the signal phase is
sensitive to displacements between images of a fraction of a
Range sphere wavelength, the interferometric technique provides a very
accurate means of determining topographic heights. Using
eqns [29]–[31] the elevation angle can be determined to be
 
lf
Doppler cone y ¼ sin1 þ a: ½32
2ppb
Baseline
vector
SAR It is given from eqns [31] and [32] that determining the
position height of a target requires knowledge of the platform position,
the range, the interferometric baseline length, the baseline
angle, and the interferometric phase. Generation of accurate
Velocity topographic maps using radar interferometry places stringent
vector requirements on the knowledge of the platform and baseline
Phase cone vectors. Figure 42 shows the interferometric-phase measure-
ments and amplitude image for data collected by the SIR-C
Target location imaging radar over the Mojave Desert in California.
It is important to appreciate the consequences of the fact
Figure 41 Target location in an SAR image could be anywhere on
the interferometric-phase measurement is made modulo 2p.
the intersection of a range sphere and Doppler cone, thereby
providing no information on the target’s elevation. 3-D information is The total-range difference between the two observations
obtained by the intersection of the phase cone with range sphere and that the phase represents in general can be many multiples of
Doppler Cone. the radar wavelength, or expressed in terms of phase, many
Azimuth

Range
Figure 42 Interferometric phase (right) and associated amplitude image (left) of the Mojave Desert, CA, generated from repeat-pass
observations using the SIR-C radar. Observe that the phase values soon on the right mimic topographic contours and hence can be converted
into topographic measurements.
72 Microwave Remote Sensing and Surface Characterization
multiples of 2p. It is this value that is required in order to
make height measurements. The standard approach for de-
termining the unique phase that is directly proportional to the
range difference is to first determine the relative phase be-
tween pixels via the so-called ‘phase unwrapping’ process.
Unwrapping of IFSAR imagery is a nontrivial process for
which a number of algorithms have been developed. Com-
plications arise in avoiding unwrapping errors in regions of
shadow, layover, and low signal return. The connected phase
field after unwrapping may still need to be adjusted by an
overall constant of 2p. The step that determines the overall
constant of 2p is referred to as absolute phase determination.
Interferometric correlation, a measure of the similarity of
the signal received at the two antennas, can be estimated
directly from the image data of the two interferometric
channels (Zebker and Villasenor, 1992). Correlation meas-
urements have values between 0 and 1, with 1 designating
perfect correlation between the channels. Sometimes it is more
convenient to refer to the amount of interferometric decorr-
elation, which is defined as one minus the correlation. The
amount of decorrelation due to the slightly different viewing
geometry is called ‘geometric decorrelation’. Thermal noise-
induced signal decorrelation is called ‘noise decorrelation’.
Shadowed regions suffer from noise decorrelation and areas
on steep slopes exhibit geometric decorrelation that increases
phase noise and can preclude useful phase measurements
altogether. Another form of decorrelation occurs when there is
a vertical distribution of scattering elements within a reso-
lution element as shown in Figure 43. Not only is the signal
decorrelated, but also the point within the resolution cell
corresponding to the interferometric-phase measurement de-
pends on the wavelength and the scatterer distribution in the
cell. This form of decorrelation is called ‘volumetric decorr-
elation’ and can be used to infer information about the ver-
tical structure of the volume. Recently, there has been a great
deal of activity using volumetric correlation to estimate tree
and canopy structure within the interferometric SAR com-
munity (Cloude and Papathanassiou, 1997; Hensley et al.,
2001a; Hensley et al., 1995; Hokeman and Varekamp, 2001;
Hagberg et al., 1995).
There is an upper value on the useful baseline length
known as the critical baseline (Rosen et al., 2000). The critical
baseline is reached when the amount of phase change per
resolution element exceeds 2p radians. This limitation is a
result of the fact that the interferometric-phase measurement
is made only modulo 2p. As the baseline approaches the
Look direction
Δ
Figure 43 Vertical structure of scatterers within a resolution element due to vegetation or other structures present in the cell causes increased
decorrelation. This form of decorrelation can be used to infer information about vertical structure within a resolution element. The increased
decorrelation results from the increased size of the range cell projected back toward the direction of the radar (shown in magenta) when
compared to a flat surface.
critical baseline, the phase values from the two antennas be-
come completely decorrelated. However, as the baseline in-
creases, the sensitivity of phase to height increases improving
the accuracy of interferometric SAR systems. Practical mapping
systems must select baselines with a balance between adequate
phase to height sensitivity to meet mapping requirements
and excessive decorrelation with corresponding processing
difficulties.
Combining the SAR-image formation process, interfero-
metric-phase measurement, unwrapping, and height de-
termination into an automated processing algorithm require a
process flow that is shown in Figure 44 (Madsen et al.,
1993b). The first processing step is decoding the byte data,
followed by range compression for each of the two interfero-
metric channels. Using the platform-motion information ob-
tained from inertial navigation unit (INU) and GPS
measurements, as well as any other baseline metrology de-
vices, the data are compensated for perturbations in aircraft
motion from a reference path and then azimuth compressed.
This generates two single-look complex images. (The process
of correction for motion away from an ideal path is referred to
as ‘motion compensation’. Motion compensation corrects for
motion on the order of fractions of a wavelength in order to
generate well-focused SAR imagery. The requirements for
motion compensation are even more demanding for IFSAR
systems and place stringent requirements on the platform and
baseline metrology systems.)
One of the single-look complex image pairs is resampled to
overlay the other. This registration must be done to a small
fraction of a pixel (typically o0.1 of a pixel) in order to avoid
phase decorrelation. Multiplying the complex pixel value in
one image by the complex conjugate of the corresponding
pixel value in the second image forms an interferogram. The
resulting interferogram is multi-looked, by spatially averaging
the complex pixels in a box about a given pixel to reduce the
amount of phase noise.
After the multi-looked interferogram has been generated,
the phase for each complex sample is computed. To generate a
continuous height map, the two-dimensional phase field must
be unwrapped. After the unwrapping process, an overall
multiple of 2p is estimated and added to the unwrapped phase
(the estimated value may be 0).
Subsequent to determining the absolute phase for each pixel
in the interferogram and possibly taking additional looks, the
three-dimensional target position can be determined. Phase
corrections are applied to the interferometric phase to account
Δ
 Microwave Remote Sensing and Surface Characterization 73

Raw signal Decode and range Azimuth

data antenna 1 compression compression
Image resample
and interferogram Looks
formation
Raw signal Decode and range Azimuth
data antenna 2 compression compression

NAV data
Calibration
data

Baseline Phenomenological
Coarse DTE measurement studies

Troposphere
model Post
Absolute phase
determination processing

3 dimensional Primary output Mosaicking

Phase unwrapping Looks location and gridding product generation

Module for which

SAR exact position in Ancillary Used if needed
signal processing processing is data by selected
algorithm dependent algorithm

Figure 44 Interferometric processing block diagram.

for tropospheric effects, and the range is corrected to account
for changes in the speed of light through the atmosphere. Using
accurate baseline and platform position information, the phase
and range information for the target position is computed.
A relief map is generated by gridding the unevenly sampled
three-dimensional target locations into a desired coordinate
system. The gridded products include the target elevations, the
SAR image (which has been orthorectified), a correlation map,
and a height-error map described below. The resulting radar-
relief map may be measuring the heights above the ground,
within the vegetation canopy, or beneath the surface in arid
regions. To convert this map into a true ground-surface DEM,
corrections based on phenomenological studies (e.g., using
scattering or semi-empirical curves to correct elevation meas-
urements based on the amount of decorrelation in the canopy)
must be incorporated either into the three-dimensional
location algorithms or into a post-processing step.
One of the unique aspects of interferometric-SAR systems is
the ability to determine the statistical-height precision, that is
the degree of height noise from pixel to pixel, estimated from
knowledge of the correlation, g (Hensley and Webb, 1994). The
amount of phase noise between the two channels is simply and
directly related to the correlation and number of looks used to
reduce phase noise. (The Cramer–Rao bound (expresses a lower
bound on the variance of estimators of a deterministic par-
ameter) used to relate the phase noise to correlation and
number of looks is only valid when the number of looks ex-
ceeds four or five. The number of looks in most interferometric
systems used to generate topographic maps usually is much
larger than four.) The Cramer–Rao bound relating the phase
variance, sf, to the correlation coefficient, g, is given by
sffiffiffiffiffiffiffiffiffiffiffiffiffi
1 1  g2
sf ¼ pffiffiffiffiffiffiffiffiffi ½33
2NL g2
where NL is the number of looks. From eqns [32] and [33] the
height error, sh, as a function of the phase noise is found to be
lrsinðyÞ
sh ¼ sf ½34
2ppbcosðy  aÞ
Equations [33] and [34] allow the generation of an ele-
vation-precision map showing the local height precision for
each post in an interferometrically derived DEM (Zebker et al.,
1994; Madsen et al., 1993a).
Differential-radar interferometry uses repeat-pass obser-
vations separated in time by seconds to years to measure
change in the surface under observation. This technique can
measure surface displacements at the millimeter level over
large areas, making it ideal for a variety of geophysical in-
vestigations. Differential interferometry has become an ac-
cepted tool in the geophysical community for mapping surface
deformation from natural sources such as earthquakes, vol-
canoes, and glaciers, and from anthropogenic sources such as
oil or groundwater pumping. Differential interferometry
works because the radar is far more sensitive to change than it
is to topography. Ideally, for surface deformation measure-
ments, one would like to have repeat passes with zero
74 Microwave Remote Sensing and Surface Characterization
baseline, and, therefore, have no topography sensitivity. In
that case, the only differential-range change observed would
be due to surface movement directed along the radar line of
sight, as illustrated in Figure 45. Thus, the phase measurement
due to surface change is simply given by
4p 4p ! ^
f¼ Dr ¼ / d ,c S ½35
l l
! 1995; Tarayre and Massonnet, 1996). Changes in the surface
where Dr ¼ / d , c^ S is the differential range shift of the sur-
face along the line-of-sight vector, c^ , which is simply the
! temporal decorrelation, which can either degrade the quality
projection of the surface-displacement vector, d, onto c^ :
In general, it is not possible to make a repeat-pass meas-
urement with a zero baseline, so the interferometric meas-
urement is in general a combination of both surface change
Figure 45 Surface deformation at the millimeter scale can be
measured over large spatial scales and fine resolution using the
technique of differential-radar interferometry. In the above figure,
SAR imagery is acquired a first time over a volcano. A subsequent
collection, which may be minutes to years later, is collected at a
second time spanning a deformation event to be monitored. In this
illustration, deformation is surface inflation on the flank of the
volcano. If the plane flew exactly the same trajectory, then the
change in range to a point on the deforming flank would be solely
due to surface motion. This surface motion can be extracted from the
interferometric phase using differential-radar interferometry. Repeat
passes with nonzero baselines need to remove the topography term
in the phase using an external DEM and knowledge of the baseline.
and a topographic measurement. The topographic-phase
component is computed and subtracted from the measured
phase using a DEM and knowledge of the interferometric
baseline. The smaller the baseline the less accurate the
DEM needs to be to remove the topographic component. At-
mosphere changes, particularly those due to tropospheric
water vapor, change the propagation speed in the atmosphere
and represents a major-error component to differential-inter-
ferometric surface-displacement measurements (Goldstein,
at the wavelength scale within a resolution element lead to
of the phase measurements or lead to the complete loss of
signal. Large disruptive events, such as earthquakes, generally
have severe decorrelation on areas of the fault where there is
surface rupture or buildings have been damaged. Most dif-
ferential-radar-interferometric measurements have been made
with satellite systems because obtaining knowledge regarding
irregular aircraft trajectories at the required accuracy (milli-
meter level) is considerably more challenging.
3.3.6.2 Geomorphological Applications
Perhaps the single biggest contribution of single-pass radar
interferometry for topographic mapping was the Shuttle Radar
Topography Mission (SRTM) that flew on the Space Shuttle in
2000 (Farr et al., 2007). This mission produced the first
globally consistent fine-resolution (30 and 90 m products)
topographic map of the Earth’s surface between 7601 latitude.
The shuttle radar consisted of a C-band (and X-band
radar supplied by the German Space Agency) with one an-
tenna in the cargo bay and second antenna deployed on a 62-
m boom. Accuracy assessments based on GPS and other
reference data showed the SRTM DEM has 8.8-m accuracy at
the 90% confidence level (Rodriguez et al., 2006). The
short C-band wavelength does not fully penetrate vegetation,
as was discussed earlier, so that the resulting DEM was a
reflective surface DEM (i.e., mapped elevations depend
on terrain cover). Techniques for exploiting or correcting
for this electromagnetic elevation bias are a continuing
area of research (Sexton et al., 2009; Carabajal and Harding,
2006; Hofton et al., 2006). This data set has become
the standard for world topography and has been used in
many scientific applications, including numerous uses in
geomorphology.
An aspect of repeat-pass differential interferometry relevant to
geomorphology is its ability to measure subcentimeter changes in
topography over large areas (Madsen and Zebker, 1999; Masson-
net and Feigl, 1998; Smith, 2009; Simons and Rosen, 2007). The
rapid infusion of differential-radar interferometric techniques into
geomorphology and geophysics has its genesis with the meas-
urement of the 1992 Landers earthquake using repeat pass C-band
radar interferometry from the ERS-1 satellite (Massonnet, 1993).
For the first time it was possible to see the displacement field from
a earthquake at fine resolution and large spatial scales. Figure 46
shows an interferogram of the Landers Earthquake, generated
using two ERS-1 radar passes that spanned the earthquake. Surface
displacement is encoded in the interferometric phase, and each
fringe or color contour represents 2.8 cm of surface displacement
 Microwave Remote Sensing and Surface Characterization 75

First pass Second pass

D1 D2

D1 Surface deformation causes

D2 Range to change

Surface deformation = D2 − D1

Phase = Surface deformation + “Topography”

Desired signal Small baseline facilitates

compensation of this term
Figure 46 Repeat-pass radar interferometry from the ERS-1 satellite shows the surface deformation produced by the Lander’s Earthquake in
California. This was the first major geophysical event studied by Massonet et al. (1993) that introduced radar-interferometic techniques to the
geophysical community. Each fringe or contour level represents about 2.8 cm of surface displacement. This marked the first time when it was
possible to see wide-scale surface morphology changes due to an earthquake at such fine resolution.

along the radar line of sight. Note that the fringe frequency in-
creases in the vicinity of the fault, and the fringes are absent in
many areas in proximity to the fault, mostly as a result of temporal
decorrelation. Displacements can be measured to millimeter pre-
cision using radar-interferometric techniques and can be com-
bined with other sensor information (e.g., GPS measurements to
gain detailed information on the processes involved in surface
modification). An extensive literature exists on the application of
this technique for studying earthquake deformation (e.g., Mas-
sonnet et al., 1993), volcano inflation and deflation (e.g.,
Lundgren and Lu, 2006; Masterlark and Lu, 2004; Zebker and
Amelung, 2000; Mouginis-Mark et al., 2000), landslides (e.g.,
Roering et al., 2009), and subsidence (e.g., Bell et al., 2008;
Fielding et al., 1998).
Note that this technique will not work on surfaces that
have changed too much at the sub-resolution scale, such as
sediment deposition and mass-wasting processes, which
would decorrelate the phase returns much like moving vege-
tation. The decorrelation phenomenon can be turned to an
advantage when change or disruption of a surface is the de-
sired signal. Areas of temporal decorrelation are easily detected
and have been used to map small landslides and damaged
buildings after an earthquake (Fielding et al., 2005) and active
lava flows (Zebker et al., 1996).
We also provide an air-borne surface-deformation example
of anthropogenically induced surface deformation using the
NASA/JPL UAVSAR radar (Hensley et al., 2009, 2010). UAVSAR
collected two repeat passes of 74 km in length over Lost
Hills, CA, on May 6 and July 25 of 2008 flying at an altitude
of 12.5 km. This area is known to be undergoing surface
deformation due to oil pumping based on satellite radar-
interferometric and GPS measurement (Fielding et al., 1998).
The area is mostly agricultural interspersed with light urban
development, and thus many areas remain highly correlated for
large temporal baselines at L-band. The two passes were sep-
arated by 80 days. Obvious from the radar interferogram, there
are two areas undergoing surface deformation centered over
light urban areas. Figure 47 shows the line-of-sight displace-
ment obtained from the unwrapped interferometric phase, with
a spatial resolution of about 7 m (36 looks). L-band has good
correlation over the urban development and in some fallow
agricultural regions. The maximal displacement is about 7 cm,
corresponding to a deformation rate of about 0.88 mm d1.
Figure 48 shows the plots of the displacement along selected
transects highlighted in Figure 47 that are about 3 km in
length. Note the asymmetric nature of the displacement profiles
along transects A and B that would be difficult to discern in
satellite imagery due to their coarser resolution.
3.3.7 Summary
The remarkable diversity of active microwave sensors and their
applicability to geomorphological applications have been
presented. Microwave sensors have the ability to inform
studies at length scales ranging from planetary scales down to
surface roughness at the centimeter scale. This tremendous
range of applicability of microwave sensors is a result of the
variety of sensor types and the large frequency range over
which these sensors operate.
The type of measurements from various sensor types can be
grouped into four basic categories and include:
• Backscatter data: Backscatter represents the amount of
microwave energy reflected from the surface back to the
sensor. The magnitude of the backscatter signal depends on
the surface composition through the dielectric constant and
76 Microwave Remote Sensing and Surface Characterization

−0.02 0 0.02 0.04 0.06 0.08 0.1

LOS displacement (m)
Figure 47 Oil pumping in the oil fields near Lost Hills, California, produces surface deformation at the rate of about 1 mm d1. This level of
surface deformation is detectable using either space-borne or air-borne radar interferometry. Black areas are regions where there was no
interferometric signal due to temporal decorrelation. Orange-boxed regions of surface deformation are shown zoomed in the images on the right.

12

Transect A
10 Transect B
Transect C
LOS displacement (cm)

0
0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Distance (km)
Figure 48 The color of the arrow corresponds to the color used to plot the displacement along that transect.
 Microwave Remote Sensing and Surface Characterization
on the wavelength-scale surface roughness within a reso-
lution element. Using polarimetric diversity, incidence-angle
diversity, or modeling, it is sometimes possible to extract
either quantitative or qualitative surface roughness and/or
composition information.
• Image data: SAR systems are capable of generating fine-
resolution imagery of a surface. Using image-interpretation
techniques and knowledge of radar-specific imaging idio-
syncrasies (e.g., foreshortening, layover, and shadow), it is
possible to identity and classify many features of geo-
morphological interest such as faults, lava flows, craters,
alluvial fans, etc.
• Topographic data: Altimeters and stereo and interferometric
SARs are capable of making direct measurements of surface
topography. Radar altimeters are nadir-pointing instruments
that are effectively profiling instruments that measure the
elevation beneath the flight track. Stereo and interferometric
SAR systems, on the other hand, measure surface elevations
over an extended swath and are effectively used for large area
fine-resolution topographic map generation. The nature of
the elevation measurement (e.g., whether bare surface or re-
flective surface elevations are being mapped) depends on the
wavelength of the sensor and how the data are processed.
• Surface-deformation data: Repeat-pass radar interferometry
is capable of making millimeter-precision surface-deform-
ation measurements over large areas at fine resolution from
either air-borne or space-borne platforms. Over the last two-
and-a-half decades these measurements have become inte-
gral to understanding the surficial and geophysical processes
that continue to shape the Earth’s surface.
Because these sensors operate over a variety of wavelengths
ranging from a few millimeters to a meter, and acquire data
over a range of spatial scales from one to thousands of kilo-
meters, these data are essential for a variety of geomorpho-
logical applications on both the Earth and other bodies in the
solar system.
As these sensors continue to proliferate and evolve, along
with computer systems and algorithms, microwave remote
sensing will increasingly make important contributions to
geomorphology.
References
Bamler, R., Hartl, P., 1998. Synthetic aperture radar interferometry. Inverse Problems
14, R1–54.
Bell, J.W., Amelung, F., Ferretti, A., Bianchi, M., Novali, F., 2008. Permanent
scatterer InSAR reveals seasonal and long-term aquifer-system response to
groundwater pumping and artificial recharge. Water Resources Research, 44. http
://dx.doi.org/10.1029/2007WR006152.
Blom, R., Elachi, C., 1981. Spaceborne and airborne imaging radar observations of
sand dunes. Journal of Geophysical Research 86, 3061–3073.
Bristow, C.S., Jol, H.M., 2003. Ground penetrating radar in sediments. Geological
Society. Special Publications, London, 211, 328.
Carabajal, C.C., Harding, D.J., 2006. SRTM C-band and ICESat laser altimetry
elevation comparisons as a function of tree cover and relief. Photogrammetric
Engineering and Remote Sensing 72, 287–298.
Cloude, S., 2010. Polarisation: Applications in Remote Sensing. Oxford University
Press, Oxford.
Cloude, S.R., Papathanassiou, K.P., 1997. Polarimetric optimization in radar
interferometry. Electronics Letters 33(13), 1176–1178.
Cochrane, C., Ghail, R., 2006. Topographic constraints on impact crater
morphology on Venus from high-resolution stereo synthetic aperture
77
radar digital elevation models. Journal of Geophysical Research 111,
E04007.
Curlander, J.C., McDonough, R.N., 1991. Synthetic Aperture Radar Systems and
Signal Processing. Wiley, New York, NY.
Farr, T.G., 1992. Microtopographic evolution of lava flows at Cima volcanic field,
Mojave Desert, California. Journal of Geophysical Research 97, 15171–15179.
Farr, T.G., Caro, E., Crippen, R., et al., 2007. The Shuttle Radar Topography
Mission. Reviews of Geophysics, 45,doi:1029/2005RG000183.
Farr, T.G., Chadwick, A.O., 1996. Geomorphic processes and remote sensing
signatures of alluvial fans in the Kun Lun Mountains, China. Journal of
Geophysical Research 101, 23,091–23,100.
Fielding, E.J., Blom, R., Goldstein, R., 1998. Rapid subsidence over oil fields
measured by SAR interferometry. Geophysical Research Letters 25,
3215–3218.
Fielding, E.J., Talebian, M., Rosen, P., Nazari, H., Jackson, J., Ghorashi, M.,
Walker, R., 2005. Surface ruptures and building damage of the 2003 Bam, Iran,
earthquake mapped satellite synthetic aperture radar interferometric correlation.
Journal of Geophysical Research, 110. http://dx.doi.org/10.1029/2004JB003299.
Franceschetti, G., Lanari, R., 1999. Synthetic Aperture Radar. CRC Press, New York,
NY.
Ford, P., Pettengill, G., 1992. Venus topography and kilometer-scale shapes.
Journal of Geophysical Research 97, 13103–13114.
Goldstein, R.M., 1995. Atmospheric limitations to repeat-track radar interferometry.
Geophysical Research Letters 22, 2517–2520.
Hagberg, J.O., Ulander, L.M.H., Askne, J., 1995. Repeat-pass SAR interferometry
over forested terrain. IEEE Transactions on Geoscience and Remote Sensing
33(2), 331–340.
Hensley, S., 2008. A combined methodology For SAR interferometric and
stereometric error modeling. Proceedings of Radarcon 2008.
Hensley, S., Chapin, E., Freedman, A., et al., 2001a. First P-band results using the
GeoSAR mapping system, Geoscience and Remote Sensing Symposium, IGARSS
’01. IEEE 2001 International 1, 126–128
Hensley, S., Gurrola, E., Harcke, L., et al., 2009. Lunar topographic mapping
using a new high resolution mode for the GSSR radar. Proceedings of Radarcon
2009.
Hensley, S., Klein, J., Rosen, P.A., Chapin, E., Madsen, S.N., Webb, F.H., 1995.
Repeat pass aircraft interferometry results at Portage Lake, Maine, and Innisfail,
Australia. Proceedings of the AIRSAR Science Workshop.
Hensley, S., Muniy, R., Rosen, P., 2001b. Interferometric synthetic aperture radar.
In: Maune, D. (Ed.), Digital Elevation Model Technologies and Applications: The
DEM users Manual. American Society for Photogrammetry and Remote Sensing
(ASPRS), Bethesda, Maryland, Ch. 6.
Hensley, S., Webb, F.H., 1994. Comparison of long valley TOPSAR data with
kinematic GPS measurements. IGARSS Proceedings, Pasadena, CA.
Hensley, S., Zebker, H., Jones, C., Michel, T., Muellerschoen, T., Chapman, B.,
2009. First Deformation Results Using The NASA/JPL UAVSAR Instrument. 2nd
Annual Asia Pacific SAR Conference, Xian, China, October.
Hensley, S., Zebker, H., Jones, C., Michel, T., Muellerschoen, T., Chapman, B.,
2010. Use of airborne SAR interferometry for monitoring deformation of large-
scale man-made features. Proceeding of the Large Scale Deformation of Man-
made Structures, Hong Kong.
Hess, L.L., Melack, J.M., Simonett, D.S., 1990. Radar detection of flooding beneath
the forest canopy: a review. International Journal of Remote Sensing 11,
1313–1325.
Hofton, M., Dubayah, R., Blair, J.B., Rabine, C., 2006. Validation of SRTM
elevations over vegetated and non-vegetated terrain using medium footprint lidar.
Photogrammetric Engineering and Remote Sensing 72, 279–285.
Hokeman, D., Varekamp, C., 2001. Observation of tropical rain forest trees by
airborne high-resolution interferometric radar. IEEE Transactions on Geoscience
and Remote Sensing 39(3), 584–594.
Kovaly, J., 1976. Synthetic Aperture Radar. Artech House, Boston, MA.
Leberl, F.W., 1990. Radargrammetric Image Processing. Artech House, Boston, MA.
Lee, J.S., Pottier, E., 2009. Polarimetric Radar Imaging: From Basics to
Applications. CRC Press, New York, NY.
Lee, J.S., Schuler, D.L., Ainsworth, T.L., 2000. Polarimetric SAR data compensation
for Terrain Azimuth slope variation. IEEE Transactions on Geoscience and
Remote Sensing 38(5), 2153–2163.
Lundgren, P., Lu, Z., 2006. Inflation model of Uzon caldera, Kamchatka, constrained
by satellite radar interferometry observations. Geophysical Research Letters, 33.
http://dx.doi.org/10.1029/2005GL025181.
Madsen, S.N., Martin, J.M., Zebker, H.A., 1993a. Analysis and evaluation of the
NASA/JPL TOPSAR across-track interferometric SAR system. IEEE Transactions
on Geoscience and Remote Sensing 33(2), 383–391.
78 Microwave Remote Sensing and Surface Characterization

Madsen, S.N., Zebker, H.A., 1999. Synthetic aperture radar interferometry: principles
and applications. In: Henderson, F., Lewis, A. (Eds.), Manual of Remote
Sensing. Artech House, Boston, MA, vol. 3, ch. 6.
Madsen, S.N., Zebker, H.A., Martin, J., 1993b. Topographic mapping using radar
interferometry: processing techniques. IEEE Transactions on Geoscience and
Remote Sensing 31(1), 246–256.
Massonnet, D., Feigl, K.L., 1998. Radar interferometry and its application to
changes in the Earth’s surface. Reviews of Geophysics 36, 441–500.
Massonnet, D., Rossi, M., Carmona, C., Adragna, F., Peltzer, G., Feigl, K., Rabaute,
T., 1993. The displacement field of the Landers earthquake mapped by radar
interferometry. Nature 364, 138–142.
Masterlark, T., Lu, Z., 2004. Transient volcano deformation sources imaged with
interferometric synthetic aperture radar: Application to Seguam Island, Alaska.
Journal of Geophysical Research, 109. http://dx.doi.org/10.1029/2003JB002568.
Mott, H., 2007. Remote Sensing with Polarimetric Radar. Wiley, Hoboken, NJ.
Mouginis-Mark, P.J., Crisp, J.A., Fink, J.H., 2000. Remote Sensing of Active
Volcanism. AGU Geophysics Monograph 116, 272.
Nerem, R., Leuliette, E., Cazenave, A., 2006. Present-day sea-level change: a review.
C. R. Geoscience 338, 1077–1083.
Nerem, R.S., Lerch, F.J., Marshall, J.A., et al., 1994. Gravity model development for
TOPEX/POSEIDON: joint gravity models 1 and 2. Journal of Geophysical
Research 99(C12), 24,421–24,447. http://dx.doi.org/10.1029/94JC01376.
Raney, K., 1999. Radar fundamentals: technical perspective. In: Henderson, F.,
Lewis, A. (Eds.), Manual of Remote Sensing. Artech House, Boston, MA, vol. 3,
ch. 2.
Robinson, C.A., 2002. Application of satellite radar data suggest that the Kharga
depression in south-western Egypt is a fracture rock aquifer. International
Journal of Remote Sensing 23, 4101–4113.
Rodriguez, E., Morris, C., Belz, E., 2006. A global assessment of the SRTM
performance. Photogrammetric Engineering and Remote Sensing 72(3),
249–260.
Roering, J.J., Stimely, L.L., Mackey, B.H., Schmidt, D.A., 2009. Using DInSAR,
airborne LiDAR, and archival air photos to quantify landsliding and sediment
transport. Geophysical Research Letters 36, L19402. http://dx.doi.org/10.1029/
2009GL040374.
Rosen, P.A., Hensley, S., Joughin, I.R., Li, F.K., Madsen, S., Rodriguez, E.,
Goldstein, R.M., 2000. Synthetic aperture radar interferometry. Proceedings of
the IEEE 88(3), 333–382.
Saatchi, S., Houghton, R., Avala, R., Yu, Y., Soares, J.-V., 2007. Spatial distribution
of live aboveground biomass in Amazon Basin. Global Change Biology 13,
816–837.
Schaber, G.G., McCauley, J.F., Breed, C.S., 1997. The use of multifrequency and
polarimetric SIR-C/X-SAR data in geologic studies of Bir Safsaf, Egypt. Remote
Sensing of Environment 59, 337–363.
Sexton, J.O., Bax, T., Siqueira, P., Swenson, J.J., Hensley, S., 2009. A comparison
of lidar, radar, and field measurements of canopy height in pine and hardwood
forests of southeastern North America. Forest Ecology Management 257,
1136–1147. http://dx.doi.org/10.1016/j.foreco.2008.11.022.
Simons, M., Rosen, P.A., 2007. Interferometric Synthetic Aperture Radar Geodesy,
Treatise on Geophysics.
Smith, L.C., 2009. Emerging applications of interferometric synthetic aperture radar
(InSAR) in geomorphology and hydrology. Annals of the Association of
American Geographers 92, 385–398.
Stiles, B.W., Hensley, S., Gim, T., et al., 2009. Determining Titan surface topography
from Cassini SAR data. Icarus 202(2), 584–598.
Tapley, I.J., 1996. An assessment of airborne synthetic aperture radar (AIRSAR) data
as an aid to iron exploration in the Hamersley Basin, Western Australia. CRC
LEME Restricted Report 12R, CSIRO Exploration and Mining Report 282R.
Tapley, I.J., 1998. Landform, regolith and geological mapping in the
Northeastern Goldfields region, Western Australia, and North Drummond Basin,
Queensland, using AIRSAR polarimetric data, CRC LEME/AMIRA Project 392.
CRC LEME Report 55, CSIRO Exploration and Mining Report 147R, Volumes 1
and 2.
Tapley, I.J., 2000. Geological mapping and terrain analysis in Australia using
AIRSAR polarimetric and interferometric radar data. Report 1 – The recognition
of geologic units within the Flinders and Olary Ranges, South Australia, CRC
LEME Report 154R, Volumes 1 and 2.
Tapley, I.J., Craig, M.D., 1995. An evaluation of AIRborne Synthetic Aperture Radar
(AIRSAR) for mapping surface and sub-surface structures in the Telfer region,
Paterson Province, Western Australia. CSIRO/AMIRA Project 392, CSIRO
Exploration and Mining Report 146R, Volumes 1 and 2.
Tarayre, H., Massonnet, D., 1996. Atmospheric propagation heterogeneities
revealed by ERS-1 interferometry. Geophysical Research Letters 23(9),
989–992.
Zebker, H.A., Amelung, F., 2000. Jonsson, Remote sensing of volcano surface and
internal processes using radar interferometry. AGU Geophysical Monograph 116,
179–205.
Zebker, H.A., Rosen, P.A., Hensley, S., Mouganis-Mark, P.J., 1996. Analysis of
active flows on Kilauea Volcano, Hawaii, using SIR-C radar correlation
measurements. Geology 24(6), 495–498.
Zebker, H.A., Stiles, B., Hensley, S., Lorenz, R., Kirk, R.L., Lunine, J., 2009. Size
and shape of Saturn’s moon Titan. Science 324(5929), 921–923.
Zebker, H.A., Villasenor, J., 1992. Decorrelation in interferometric radar echoes. IEEE
Transactions on Geoscience and Remote Sensing 30, 950–959.
Zebker, H.A., Werner, C.L., Rosen, P.A., Hensley, S., 1994b. Accuracy of
topographic maps derived from ERS-1 interferometric radar. IEEE Transactions
on Geoscience and Remote Sensing 32, 823–836.
Zebker, H.A., Wye, L.C., Janssen, M.A., Cassini Radar Team, 2008. Titan’s surface
from reconciled Cassini microwave reflectivity and emissivity observations.
Icarus 194(2), 704–710.

Biographical Sketch

Scott Hensley received his BS degrees in mathematics and physics from the University of California at Irvine and a
PhD in mathematics from Stony Brook University, where he specialized in the study of differential geometry. In
1992, Dr. Hensley joined the staff of the Jet Propulsion Laboratory where he studied advanced radar techniques
for geophysical applications. He has worked on most of the SAR systems developed at JPL over the past two
decades, including the Magellan and Cassini radars. He was the GeoSAR chief scientist, a simultaneous X-band
and P-band air-borne radar interferometer for mapping above- and beneath-the-vegetation canopies that is now
commercially operated by Earthdata International. He led the SRTM Interferometric Processor Development Team
for a shuttle-based interferometric radar used to map the Earth’s topography between 7601 latitude. Recently, he
began working with the Earth-based Goldstone Solar System Radar to generate topographic maps of the lunar
surface. He was principal investigator and is currently the chief scientist for the NASA UAVSAR program that uses
an electronically scanned active-array L-band fully polarimetric-designed radar for repeat-pass applications.
 Microwave Remote Sensing and Surface Characterization 79

Tom Farr received BS and MS degrees from Caltech, and a PhD from the University of Washington, all in Geology.
After a short time as an engineering geologist, he joined the Radar Sciences Group at the Jet Propulsion La-
boratory, where he has been since 1975. At JPL, he helped develop the first geologic applications of imaging radar
using aircraft, satellites, and the Space Shuttle. He was the deputy project scientist on the Shuttle Radar Topography
Mission, which used interferometric radar to produce a near-global map of the Earth. He has also been a science
investigator on European and Japanese satellite programs, and has assisted in the interpretation of radar images
from Venus and recently from Saturn’s moon Titan. His scientific research includes the use of remote sensing and
digital topographic data for study of landscapes on Earth and other planets and how they are formed and
modified by climate and tectonic or volcanic activity.
 3.4 Remote Sensing of Land Cover Dynamics
TR Allen, Y Wang, and TW Crawford, East Carolina University, Greenville, NC, USA
r 2013 Elsevier Inc. All rights reserved.

3.4.1 Introduction 80
3.4.2 Remote Sensing of Land Cover 81
3.4.2.1 Discrete Information and Hard Classification 81
3.4.2.2 Landscape Metrics and Biophysical Change 82
3.4.2.3 Soft and Fuzzy Classification 82
3.4.2.4 Change Detection 82
3.4.2.5 Spatial and Temporal Scale 83
3.4.2.6 Surface Dynamics and Land Cover 83
3.4.2.7 Impervious Surfaces and Runoff 84
3.4.3 Case Studies 84
3.4.3.1 Coastal Impervious Runoff 84
3.4.3.2 Shoreline Change Analysis using a Multitemporal Radar 88
3.4.3.3 Dynamic Feature Extraction: Riverine Flood Inundation 88
3.4.4 Land-Cover Change Modeling 91
3.4.4.1 Emergence and Milestones 91
3.4.4.2 Proximate and Ultimate Drivers of Land Change 95
3.4.4.3 Modeling Approaches 96
3.4.4.3.1 Spatial modeling approaches 96
3.4.4.3.1.1 Multicriteria evaluation 97
3.4.4.3.1.2 Statistical regression 97
3.4.4.3.1.3 Dynamic spatial simulation 98
3.4.5 Future Research Directions 98
References 99

Abstract

Our understanding of land-cover dynamics and geomorphology has advanced with the development of remote-sensing
science and technology. Remote sensing provides data that drive biophysical models of Earth-surface processes, as well as
mapping discrete land cover and land-use change information and landform maps. Land cover remote sensing has also
spurred land-change science, an emergent paradigm linking anthropogenic and natural-system change dynamics using
predictive models. This chapter reviews the state of remote sensing applied to land-cover dynamics, provides selected case
studies, and surveys future challenges and advances in light of increasing anthropogenic changes on the landscape.

3.4.1 Introduction environments including buildings, transportation, and other

The significance of land-cover change to the Earth’s surface,
climate, and life-sustaining systems cannot be understated as a
concern confronting global civilization. Past and ongoing an-
thropogenic and natural-system changes have wrought grand
challenges to ecological integrity and global biogeochemical
cycles. Fundamental biotic and abiotic processes linking the
Earth’s surface to climate change are being studied for their
potential to contribute feedbacks to greenhouse warming,
ocean acidification, and climatic variability and extremes. Many
natural environments have been converted from natural vege-
tation cover into cropland or altered by construction of built
Allen, T.R., Wang, Y., Crawford, T.W., 2013. Remote sensing of land cover
dynamics. In: Shroder, J. (Editor in Chief), Bishop, M.P. (Ed.), Treatise on
Geomorphology. Academic Press, San Diego, CA, vol. 3, Remote Sensing
and GIScience in Geomorphology, pp. 80–102.
impervious surfaces. The resulting microclimates, soil–atmo-
spheric interactions, and hydrologic changes now manifest
various geomorphic state and rate changes including erosion,
transportation, and deposition. Less visible, but significant
changes in groundwater hydrology and water quality are ex-
hibited in water-resource developments (dams and reservoirs,
aqueducts, and altered hydro-regimes), extensive soil drainage,
and even residential septic systems and urban storm-water
management. In sum, anthropogenic influences are profoundly
evident on the land cover of the Earth. Geomorphology and
Earth-surface processes, including biotic, soil, and hydrologic
processes, must be factored into understanding of the inter-
actions of human beings and the natural environment. Scien-
tific analyses of these factors have increasingly sought to apply
remote sensing and land-cover change to understand anthro-
pogenic impacts and effects.
Applying systems approaches to investigation of land-
cover dynamics requires information on surface processes,

80 Treatise on Geomorphology, Volume 3 http://dx.doi.org/10.1016/B978-0-12-374739-6.00046-4


landforms, and land use/land cover. The purpose of this
chapter is to review the relevance, limitations, and oppor-
tunities of remote sensing of land-cover dynamics for geo-
morphology, including recent methodological developments
and challenges. Using examples drawn from studies of land-
cover change and dynamic surface processes, we illustrate
research that captures the state of the art and highlights the
near-term challenges and potential, as new remote-sensing
technologies emerge and our cumulative understanding of
land-cover dynamics increases.
3.4.2 Remote Sensing of Land Cover
Remote-sensing science and technology is very effective for the
production of spatially continuous biophysical and thematic
information that is valuable for geomorphology. Remote
sensing provides for primary data acquisition of continuous
phenomena (e.g., albedo, elevation, surface temperature) as
well as derivative discrete information (e.g., land cover, vege-
tation, or landforms) using space-borne or aerial platforms.
Despite the wide variety of observations and derivative infor-
mation, thematic land-cover products vary in detail depending
on four key sensor characteristics including the spatial extent
(image swath and length dimensions) and spatial resolution
(pixel size or field-of-view distance), radiometric resolution
(recording range and sampling of reflectance values), spectral
resolution (number and width of reflectance wavebands in
the electromagnetic spectrum), and temporal resolution (fre-
quency of observation). Land-cover information is primarily
derived from satellite-based, passive sensors that measure re-
flected solar illumination from the visible to infrared regions
of the spectrum (e.g., Landsat, Ikonos, Quickbird, WorldView
2, moderate resolution imaging spectrometer (MODIS),
ASTER, and SPOT). Land-cover maps are now produced using
image classification methods (i.e., converting spectral pixel
data into discrete information classes), with robust statistics
available for estimating classification accuracy and error.
Nonetheless, there are trends in the production of selected,
fine-scale land-cover products, such as impervious surfaces,
building and infrastructure, shorelines and coastal landforms,
and applications utilizing ultrahigh spatial resolution satellite
or airborne sensors. The latter products, however, remain
costly, less common, and in the domain of isolated projects or
relatively high-value areas or features (e.g., valuable shoreline
erosion monitoring or high-intensity urban development).
Sensors with a spatial resolution of tens of meters, multi-
spectral imaging (4–6 spectral bands, primarily in the visible
and near-infrared regions), temporal resolution of 5–16 days
between nadir-viewing, and radiometric resolution of 8–10
bits produce imagery that remain the most prevalent data
sources for the production of land-cover information. These
data products offer an attractive combination of regional ex-
tent, low cost, and relatively frequent observations. Coupled
with the availability of robust, inexpensive, and proven image
analysis and classification software, satellite-derived land-
cover information can be routinely generated for geo-
morphological studies. Typical remotely sensed products for
analysis of land-cover dynamics are discrete land-cover maps
(usually having 4–20 thematic classes, but up to B30) or sets
Remote Sensing of Land Cover Dynamics 81
of derived multispectral indices (linearly transformed spectral
data) representing various biophysical characteristics of the
land surface that are useful in monitoring efforts.
Land cover is now well understood to affect many en-
vironmental processes. The linkages between land cover and
environmental processes are superbly exhibited in coastal
research linking watersheds and receiving waters. Coastal eu-
trophication and aquatic habitat degradation, for instance, are
commonly attributed to nonpoint source nutrient pollution.
Satellite remote sensing of land cover provides for routine
characterization for coastal land cover and continues to in-
form coastal resource management (Thomas, 1995). The Na-
tional Oceanic and Atmospheric Administration (NOAA)
Coastal Change Analysis Program (CCAP), for example, uses
Landsat Thematic Mapper (TM) and Enhanced Thematic
Mapper (ETM þ ) satellite data to map coastal land-cover
change (Dobson and Bright, 1991; Jensen et al., 1993a).
Klemas et al. (1993) describe how the classification scheme of
CCAP distinguishes ecologically important features such as
different coastal wetland types. The CCAP program, however,
does not provide a historical perspective based on the use of
the Landsat multispectral scanner (MSS) image archive or
presatellite historical trends. Archived Landsat MSS data have
been analyzed in conjunction with new, higher resolution
sensors (cf., Lee and Marsh, 1995; Jensen et al., 1993b, 1995).
Although the MSS data do not provide highly detailed classi-
fications, the North American Landscape Characterization
(NALC), an archival project of the U.S. Geological Survey
(Lunetta et al., 1998), provides MSS data that are cloud free
and of high radiometric quality. In conjunction with other
Landsat MSS archives, Landsat TM data and robust historical
analyses of land-cover change have been performed. In a study
of urbanization (1972–82) in Montreal, Charbonneau et al.
(1993) found that automated image classification of a Landsat
MSS time series was 5% to 30% more accurate than traditional
photo-interpretation. Where historical dynamics are rapid and
ground verification is weak, such as dynamic coastal en-
vironments, satellite time series are ever more important. Shao
et al. (1998), for example, demonstrated integration of geos-
patial techniques for assessing shrub thicket distributions in
naturally dynamic barrier islands of the Virginia coast. Clearly,
the integration of geospatial information technology (remote
sensing, GIS, and spatial analysis) and landscape ecology is
still maturing. To more fully understand landscape change,
empirical studies and simulation methods offer a first step to
benchmarking the utility and accuracy of using spatial-analysis
tools for information extraction. Given the geomorphic sur-
face-process dynamism, or degree of permanence of a land-
form, remote sensing offers numerous image-classification
options that can be matched to the target characteristics.
3.4.2.1 Discrete Information and Hard Classification
Thematic land-cover maps are a staple remote-sensing product
relevant to geomorphology. A handful of national and global
land-cover classification programs are representative of a wide
array of products that typically include a thematic land-cover
classification derived from the hierarchical scheme of Anderson
et al. (1976) and provide a standardized system for a wide
range of uses. This seminal system has subsequently evolved
82 Remote Sensing of Land Cover Dynamics
into the Multi-Resolution Landscape Characterization (MRLC)
Consortium formed by the U.S. Geological Survey, Environ-
mental Protection Agency (EPA), NOAA, and U.S. Forest Service
(USFS). The Global Land Cover Characteristics Database
(GLCC) (Loveland et al., 2000) provides a 1 km spatial reso-
lution dataset derived from Advanced Very High Resolution
Radiometer (AVHRR) satellite data over April 1992 to March
1993, with seasonal land-cover classes compiled regionally.
A parallel, national land-cover grid for the U.S. was similarly
compiled (Loveland et al., 1991). In the European Union, the
CO-ordination of Information on the Environment (CORINE)
program has also produced regional NOAA AVHRR-derived
land-cover data. Applications of CORINE data have been wide-
ranging for land cover and agriculture, including the combined
use of thematic cover types and continuous Normalized Dif-
ference Vegetation Index (NDVI) time series (Genovese et al.,
2001). In the U.S., efforts to characterize state and regional
land-cover trends for the conservation and preservation of
species and community biodiversity prompted a wide primarily
state-level land-cover program for GAP Analysis Projects in the
1990s (Scott et al., 1996). Other regional land-cover mapping
projects have been implemented by numerous groups. The
Nature Conservancy (TNC) developed a highly floristic vege-
tation classification system (Grossman et al., 1998). The Albe-
marle-Pamlico estuarine study also developed a land-use/land-
cover map from Landsat data as a prelude to development of a
National Estuary Program site for the U.S. Environmental
Protection Agency and State of North Carolina (Albemarle-
Pamlico Estuary Study (APES), 1994). In the intervening years
since land-cover data have been created, limitations have also
been revealed with so-called hard classification systems, in-
cluding the necessity of dealing with mixture classes and in-
herently ‘soft’ information classes such as transitional
vegetation or ephemeral land use and cover types.
3.4.2.2 Landscape Metrics and Biophysical Change
Thematic land-cover data are highly useful for the landscape
study of ecological and geomorphological relationships (e.g.,
ecogeomorphology or biogeomorphology). Spatial pattern
metrics of the distribution of land cover or habitats have been
increasingly applied for landscape ecological or geobotanical
studies. Numerous indices of landscape spatial pattern have
been the subject of research on land-cover dynamics in land-
scape ecology (e.g., McGarigal and Marks, 1993; Frohn, 1997;
Gustafson, 1998). These indices primarily measure com-
positional and spatial patterns of thematic land cover or eco-
logical classes. Predictor variables drawn from terrain (e.g.,
digital elevation models) and biophysical remote-sensing in-
dices such as the NDVI are commonly applied to predict a
particular environmental resource or diagnostic indicator of
environmental change, such as alpine treeline. A wide diversity
and redundancy among the metrics, however, complicates the
task of deriving meaning from quantitative measures of land-
scape patterns. Further, several metrics are scale-dependent;
hence, the application of semi-variograms to assess scale de-
pendencies of vegetation pattern may be warranted, as illus-
trated by Bian and Walsh (1993). Yet, scale dependency has
only seldom been compared with measures of spatial pattern
and scale using thematic land cover and landscape pattern
metrics (cf., Gustafson, 1998; Uemaa et al., 2005).
On the coast, remote sensing may identify trends and
patterns of urbanization, agricultural intensification, wetland
loss, or estuarine sedimentation useful to environmental
managers. Among possible change-analysis tools, Markov
models have been used to assess temporal dynamics in eco-
systems (Hobbs, 1994) and simulate historical land-cover
changes (Turner, 1988.) They may also be directly derived
from remotely sensed change maps using GIS (Allen and
Walsh, 1993; Li and Reynolds, 1997; Savitzky et al., 1998).
Forest fragmentation, for example, can be characterized by
increasing urbanization or agricultural expansion eroding
forest boundaries and increasing patch isolation. Forest
patch and urbanization patterns have been evaluated in the
northeastern U.S. using landscape-pattern analysis along a
spatial-urbanization gradient (Medley et al., 1995). In sum,
spatial-pattern metrics of remotely sensed land cover have
become widely used for inferring and predicting surface
process influences on habitats and land-cover change.
3.4.2.3 Soft and Fuzzy Classification
Although spatial metrics most commonly exploit pattern an-
alysis of discrete land-cover classification, many instances
occur in geomorphology and landscape ecology where the
phenomena are continuously distributed or exhibit an in-
herent uncertainty as to thematic classification. Alternatives
have emerged to hard classification schemes with the advent
of techniques such as fuzzy logic, which allow a measure of
pixel similarity (or class membership probability) to be
ascribed to every class (Mather, 1999). For a landscape
undergoing some continuous change, process, or disturbance,
the assignable information class may not be appropriate for a
hard classification, and no objects can be identified or
boundaries delineated. Solutions such as partial membership
or ‘fuzzy’ classification may be more appropriate in these
cases, where fuzziness may be ascribed to limitations of the
resolution of landscape observation or to mixtures of various
spectral ‘end-members’ in a pixel. In such situations, pixels can
be decomposed using methods of spectral unmixing. Some
landscapes may lend themselves to discrete hard classification,
whereas other landscapes (or landscapes observed at different
scales) require a soft approach using linear spectral unmixing,
spectral-angle mapping, or reference vectors, which describe
the boundaries between classes in spectral-feature space
(Foody and Mathur, 2004).
Geomorphic examples of fuzzy classification techniques
are most prevalent for inherently continuous or gradational
features and processes. Fuzzy classification has been applied,
for instance, to soil and landform classification (Brabyn, 1997;
Irvin et al., 1997), soil patterns (Odeh et al., 1992), and
geomorphic terrain position (Dragut and Blaschke, 2006;
Arrell et al., 2007; Schmidt and Hewitt, 2004).
3.4.2.4 Change Detection
Oftentimes, the focus of geomorphologic research is the dy-
namic change or trend rather than the analysis of a static

distribution or relationships. In such research, remote-sensing
change detection offers an array of potential specialized
techniques. In addition, change detection may apply discrete
or continuous approaches to surface dynamics to incorporate
temporal complexity. Change detection can be conducted for
thematic or soft classification, land-cover data as well as more
specialized, hybrid methods. Technical advances in change-
detection algorithms and sensor and platform design are re-
ceiving major attention with the advent of global environ-
mental change, environmental monitoring and assessment,
and operational needs for natural and technological disasters.
The variety of tools for change detection range from relatively
visual and qualitative interpretation techniques such as
straight-forward, multitemporal composites (e.g., write-func-
tion memory displays) and multitemporal principal com-
ponents analysis (PCA), raster GIS and map algebraic,
postclassification change detection, to abstract spatiotemporal
change vectors for both high spatial and temporal resolution
imagery. Intermediate techniques include band ratioing and
composite indices (e.g., time-series NDVI). The subject of
change detection in remote-sensing research is a major topic.
Among several developments in the change-detection research
arena, ‘phenology,’ the periodic and seasonal life cycle of
plants and animals, imparts temporal variation between re-
peat image acquisitions, particularly on an interannual basis.
Although this poses a challenge to the tracking and study of
trends in hard classifications (Moody and Strahler, 1994),
seasonal phenology such as leaf-out, green-up, and senescence
also provides for potential improvement of forest-cover clas-
sification (Dymond et al., 2002) or seasonal vegetation vari-
ability (Reed et al., 1994). Such temporal spectral changes are
also the foundation of widely used indices, such as the Tas-
seled Cap for Landsat-derived brightness, greenness, and
wetness (Landsat TM/ETM þ ) (Crist and Ciccone, 1984).
Sensor temporal frequency of coverage has also been evalu-
ated for its potential impact on change detection. Lunetta et al.
(2004) concluded that a minimum of 3–4 years of temporal
data-acquisition frequency is necessary to monitor land-cover
change in the northern Piedmont of North Carolina. Another
technique originating in the application of geometric statistics
using multispectral imagery, change-vector analysis (CVA), has
been applied to continuously changing phenomena, ranging
from forest dieback and regeneration (Allen and Kupfer, 2000)
to urbanization (Chen et al., 2003). Emerging platforms with
more flexible orbital-path controls and off-nadir viewing an-
gles also provide for greater temporal frequency of coverage
for monitoring change.
3.4.2.5 Spatial and Temporal Scale
The components of land-cover change necessary for the de-
termination of land-use dynamics are complex. The first dy-
namic, typical of traditional change detection by map overlay
tabulation, measures aspatial compositional change by quanti-
fying areal amounts and relative proportions of cover types,
without regard to spatial configuration or pattern. Charbon-
neau et al. (1993) reported primarily on the areal growth of
urban land cover in Montreal (65% increase in 1972–82).
Second, land-cover changes exhibit explicit spatial patterns
Remote Sensing of Land Cover Dynamics 83
such as the number, density, and complexity of patches,
corridors, or mosaic patterns. These are measurable charac-
teristics that may be obtained using a variety of GIS and
pattern statistics software packages (cf., Turner et al., 1991;
Baker and Cai, 1992; McGarigal and Marks, 1993). Because
transition probabilities, such as Markov models, are based on
temporal changes and do not include spatial information
(Barringer and Robinson, 1981), spatial-scale dependence
must be addressed separately. Thus, the third dynamic pertains
to the scale of variation, or scale dependence, of landscape
elements such as grain-size effects on habitat, ecologic trans-
port and flows, and spatial interactions (Forman, 1997). An-
alysis of semivariograms could indicate the presence of spatial
trends in grain size and test for an ‘ergodic’ trend, wherein
spatial patterns across a geographic region (e.g., urbanization
front) provide a surrogate or a substitution for temporal
evolution of landscape structure. The assumption or explor-
ation of ergodicity is a fundamental concept of geomorph-
ology, tracing its lineage as early as William Morris Davis’ Cycle
of Erosion. As a more recent example, ‘space-for-time substi-
tution’ has been used to conceptually model the transgression
of coastal wetlands in response to climate change and sea-level
rise (Michener et al., 1997). In space–time dimensions, a
landscape may evolve dynamically through various stages of
spatial, compositional, and scale-dependent structure. Meas-
urement and modeling of the scale dependence may use semi-
variogram analysis of either thematic land-cover data or con-
tinuous biophysical information such as the NDVI (Frank,
1988). For example, toward the rural southern terminus of the
Delmarva Peninsula, one finds a persisting agricultural land-
scape, whereas northward, landscape structure reflects in-
creasing urbanization and forest fragmentation that should be
evident in NDVI semi-variograms.
3.4.2.6 Surface Dynamics and Land Cover
Numerous dynamic geomorphological processes and features
are investigated using remote sensing, particularly erosion,
deposition, flooding, and natural hazards. Coastal geo-
morphology and natural hazards, for instance, offer relatively
abundant examples of remote sensing and geomorphology.
Landsat imagery is commonly used for the extraction of land-
cover data for modeling landslide susceptibility (e.g., Choi
et al., 2010; Pande et al., 2009) and mapping groundwater
recharge zones for aquifers or karst terrain (Nagarajan and
Singh, 2009). Dynamic fluvial landforms and water levels
are also resolvable using airborne light detection and ranging
(LiDAR) and analysis of signal intensity (Hofle et al., 2009).
LiDAR data are classified into floodplain, channel, and
flood-risk zones specific to emergency management, such as
Federal Emergency Management Agency (FEMA) flood-hazard
risk maps. Flood dynamics and vegetation distribution
have been studied using the cloud-penetrating capability of
synthetic aperture radar (SAR) satellite data (e.g., Martinez and
Le Toan, 2007). Regional climates exhibiting prolonged sea-
sonal cloud cover continue to pose challenges to remote
sensing, particularly in tropical and seasonal wet–dry and
monsoonal climates. Commonly, a multisensor approach
incorporating aerial as well as satellite data is necessary
84 Remote Sensing of Land Cover Dynamics
(e.g., tropical dry deforestation; Galicia et al., 2008). Visual
cues from direct observation of surface conditions are also
feasible for Aeolian geomorphologic modeling (cf., Webb
et al., 2009).
3.4.2.7 Impervious Surfaces and Runoff
Geospatial technologies are accurate and efficient means for
characterizing urban features, impervious surfaces, and pro-
viding a variety of useful information for storm-water man-
agement and watershed planning (Brown, 1997; Brabec et al.,
2002). The advent of high-resolution satellite imagery and the
increasing environmental problems associated with impervi-
ousness have attracted the attention of remote-sensing re-
searchers for improved characterization and quantification of
impervious surfaces (Weng, 2008). Impervious surfaces affect
energy balance, habitat fragmentation, and especially hydrol-
ogy (Goetz et al., 2003). Radar imagery has been found to be a
successful tool for characterizing urban features, topography,
and impervious surfaces by several studies (Hepner et al.,
1998) and holds promise for subsurface hydrologic mapping
(Smith, 2002; Becker, 2006). Coastal imperviousness has been
mapped using LiDAR and aerial photography (Hodgson et al.,
2003), as well as IKONOS multispectral data (Goetz et al.,
2003; Sawaya et al., 2003). Urban land-cover and land-use
classification has been historically challenging, but impervi-
ousness mapping in complex urban areas is a priority (Ridd,
1995; Bauer et al., 2002). The literature points to the need for
advancing various techniques using fine-resolution grid-based
models.
A wide variety of spatial rainfall-runoff models have
evolved with the advent of widely available digital data such
as land cover, digital elevation models (DEMs), soils, and
gridded rainfall products from radar or interpolated me-
teorological-station precipitation. A well-known suite of
TOPographically based hydrologic MODels has been derived
from ‘TOPMODEL’ (Beven, 1997). The Terrain Analysis Using
Digital Elevation Models (TAUDEM; Tarboton, 1997) has also
been widely applied and improved, such as characterizing
drying in catchments or transport-limited accumulation.
Another example includes the Depth to water, Recharge
rate, Aquifer characteristics, Soil media, Topography, Impact
of vadose zone and hydraulic Conductivity (DRASTIC)
model for predicting vulnerability of groundwater pollution.
DRASTIC has been ported to GIS to build an index value
for the potential of groundwater pollution from surface con-
tamination. The acronym derives from seven input layers
representing DRASTIC. Similar to the AGNPS model (Agri-
cultural Nonpoint Source Pollution) approach, this approach
incorporates topography, vegetation, and soil characteristics
to evaluate field- to watershed-scale runoff characteristics
(Choi and Blood, 1999). The model is demonstrative of a
software tool that has been widely accepted and applied for
aquifer vulnerability. Several other models are currently pro-
moted by the NOAA Coastal Services Center (NOAA
CSC, 2010) and germane to any coastal toolset, such as
the Impervious Surface Analysis Tool (ISAT). Furthermore,
the Nonpoint Source Pollution and Erosion Comparison Tool
(N-SPECT) has also been developed at NOAA CSC for
predicting nonpoint source pollution utilizing integrated
DEMs, hydrology, land cover, and soil-runoff parameters.
Such models are approaching the local scale, but still remain
coarse and more appropriate to regional planning that inte-
grates multiple thematic-surface features and addresses both
groundwater and surface-water interactions. These techniques
have evolved during the prior era of predominantly medium-
resolution land-cover data (e.g., Landsat). The N-SPECT and
ISAT software, for instance, principally use the NOAA CCAP
land-use/land-cover data. The application of newly available
LiDAR and ultrahigh resolution satellite data provides chal-
lenges and opportunities for the further advancement of these
techniques.
3.4.3 Case Studies
3.4.3.1 Coastal Impervious Runoff
Storm-water runoff arising from impervious surfaces signifi-
cantly degrades environmental quality, alters stream hydrol-
ogy and habitats, poses hazards to human health, often
reduces esthetics, and commits a community to capital in-
vestment and maintenance. Waters receiving storm-water
runoff accumulate numerous pollutants, including nitrogen,
phosphorus, pesticides and herbicides, oil, lead, grease,
copper, mercury, and pathogens (bacteria and viruses.) Non-
point-source pollution of surface waters is also affected by
urbanization and is expected to increase unless runoff can be
decreased by diverse methods (Carpenter et al., 1998). Con-
comitant to expanded estuarine and water-quality monitoring,
North Carolina is among several states now widely utilizing
high-resolution LiDAR laser altimetry that has been collected
in partnership with the FEMA for floodplain mapping. With
spatial resolution finer than 5 m, LiDAR-derived bare-surface
elevation grids provide for improved broad area DEMs for
floodplain delineation and environmental modeling. Inte-
grating these data into a processing workflow provides for
drainage basin delineation and calculation of useful deriva-
tives, including slope, flow direction, and flow accumulation
(e.g., ESRI Hydrologic Analyst tool and TauDEM extensions
for ArcGIS; Tarboton, 2004). These products can be further
augmented with land-cover data and associated runoff and
water-quality models. GIS-based spatial models have demon-
strated their ability to provide for accurate runoff, ground-
water infiltration, and vulnerability analysis (e.g., GIS-based
DRASTIC groundwater model; Babiker et al., 2005), and such
spatial data are accepted for storm-water permit applications
in model ordinances. In addition to floodplain delineation
(e.g., 100-year recurrence interval flood zones), these data are
used for other hydrologic modeling applications to urban
runoff, water quality and conservation planning, and storm-
surge inundation.
A specific coastal example is drawn from a study seeking
to determine the causes of isolated residential street flooding
and to prescribe solutions for storm-water management
and coastal-water protection in the Town of Southern Shores,
located on a barrier island of the northern Outer Banks,
North Carolina. Despite deep, sandy soils with high permea-
bility and infiltration capacity, extensive urbanization resulting

in impervious roofs, driveways, and streets lead to a potential
exceedance of soil infiltration by moderate rainfall. Subsurface
peat deposits, primarily in interdune swales, were also posited
to cause local ponding and perched water tables. Flood-prone
areas (characteristically with connective impervious surfaces)
were mapped by residents and town personnel. The distri-
bution of contributing subwatersheds and underlying peat
were largely unknown, except for field topographic and geo-
morphic interpretation. Resistivity mapping and reconnais-
sance investigation with soil-moisture meters, penetrometers,
and augering qualitatively confirmed the interpreted distri-
bution of peat in swales, but extensive mapping was required
for subwatershed delineation, impervious cover mapping, and
analysis of flow accumulation.
The study design provided for a series of remote sensing,
hydrologic GIS modeling, and impervious surface analysis
Figure 1 Raster flow-accumulation grid derived from bare-earth LiDAR DEMs with superimposed subwatershed delineation and chronic flood-
prone sites in a coastal barrier-island community, Town of Southern Shores, Dare County, Outer Banks, North Carolina.
Remote Sensing of Land Cover Dynamics 85
within developed subwatershed units. First, available LiDAR
DEMs from the NC Floodplain Mapping Program (NCFMP)
were acquired and merged. These bare-earth DEMs (50 ft
resolution with ‘hydro-correction’) were analyzed using Tau-
DEM, including verifying the absence of sinks and discovering
disconnected topographic lows. Flow accumulation was cal-
culated using the D8 flow-direction algorithm. Figure 1 shows
the distribution of flow accumulation (cell counts) and de-
rived subwatersheds. Flood-prone areas and pixel centroids
(points) are also illustrated, highlighting the strong coinci-
dence of flood-prone areas and topographic flow accumu-
lation for several sites. To characterize impervious cover extent
and runoff contribution within subwatersheds, the NOAA
ISAT tool was applied using 2001 CCAP land cover for the
subwatershed using North Carolina imperviousness-land-
cover coefficients. Figure 2 shows the CCAP land cover and
Flooding point
Flood-prone areas
Drainage basins
Flow accumulation
High: 12071
Low: 180
0 100 200 400 m
86 Remote Sensing of Land Cover Dynamics

Problems
Percent impervious
Problems 0.0−2.2
Percent ISAT 2.3−3.7
< 10% 3.8−7.6
10−25% 7.7−16.2
> 25% 16.3−30.2

(a) (b) (c)

Figure 2 Comparison of (a) Coastal Change Analysis Program (CCAP) land cover and (b) derived ISAT impervious classes and (c) coastal
imperviousness using high-resolution LiDAR DEMs and derived subwatersheds with very high-resolution IKONOS and municipal jurisdiction
impervious data.

two cartographic classifications of the developed impervious-

ness percentages. Figure 2(b) depicts the default manual
classification intervals, with relatively modest impervious
cover over the small basins, and only six basins with the
highest class of imperviousness (425%); yet, none of these
exhibit runoff-induced flooding problem sites. Out of 31
flood-prone sites, ten occur within the subwatersheds with
moderate impervious cover (10–25%), whereas the remainder
are situated in basins with relatively low impervious cover
(o10%). A quantile classification of basin imperviousness
shown in Figure 2(c), however, indicates that two-thirds of
the flood-prone sites are found within basins with impervious
cover 43.8%. Despite having a relatively low impervious
cover, some watersheds are prone to urban flooding, and yet,
higher spatial-resolution data are needed to infer localized
effects arising from imperviousness or the overriding effects of
Figure 3 Coastal impervious surface cover extracted from a hybrid
shallow peat deposits.
classification of IKONOS imagery including vegetation index
In order to identify finer-scale impervious cover linked
thresholding and ISODATA unsupervised classification.
to flooding, IKONOS satellite imagery and town cadastral
data were acquired. GeoEye’s IKONOSTM 1-m panchromatic
and 4-m multispectral imagery were georeferenced and provided a high-accuracy delineation of impervious cover,
merged for a composite 1-m resolution pan-sharpened image, illustrated in a sample of the town extent in Figure 3. A
and analyzed using TRIMBLE eCognitionTM object-based notable improvement over prior analyses using only per-
image-classification software. The context-based algorithm pixel maximum-likelihood classification, the distribution and

pattern of impervious classes clearly avoids commission errors
over extensive bare-sand areas, despite the sometimes strong
spectral similarity to impervious areas. By mapping the flood-
prone areas and point-of-maximum flow accumulation within
these zones (Figure 4), it is also possible to further infer
the effective imperviousness of the upslope accumulation
areas (shaded salmon color). Figure 4 reveals that the
flood-prone sites do not always arise from extensive sub-
watersheds. Further, when a random sample of sites in basins
NOT prone to flooding is drawn and compared, the tally of
total impervious area of the smaller, upslope contributing area
is able to predict the location of flood-prone sites with 67%
accuracy using discriminant analysis. As the town of Southern
Shores grapples with flood hazards and their mitigation, they
are considering the implementation of storm-water manage-
ment rules and new collection fees to raise capital for costly
Figure 4 Flood-prone and stratified-random sample site locations and flow-accumulation areas.
Remote Sensing of Land Cover Dynamics 87
engineering solutions. Figure 5, however, illustrates the crux
of the dilemma and a possible geomorphically informed
alternative. Figure 5 shows a chronic flooding site, a LiDAR-
derived subwatershed, and a high-resolution digital ortho-
photo. The small subwatershed consists of two home sites
affected by flooding and vehicular inaccessibility in a cul-
de-sac; yet, the subwatershed shows three additional neigh-
boring parcels that likely contribute runoff. The fine-scale
subwatershed reveals the high imperviousness of this site’s
upslope area, despite the relatively moderate parcel impervi-
ous cover. Future research in geospatial applications to sus-
tainable design and planning offers the possibility of more
efficient applications, including spatially targeted storm-water
best management practices (BMPs), rainwater collection cis-
terns, pervious pavers, and the use of remote sensing and GIS
to target subwatersheds for retrofit solutions.
Flood-prone and sample sites,
contributing areas, and sub-basins
Sample
Flood-prone
Sub-basin
Elevation (m)
NAVD 1988
High : 20.0
Low : 0.0
88 Remote Sensing of Land Cover Dynamics
Figure 5 Flood-prone swale site with highly connected, multiple
parcel, high-connectivity impervious surface area and low-infiltration
soils with subsurface peat. The flood-prone cul-de-sac is delineated
with hachures, flow accumulation low point, and DEM-derived
subwatershed transparently superimposed on a 1:100 color digital
orthophoto.
3.4.3.2 Shoreline Change Analysis using a Multitemporal
Radar
Shorelines and coastal ecosystems exhibit dynamics that are of
high interest to society and geomorphologic research. Remote
sensing of coastal areas provides synoptic observations and
temporal monitoring capabilities. Cloud cover, however,
hampers optical and multispectral coastal remote sensing in
several ways, such as by limiting the observation of shore
features at optimal repeating times (e.g., low tides), acquiring
images during urgent situations such as immediately pre- and
postdisaster (e.g., hurricanes), or repeating observations on
annual calendar dates for environmental monitoring. Radar
sensors, however, utilize microwave radiation emission and
backscatter to actively image the landscape and have ‘all-
weather’ imaging capabilities that overcome cloud-cover con-
straints. For example, Wang and Allen (2008) developed an
edge-extraction filter model of shoreline delineation and
quantified shoreline changes using multidate SAR imagery
over a 13-year period. On the Pamlico Peninsula of North
Carolina, shoreline changes vary from statistically none, to a
maximum landward migration rate of B11 m year1 between
December 1994 and December 2006. Subsequently, Wang
et al. (2009) reported almost no shoreline changes between
December 2006 and June 2008 for their study sites using
Advanced Land Observation Satellite/Phrased Array Type L-
Band SAR (ALOS/PALSAR) data. One of the revealing insights
is that multisensor imagery can be used effectively to assess
short- and long-term changes. In the case of relatively short
time-periods to capture relative erosion, deposition, or
shoreline stability, SAR has the advantage of being unaffected
by cloudiness, providing for a rich source of observations
compared with patchy or wholly cloud-obscured optical
imagery.
The SAR data in the Wang et al. (2009) study were multi-
look (ML), geo-referenced, amplitude-only imagery. The
pixel size was 12.5 12.5 m for 1994 JERS–1 SAR data,
6.25  6.25 m for the fine-beam single-polarization mode of
PALSAR data, or 12.5 12.5 m for the fine-beam double-po-
larization mode of PALSAR data. The multilook process re-
duces speckle of SAR imagery and simplifies the use of the
data, but unfortunately, the multilook fusion procedure de-
grades the spatial resolution and removes the phase infor-
mation, compared with single-look complex (SLC) data, a
more standard and widely available SAR data processing
product. For instance, the ML and SLC PALSAR are defined as
level 1.5 and level 1.1, respectively. The PALSAR data at level
1.5 can be used to delineate large changes, such as dozens of
pixels. Fine-scale changes, however, within 71 pixels, cannot
be discerned statistically with confidence. Finally, three SLC
data sets with near anniversary dates (24 September 2007, 26
September 2008, and 14 August 2009) were available (the
anniversary dates minimize the impact of seasonal variation of
individual land-use and land-cover (LULC) types on the co-
herence of data pairs. The coherence is one of the parameters
used in geo-referencing the pairs.). Therefore, one can reliably
use the SLC datasets to delineate changes in the study area
between a relatively short time span of 2007 and 2009.
After reanalyzing the ML and SLC SAR data between 2007
and 2009, we conclude that in most locations, ML and SLC data
reveal no major shoreline changes. In a small number of isol-
ated places, however, the ML data show no changes, whereas
the SLC data show erosion or accretion. For example, Figure 6
shows Stumpy Point Bay, North Carolina, where no shoreline
changes are detected from 2007 (Figure 6(a)), 2008
(Figure 6(b)), to 2009 (Figure 6(c)) using ML datasets. The
distance measurements, shown along solid reference lines on
Figure 6(a)–6(c), are 4448.6, 4448.5, and 4446.9 m, respect-
ively. The absolute difference is B2 m, which is about 4.5% of
the entire distance measured. Because each pixel size is
12.5 12.5 m, the difference cannot be determined or there
was no detectable shoreline change. However, using the SLC
data pair of 2007 (Figure 7(a)) and 2009 (Figure 7(b)), a slight
landward movement occurred near the tip of the peninsula of
the Stumpy Point Bay. The estimated erosion rate here is
1–3 m year1 between 2007 and 2009. One cannot observe the
change at all using the ML data (Figure 7(c) and 7(d)); how-
ever, because the total change of the two-year period is on the
order of or less than half of the pixel size of the ML data, the
delineation of small changes can be attributed to the fine spatial
resolution of the SLC data as compared with the ML data. In
addition, due to the use of phase information in geo-refer-
encing the SLC pair, the RMS error can be as low as 1/10 of a
pixel size. There is no phase information in level 1.5 data. Thus,
the RMS error can be easily up to or more than 71 pixel.
3.4.3.3 Dynamic Feature Extraction: Riverine Flood
Inundation
Delineation of shoreline changes and floodwater/dry land
boundaries present a challenge for remote sensing that re-
quires ultrafine-scale feature extraction. The case studies that
follow focus on shoreline changes of the North Carolina coast

(a)
(b)
(c)
Figure 6 Multilook and multitemporal amplitude data of ALOS/
PALSAR show the Stumpy Point Bay, North Carolina. Acquisition
dates are (a) 24 September 2007, (b) 26 September 2008, and (c) 14
August 2009. The images are geo-rectified. The cross provides geo-
referencing information, and it is the starting point for the measured
line segment.
using multitemporal SAR data and the delineation of water/
nonwater boundaries at different river stages near Greenville,
NC, using a DEM inundation-overlay model.
Impacts of global climate change and sea-level rise on
coastal zones will accelerate, and coastal vulnerability will
concomitantly increase (Intergovernmental Panel on Climate
Change (IPCC), 2007). Rapid population growth in coastal
communities increases demands on coastal resources and
natural hazards pose additional threats to the communities.
Remote Sensing of Land Cover Dynamics 89
As one moves landward, people and communities on
coastal floodplains are also impacted by the change in water
levels in streams. Owing to a gently sloping coastal plain, a
large increase of surface water in streams could produce
a massive flood event. The ability to delineate the water/
nonwater boundary in a timely manner and accurately
during a flood event could mitigate dire consequences.
Thus, several efforts have been made toward resolving this
problem, including the creation of nationwide flood maps
(i.e., flood insurance rate maps (FIRMs) and recently digital
FIRMs for some states) by FEMA. The flood maps are widely
used in flood-hazard mitigation because the floodway, flood-
way fringe, boundaries of a 100-year floodplain, and a 500-
year floodplain are delineated. Given the variability of pre-
cipitation and climate change, however, the 100-year or
the 500-year floodplain can vary considerably. In addition,
the flood maps are static products that are the result of the
integration of hydraulic and hydrological modeling, para-
meterization of local topography and LULC types, and stream
flow/discharge. Changes between LULC types and impervi-
ousness on the floodplain can alter the local topography,
drainage, and runoff flow pattern and speed. Such temporal
variations cause variability in flood mapping results. Finally, it
is important to note that flood maps are legal documents.
Therefore, their compilation and revision are a very long and
costly process. More importantly, during the course of a flood
event, the flood maps may not provide timely and accurate
information on the location of floodwater/dry-land bound-
aries. Other alternatives to determine the boundary locations
must be sought.
Mapping the extent of a flood using remotely sensed
data, DEMs, river-gauge data, and supporting ground obser-
vations has received increased attention because of data
availability and the potential to assess flood recurrence inter-
vals, delineate floodplains, and respond to emergencies
(e.g., Colby et al., 2000; Wang et al., 2002; Wang, 2004;
Bourgeau–Chavez et al., 2005; Zheng and Wang, 2007; Lang
et al., 2008). Wang et al. (2002) summarized the advantages
and limitations of using the datasets individually or in various
combinations. The temporal variability of LULC types is
relatively high on and near floodplains because of urban ex-
pansion and floodplain development. Updated spatial infor-
mation about individual LULC types, communities, and
infrastructure is critical for assessing future flooding. Multi-
temporal and high-resolution imagery is best suited for pro-
viding this information. Furthermore, multiple governmental
agencies are obtaining and releasing topographic data and
multitemporal DEMs. For example, the United States Geo-
logical Survey (USGS) is currently carrying out a nationwide
LiDAR-DEM initiative to create the next generation of DEMs
for the nation (http://lidar.cr.usgs.gov/). Other topographic
data can be obtained from the NASA Shuttle Radar Topo-
graphy Mission (SRTM; http://www2.jpl.nasa.gov/srtm/) and
other optical-image sensors such as the Advanced Spaceborne
Thermal Emission and Reflection Radiometer (ASTER; http://
www.ersdac.or.jp/GDEM/ or https://lpdaac.usgs.gov/). Finally,
the surface-water heights of streams are continuously moni-
tored in the U.S. at tens of thousands of river-gauge stations
run by the USGS and the EPA. These measurements and other
data are available and downloadable through the Internet.
90 Remote Sensing of Land Cover Dynamics
Range direction
Flight direction
Land
(a)
800
Meters
0 800
(c)
Figure 7 PALSAR data for the peninsula near the Stumpy Point Bay, NC. Images in (a) and (b) are slant range, single-look complex data. Their
acquisition dates were 24 September 2007 and 14 August 2009. The cross indicates the landward migration of about one to two pixels from
2007 to 2009. The shoreline change is not noticeable on the geo-rectified multilook data from 2007 (c) and 2009 (d) for the same area.
These field data can be used in combination with a simple
and heuristic DEM inundation-overlay model (Zheng and
Wang, 2007) that maps flood extent. Such simple models can
efficiently and accurately simulate real-time flooding, and the
spatial information considerably facilitates the use of existing
floodplain maps. The approach includes: (1) the delineation
of a stream centerline using a DEM and high-resolution
imagery; (2) the interpolation of a river’s surface-water height
along the centerline using the measured surface-water heights
at the upstream and downstream gauge stations; (3) de-
lineation of the regular stream surface based on the regular
flow or surface-water heights along the channel; and (4)
delineation of the spatial extent using the elevated and
measured surface-water height to simulate a flood event.
Advantages include accounting for topographic changes using
multitemporal DEMs and overlaying flood extent over multi-
temporal remotely sensed data.
For example, Figure 8 shows DEMs arranged from the
finest to coarsest cell resolution, and depicts a small area near
Greenville, NC. By varying the scale of the DEM cell reso-
lution, the sensitivity of the inundation model to DEM error
can be assessed. The DEMs are displayed in grayscale, with
dark tones representing lower elevations and lighter tones
representing higher elevations. In general, lower altitudes
are located in the north and higher altitudes in the south.
The Tar River is also identified (Figure 8(a) and 8(b)). From
Water
(b)
(d)
Figure 8(a)–8(f), one can note the pixilation or loss of detail
in portraying local topography. It is important to note the
amount of lost detail in the SRTM DEM. Thus, one should be
cautious in using the DEM-inundation-overlay model and
attentive to scale-dependent representation and error propa-
gation when modeling processes such as inundation.
Spatial and temporal changes in inundation arising from
Land-Use and Land-Cover Change (LULCC) are also signifi-
cant to riverine and coastal-flood hazards. Figure 9(a) depicts
an overlay of the simulated surface-water coverage of Tar River
at regular flow over the 1998 USGS false-color IR digital
orthophoto quartile quads (DOQQ). The same surface-water
coverage is then overlaid on the 2006 USDA natural-color
orthophotos. Significant development occurred during the
1998–2006 period, which altered the LULC types in the
floodplain and watersheds. This information on landscape
change could be critical to forecasting and emergency re-
sponse to a future flood event, including terms of runoff
modeling and flood-impact assessment. To illustrate its im-
portance, the surface-water heights at flood stage on the Tar
River and the 1999 flood after Hurricane Floyd (Wang et al.,
2002) were simulated and overlaid on the 2006 image (Fig-
ure 10). The water is near the door-steps of the houses when
the river is at its flood stage (Figure 10(a)). Consequently,
most of the area is flooded, and a newly developed area is
completely submerged when the water rises to the level of the

(a)
(c)
(e)
Figure 8 Various DEMs near Greenville, North Carolina, arranged from fine to coarse spatial resolutions illustrate scale and resolution
sensitivity. (a) USGS DEM of 3  3 m, (b) NC floodplain mapping program DEM of 6  6 m, (c) USGS DEM of 10  10 m, (d) NC floodplain
mapping program DEM of 15  15 m, (e) USGS NED of 30  30 m, and (f) SRTM DEM of 1  1 arcsecond.
1999 flood (Figure 10(b)). The ground photo (Figure 11)
taken after the 1999 flood looking toward the north illustrates
the disastrous result. In the photo, the flood water line was
clearly visible, and its elevation value is much higher than that
of the ground where the development was underway (com-
pleted as shown in a 2006 image).
3.4.4 Land-Cover Change Modeling
3.4.4.1 Emergence and Milestones
Land-cover mapping and analysis enabled by geospatial tech-
nologies is a major component of integrative land-change sci-
ence. This field has emerged as a cornerstone of global
environmental change and attempts to understand the human
and biophysical factors influencing land-cover dynamics,
defined here as temporal variations in land cover and land use
(Rindfuss et al., 2004). ‘Land-change science,’ as it may be
termed, is of interest for understanding spatial and temporal
dynamics and promotes investigations of human–environment
Remote Sensing of Land Cover Dynamics 91
(b)
(d)
(f)
linkages and systems that require integrative contributions
from social, natural, and geographical information sciences.
Goals include the identification and understanding of forcing
factors, spatial and temporal dynamics, and the consequences
of anthropogenic and natural change. Expertise in geospatial
techniques and information theory is a prerequisite for
integration of social and environmental data or the linking
of ‘people-to-pixels’ at relevant spatial and temporal scales
(Liverman et al., 1998; Rindfuss et al., 2004). Furthermore,
investigation of complex anthropogenic influences on land
cover that affect surface processes and geomorphology re-
quires a multidisciplinary approach, careful research design,
and expertise spanning social science, physical geography, and
geospatial techniques.
Major milestones in the development of land cover-related
science include the LULCC program that was initiated in 1994
as a core project of the International Geosphere–Biosphere
Programme to investigate the questions: How do human and
biophysical forces affect land cover, and what are the environ-
mental and social impacts of land-cover dynamics? Affiliated
LULCC projects have applied remote-sensing approaches to a
92 Remote Sensing of Land Cover Dynamics

Tar r
iver a
t reg
ular
flow
stag
e

(a)

Tar riv
er at
regula
r flow
stage

(b)

Figure 9 Overlays of modeled surface-water coverage at the regular flow of Tar River on the 1998 USGS false IR DOQQ image (a) and the
2006 USDA natural color image (b). Water appears black. Development has occurred on the south side of the river.
 Remote Sensing of Land Cover Dynamics 93

Water from Tar River at

flood stage

(a)

Water from Tar River at surface

height after Hurricane Floyd in 1999

100

meters

0 100

(b)

Figure 10 Modeled surface-water coverage for 2006 natural color image (a) Tar River, NC, during flood stage, and (b) flood flow during 1999
Hurricane Floyd. Water is shown in black.
94 Remote Sensing of Land Cover Dynamics
Figure 11 Ground photo of the Tar River floodplain after the recession of the 1999 Hurricane Floyd flood. The flood-water line is clearly visible,
with its elevation much higher than that of the ground.
variety of internationally ranging case studies having global
implications, with notable emphases on tropical deforestation
(Galicia et al., 2008), agricultural extensification–intensifica-
tion, desertification, and grassland dynamics. A comprehen-
sive synthesis of LULCC results since its inception is reported
in Lambin and Geist (2006). Related in scope and focus to
LULCC, in 1996, the U.S. National Aeronautics and Space
Administration (NASA) launched its LCLUC program, which
addresses science questions similar to LULCC, but has an
additional focus on remote sensing applications using NASA
or other earth-observation data products. NASA LCLUC
also regularly sponsors funded regional and thematic-based
research. A synthesis of LCLUC results since its inception is
reported in Gutman et al. (2004).
Recognizing the significance of land-cover change, the
National Research Council (2001) identified ‘land-use dy-
namics’ as one of eight ‘grand challenges’ in its prescriptive
publication Grand Challenges in Environmental Sciences. Al-
though stated as a stand-alone problem domain, it is worth
noting that other identified challenges have clear connections
and feedbacks with land use and cover change. These include
biogeochemical cycles, biodiversity and ecosystem func-
tioning, climate variability, hydrologic forecasting, infectious
disease and the environment, and institutions and resource
use. The impetus and relevance to geomorphology and
Earth-surface processes in these research thrusts are broad. It
has been recognized, for instance, that changes in land cover
affect biodiversity (e.g., tropical rain-forest deforestation;
Michalski et al., 2010), and yet biodiversity owes much of its
community and landscape heterogeneity to underlying topo-
graphic and physiographic diversity. Hence, geomorphologists
and ecologists recognize that interactions of biotic and geo-
morphic change from sources to sinks, including environ-
ments ranging from alpine tundra (French, 1986), montane
forests (Allen and Kupfer, 2000), to coasts (Efe and Tagil,
2008). Processes stemming from land-cover change, such as
erosion, are also known to cause riverine and estuarine sedi-
ment pollution (Kusimi, 2008). Land cover derived from the
classification of remotely sensed data (digital satellite data or
aerial photography), in particular, governs many parameters
used in modeling environmental processes, including the
Universal Soil Loss Equation (USLE) and soil-runoff curve
numbers found in many models of soil erosion and runoff
potential (Renard et al., 1997). Increasingly fine spatial-reso-
lution sensors have promoted the study of smaller catchments
(Jurgens and Fander, 1993; Schonbrodt et al., 2010, Seeber
et al., 2010).
Table 1 Strategic goals from Geography for a Changing World,
USGS Circular 1281
1. Characterize and quantify land-surface status and trends to provide a
framework for understanding change patterns and processes from
local to global scales.
2. Identify local, regional, national, and global drivers of land change to
forecast plausible land-change scenarios over the next 20–50 years.
3. Understand past, present, and future environmental consequences of
land change to support better management of their effect on people,
environment, economy, and resources.
4. Improve the scientific basis for vulnerability and risk assessment,
mitigation, response, and recovery related to the human and
environmental dynamics of land change.
5. Develop credible and accessible geographic research, tools, and
methods to support decision-making related to the human and
environmental consequences of land change.
6. Develop and test hypotheses about the use of geographic regions to
understand the human and environmental dynamics of land change.
7. Observe the Earth at all scales using remote sensing to understand
the human and environmental dynamics of land change.
8. Provide timely, intelligent access to new and archived USGS geographic
data needed to conduct science and support policy decisions.
9. Develop innovative methods of modeling and information synthesis,
fusion, and visualization to improve our ability to explore geographic
data and create new knowledge.
U.S. Geological Survey (USGS) (2005) published a science
strategy for geographic research titled Geography for a Changing
World. Although ‘geography’ has connotations beyond the sole
topic of land-cover change, this strategic plan parallels related
programmatic developments of land-change science described
above. The science strategy articulated nine goals each with
associated operational objectives (Table 1). A related USGS
initiative was the establishment of the Land Cover Institute in
2006 that functions ‘‘to address land-cover topics from local to
global scales, and serves as a facilitator for land-cover and
land-use science, applications, and production functions’’
(U.S. Geological Survey (USGS), 2010).
Initiatives of the U.S. National Science Foundation (NSF)
have contributed to the development of land-change science
that investigates geomorphology science questions. Beginning
in 1999, NSF’s Biocomplexity Program has funded inter-
disciplinary research dealing with complex problems in-
cluding land use and cover dynamics. The year 2000 program
announcement stated that ‘‘in order to understand land-cover
change processes, we need to better understand the impact of
the constructed environment, human activities, and social
 Remote Sensing of Land Cover Dynamics 95

structures on ecosystem functions.’’ The Biocomplexity pro-
gram ceased following the 2006 award cycle, but elements of
the program that support land-change science projects have
continued to be present with the 2007 establishment of NSF’s
Dynamics of Coupled Natural and Human Systems Program.
NSF’s Geomorphology and Land Use Dynamics Program has
funded research since 2003 that examines natural and human
processes shaping geomorphic landscapes over a variety of
space and time scales.
3.4.4.2 Proximate and Ultimate Drivers of Land Change
To date, no unifying formal theory or theories of land use and
cover change exist. Instead, a growing body of regionally
varying case studies has investigated causative factors and as-
sociated consequences of change. Although specific contexts
necessarily vary, a commonality in much of the land-change
research is a basic framework that specifies the roles of both
proximate and underlying drivers (Turner et al., 1993; Geist
and Lambin, 2001; Xu and Wilkes, 2004). As an example,
Figure 12 illustrates forcing factors relevant to urban and
suburban development in the contemporary U.S. Such chan-
ges in urbanizing watersheds may promote the increase in
soil erosion and runoff in small watersheds, contributing to
‘flashy’ hydrology, incision in urban watersheds and down-
stream sediment pollution and exacerbated flood inundation.
Land-cover change impacts on storm hydrographs are
well documented in hydrologic models under current and
projected climate change (Praskievicz and Chang, 2009).
Proximate causes are behaviors that directly alter land uses
and associated land covers. For example, the construction of
new roads, housing subdivisions, and shopping centers
are proximate causes in that they convert natural cover types
(or prior human altered uses) into newly anthropogenized
covers. The pattern of such change is also of interest, as land-
cover fragmentation from increasing urbanization increases

Infrastructure expansion Demographic factors

Transport Fertility and mortality
• Roads, sidewalks, paths In and out-migration
Public service Population density
• Water lines, electrical grids Population distribution
sanitation, govt. buildings
Life cycle features

Residential expansion Economic factors

Single family homes
Market growth
Apertments, condominiums
Economic structure
Proximate causes

Golf course communities

Underlying causes
Urbanization
Retirement communities
Special variables
Institutional housing
• Price increases, comparative
cost advantages
Recreation expansion
Hotels/motels Policy/institutional factors
Seasonal vacation houses Formal policies
Camp grounds, RV parks Planning and land use regulations
Golf course, restaurants, shops Policy climate
Parks Non-profit organizations

Commercial expansion Cultural factors

Office and industrial Attitudes, values, beliefs
Wholesale and retail • Environment protection
• Pro-growth, anti-growth
Individual and household behavior
• Place utility, sense of place

Other factors
Predisposing environmental factors (soil type, topography, etc.)
Biophysical drivers (triggers, floods, hurricanes, sea-level rise)
Social trigger events (war, economic shocks, abrupt policy shifts)

Figure 12 Hypothesized causes of land-use change impacting land cover associated with urban/suburban development in the United States.
96 Remote Sensing of Land Cover Dynamics
stream flashiness (McMahon et al., 2003). Such tangible al-
terations are detectable on the ground and via remote obser-
vation using standard data products such as Landsat, MODIS,
and other sensors. They generally involve complex interactions
and typically are more difficult to observe. A case in point may
be the transitional process of agricultural land subdivision for
residential housing. Agricultural land cover can be mistaken
for new or imminent suburbs. For instance, bare agricultural
land recently ploughed may be spectrally similar to newly
graded residential lots. Emergent crops with high greenness
may mimic newly sodded or seeded residential lawns. The
process of land subdivision or ‘parcelization’ can thus
be difficult to discern in early phases as it is happening.
Nonetheless, mediating the link between underlying and
proximate causes are ‘other factors’ that can be viewed as in-
fluences that may be site-specific and contextual in nature and
may vary across temporal and spatial scales.
Key challenges of land-change science include: (1) obser-
vation, measurement, and detection of changing use and cover;
(2) establishment of empirical links between underlying and
proximate causes and their contributions to changing use and
cover; and (3) development of conceptual theories to explain
land-cover dynamics. The first challenge clearly features
the role of remote sensing and related geospatial science
approaches. The subsequent challenges involve contributions
from allied social-science perspectives as well as scientific
modeling techniques. Importantly, Figure 12 indicates a po-
tentially complex array of causative drivers and negates a
more simplistic monocausal explanation. A singular focus
on population growth exemplifies a simplistic monocausal
explanation. Although population growth can surely be a sig-
nificant driver, Lambin et al. (2001) identified it as one
of several other ‘myths,’ or what might be considered biases
regarding issues and problems in understanding land use and
cover change (Table 2; see also Heilig, 1994). Major challenges
still remain for the linkage of human interactions and the Earth
system, particularly in the acquisition of socio-economic data
and its linkage to satellite imagery and the forecasting of future
human activities and policies (Liverman and Cuesta, 2008).
3.4.4.3 Modeling Approaches
Modeling of land-cover changes involves manipulating digital
geospatial representations of human and biophysical patterns
and processes. Description and qualitative narrative
Table 2 Myths (or biases) of land use and cover change research
Myth (or bias)
Population growth is the major cause of change.
Change is comprehensive land conversion.
Tropical forest is the most pressing problem of land change.
Agricultural expansion in developing world is among the most pressing
problem of land change.
Prior ‘unaltered’ land is assumed pristine.
Change is permanent.
Change is homogeneous.
Local factors primarily drive change.
Source: Adapted from Heilig, G.K., 1994. Neglected dimensions of global land-use change: reflections and data. Population and Development Review 20(4), 831–859, and Lambin,
E.F., Turner, II B.L., Geist, H.J., et al., 2001. The causes of land-use and land-cover change: moving beyond the myths. Global Environmental Change 11, 261–269.
methodologies are sometimes used to first build a conceptual
model before quantitative predictive models are constructed.
Before modeling, a remotely sensed time-series data set is
typically required to quantify trends of change describing the
types, magnitudes, and pattern of change. Additionally, spa-
tial-pattern metrics can be used to describe the change tra-
jectories of spatial configurations and patch dynamics
(McGarigal and Marks, 1993). Human-dimensions data col-
lected via detailed social surveys and key informant interviews
are required to describe and understand land-use decisions of
relevant decision makers (i.e., households, firms, land man-
agers, government agencies) and to link these agents to their
respective land parcels (Rindfuss et al., 2004).
A diversity of modeling approaches exists that can be
characterized along different dimensions (Lambin and Geist,
2006). One distinction is between spatial and nonspatial
models. Spatial, or spatially explicit models, represent change
at some level of spatial detail. ‘Spatially explicit’ models
simulate change in disaggregated, relatively fine spatial units
such as individual pixels within a digital geospatial context.
Clearly, finer spatial resolution yields more detailed model
output, but at an added computational cost. Models may also
be either dynamic or static.
3.4.4.3.1 Spatial modeling approaches
Temporally static modeling approaches typically are in the
form of regression models and are useful to acquire know-
ledge about the driving factors of land change. They require at
least two observation dates (although more can also be used),
with which to observe change as the dependent variable to be
modeled. They are static in that no dynamic and iterative
updating is implemented, either spatially or temporally, to
reflect change occurring within the modeled time window.
Conversely, dynamic models are constructed to incrementally
represent change at specific locations and at incremental time
steps throughout the modeled time window. A geomorpho-
logic example of a static spatial model using linear regression
is the application of land cover, soil type, and soil-runoff curve
numbers to predict soil erosion. Applications of this approach
may also be spatially explicit (e.g., soil erosion, rainfall infil-
tration, or runoff per pixel) or alternatively, aggregated spatial
units such as catchments, hillslopes, or land-use zones.
Descriptive models aim to represent the actual functioning
of land use and cover change in a manner that best matches
empirically observed data. Prescriptive models, in contrast,
Reality
Change is due to multicausal factors.
Change includes incremental land modification.
Many other important types of changes exist.
Urbanization and suburbanization in developed and developing world is
important.
Land has been altered for millennia.
Change can be reversible.
Change is heterogeneous and fragmented.
Remote influences/globalization are important.

aim to represent optimized land use and change patterns ac-
cording to a defined set of goals and objectives. A further
distinction can be made between deductive and inductive
models. This distinction shares similarities to the descriptive/
prescriptive distinction. Deductive modeling postulates theory
a priori governing change to predict pattern from process,
albeit articulated in the theory. Theory derived from Von
Thunen, for example, has been used to model agricultural
expansion, deforestation, and wetland conversion. Inductive
modeling is more common within the land-change science
community, where model specifications are commonly based
on statistical correlations between land change and a suite of
explanatory variables. Other inductive approaches include
rule-based models informed by empirical observations and
neural-network techniques. Although most studies rely heavily
on inductive approaches to develop quantitative models,
they typically rely on theory and understanding of land-use
decision processes to guide the set of factors to include in
model representation. A rule-based descriptive geomorpho-
logic model example could be sea-level rise resulting in in-
undation and shifting floodplains (Brown, 2006). In this
sense, land-change modeling commonly blends inductive and
deductive approaches drawing from a range of physical and
social-scientific disciplines, such as anthropology, ecology,
economics, geography, political science, and geography for
both theoretical and methodological guidance.
3.4.4.3.1.1 Multicriteria evaluation
Multicriteria evaluation (MCE) modeling is a prescriptive form
of suitability-site modeling that is used in numerous GIS ap-
plications (Malczewski, 2006). Geomorphologic examples of
MCE modeling include management of groundwater-recharge
zones (Murthy and Mamo, 2009), erosion-risk assessment
for railway corridors (Barbieri et al., 2002), site-suitability
modeling for sanitary landfills (Mariolakos et al., 1997), and
landslide-hazard assessment (Barredo et al., 2000). As applied
to land-change science, MCE incorporates multiple GIS data
layers hypothesized to represent relevant factors (criteria) in-
fluencing human behavior and future land-use decisions. For
example, factors might include current zoning, future land-use
plan, terrain slope angles, important natural resource areas,
hazard status, road accessibility, and proximity to existing
development. Central to MCE is the assignment of quantita-
tive scores reflecting suitability for a particular land use in the
future. Typically, a numeric scheme of 1 (least suitable) to 9
(most suitable) is applied to each spatial unit in each GIS layer
(vector polygons or raster grid cells) with a weighting scheme
for each input layer. The idea is that criteria (i.e., GIS layers) do
not share the same importance regarding suitable future land
use. A weighted-sum for each spatial land unit is obtained via
GIS overlay that quantifies the MCE output, which is a
quantitative surface with high scores indicating high suit-
ability for conversion to a particular land use (i.e., develop-
ment) and vice versa. Multiple outputs can be created by using
varying rankings and weighting schemes to assess output
stability.
The use of subjective assessment regarding both the suit-
ability scoring and weighting schemes is a common criticism
of the MCE technique. One advantage is that it includes
local stakeholder input. Including stakeholder input, however,
Remote Sensing of Land Cover Dynamics 97
comes with the cost of additional local engagement with local
planners and stakeholders. Another criticism of the MCE ap-
proach is that it quantifies a gradient of suitable future land
use, but it does not actually produce future land-use maps for
defined time periods. Methods requiring projections of land
consumption or population growth must be implemented in
concert to project amounts of land to allocate for future
conversion. Rule sets must also be implemented to resolve
situations where competing uses have similarly high suit-
ability. These decisions partly depend on the use of the model
and the resolution of data available during the modeling
process. The LUCIS Model (Land Use Conflict Identification
Strategy) developed by Carr and Zwick (2007) is an example
of a recent and significant MCE model that draws from a rich
tradition of similar modeling techniques.
3.4.4.3.1.2 Statistical regression
Statistical-regression techniques offer another approach to
modeling land-cover dynamics. Spatial-regression appli-
cations of land cover in geomorphology are wide ranging,
including mapping applications such as bathymetry (pre-
dicting depth from reflectance among different classified bot-
tom types), predicting landslide activity as a function of
topographic variables and land cover (Biswajeet et al. (2011)),
relationships between woody debris, fluvial channels, and
land cover (Wing and Skaugset, 2002), or stream sediment
loads (Phippen and Wohl, 2003). Logistic regression and
multinominal regression have been considerably used for such
purposes within the research community of land-change sci-
ence (Lesschen et al., 2005). Logistic regression is described
below, although similarities extend to multinominal re-
gression (see Lesschen et al., 2005).
Geomorphologists seeking to predict the occurrence of
patterns or processes, but having categorical or dichotomous,
spatial-variable predictors may use logistic regression models.
Pradhan and Lee (2010), for example, incorporated multi-
nominal land-cover classes into a logistic-regression model to
delineate landslide-hazard zones on Penang Island, Malaysia.
The dependent variable in logistic regression, y, is the presence
(1) or absence (0) of land conversion at the individual pixel
level. This observed change is for a defined historical period
that includes initial and terminal dates. Observed change is
measured via change-detection techniques using multi-
temporal satellite data. A vector, X, serves as a set of spatially
explicit predictive variables and p is the response probability,
where p ¼ Pr(Y ¼ 19X). A random sample of pixels can be se-
lected for initial modeling. A second random sample can be
selected to validate the resulting model(s) using standard
techniques such as error matrices, kappa coefficients, and the
receiver operator characteristic (ROC) method. The resulting
logistic-regression equation can be implemented within a GIS
to calculate and map a probability surface (p) of development
ranging from 0 (low) to 1.0 (high).
The resulting probability surfaces are used to project future
land change. A key input for models of anthropogenic land
change is observed population growth and land consumption
for the observed historical data that provide information on the
amount of land consumed (i.e., converted to human use) per
capita. Using the regression-derived probability surfaces, pixels
are ranked according to the highest probability of change and
98 Remote Sensing of Land Cover Dynamics
selected in descending rank order to be converted up to a vol-
ume consistent with the estimated conversion amounts.
Some limitations and criticisms of this regression-based
approach include the following: (1) regression parameter
estimates from historical data are assumed to remain sta-
tionary across space and through time; (2) empirically derived
models based on historical data exhibit poor accuracy
performance as assessed by quantitative accuracy assessment
techniques, and accuracies may vary geographically; and
(3) the temporal scale of the original regression models
(e.g., a 20-year time step) may not match the temporal scales
of the projected output (e.g., projecting out to 40 years).
3.4.4.3.1.3 Dynamic spatial simulation
Cellular automata (CA) and agent-based modeling are rapidly
emerging dynamic spatial-simulation modeling techniques.
CA models typically operate using a gridded raster landscape,
with individual grid cells having discrete land-use/cover cat-
egories (i.e., residential, commercial, agriculture, natural, etc.).
A set of locally based transformation rules are applied in
discrete time steps (i.e., annual updates) to model land-use
change. For example, for an individual ‘undeveloped’ cell, if a
defined number of neighboring cells are in a ‘developed’ state,
then the cell is converted from undeveloped to developed at
the next annual iteration. Topographic and landform features
can be used as raster inputs to CA models affecting land-use
changes (e.g., elevation, slope, aspect, and curvature; discrete
landforms such as floodplains, valleys, and ridges; or higher
order topographic derivatives such as view-sheds and sub-
watersheds). A variety of rules may also be incorporated into
such models, such as diffusive growth emanating from existing
roads or the developed fringe. Stochastic (random) growth
such as ‘leapfrog development’ can be simulated by using
appropriate model parameters. The SLEUTH model (Clarke
and Gaydos, 1998) developed by Keith Clarke at UC Santa
Barbara is perhaps the most well known of such CA modeling
approaches that has been widely implemented.
In addition to the use of terrain or landforms as predictors
of human land-use change, spatial simulation models are in-
creasingly being applied for investigating environmental pro-
cesses and terrain evolution. For instance, Saco et al. (2006)
developed an enhanced form of the SIBERIA model that
incorporates landform evolution to predict the spatial organ-
ization and depth of soils. Three-dimensional landscape-
pedogenesis models have also emerged such as matrices
ARMOUR 3D (mARM3d) to simulate soil evolution as a
function of soil profile for points on a spatial grid. Markov-like
transition matrices provide for the examination of depth-
dependent and independent soil-forming factors (Cohen,
2010). Vegetation-geomorphic patterns have also been simu-
lated using spatial models, including montane and subalpine
forests (Brown, 1994; Horsch, 2003), and patterns of South-
ern Appalachian forest regeneration following invasive
pathogen introduction (Allen and Kupfer, 2000). Spatial de-
pendence has also developed as a valuable consideration in
the prediction of pattern and process. Some spatial-simulation
models incorporate scale dependency explicitly, such as using
semivariograms and geostatistics (e.g., Brown et al., 2002).
Agent-based models are another form of dynamic spatial
simulation and are thoroughly reviewed in Parker et al. (2002).
Similar to CA, agent-based models typically operate on a
gridded, raster landscape. They differ by simulating a large
number of virtual agents (i.e., homebuyers, developers, busi-
nesses) that are imbued with decision-making behavior
that alters land use in discrete, typically annual, time steps.
Decision-making behavior is represented by a rule set
hypothesized to represent real-world processes changing the
landscape. Organisms have become a subject of simulation
using agent-based models, because behaviors can be modeled
in coupled external routines (i.e., model code) that in-
corporate economic strategies such as utility maximization or
bounded rationality. Empirical survey data are frequently
collected from actual agents (i.e., a survey of landowners) to
reflect empirical realities. Interactions between prehistoric
human populations and basic processes of food acquisition
(hunting, gathering, and agriculture), for example, have been
simulated and linked to landform evolution (Wainwright,
2008). Developments in anthropogenic factors modifying
landscapes have further been demonstrated for human im-
pacts on wildfire (the Special Protection Area SIMulator
model, SPASIMv1) and simulation of changes in erosion and
deposition over century scales (CYBEREROSION model;
Wainright and Millington, 2010).
Both approaches to dynamic spatial simulation provide a
virtual laboratory to help replicate and explain observed
change, and explore and probe alternative futures and scen-
arios. Difficulties with these models include complexities of
both model calibration (i.e., setting appropriate model par-
ameters) and model validation (i.e., checking results against
reference data). Other challenges include obtaining the
technical expertise required for software coding (i.e., Java,
Repast, C þ þ ).
3.4.5 Future Research Directions
A treatment of remote sensing of land-cover dynamics has
been presented that accounts for new developments and
challenges in land-change science and geospatial technologies.
Selected examples of new sensors and methodologies have
also been presented. Concluding remarks can thus be sum-
marized as a situation of coevolution of remote-sensing sci-
ence with scientific paradigms and geographic information
technology.
The rapid advancement of remote sensing is widely known,
with sensor technology and information extraction being
dominant themes. Historically, the first impetus was on the
increasing continuum of information extraction, classification,
and change detection. Each increase in this continuum rep-
resents a higher order of information. Most of the infor-
mation-extraction techniques also engender products that are
informative to science. For instance, we quantified land-cover
change and inferred the increasing impact of land-use con-
version on sediment yield, pollution, or habitat degradation.
Having quantified change and verified the relationships be-
tween cause and effect and understanding processes, we can
thus predict and simulate alternatives in order to further
understand Earth-surface systems coupled with human alter-
ations. Land-change science has emerged concomitant to
more widely available LULC data. Although the use of hard

classification still predominates, soft and objected-based
classification techniques are attracting research and appli-
cation interest. The proliferation of available imagery and
growing time series for new change-detection algorithms has
also prompted the expansion of biophysical analysis, such as
the assessment of phenology in global change phenomena
and operational use of remote sensing data for near-real-time
use in hazards and agriculture. Dynamic geomorphic en-
vironments such as coasts and urban environments are at-
tracting greater interest in these techniques. Advances in
sensors and platforms provide for new opportunities, such as
illustrated in coastal imperviousness mapping, LiDAR-derived
digital terrain modeling (including subwatershed delineation
and hydrologic modeling), and their integration for predicting
flooding from urban runoff. Passive space-borne SAR sensors
were also illustrated for time-series analysis of shorelines, in-
cluding coastal erosion and riverine flooding.
Many challenges and great potential remain for the ex-
panded application of remote sensing to a wide array of
geomorphologic research. New technologies are allowing for
questions that could not be addressed in recent decades to be
handily studied by primarily remote-sensing investigations. In
other cases, new observational data have revealed limitations
that require more research and modeling and technology so-
lutions, such as addressing the effects of differential illumin-
ation in topographically complex environments (Colby, 1991;
Allen, 2000; Bishop and Colby, 2002). The use of new data is
another example, as LiDAR is becoming essential in the
hydrologic and coastal geomorphology subdisciplines. Yet,
LiDAR for some environments is relatively straightforward
(e.g., topographic LiDAR for beach mapping) versus highly
vegetated, urban, or aquatic environments. The tropics and
perennially cloudy regions also pose great challenges to rou-
tine earth observations from passive sensors, which has
spawned a greater interest in active sensors such as SAR.
Land-change science provides a road map and rubric for
integrating remote-sensing derived products with geomorph-
ology. We might expect that land-change science will further
exploit the ever-growing time series of imagery, perhaps ex-
panding beyond discrete, thematic land-cover conditions and
typologies. Beyond the principal human dimensions of en-
vironmental change, such as land use and urban growth
modeling, geomorphologists can join in biocomplexity re-
search programs, further investigating the stressor or receptor
systems in environmental processes. Land-cover change studies
in arctic and alpine environments prompt many studies of
glacio-hydrological and geomorphic significance, for example.
The resulting changes in mountain hydrology, although having
human impacts on water resources, also portend significant
changes to stream power, load, and fluvial processes docu-
mented elsewhere in this volume. These applications have ini-
tiated the use of time-dependent classes, phases, and sequential
Markovian analyses of change. Second, biophysical remote
sensing is increasingly able to integrate and apply ‘data-fusion’
techniques for combining multisensor information, such as
passive and active sensors, airborne and satellite, and spectral
and topographic data. The coastal imperviousness case study,
for example, demonstrated the integration of LiDAR and IKO-
NOS satellite data, but the study also used ground-based re-
sistivity mapping by means of an OhmMapperTM system. This
Remote Sensing of Land Cover Dynamics 99
example also suggests potential increased application of remote
sensing at ever finer spatial resolutions, taking advantage of
ultrahigh resolution sensors using objected-based image clas-
sification. The coastal SAR image analysis case points to a par-
allel development occurring in time series and multilook-angle
image analysis for feature extraction.
Feature-extraction techniques are also poised to continue
their rapid advancement. High spatial and temporal reso-
lution datasets of SAR, and the taskable, off-nadir viewing
geometry of some satellites allow improved research on shore
erosion processes and disasters. Thus, one can map and
quantify trends of shoreline changes, discover how shoreline
changes and erosion rates may affect infrastructure, security,
and social and economic vulnerabilities, and disseminate the
geospatial data and shore information to stakeholders in
support of improved coastal policy and sustainability.
As these examples and others demonstrate, the remote-
sensing research community will have numerous opportun-
ities to investigate the information content of ever-increasing
sophisticated sensors that produce spatiotemporal data. The
development and evaluation of new information extraction
techniques and models will also be critical and will permit
new scientific applications. Land-cover dynamics will become
ever more important to geomorphology in order to under-
stand landscape evolution and estimate anthropogenic forcing
on erosion, natural hazards, and ecological and hydrological
systems.
References
Allen, T.R., 2000. Topographic normalization of Landsat Thematic Mapper data in
three mountain environments. Geocarto International 15(2), 13–19.
Allen, T.R., Kupfer, J.A., 2000. Application of spherical statistics to change vector
analysis of Landsat data: southern Appalachian Spruce-fir forests. Remote
Sensing of Environment 74, 482–493.
Allen, T.R., Walsh, S.J., 1993. Characterizing multitemporal alpine snowmelt
patterns for ecological inferences. Photogrammetric Engineering & Remote
Sensing 59(10), 1521–1529.
Anderson, J.R., Hardy, E.E., Roach, J.T., Witmer, R.E., 1976. A Land Use and Land
Cover Classification System for Use With Remote Sensor Data. US Geological
Survey Professional Paper 964, Washington, 41 pp.
APES Albemarle-Pamlico Estuary Study, 1994. Land Use/Land Cover – TM
(APES 1987). North Carolina Center for Geographic Information and
Analysis (NC CGIA). http://www.geodata.gov/metadata/ (last accessed 12 May
2010).
Arrell, K.E., Fisher, P.F., Tate, N.J., Bastin, L., 2007. A fuzzy c-means classification
of elevation derivatives to extract the morphometric classification of landforms in
Snowdonia, Wales. Computers & Geosciences 33, 1366–1381.
Babiker, I.S., Mohamed, M.A., Hiyama, T., Kato, K., 2005. A GIS-based DRASTIC
model for assessing aquifer vulnerability in Kakamigahara Hights, Gifu
Prefecture, central Japan. Science of the Total Environment 345(1-3), 127–140.
Baker, W.L., Cai, Y., 1992. The r.le programs for multiscale analysis of landscape
structure using the GRASS geographical information system. Landscape Ecology
7, 291–302.
Barbieri, M., Bonachea, J., Corsini, A., Giusti, C., 2002. A GIS based model for
geomorphological impact assessment of a railway track (province of Modena,
Italy). Geografia Fisica e Dinamica Quaternaria 25, 11–22.
Barredo, J.I., Benavides, A., Hervás, J., van Westen, C.J., 2000. Comparing
heuristic landslide hazard assessment techniques using GIS in the Tirajana
basin, Gran Canaria Island, Spain. International Journal of Applied Earth
Observation and Geoinformation 2, 9–23.
Barringer, T.H., Robinson, V.B., 1981. Stochastic models of cover class dynamics.
Proceedings of Fifth International Symposium on Remote Sensing of
Environment, Ann Arbor, MI, pp. 125-144.
100 Remote Sensing of Land Cover Dynamics
Bauer, M.E., Doyle, J.K., Heinert, N.J., 2002. Impervious surface mapping using
satellite remote sensing. Proceedings of IEEE Geoscience and Remote Sensing
Symposium, Toronto, Canada, June 24–28, cd-rom, 3 pp.
Becker, M.W., 2006. Potential for satellite remote sensing of ground water. Ground
Water 44(2), 306–318.
Beven, K., 1997. TOPMODEL: a critique. Hydrologic Processes 11, 1069–1085.
Bian, L., Walsh, S.J., 1993. Scale dependencies of vegetation and topography in a
mountainous environment of Montana. The Professional Geographer 45(1), 1–11.
Bishop, M.P., Colby, J.D., 2002. Anisotropic reflectance correction of SPOT-3 HRV
imagery. International Journal of Remote Sensing 23(10), 2125–2131.
Biswajeet, P., Mansor, S., Pirasteh, S., Buchroithner, M.F., 2011. Landslide hazard
and risk analyses at a landslide prone catchment area using statistical based
geospatial model. International Journal of Remote Sensing 32(14), 4075–4087.
Bourgeau–Chavez, L.L., Smith, K.B., Brunzell, S.M., Kasischke, E.S., Romanowicz,
E.A., Richardson, C.J., 2005. Remote monitoring of regional inundation patterns
and hydroperiod in the greater Everglades using Synthetic Aperture Radar.
Wetlands 25(1), 176–191.
Brabec, E., Schulte, S., Richards, P.L., 2002. Impervious surfaces and water quality:
a review of current literature and its implications for watershed planning. Journal
of Planning Literature 16(4), 499–514.
Brabyn, L.K., 1997. Classification of macro landforms using GIS. ITC Journal
1997(1), 26–40.
Brown, D., 1997. Using GIS technology in the development and maintenance
of a stormwater utility. GIS/LIS Annual Conference Proceedings of American
Society for Photogrammetry and Remote Sensing. Cincinnati, OH. (http://
stormwaterfinance.urbancenter.iupui.edu/PDFs/GISLISv3.pdf).
Brown, D.G., 1994. Predicting vegetation types at treeline using topography and
biophysical disturbance variables. Journal of Vegetation Science 5, 641–656.
Brown, D.G., Goovaerts, P., Burnicki, A., Li, M.-Y., 2002. Stochastic simulation of
land-cover change using geostatistics and generalized additive models.
Photogrammetric Engineering and Remote Sensing 68(10), 1051–1061.
Brown, I., 2006. Modelling future landscape change on coastal floodplains using a
rule-based GIS. Environmental Modelling & Software 21(10), 1479–1490.
Carpenter, S.R., Caraco, N.F., Correll, D.L., Howarth, R.W., Sharpley, A.N., Smith,
V.H., 1998. Nonpoint pollution of surface waters with phosphorus and nitrogen.
Ecological Applications 8(3), 559–568.
Carr, M., Zwick, P., 2007. Smart Land Use Analysis: The LUCIS Model. ESRI Press,
Redlands, CA.
Charbonneau, L., Morin, D., Royer, A., 1993. Analysis of different methods for
monitoring the urbanization process. Geocarto International 8(1), 17–25.
Chen, J., Gong, P., He, C., Pu, R., Shi, P., 2003. Land-use/land-cover change
detection using improved change-vector analysis. Photogrammetric Engineering
& Remote Sensing 69(4), 369–379.
Choi, J., Oh, H.J., Won, J.S., Lee, S., 2010. Validation of an artificial neural
network model for landslide susceptibility mapping. Environmental Earth
Sciences 60(3), 473–483.
Choi, K.S., Blood, E., 1999. Modeling developed coastal watersheds with the
agricultural non-point source model. Journal of the American Water Resources
Association 35(2), 233–244.
Clarke, K., Gaydos, L., 1998. Loose-coupling a cellular automaton model and GIS:
long term urban growth prediction for San Francisco and Washington/Baltimore.
International Journal of Geographical Information Science 12(7), 699–714.
Cohen, S., 2010. The mARM3D spatially distributed soil evolution model:
three–dimensional model framework and analysis of hillslope and landform
responses. Journal of Geophysical Research 115, 16.
Colby, J.D., 1991. Topographic normalization in rugged terrain. Photogrammetric
Engineering and Remote Sensing 57(5), 531–537.
Colby, J.D., Mulcahy, K.A., Wang, Y., 2000. Modeling flooding extent from
Hurricane Floyd in the coastal plains of North Carolina. Global Environmental
Change Part B: Environmental Hazards 2(4), 157–168.
Crist, E.P., Ciccone, R.C., 1984. A physically-based transformation of Thematic
Mapper data – the TM Tasseled Cap. IEEE Transactions in Geoscience and
Remote Sensing GE-22, 256–263.
Dobson, J.D., Bright, E.A., 1991. CoastWatch – detecting change in coastal
wetlands. Geological Information Systems 1, 36–40.
Dragut, L., Blaschke, T., 2006. Automated classification of landform elements using
object-based image analysis. Geomorphology 81, 330–344.
Dymond, C.C., Mladenoff, D.J., Radeloff, V.C., 2002. Phenological differences in
tasseled cap indices improve deciduous forest classification. Remote Sensing of
Environment 80, 460–472.
Efe, R., Tagil, S., 2008. The use of multitemporal and multispectral Landsat data to
determine change detection around Tuz Lake on Seyhan Delta. Fresenius
Environmental Bulletin 17(5), 577–583.
Foody, G.M., Mathur, A., 2004. Toward intelligent training of supervised image
classifications: directing training data acquisition for SVM classification. Remote
Sensing of Environment 93, 107–117.
Forman, R.T.T., 1997. Land Mosaics: The Ecology of Landscapes and Regions.
Cambridge University Press, Cambridge, 632 pp.
Frank, T.D., 1988. Mapping dominant vegetation communities in the Colorado Front
Range with Landsat Thematic Mapper and digital terrain data. Photogrammetric
Engineering & Remote Sensing 54(12), 1727–1734.
French, N.R., 1986. Hierarchical conceptual model of the alpine geosystem. Arctic
and Alpine Research 18, 133–146.
Frohn, R.C., 1997. Remote sensing for landscape ecology: new metric indicators for
monitoring. Modeling, and Assessment of Ecosystems. Lewis Publishers, Boca
Raton, FL, 99 pp.
Galicia, L., Zarco-Arista, A.E., Mendoza-Robles, K.I., Palacio-Prieto, J.L., Garcia-
Romero, A., 2008. Land use/cover, landforms and fragmentation patterns in a
tropical dry forest in the southern region of Mexico. Singapore Journal of
Tropical Geography 29(2), 137–154.
Geist, H.J., Lambin, E.F., 2001. What Drives Tropical Deforestation?. LUCC
International Project Office, Louvain-la-Neuve, Belgium, LUCC Report Series No.4.
Genovese, G., Vignolles, C., Negre, T., Passera, P., 2001. A methodology for a
combined use of normalized vegetation index and CORINE land cover data for
crop yield monitoring and forecasting: A case study on Spain. Agronomie 21(1),
91–111.
Goetz, S.J., Wright, R.K., Smith, A.J., Zinecker, E., Schaub, E., 2003. IKONOS
imagery for resource management: Tree cover, impervious surfaces, and riparian
buffer analysis in the mid-Atlantic region. Remote Sensing of Environment 88,
195–208.
Grossman, D.H., Faber-Langendoen, D., Weakley, A.S., et al., 1998. International
classification of ecological communities: terrestrial vegetation of the United
States. Volume I. The National Vegetation Classification System: development,
status, and applications. The Nature Conservancy, Arlington, VA.
Gustafson, E.J., 1998. Quantifying landscape spatial pattern: what is the state of the
art? Ecosystems 1, 143–156.
Gutman, G., Janetos, A.C., Justice, C.O. (Eds.), 2004. Land Change Science:
Observing, Monitoring and Understanding Trajectories of Change on Earth’s
Surface. Kluwer Academic Publishers, Dordrecht.
Heilig, G.K., 1994. Neglected dimensions of global land-use change: reflections and
data. Population and Development Review 20(4), 831–859.
Hepner, G.E., Houshmand, B., Kulikov, I., Bryant, N., 1998. Investigation of the
integration of AVIRIS and IFSAR for urban analysis. Photogrammetric
Engineering and Remote Sensing 64(8), 813–820.
Hobbs, R.J., 1994. Dynamics of vegetation mosaics: can we predict responses to
global change? Ecoscience 1, 346.
Hodgson, M.E., Jensen, J.R., Tullis, J.A., Riordan, K.D., Archer, C.M., 2003.
Synergistic use of lidar and color aerial photography for mapping urban parcel
imperviousness. Photogrammetric Engineering and Remote Sensing 69(9),
973–980.
Hofle, B., Vetter, M., Pfeiffer, N., Mandlburger, G., Stotter, J., 2009. Water surface
mapping from airborne laser scanning using signal intensity and elevation data.
Earth Surface Processes and Landforms 34(12), 1635–1649.
Horsch, B., 2003. Modelling the spatial distribution of montane and subalpine
forests in the central Alps using digital elevation models. Ecological Modelling
168(3), 267–282.
Intergovernmental Panel on Climate Change (IPCC), 2007. Climate Change of 2007:
Impacts, Adaptation and Vulnerability. Working Group II Contribution to the
Intergovernmental Panel on Climate Change Fourth Assessment Report. 23 p.
Irvin, B.J., Ventura, S.J., Slater, B.K., 1997. Fuzzy and isodata classification of
landform elements from digital terrain data in Pleasant Valley, Wisconsin.
Geoderma 77, 137–154.
Jensen, J.R., Cowen, D., Althausen, J.D., Narumalani, S., Weatherbee, O., 1993a.
An evaluation of CoastWatch change detection protocol in South Carolina.
Photogrammetric Engineering & Remote Sensing 59(6), 1039–1046.
Jensen, J.R., Cowen, D.J., Althausen, J.D., Narumalani, S., Weatherbee, O., 1993b.
The detection and prediction of sea level changes on coastal wetlands using
satellite imagery and a geographic information system. Geocarto International
8(4), 87–98.
Jensen, J.R., Rutchey, K., Koch, M.S., Narumalani, S., 1995. Inland wetland change
detection in the Everglades Water Conservation Area 2A using a time series of
normalized remotely sensed data. Photogrammetric Engineering & Remote
Sensing 61(2), 199–209.
Jurgens, C., Fander, M., 1993. Soil erosion assessment and simulation by means
of SGEOS and ancillary digital data. International Journal of Remote Sensing
14(15), 2847–2855.

Klemas, V.V., Dobson, J.E., Ferguson, R.L., Haddad, K.D., 1993. A coastal land
cover classification system for the NOAA Coastwatch Change Analysis Project.
Journal of Coastal Research 9(3), 862–872.
Kusimi, J.M., 2008. Stream processes and dynamics in the morphology of the
Densu river channel in Ghana. International Archives of the Photogrammetry,
Remote Sensing, and Spatial Information Sciences 37, 1177–1182.
Lambin, E.F., Turner, II B.L., Geist, H.J., et al., 2001. The causes of land-use and
land-cover change: moving beyond the myths. Global Environmental Change 11,
261–269.
Lambin, E.R., Geist, H.J. (Eds.), 2006. Land-Use and Land-Cover Change, Local
Processes and Global Impacts. Springer, Berlin.
Lang, M.W., Pittman, K.W., Kasischke, E.S., Prince, S.D., 2008. Assessment of
C-band synthetic aperture radar data for mapping coastal plain forested wetlands
in the Mid-Atlantic Region. Remote Sensing of Environment 112, 4120–4135.
Lee, C.T., Marsh, S.E., 1995. The use of archival Landsat MSS and ancillary data in
a GIS environment to map historical change in an urban riparian habitat.
Photogrammetric Engineering & Remote Sensing 61(8), 999–1008.
Lesschen, J.P., Verburg, P.H., Staal, S.J., 2005. Statistical Methods for Analysing
the Spatial Dimension of Changes in Land Use and Farming Systems. LUCC
Focus 3 Office, Wageningen University, Wageningen, The Netherlands, LUCC
Report Series No. 7.
Li, H., Reynolds, J.F., 1997. Modeling effects of spatial pattern, drought, and
grazing on rates of rangeland degradation: a combined Markov and cellular
automaton approach. In: Quattrochi, D.A., Goodchild, M.F. (Eds.), Scale in
Remote Sensing and GIS. Lewis Publishers, Boca Raton, FL, pp. 211–230.
Liverman, D., Moran, E.F., Rindfuss, R.R., Stern, P.C. (Eds.), 1998. People and
Pixels: Linking Remote Sensing and Social Science. National Academy Press,
Washington, DC.
Liverman, D.M., Cuesta, R., 2008. Human interactions with the Earth system: people
and pixels revisited. Earth Surface Processes and Landforms 33(9), 1458–1471.
Loveland, T.R., Merchant, J.W., Ohlen, D.O., Brown, J.F., 1991. Development of a
land-cover characteristics database for the conterminous U.S. Photogrammetric
Engineering and Remote Sensing. 57(11), 1453–1463.
Loveland, T.R., Reed, B.C., Brown, J.F., Ohlen, D.O., Zhu, J, Yang, L., Merchant,
J.W., 2000. Development of a global land cover characteristics database and
IGBP DISCover from 1-km AVHRR data. International Journal of Remote
Sensing. 21(6/7), 1303–1330.
Lunetta, R.S., Johnson, D.M., Lyon, J.G., Crotwell, J., 2004. Impacts of temporal
frequency on land-cover change detection monitoring. Remote Sensing of
Environment. 89, 444–454.
Lunetta, R.S., Lyon, J.G., Guindon, B., Elvidge, C.D., 1998. North American
landscape characterization dataset development and data diffusion.
Photogrammetric Engineering & Remote Sensing 64(8), 821–829.
Malczewski, J., 2006. GIS-based multicriteria decision analysis: a survey of the
literature. International Journal of Geographical Information Science 20(7), 703–726.
Mariolakos, I., Fountoulis, I, Logos, E., 1997. The crucial role of the neotectonic
deformation at the landfill site selection. The case study of Avlona (Attiki,
Greece). Proceedings, International Association for Engineering Geology 2,
Balkema, Rotterdam, pp. 2007–2010.
Martinez, J.M., Le Toan, T., 2007. Mapping of flood dynamics and spatial
distribution of vegetation in the Amazon floodplain using multitemporal SAR
data. Remote Sensing of Environment 108(3), 209–223.
Mather, P.M., 1999. Land cover classification revisited. In: Atkinson, P.M., Tate, N.J.
(Eds.), Advances in GIS Analysis. Wiley and Sons, Chichester, pp. 7–16.
McGarigal, K., Marks, B.J., 1993. FRAGSTATS: Spatial Pattern Analysis Program for
Quantifying Landscape Structure, Forest Sciences Department, Oregon State
University, Corvallis, OR.
McMahon, G., Bales, J.D., Coles, J.F., Giddings, E.M.P., Zappia, H., 2003. Use of
stage data to characterize hydrologic conditions in an urbanizing environment.
Journal of the American Water Resources Association 39(6), 1529–1546.
Medley, K., McDonnell, M.J., Pickett, S.T.A., 1995. Forest-landscape structure along
an urban-to-rural gradient. Professional Geographer 47(2), 159–168.
Michalski, F., Norris, D., Peres, C.A., 2010. No return from biodiversity loss.
Science 329, 1282.
Michener, W.K., Blood, E.R., Bildstein, K.L., Brinson, M.M., 1997. Climate change,
hurricanes and tropical storms, and rising sea level in coastal wetlands.
Ecological Applications. 7(3), 770–801.
Moody, A., Strahler, A.H., 1994. Characteristics of composited AVHRR data and
problems in their classification. International Journal of Remote Sensing 15(17),
3473–3491.
Murthy, K.S.R., Mamo, A.G., 2009. Multi-criteria decision evaluation in groundwater
zones identification in Moyale–Teltele subbasin, South Ethiopia. International
Journal of Remote Sensing 30, 2729–2740.
Remote Sensing of Land Cover Dynamics 101
Nagarajan, M., Singh, S., 2009. Assessment of groundwater potential zones using
GIS techniques. Journal of the Indian Society of Remote Sensing 37(1), 69–77.
National Research Council, 2001. Grand Challenges in Environment Sciences.
National Academies Press, Washington, DC.
NOAA CSC, 2010. National Oceanographic and Atmospheric Administration, Coastal
Services Center. Nonpoint Source Pollution and Erosion Comparison Tool (N-
SPECT) and Impervious Surface Analysis Tool (ISAT). http://www.csc.noaa.gov/
digitalcoast/ (last accessed 31 May 2010).
Odeh, I.O.A., McBratney, A.B., Chittleborough, D.J., 1992. Soil pattern recognition
with fuzzy-c-means: application to classification and soil–landform
interrelationships. Soil Science Society of America Journal 56(2), 505–516.
Pande, R.K., Bunnan, D., Singh, R., 2009. Landslide hazard zonation in Hanuman
Chatti area of Uttarakhand, India. Disaster Prevention and Management 18(4),
410–417.
Parker, D.C., Berger, T., Manson, S.M. (Eds.), 2002. Agent-Based Models of Land-
Use and Land-Cover Change. Indiana University, LUCC Focus 1 Office, LUCC
Report Series No. 6.
Phippen, S.J., Wohl, E., 2003. An assessment of land use and other factors
affecting sediment loads in the Rio Puerco watershed, New Mexico.
Geomorphology 52(3–4), 269–287.
Pradhan, B., Lee, S., 2010. Landslide susceptibility assessment and factor effect
analysis: backpropagation artificial neural networks and their comparison with
frequency ratio and bivariate logistic regression modelling. Environmental
Modeling & Software 25(6), 747–759.
Praskievicz, S., Chang, H., 2009. A review of hydrologic modeling of basin-scale
climate change and urban development impacts. Progress in Physical Geography
33(5), 650–671.
Reed, B.C., Brown, J.F., VanderZee, D., Loveland, T.R., Merchant, J.W., Ohlen, D.O.,
1994. Variability of land cover phenology in the United States. Journal of
Vegetation Science v. 5, 703–714.
Renard, K.G., Foster, G.R., Weesies, G.A., McCool, D.K., Yoder, D.C., 1997.
Predicting Soil Erosion by Water: A Guide to Conservation Planning With the
Revised Universal Soil Loss Equation (RUSLE). U.S. Department of Agriculture,
Washington, D.C, Agricultural Handbook 703.
Ridd, M.K., 1995. Exploring a V-I-S (vegetation-impervious surface-soil) model for
urban ecosystem analysis through remote sensing: comparative anatomy for
cities. International Journal of Remote Sensing 16(12), 2165–2185.
Rindfuss, R.R., Walsh, S.J., Turner, II B.L., Fox, J., Mishra, V., 2004. Developing a
science of land change: challenges and methodological issues. Proceedings of
the National Academy of Sciences of the United States of America 101(39),
13976–13981.
Saco, P.M., Willgoose, G.R., Hancock, G.R., 2006. Spatial organization of soil depths
using a landform evolution model. Journal of Geophysical Research 111, 16.
Savitzky, B.G., Ogneva-Himmelberger, Y., Chowdhury, R., 1998. The use of multi-
temporal image classification in modeling land use and land cover change: a
case study in the southern Yucatn Peninsular region. Proceedings of Annual
Meeting of the American Society for Photogrammetry and Remote Sensing,
Tampa, FL (cd-rom.).
Sawaya, K.E., Olmanson, L.G., Heinert, N.J., Brezonik, P.L., Bauer, M.E., 2003.
Extending satellite remote sensing to local scales: land and water resource
monitoring using high-resolution imagery. Remote Sensing of Environment 88,
144–156.
Schmidt, J., Hewitt, A., 2004. Fuzzy land element classification from DTMs based
on geometry and terrain position. Computers and Geosciences 121, 243–256.
Schonbrodt, S., Behrens, T., Imbery, S., Scholten, T., 2010. Soil erosion modeling
in terraced landscapes – examples from the Three Gorges area, China. 19th
World Congress of Soil Science, Soil Solutions for a Changing World. Brisbane,
Australia 1–6, 228–231,August.
Scott, J.M., Tear, T.H., Davis, F.W. (Eds.), 1996. Gap Analysis: A Landscape
Approach to Biodiversity Planning. American Society for Photogrammetry and
Remote Sensing, ASPRS Distribution Center. Annapolis Junction, Maryland.
Seeber, C., Hartmann, H., Xiang, W., King, L., 2010. Land use change and causes
in the Xiang catchment, Three Gorges Area, derived from multispectral data.
Journal of Earth Science 21(6), 846–855.
Shao, G., Young, D.R., Porter, J.H., Hayden, B.P., 1998. An integration of remote
sensing and GIS to examine the responses of shrub thicket distributions to
shoreline changes on Virginia barrier islands. Journal of Coastal Research
14(1), 299–307.
Smith, L.C., 2002. Emerging applications of interferometric synthetic aperture radar
(InSAR) in geomorphology and hydrology. Annals of the Association of
American Geographers 92(3), 385–398.
Tarboton, D.G., 2004. Terrain Analysis Using Digital Elevation Models (TauDEM.)
http://hydrology.neng.usu.edu/taudem/ (accessed May 31, 2010).
102 Remote Sensing of Land Cover Dynamics

Tarboton, D.G., 1997. A new method for the determination of flow directions and
contributing areas in grid digital elevation models. Water Resources Research
33(2), 309–319.
Thomas, J.P., 1995. Remote sensing and relating coastal development to living
marine resources and their habitats. Natural Areas Journal 15, 21–36.
Turner, M.G., 1988. A spatial simulation model of land use changes in a piedmont
county in Georgia. Applied Mathematics and Computation 27, 39–51.
Turner, S.J., O’Neill, R.V., Conley, W., Conley, M.R., Humphries, H.C., 1991. Pattern
and scale: statistics for landscape ecology. In: Turner, M.G., Gardner, R.H. (Eds.),
Quantitative Methods in Landscape Ecology. Springer-Verlag, New York, pp. 17–49.
Turner, II B.L., Hyden, G., Kates, R.W. (Eds.), 1993. Population Growth and
Agricultural Change in Africa. University of Florida Press, Gainesville.
Uemaa, E., Roosaare, J., Mander, U., 2005. Scale dependence of landscape metrics
and their indicatory value for nutrient and organic matter losses from
catchments. Ecological Indicators 5(4), 350–369.
U.S. Geological Survey (USGS), 2005. Geography for a Changing World: A Science
Strategy for the Geographic Research of the U.S. Geological Survey, 2005–2015,
Circular 1281. U.S. Department of Interior, U.S. Geological Survey, Reston, VA.
U.S. Geological Survey (USGS), 2010. The USGS Land Cover Institute (LCI). Web
access: http://landcover.usgs.gov/ (last accessed May 11, 2010).
Wainright, J., Millington, J.D.A., 2010. Mind, the gap in landscape-evolution
modelling. Earth Surface Processes and Landforms 35(7), 842–855.
Wainwright, J., 2008. Can modeling enable us to understand the role of humans in
landscape evolution? Geoforum 39(2), 659–674.
Wang, Y., Colby, J.D., Mulcahy, K.A., 2002. An efficient method for mapping flood
extent in a coastal floodplain by integrating Landsat TM and DEM data.
International Journal of Remote Sensing 23(18), 3681–3696.
Wang, Y., 2004. Seasonal change in the extent of inundation on floodplains
detected by JERS–1 synthetic aperture radar data. International Journal of
Remote Sensing 25(13), 2479–2508.
Wang, Y., Allen, T.R., 2008. Estuarine shoreline change detection using Japanese
ALOS PALSAR HH and JERS–1 L–HH SAR data in the Albemarle–Pamlico
Sounds, North Carolina, USA. International Journal of Remote Sensing 29(15),
4429–4442.
Wang, Y., Liao, M., Wang, C., 2009. Applications and analyses of satellite-borne
L-band synthetic aperture radar data in coastal environments. Geography
Compass 3(4), 1465–1482.
Webb, N.P., Phinn, S.R., McGowan, H.A., 2009. Visual assessment of the Australian
Land Erodibility Model. Journal of Arid Environments. 73(6–7), 678–682.
Weng, Q. (Ed.), 2008. Remote Sensing of Impervious Surfaces. CRC Press, Taylor
& Francis, Boca Raton, FL, 454 pp.
Wing, M.G., Skaugset, A., 2002. Relationships of channel characteristics, land
ownership, and land use patterns to large woody debris in western Oregon
streams. Canadian Journal of Fisheries and Aquatic Sciences 59(5), 796–807.
Xu, J.C., Wilkes, A., 2004. Biodiversity impact in northwest Yunnan, southwest
China. Biodiversity and Conservation 13, 959–983.
Zheng, T., Wang, Y., 2007. Mapping flood extent using a simple DEM-inundation
model. The North Carolina Geographer 15, 1–19.

Biographical Sketch
Dr. Thomas R. Allen is Associate Professor of Geography at East Carolina University. Allen holds a BS degree in
geography from Old Dominion University (Norfolk, VA, USA) and a PhD from the University of North Carolina
at Chapel Hill. His research focuses on applied environmental applications of remote sensing and geographic
information systems, emphasizing coastal, biogeographic, and geomorphologic applications and environmental
hazards. He is a past Fulbright Scholar (Finland) and currently chairs the Coastal and Marine Specialty Group of
the Association of American Geographers.

Dr. Yong Wang obtained his BS in electric engineering from the Northwestern Polytechnic University, China, and
MA and PhD in geography with a focus on synthetic aperture radar (SAR) and remote sensing from the University
of California, Santa Barbara, CA, USA. Current research interests include the application of remote sensing and
geographic information systems to coastal areas and natural hazards, and development of imaging and appli-
cation algorithms for air- and space-borne SAR. He is a professor of Geography, East Carolina University, NC,
USA.

Dr. Thomas W. Crawford is Associate Professor of Geography at East Carolina University. Crawford holds a BS
degree in Mathematical Economics from Wake Forest University (Winston-Salem, NC, USA) and a PhD in
Geography from the University of North Carolina at Chapel Hill. His research focuses on GIScience and applied
human-environment applications of geospatial technologies including land-use analysis, coastal development,
environmental hazards, and spatial accessibility. He is currently the president of the North Carolina Geographical
Society and is a member of the Association of American Geographers.
 3.5 Near-Surface Geophysics in Geomorphology
S Kruse, University of South Florida, Tampa, FL, USA
r 2013 Elsevier Inc. All rights reserved.

3.5.1 Introduction 104

3.5.2 Gravity 105
3.5.2.1 Strengths 106
3.5.2.2 Limitations 106
3.5.2.3 Gravity Case Studies 106
3.5.2.3.1 Faulting 106
3.5.2.3.2 Volcanic processes 106
3.5.2.3.3 Topography and weathering 107
3.5.2.3.4 Slope processes 109
3.5.2.3.5 Coastal processes and sea-level change 109
3.5.2.3.6 Glacial and periglacial processes 109
3.5.3 Magnetics 109
3.5.3.1 Strengths 109
3.5.3.2 Limitations 110
3.5.3.3 Case Studies 110
3.5.3.3.1 Faulting 110
3.5.3.3.2 Volcanic processes 110
3.5.3.3.3 Topography and weathering 110
3.5.3.3.4 Glacial and periglacial processes 110
3.5.4 Resistivity and EM Methods 110
3.5.4.1 Strengths 112
3.5.4.2 Limitations 112
3.5.4.3 Case Studies 112
3.5.4.3.1 Faulting 112
3.5.4.3.2 Volcanic processes 112
3.5.4.3.3 Topography and weathering 112
3.5.4.3.4 Slope processes 112
3.5.4.3.5 Coastal processes 113
3.5.4.3.6 Fluvial processes 113
3.5.4.3.7 Glacial and periglacial processes 113
3.5.5 Ground-Penetrating Radar 113
3.5.5.1 Strengths 115
3.5.5.2 Limitations 115
3.5.5.3 Case Studies 115
3.5.5.3.1 Faulting 115
3.5.5.3.2 Volcanic processes 115
3.5.5.3.3 Topography and weathering 115
3.5.5.3.4 Slope processes 117
3.5.5.3.5 Coastal processes and sea-level change 119
3.5.5.3.6 Aeolian processes 119
3.5.5.3.7 Fluvial processes 119
3.5.5.3.8 Glacial and periglacial processes 119
3.5.6 Seismic Methods 119
3.5.6.1 Strengths 121
3.5.6.2 Limitations 121
3.5.6.3 Case Studies 121
3.5.6.3.1 Faulting 121
3.5.6.3.2 Volcanic processes 122
3.5.6.3.3 Topography and weathering 122

Kruse, S., 2013. Near-surface geophysics in geomorphology. In: Shroder, J.

(Editor in Chief), Bishop, M.P. (Ed.), Treatise on Geomorphology. Academic
Press, San Diego, CA, vol. 3, Remote Sensing and GIScience in
Geomorphology, pp. 103–129.

Treatise on Geomorphology, Volume 3 http://dx.doi.org/10.1016/B978-0-12-374739-6.00047-6 103

104 Near-Surface Geophysics in Geomorphology

3.5.6.3.4 Slope processes 122

3.5.6.3.5 Coastal processes and sea-level change 122
3.5.6.3.6 Fluvial processes 122
3.5.6.3.7 Glacial and periglacial processes 122
3.5.7 Combining Geophysical Methods 122
3.5.8 Discussion and Conclusions 123
3.5.8.1 Practical Considerations 123
3.5.8.2 Integrating Geophysics, Remote Sensing, and Geographic Information Systems 125
References 125

Abbrevations GOCE gravity field and steady-state ocean

CMP common mid-point circulation explorer
CSAMT controlled source audio-magnetotellurics GRACE gravity recovery and climate experiment
DEM digital elevation model InSAR interferometric synthetic aperture radar
dGPS differential global positioning system LiDAR light detection and ranging
EM electromagnetic MT magnetotelluric
ERS-1 European remote-sensing satellite 1 Seasat sea satellite
FEM frequency-domain electromagnetics TEM, TDEM time–domain electromagnetics
Geosat geodetic satellite TLS terrestrial laser scanning

Abstract

Near-surface geophysical methods can provide information on subsurface structure and stratigraphy that is critical to
understanding surficial processes. Gravity, magnetics, resistivity, electromagnetics, ground-penetrating radar, and various
seismic methods are applied across a range of process domains, including faulting, volcanism, topography and weathering,
hillslope processes, coastal and sea-level change, aeolian and fluvial processes, and glacial and periglacial processes. Gravity
and magnetic methods have long been used to image faults and other tectonic and volcanic features, and they have also
been used to document weathering patterns. With the development of multi-electrode resistivity systems, resistivity pro-
filing has become a staple tool to map the structure and water content of slope deposits and volcanic features. Ground-
penetrating radar studies have been crucial to understanding of geologic features as varied as the internal structure of
aeolian dunes, coastal and fluvial deposits, the structure and volume of glaciers, the spatial distribution of sinkholes, and
the geometry of tephra and lahar deposits. Key to successful studies are calibrations against direct observations and/or uses
of multiple complementary methods. Increased efficiencies in geophysical data acquisition and positioning in recent years
have made possible very-high-resolution three-dimensional (3-D) or quasi-3-D imaging of subsurface structures. However,
there is still a mismatch in the typical spatial scales of large remote-sensing-based studies and smaller geophysical surveys.
Bridging this disconnect with new geophysical acquisition techniques and new instrumentation such as terrestrial laser
scanning should improve our understanding of the role that subsurface structure plays in the evolution of topography.

3.5.1 Introduction
Near-surface geophysical methods are invaluable to geo-
morphology investigations where knowledge of subsurface
structure is important for understanding surface processes and
tectonics. The internal structure of landforms generally pro-
vides insight into the evolution of paleo-topography and
paleo-landforms. For example, gravity and magnetic surveys
have been used for over a century both to understand mech-
anisms of isostatic compensation of topography and to esti-
mate fault offsets. Much more recently (in the past 20 years),
ground-penetrating radar imaging of the internal structure of
various types of dunes has significantly advanced our know-
ledge of dune formation and evolution. In both cases, near-
surface structures could simply not be adequately resolved
from direct observations, and surficial or borehole data.
Geophysical techniques offer the only practical means of
imaging structures between boreholes and outcrops, and ex-
tending imaging into the third dimension.
There is some confusion about the meaning of ‘near
surface’ in geophysics. To many practitioners, near surface
implies material in the uppermost B100 m. This is a use-
ful definition because numerous methods are only effective
at depths less than B100 m, and, as a result, are simply not
used in kilometer-scale or whole-Earth studies. These methods
include high-frequency electromagnetic (EM) methods,
ground-penetrating radar (except on ice), 12-V battery-
powered resistivity systems, and hammer-sourced seismic
refraction. However, some methods applicable to studying the
uppermost 100 ms can, with appropriate survey design, cap-
ture information on much deeper structure. In particular,
gravity and magnetic measurements are sensitive to hetero-
geneities from a few meters to hundreds of kilometers or more
in scale.

The key to successful use of near-surface geophysics in
geomorphology involves verifying the interpretation of geo-
physical data by using multiple sources of information. This
requires collection of in situ data for calibration or validation
via cores, trenches, outcrops, or geologic field observations, or
the use of multiple geophysical methods that yield comple-
mentary information. Successful studies increase data quality
by attempting to achieve high spatial resolution and by
choosing targets that produce strong geophysical signals (the
result of either careful reconnaissance or good luck). Pro-
ductive strategies include field collection of rock and sediment
samples and subsequent laboratory analyses to determine
relevant physical properties, and development and modifi-
cation of data-processing routines to reduce site-specific noise
and enhance target signal. Most positive factors add cost, in
terms of either data acquisition (e.g., high-resolution airborne
studies and multiple methods) or post-processing labor
(careful merging of separate data sets and development and
testing of nonstandard processing algorithms).
Because there is no forum in geophysics for describing sites
where methods, that might be expected to work, in fact do not
work, it is difficult to estimate what fraction of attempts to
use a given geophysical method yields results productive
enough to be included in geomorphological research and
publications.
This chapter focuses on the most commonly used geo-
physical techniques in practical terms, based upon the use of
simple physics, relationships between measured parameters
and soil/rock properties, survey requirements, and inherent
strengths and limitations. Conditions both likely and unlikely
to lead to productive results are discussed. The emphasis on
recent studies gives the reader an easy path to data sets that
exemplify what can be achieved with current equipment and
acquisition and processing methods. The uses of methods are
described on large and small scales, reflecting common usage.
Examples are presented that utilize multiple techniques in
successful combinations.
3.5.2 Gravity
The strength of the Earth’s gravity field at a given point on the
surface is sensitive to a weighted sum of density variations in
the near and far subsurface. Thus, gravity surveys can be useful
in settings where high-density materials, such as igneous or
metamorphic basement, are in contact with much lower-
density materials, such as unconsolidated sediments or ice.
Depth to basement is perhaps the most common target of
gravity surveys. Measuring the gravity field at a given point
simply involves placing a gravity meter on the ground at the
desired site and following the necessary steps to acquire an
instrument reading. Acquiring high-quality gravity data,
however, requires care and ancillary data.
Some key aspects of the gravity method can be illustrated
with the analytical expression for the extremely simplified case
of a buried spherical anomaly. In this case
G4pr 2 Dr
gsphere ¼ ½1
3ðx2 þ z2 Þ
Near-Surface Geophysics in Geomorphology 105
−
+
Gravity anomaly
Figure 1 Schematic illustration of gravity anomalies over a 2-D
positive-density anomaly (red) and negative anomaly, such as a void
(orange area). (Top) Cross section of subsurface structure. (Bottom)
Gravity anomalies measured on a profile over these features. Red line
shows gravity anomalies that would be measured in the absence of
the orange void. The Red positive-density anomaly is assumed to
extend to great depth and past the end of the profile. Note the deep
positive anomaly produces a long-wavelength signal with a high over
the mass. The black line represents values that would be measured
over the combined features. Note that the effect of the shallow
negative-density anomaly is to depress gravity readings across a
shorter-wavelength region centered over the density anomaly. The
schematic emphasizes that gravity is sensitive to a range of depths
and that deeper sources produce broader anomalies.
where gsphere is the gravity anomaly measured at a point at the
surface that is a horizontal distance, x, from the center of the
sphere, z is the depth of the sphere center, G is the gravi-
tational constant, r is the radius of the sphere, and Dr is the
density contrast between the sphere and the host rock. Gravity
anomalies are intuitive: excess mass (positive Dr) produces
positive density anomalies that peak over the source mass –
the opposite for mass deficits (Figure 1). The gravity anomaly
weakens as 1/distance2 ( ¼ 1/(x2 þ z2)) to the excess mass.
Deeper targets produce lower-amplitude anomalies and
longer-wavelength anomalies. It is important to note that the
horizontal distance x must be on the order of z or greater
before gsphere decays significantly. So, the greater the target
depth, the longer the survey dimensions needed to capture the
gravity signal. In general, a survey dimension of Z5 times the
target depth is desirable.
Gravity signals are also inherently nonunique. The density
contrast Dr and the target dimension r can trade off in eqn [1]:
small, high-contrast anomalies can produce identical signals
to larger, lower-contrast features. Because the gravity anomaly
measured at any one point in the subsurface represents the
sum of the effects of all local and regional density anomalies
(similar to the sum of many gspheres), an infinite number of
density patterns could produce any given gravity profile. In-
terpretation of gravity data is therefore only useful when other
constraints are applied. Well data, seismic profiles, and field
observations can show where density contrasts do exist.
106 Near-Surface Geophysics in Geomorphology
Simplifying assumptions are generally made about density
patterns, including a limited number of layers of features.
Practitioners generate a series of trial-and-error models of
Earth density structures, honing in on plausible geometries
that are compatible with the observations. The method is
called ‘forward modeling’, and, to keep the modeling process
tractable, most surveys are fit with two-dimensional (2-D)
models that make simple assumptions about the extent of
features in the third dimension. Consequently, gravity surveys
should be run perpendicular to the expected strike of features
of interest. Advanced software can automate the process of
finding a geometry that best fits the observations, but users are
still required to limit the number of unknown variables in the
problem by making significant assumptions about the starting
model. This latter process is called ‘inverse modeling’.
Modern gravimeters are capable of measuring variations in
the Earth’s gravity field with a precision of 0.001–0.01 mGal
(1 mGal ¼ 10–5 m s–2). To illustrate the sensitivity, a spherical
void in limestone (density 2000 kg m–3) at 10-m depth, with
radius 2.6 m, produces an anomaly of 0.01 mGal directly over
the sphere. A significant number of corrections are needed to
reduce a gravity measurement to a value that reflects only
subsurface density anomalies. Data need to be corrected for
elevation (gravity weakens with increasing distance from the
center of the Earth) and the effects of the gravitational at-
traction of local and regional topography. These corrections
can be large compared to target anomalies. A difference in
5 cm in elevation corresponds to a correction on the order of
the instrument sensitivity (0.01 mGal); hence, gravity meas-
urements must typically be made at benchmarks or at ele-
vations surveyed with differential global positioning system
(dGPS) equipment. Instrument drift and the absolute value of
instrument readings have to be calibrated by acquisition of
data at known base stations. In practice, several minutes are
required for each high-quality gravity-plus-elevation meas-
urement, and including trips to base stations, at most a few to
tens of data points can typically be collected per day.
Aerogravity surveys can acquire data at much higher rates
and can be extremely useful for reconnaissance surveys in
inaccessible settings. The high cost of such surveys and data-
processing issues, however, limits widespread applicability.
Satellites such as Seasat (Sea Satellite), Geosat (Geodetic Sat-
ellite), ERS-1 (European Remote-Sensing Satellite 1), GRACE
(Gravity Recovery and Climate Experiment), and GOCE
(Gravity Field and Steady-State Ocean Circulation Explorer)
have measured the gravity field over continent and ocean-
basin scales.
3.5.2.1 Strengths
Gravity is widely used as a reconnaissance tool, because in-
formation about relatively deep structure (hundreds of meters
to kilometers or more) can be derived from surficial obser-
vations at much lower cost than seismic methods. Small-scale
near-surface features (uppermost tens of meters) produce
smaller gravity anomalies, requiring extreme care in elevation
measurements and data processing. These high-resolution
surveys are commonly referred to in the literature as micro-
gravity surveys, as they require precision on the order of
microgals (0.001 mGal). Despite the labor required, surveys
targeting these smaller anomalies are yielding fruitful new
information in studies of weathering processes and volcano
evolution.
3.5.2.2 Limitations
The nonuniqueness of the gravity method limits definitive
interpretation of results in the absence of other direct or
sensor observations. Gravity is best used in combination with
other geophysical measurements. Automobile traffic causing
ground vibrations can render data acquisition difficult in
daytime in urban environments. Many local surveys are ref-
erenced to a local base station and cannot be easily in-
corporated into larger databases.
3.5.2.3 Gravity Case Studies
3.5.2.3.1 Faulting
Gravity profiles across faults, particularly those that juxtapose
basement against less dense sediments, have been a staple of
tectonic studies for decades (e.g., Telford et al., 1990). Gravity
data are typically used in conjunction with field observations,
and with magnetic and/or seismic data. Here, we illustrate
gravity methods with results of a combined magnetic, gravity,
and light detection and ranging (LiDAR) topography analysis
by Blakely et al. (2009) for faulting on scales of tens of kilo-
meters (Figure 2). This study demonstrates the utility of
gravity mapping in conjunction with magnetic surveys and
common interpretation techniques. The authors compute
isostatic residual gravity anomalies (i.e., they assume that
topography is isostatically compensated and that remaining
gravity anomalies reflect middle and upper crustal density
heterogeneities). With gravity and magnetic observations, the
attitude of active faults mapped at the surface from LiDAR and
field observations can be projected downward to many kilo-
meters depth. Figure 2(a) and the corresponding cross section
on Figure 3 demonstrate classical techniques of (1) compiling
a gravity map to identify regional strike directions and (2)
developing a cross-strike 2-D model that accounts for geologic
observations, and gravity and magnetic data. It is the contrast
between higher-density Crescent Formation basalts and lower-
density surrounding sediments that produces an B75-mGal
gravity signature that can be modeled. Figure 3(c) illustrates
the complexities inherent in forward modeling of gravity data,
such as the assumptions that must be made about density
contrasts of rocks at depth, as well as the geometry of contacts.
Most authors present a best-fitting model that is based on a
conceptual structural model, as done by Blakely et al. (2009)
for Figure 3(c) from Tabor and Cady (1978). It can be illu-
minative when authors also illustrate a range of models, and
the relative quality of their fits.
3.5.2.3.2 Volcanic processes
Gravity, combined with geological observations, can provide
useful constraints on isostatic compensation mechanisms for
volcanic edifices, which influence and are in turn influenced
by erosion processes and deformation histories. Gravity data
complement surface geological information for understanding
 Near-Surface Geophysics in Geomorphology 107

−123°30′ −123° −122°30′ −123°30′ −123° −122°30′

48°

48°45′

48°30′

Susceptibility, SIU × 103 10 km

−334
−227
−184
−136

−40
−28
−16
141
396
−56

>30

12
20
29
−5
34

4
20−30
10−20
nT 0−10 mGal
(a) (b)

Figure 2 Example of combined magnetic and gravity survey. Magnetic and gravity data over the Saddle Mountain Fault area in northwest
Washington, USA. (a) Aeromagnetic anomalies. Colored dots show magnetic susceptibility measurements of outcrop samples. (b) Isostatic
residual gravity anomalies. Red dashed line indicates location of geologic cross section shown in Figure 3. White barbed lines are thrust faults.
White solid lines map outlines of exposures of the basaltic Crescent Formation between sedimentary deposits. White dotted line shows mapped
contact between lower and upper members of the Crescent Formation (see Figure 3). Coincidence of geophysical signatures and mapped faults
indicates fault structures extend to depth. Figure 3 shows forward models of fault structure. Modified from Blakely, R.J., Sherrod, B.L., Hughes,
J.F., Anderson, M.L., Wells, R.E., Weaver, C.S., 2009. Saddle Mountain fault deformation zone, Olympic Peninsula, Washington: western
boundary of the Seattle uplift. Geosphere 5(2), 105–125, with permission from GSA.

active volcano evolution (e.g., Brown et al., 1987). For ex-
ample, surveys on Concepción volcano in Nicaragua show a
remarkably low bulk density (1700 kg m–3) for the edifice as a
whole. This is caused by thick deposits of low-density scoria
that are presumed to be the underlying cause of the frequent
occurrences of lahars (Saballos et al., 2010).
Recent time-lapse gravity studies at volcanoes have gener-
ated new constraints on processes underlying surface de-
formation and unrest. Typically, surface deformation is
initially noted from surface observations or interferometric
synthetic aperture radar (InSAR). Because microgravity data
can be used to constrain mass changes at depth, in several
studies, these data are key to identification of magma as the
driving force for surface deformation (e.g., Rymer, 1994;
Budetta et al., 1999; Carbone et al., 2003; de Zeeuw-van
Dalfsen et al., 2005; Tizzani et al., 2009). Such interpretation
of volcano gravity surveys requires corrections beyond
those typically considered, including temperature and water-
table effects (Battaglia et al., 2008). Because deformation
processes at volcanoes are so rapid, investigations are
currently limited by the high costs of automatic gravity data
acquisition with high temporal resolution (Williams-Jones
et al., 2008).
3.5.2.3.3 Topography and weathering
Temporal changes in topography are observed on scales ran-
ging from sinkholes to continents. Gravity data can provide
insights into the processes driving changes on all these scales.
For instance, microgravity surveys are used to estimate mass
deficiency associated with sinkhole formation and ground
subsidence on lateral scales of meters to tens of meters (e.g.,
Calhoun et al., 1989). Branston and Styles (2003) showed
remarkably good agreement between time-lapse microgravity
surveys and resistivity surveys in a study targeting the source of
subsidence in an area of historical salt mining. The combined
data sets show a zone of low-density ground 3–4 m below
the subsiding surface. Interestingly, they documented that the
zone of low density is migrating upward over a 3-year period.
On intermediate length scales, circular gravity anomalies,
either positive or negative, can be indicators of impact struc-
tures (e.g., McCall, 2009; Dentith et al., 1999). On scales
of hundreds of kilometers or more, gravity data are key
to understanding the causes of continental-scale surface
dynamics. Gravity signatures have been used for many years
to distinguish mantle processes (e.g., Forte et al., 2010),
lithospheric flexure (e.g., Prezzi et al., 2009), and glacial iso-
static adjustment (e.g., Steffen et al., 2009).
108 Near-Surface Geophysics in Geomorphology

1000
NW SE

Magnetic Anom. (nT)

500
Observed
0

−500
Calculated
−1000
(a)

25
Gravity Anom. (mGal)

0 Calculated

−25
Observed

−50
(b)

Mapped faults Flight path

Δ = −0.12,  = 0.006
0 Δ = 0.05,  = 0

Δ = 0.20,  = 0.014 Δ = −0.01,  = 0.00

Depth (km)

Δ = 0.05,  = 0 Δ = 0.25,  = 0.088

10
Δ = 0.05,  = 0.025
Δ = 0.25,  = 0.050

Δ = 0.20,  = 0.020
20 Δ = 0.05,  = 0
Vertical exaggeration 1:1
(c)

Hurricane ridge fault Saddle Mt. fault

0
Depth (km)

Pliocene and younger

10 sedimentary deposits
Oligocene−Miocene
sedimentary rocks
Eocene crescent formation,
upper member
20
Eocene crescent formation,
lower member

0 10 20 30 40 50 60
(d) Distance (km) Eocene−Oligocene
sedimentary rocks

Figure 3 Successful forward models of gravity and magnetic data for the profile shown with red dashed line in Figure 2. Although other models
could be generated that fit the data, this model is shown because it agrees with the proposed regional tectonic structure (Tabor and Cady, 1978).
The models assume density and susceptibility structure shown in (c). (a) Model and observed magnetic anomaly. (b) Model and observed gravity
anomaly. (c) Model densities and susceptibilities. (d) Geologic interpretation of model. Dr ¼ density contrast relative to 2670 kg m–3.
w ¼ magnetic susceptibility in SI units multiplied by 1000. Note that some contacts act as sources of both magnetic anomalies and gravity
anomalies, while others only act as sources for one or the other. The overall attitude of beds proposed fits the form and amplitude of both sets
of anomalies. Reproduced from Blakely, R.J., Sherrod, B.L., Hughes, J.F., Anderson, M.L., Wells, R.E., Weaver, C.S., 2009. Saddle Mountain fault
deformation zone, Olympic Peninsula, Washington: western boundary of the Seattle uplift. Geosphere 5(2), 105–125, with permission from GSA.

3.5.2.3.4 Slope processes
Gravity surveys can constrain subsurface structures that are
related to vertical motions. For example, Gottsmann et al.
(2008) used gravity data to construct a model that explains the
position of a landslide in the Central Volcanic Complex of
Tenerife.
3.5.2.3.5 Coastal processes and sea-level change
Gravity data are key to distinguishing origins of local observed
sea-level changes. Local changes represent a combined re-
sponse to glacial isostatic adjustment, tectonic motions,
changes in ocean water temperature and salinity, and eustasy
(e.g., Milne and Mitrovica, 2008).
3.5.2.3.6 Glacial and periglacial processes
Large-scale (tens to hundreds of kilometers) airborne and
satellite gravity data are used to extract information on sub-
glacial geology (e.g., Bell et al., 1998), glacial isostatic ad-
justment (e.g., Tamisiea et al., 2007), and mass changes in ice
sheets (Riva et al., 2009). Recent satellite gravity data are of
sufficient quality that the relative magnitudes of the latter two
can be distinguished over Antarctica (Riva et al., 2009).
Gravity data have also been used as an ancillary technique for
assessing the existence and thickness of permafrost (Vonder
Muehll et al., 2000), although other geophysical methods are
typically more effective.
3.5.3 Magnetics
Magnetic methods share fundamental similarities with gravity
methods. In magnetic surveys, as in gravity surveys, the total
strength of the field is measured at points on or above the
surface, except that the measurements are sensitive to rock
magnetization rather than rock density. Igneous rocks typi-
cally have much higher intensities of magnetization than
sedimentary rocks; so, magnetic data are useful for assessing
sediment thickness over basement. Igneous dikes and sills
produce distinctive anomalies; hence, magnetic data can be
particularly useful for understanding pluton emplacement and
associated host rock deformation and erosion patterns.
Interpretation of magnetic anomalies is less intuitive than
that of gravity anomalies for several reasons. Magnetization is
inherently dipolar, and the direction of target rock magnet-
ization may be unknown. If rock magnetization is remnant,
the rock locks in the direction of the magnetic field that
existed at the time it cooled through the Curie temperature,
around 300 1C. If rock magnetization is induced, then the
magnetization direction parallels that of the present-day
Earth’s field, and the strength of the rock’s magnetization is a
function of the rock’s magnetic susceptibility. Adding further
complexity, the measured total field strength (Earth’s main
field plus anomalous field produced by study targets) depends
on both the direction and strength of the main field and the
anomalous field at the time the measurement was made.
Identical buried structures produce different magnetic anom-
alies depending on their strike, and the inclination and dec-
lination of the Earth’s field at the time of measurement
(Figure 4). Drift corrections must be made when collecting
Near-Surface Geophysics in Geomorphology 109
−
+
Magnetic anomaly
Figure 4 Schematic illustration of magnetic anomalies over a 2-D
target with higher magnetic susceptibility than background (red area),
and a target with lower susceptibility than background (orange). Red
line shows the magnetic anomalies that would be measured on an
E–W traverse across target structures striking N–S in an environment
with an Earth’s field inclination of 601. Solid black line shows the net-
magnetic anomaly from both targets. Dashed black line shows
magnetic anomalies if the scenario were rotated 901, to an N–S
profile. The schematic emphasizes that the form of the magnetic
anomalies is more complicated than gravity anomalies (compare with
Figure 1) and the form depends on the orientation of the profile and
the local direction of the Earth’s field.
magnetic data, not typically for instrument response drift, but
for diurnal variations in the Earth’s field.
Following data acquisition and processing, users typically
generate models that can fit the observations. As with gravity,
these models are inherently nonunique and must be con-
strained wherever possible by geologic observations. For
studies on scales of kilometers to hundreds of kilometers,
magnetic data are commonly interpreted in conjunction
with gravity data, and forward models are sought that fit both
data sets.
It should be noted that this chapter discusses field-based
surveys of the total magnetic field strength, and not magnetic
properties of samples measured in the laboratory. Laboratory
measurements associated with rock magnetization, however,
are used in geomorphology, including paleomagnetic meas-
urements that are used to reconstruct the timing of the for-
mation of volcanic deposits. Laboratory measurements of
anisotropy in magnetic susceptibility from field samples are
used to understand development of slip zones in landslides
(Herrero-Bervera et al., 2002; Mamtani and Sengupta, 2009).
Finally, magnetic susceptibility measurements of core samples
are common components of sedimentological studies in
coastal settings and for paleoenvironmental reconstructions.
3.5.3.1 Strengths
Unlike in gravity surveys, high-precision elevation measure-
ments are not required for ground-based acquisition of
110 Near-Surface Geophysics in Geomorphology
magnetic data. With modern magnetometers, data can be
collected with instruments carried by a single person at
walking speeds.
3.5.3.2 Limitations
Ground-based magnetic surveys can be very noisy, as natural
lithologic variability, particularly in igneous rocks, can gener-
ate large-amplitude, short-wavelength magnetic anomalies.
On the other hand, the short-wavelength variability can be
exploited as an interpretation tool, as some geologic units will
produce much more erratic magnetic data than others. In any
developed setting, cultural noise can be a significant problem
and agricultural areas can be problematic where metal is used
in irrigation systems. Finally, because of the inherent non-
uniqueness of magnetic data, other in situ data or sensor ob-
servations are essential for meaningful interpretation.
3.5.3.3 Case Studies
3.5.3.3.1 Faulting
To illustrate an effective magnetic survey, we return to the
Blakely et al. (2009) study of active faulting on the Olympic
Peninsula of Washington, USA (Figures 2 and 3). Blakely et al.
(2009) combined aeromagnetic data with more detailed sur-
face-based data acquisition, including a canoe-mounted pro-
file across a lake. The magnetic method exploits the magnetic
susceptibility contrast between basalts and sediments, and
between individual basaltic units. Direct measurements of
susceptibilities at outcrops (Figure 2) provide added con-
fidence to the magnetic forward modeling (Figure 3). It is
interesting to note that the magnetic data provide comple-
mentary information to the gravity data. For example, in the
forward model, magnetic anomalies respond to high sus-
ceptibilities in the uppermost basalt unit, whereas gravity
anomalies respond to strong density contrasts bounding the
entire basaltic package (Figure 3).
3.5.3.3.2 Volcanic processes
Magnetic surveys are effective at delineating the boundaries of
volcanic units. Because the geometry of deposits sheds light on
their deposition and emplacement mechanisms, magnetics
can be useful for understanding these processes (e.g., Diez
et al., 2009; Wetmore et al., 2009), as well as the resultant
surface topography (e.g., Morgan et al., 2003b). Zones of low
magnetization can be associated with demagnetization caused
by heat or hydrothermal activity (e.g., Okuma et al., 2009).
Identifying areas affected by hydrothermal activity may be of
practical importance because hydrothermally altered rocks can
form weakened zones with potential to form debris flows (e.g.,
Finn et al., 2007). Debris flow deposits or volcanic fans will
also have lower magnetizations than intact rock (Okubo et al.,
2009).
Thermally active volcanic zones can be associated with
temporally variable magnetic fields. Thus, monitoring of
magnetic field strength may be useful for identifying thermally
active sites (Kanda et al., 2010). Various temporal relationships
between magnetic field changes and ground deformation have
also been observed (Hurst et al., 2004; Zlotnicki et al., 2009).
With recent improvements in the resolution of surveys with
very close line spacings, time-lapse aeromagnetic methods
(repeated surveys) now can detect the subtle changes associ-
ated with volcanic activity (e.g., Ueda, 2007; Nakatsuka et al.,
2009).
3.5.3.3.3 Topography and weathering
Many weathering and soil-formation processes enhance
magnetic signals through the formation of magnetic minerals
(Evans and Heller, 2003). Surface measurements of magnetic
signal amplitude and spatial variability can be used to dis-
tinguish variability in weathering processes and rates, as well
as to characterize underlying variability in bedrock lithology
(e.g., Dimalanta et al., 2009). In addition, magnetic surveys
can provide insight on the internal structure of impact craters.
For example, Shah et al. (2009) used core-derived magnetic
properties to identify basement-derived megablocks as the
source of significant magnetic anomalies over the Chesapeake
Bay impact structure.
3.5.3.3.4 Glacial and periglacial processes
Magnetic surveys, especially aeromagnetic surveys, are used to
map subglacial lithology (e.g., Ferraccioli et al., 2009). Bell
et al. (1998) used aeromagnetic data, along with airborne
radar and gravity, to demonstrate that underlying geology in-
fluences ice-stream dynamics in the West Antarctic ice sheet.
3.5.4 Resistivity and EM Methods
The resistivity, r, of ground (r ¼ 1/conductivity) represents the
combined electrical response of matrix and pore waters.
Measurements of ground resistivity can be useful for detecting
lithologic boundaries, degree of saturation, and variations in
pore-water conductivity. Resistivities are also mildly tempera-
ture dependent, a characteristic that is exploited in volcano
studies and must be corrected for in other settings. Terrain
resistivities can be quite sensitive to the degree of fracturing
and weathering. Saturated fractured rocks are typically better
conductors than their intact equivalents; hence, their resistivity
is lower. Many clays that are a by-product of weathering
have high conductivities relative to their protoliths. Sediment–
basement contacts typically exhibit significant resistivity
contrasts.
Two different types of methods are used to measure terrain
resistivities. The first method sends a direct current through
the ground; this is called resistivity sounding, resistivity pro-
filing, electrical resistivity tomography (ERT), or electrical re-
sistivity imaging (ERI). With the second method, currents in
the ground are induced by time-varying magnetic fields; this
is known as EM methods. This second category includes the
magnetotelluric (MT) method, which exploits naturally oc-
curring time-varying magnetic fields generated over a broad
frequency range by lightning storms and other natural phe-
nomena. Other EM methods require instruments with an ac-
tive source that generates currents in the ground by creating
alternating currents in a transmitter coil. These AC currents
spawn their own time-varying magnetic field, which in turn
drives currents in the ground. The secondary magnetic field
associated with the ground currents is detected in a second
 Near-Surface Geophysics in Geomorphology 111

passive coil. These methods include controlled source
audio-magnetotellurics (CSAMT) and frequency-domain
electromagnetics (FEM or simply EM). Time-domain electro-
magnetics (TEM or TDEM) work slightly differently, as they
induce currents to flow in the ground at the moment a con-
stant current in an instrument coil is shut off.
Measurements of terrain resistivity via ERT are relatively
labor intensive. A known current is driven through the ground
between a pair of electrodes (metal stakes) and the potential
is measured between nearby electrode pairs. The method is
automated by modern instruments pre-programmed to make
selected combinations of measurements with pairs of elec-
trodes within a typically linear array of electrodes (25 to about
100) set in the ground (Figure 5). A DC current must be
established between a pair of current electrodes, and reestab-
lished in the opposite direction to avoid charge buildup on
electrodes. Because it takes a few seconds to establish each set
of DC currents, completing surveys with many tens of elec-
trodes can take up to several hours after the electrode array is
in place.
The separation of electrodes is related to the overall depth
of current flow; the greater the separation, the greater is the
depth sensitivity of the measurement. Depths of hundreds of
meters can be imaged when power supplies are large enough
to drive current that will produce measureable potential dif-
ferences over large scales. Twelve-volt (12 V) battery-driven
systems can achieve penetration depths of several tens of
meters or more, depending on terrain conductivity.
The potential that arises between a given pair of electrodes
as current flows depends on the electrode positions and on the
subsurface resistivity structure. To recover the resistivity struc-
ture requires inverting the potential measurements for a
ground resistivity pattern that best fits the readings. The in-
version is typically done with commercial software. Common
practice with modern systems is to acquire and invert data
along profiles, assuming 2-D structure. However, 3-D data can
in theory be acquired and inverted with the same systems.
A variety of traditional survey geometries (different spatial
patterns of current and potential electrodes) are commonly
used (Wenner, dipole–dipole, etc.), although a growing body
of literature suggests that resolution can be maximized with
nontraditional geometries (Stummer et al., 2004; Wilkinson
et al., 2006). For most survey geometries, the maximum depth
of resolution will be approximately 1/5 of the longest spacing
between electrodes in any current/potential combination. This
high ratio of survey length to penetration depth makes resist-
ivity surveys vulnerable to off-profile heterogeneities as current
flows radially out from and into electrodes. As almost all results
reported in the literature involve 2-D inversions, it is useful to
be aware of potential 3-D features that could affect a survey.
Surveys produce noisy data in settings where it is difficult
to make good contact between electrodes and ground. Dry
surface sands yield notoriously noisy data and require spray-
ing saltwater around electrodes to lower contact resistances.
EM methods are typically much less labor intensive than
ERT, and are more commonly used in reconnaissance surveys.

−
Depth

Low resisitivity High resisitivity

Figure 5 Schematic illustration of resistivity inversion from data collected over 2-D subsurface heterogeneities. In this model, the yellow area is
less resistive (more conductive) than background and the cylindrical orange target is less conductive than the background. This schematic
illustrates features of resistivity profiles, where sharp contrasts appear as gradients, and the resolution is better near the surface compared to the
resolution at greater depth. For example, the dip angle of the semi-vertical contact and the lower extent of the cylindrical target are not correctly
reproduced in the inversion process. The true resistivities of smaller features (such as the orange target) are not typically recovered. The nature
of such artifacts depends also on the inversion algorithms used.
112 Near-Surface Geophysics in Geomorphology
Because current is induced in the ground by alternating
magnetic fields, no direct contact between instrument and
ground is required. Airborne EM surveys are used, particularly
in volcanological studies.
Depth of investigation in EM surveys is generally inversely
proportional to the EM frequencies used. MT recordings over
many hours or days can capture low frequencies and gather
information about features that are kilometers in depth. The
highest-frequency portable alternating current EM systems
record at B10 000 kHz and sample only the uppermost few
meters. The geometry and size of the receiver coils that sense
the secondary magnetic fields vary with manufacturer design,
but are generally inversely proportional to instrument fre-
quency. To capture a range of depths may require the use of
multiple instruments. With time-domain electromagnetics,
the secondary currents generated in the ground flow initially at
shallow depths, and propagate deeper with time. Thus, the
resistivity structure over a range of depths can be extracted
from the progressively deeper current flow at a single instru-
ment site.
Solving even the forward problem of induced current flow
for either frequency-domain or time-domain instruments is
difficult for structures more complicated than horizontal
layering. For shallow investigations, data interpretation is
typically done using the horizontal layering assumption at
each reading location. For airborne data, more advanced 2-D
and 3-D inversion schemes are sometimes used.
In comparison with resistivity profiling, EM method survey
footprints can be quite small. A time-domain EM sounding at
a single site can sample to a depth many times of the di-
mension of the current coil laid out on the ground. However,
unlike DC resistivity, EM methods are incompatible with
nearby sources of cultural noise.
3.5.4.1 Strengths
Ground resistivities are sensitive to changes in porosity and
water saturation, as well as lithology. Resistivity and EM
methods are thus well suited to investigating phenomena that
influence surficial processes, such as the degree of sediment
compaction, presence of fractures, and degree of water satur-
ation. These methods work in high-conductivity environments
(brackish water and clays) where ground-penetrating radar
fails.
3.5.4.2 Limitations
Because numerous factors influence the electrical behavior of
ground, it can be difficult to differentiate pore water, matrix,
and temperature effects. Quantitative resistivity measurements
at one site cannot typically be extrapolated to other sites. For
meaningful interpretation, local in situ measurements are
critical. Inversion algorithms have inherent limitations, as
ground resistivity profiles are smoother than reality, so sharp
contacts are smeared. Furthermore, resistivity contrasts of
small anomalies are significantly underestimated (e.g., Day-
Lewis et al., 2006) and 3-D structures with short dimensions
perpendicular to a 2-D profile are improperly characterized in
1-D and 2-D inversions.
3.5.4.3 Case Studies
3.5.4.3.1 Faulting
Where hanging wall and footwalls have distinct lithologies,
resistivity profiles across faults can help constrain fault attitude
(Diaferia et al., 2006; Lee et al., 2009). On some faults, fault
gouge or other materials within the fault zone have a distinct
low-resistivity signature (e.g., Diaferia et al., 2006; Suski et al.,
2010). Pseudo-3D surveys, which involve a compilation of
multiple parallel resistivity profiles, can help confirm the
identification of fault structure and better establish fault atti-
tudes (e.g., Vanneste et al., 2008).
3.5.4.3.2 Volcanic processes
Hydrothermally active zones are conductive (low resistivity)
and many investigators have exploited this property to better
understand fluid circulation in volcanic processes. Examples
of recent ERT studies include Barde-Cabusson et al. (2009)
and Finizola et al. (2009). Time-domain electromagnetics
studies yield hydrostratigraphy to hundreds of kilometers
depth (MacNeil et al., 2007; Srigutomo et al., 2008). Yamaya
et al. (2009) and Kanda et al. (2010) used audio-frequency
magnetotellurics, the latter combining the EM with magnetics.
Finn et al. (2007) used helicopter-borne EM data together with
magnetic data and mapping to evaluate the extent of hydro-
thermally altered rocks with avalanche potential.
3.5.4.3.3 Topography and weathering
Resistivity profiles have been commonly used for detecting
approximate depth to bedrock for geomorphological and
hydrogeological investigations. Bedrock typically has much
higher resistivity than regolith. Recent studies emphasize new
methods for extracting information on this type of contact
from 2-D surveys (e.g., Hsu et al., 2010). Resistivity’s sensitivity
to sediment/rock moisture has been used to study relation-
ships between sediment moisture and weathering (Mol and
Viles, 2010). Braun et al. (2009) combined EM (Barbiéro et al.,
2007) and resistivity surveys with core sample mineralogy and
geochemistry to derive a long-term mass balance for chemical
weathering of regolith in gneissic terrain (Figure 6). This study
illustrates the utility of reconnaissance surveying with EM
before running more labor-intensive resistivity profiles, and
the importance of field-data acquisition for interpreting re-
sistivity data. The authors estimated regolith thickness based
on a resistivity threshold (Figure 6(b)), in this case 400 O m.
Braun et al. (2009) further assessed the uncertainties in re-
sistivity-based thickness estimates both with simple synthetic
models and by comparing profiles and boreholes.
3.5.4.3.4 Slope processes
Material in landslides, such as intact blocks (Sass et al., 2008)
or earthflows (Schmutz et al., 2009), can be distinguished
from substratum with resistivity methods (Naudet et al.,
2008). Bichler et al. (2004) and Jomard et al. (2009) resolved
stratigraphic units in landslide deposits many tens of meters
deep. Mauritsch et al. (2000) applied EM and resistivity
methods, together with seismic refraction on an unstable zone
of a large landslide. They found the combined methods could
not resolve a distinct slip surface, although EM data are in
good agreement with seismic refraction data in determining
 Near-Surface Geophysics in Geomorphology 113

9
23.7 m 9

8 10
16.2 m 8 19.5 m 10

11
7 BH6
22.5 m 11
13.2 m 7

6 12 17.4 m
5 BH5 16.8 m 12
Weathered rock 6
4 Fresh rock 19.5 m 5
N

16.2 m 4
3 BH1 14.1 m 3 m
0
BH12 m 14.2 m 2 50
2 00

30 m
5 13.5 m 1
500 m
1 30 m
ERT calculated resistivity (Ohm.m) 500 m
Estimated regolith thickness (m)
10 26 66 168 429 100 815 206
1 2 7

Figure 6 Example of resistivity profiles targeting regolith thickness. Resistivity profiles were run on the small, subhumid, forested Mule Hole
watershed on the margin of the Karnataka Plateau, India. Regolith thicknesses are estimated assuming a threshold resistivity (400 ohm-m)
between weathered and fresh rock. Uncertainties associated with this method were then assessed by comparison of boreholes (marked with BH)
and by modeling of resistivity signatures of simple synthetic models. Modified from Braun, J.J., Descloitres, M., Riotte, J., et al., 2009. Regolith
mass balance inferred from combined mineralogical, geochemical and geophysical studies; Mule Hole gneissic watershed, south India.
Geochimica et Cosmochimica Acta 73(4), 935–961, with permission from Springer.

thickness of overburden in some areas. Helicopter-borne EM
surveys over landslides in Japan showed positive correlations
between zones of high conductivity and locations of previous
landslides (Konishi, 1998).
3.5.4.3.5 Coastal processes
Most resistivity surveys in coastal settings have hydro-
geological targets. As saltwater is a much better conductor than
freshwater, resistivity is an excellent technique for mapping
the freshwater/saltwater interface. A number of studies, how-
ever, have used resistivity methods to target coastal stratig-
raphy and to link coastal hydrogeology and stratigraphy. For
example, Schneider and Kruse (2005) used ERT and EM sur-
veys to suggest that rapid coastal erosion has left a barrier
island out of equilibrium with its groundwater system. Voett
et al. (2009) used ERT to trace the lateral extent of Holocene
tsunami deposits. Massey and Taylor (2007) used ERT surveys
as an integral part of a paleoenvironmental reconstruction in
the southwest coast of England.
3.5.4.3.6 Fluvial processes
Resistivity data can also be used to extrapolate fluvial archi-
tectural elements between boreholes or trenches (Bersezio
et al., 2007; Hickin et al., 2009). Hickin et al. (2009) com-
bined resistivity with ground-penetrating radar, using resist-
ivity as a proxy for sediment grain size. Froese et al. (2005)
used a combination of resistivity, ground penetrating radar
(GPR), and cores to construct a sediment mass balance for a
large river system.
3.5.4.3.7 Glacial and periglacial processes
Resistivity data can be used in a manner similar to magnetic
data for extracting information on subglacial geology (e.g.,
Ahmad et al., 2009). Massive ice and ice-rich sediments can be
distinguished in rock glaciers with resistivity (e.g., Ribolini
et al., 2007). Kristensen et al. (2009) used resistivity profiling
to help distinguish between glaciogenic sediments, buried
glacier ice, and redeposited marine muds on the edge of surge
moraines.
Resistivity profiles can be useful for detecting the presence
of permafrost (e.g., Vonder Muehll et al., 2002), as well
as for characterizing permafrost and rock glaciers (e.g., Maurer
and Hauck, 2007). Ice and permafrost tend to have higher
resistivities than unfrozen sediments or bedrock. Hauck
et al. (2003) showed that one can address the problem of
high-resistivity surfaces in dry permafrost environments by
attaching saltwater-soaked sponges to the electrodes. They
demonstrated the resolution limits of the resistivity method in
these extremely resistive glacial settings. Because resistivity
surveys can help distinguish ice–debris mixtures from massive
ice, they provided insight on the evolution of rock glaciers
(Kneisel, 2004; Kneisel and Kääb, 2007; Ribolini et al., 2010).
Recent research has been conducted using 2–5-m electrode
spacing (multielectrode surveys), achieving depths of pene-
tration on the order of 10–40 m (Hauck et al., 2003; Ribolini
et al., 2010; Kneisel and Kääb, 2007).
3.5.5 Ground-Penetrating Radar
Most commercial ground-penetrating radar systems transmit
an EM pulse from one antenna and record returns at a second
adjacent antenna. The antennas are typically placed on or
just above the ground surface. The receiving antenna records
pulses that reflect off of contacts or scatter (diffract) off of
point-like discontinuities in the subsurface. Radar pulses are
reflected or scattered where there are abrupt changes in radar-
wave velocity. Wave velocity is controlled by the ground
114 Near-Surface Geophysics in Geomorphology

through which the wave is traveling, and depends primarily on
the electric permittivity (dielectric constant) in the medium.
The permittivity varies with porosity, lithology, and, to a large
extent, the degree of water saturation. In ice, permittivity varies
with ice structure, water content, and rock debris.
As transmitter and receiver are moved across the surface,
and the pattern of returns recorded produces a filtered
and somewhat distorted view of subsurface discontinuities
(Figure 7). The filtering of the subsurface in the GPR profiles
comes from several sources: (1) most systems produce pulses
that are tri-lobed (positive–negative–positive), so that a single
contact to an untrained interpreter may appear as two parallel
contacts (Figure 7); (2) resolution of scattering point-like
objects typically decreases with depth (e.g., Irving and Knight,
2003; Kruse et al., 2010); (3) the GPR wave is polarized, so
returns from dipping contacts can be sensitive to antenna
orientation (e.g., Roberts and Daniels, 1996); (4) strongly
reflecting horizons can produce multiples, in which energy
travels twice round-trip between the surface and the reflecting
horizon; and (5) although layers thinner than B14 wavelength
may be detected, their thickness cannot typically be resolved
(Guha et al., 2005). In general, the attitude of bedding can be
determined while individual contacts within layered units
generally cannot be resolved.
The distortion of the subsurface image stems from the fact
that arrivals recorded at the receiver are plotted on a vertical
trace; so, returns appear to have originated directly below the
receiver. Figure 7 illustrates some of the complexities that arise
as true dips of beds are distorted, very steeply dipping contacts
Time
cannot be imaged directly because no energy is reflected back
to the surface, and point-like objects and sharp corners appear
as inverted hyperbolas. These distortions can be largely cor-
rected for by using migration (a reflection seismic data pro-
cessing technique). Many published GPR images, however,
have not been corrected using migration.
Because wave velocity varies with water content, a radar
profile can look quite different in the wet season than in the
dry season, and radar-wave velocities have to be established on
the day profiles are acquired. Knowledge of the velocity
structure at time of survey is essential for converting the arrival
times (Figure 7) into the depths of the corresponding re-
flectors. Three methods are used to determine velocities. First,
special surveys can be conducted where antenna separation is
expanded around a common mid-point (CMP surveys). Vel-
ocities can be readily determined from the increase in arrival
times with increasing antenna separation, in environments
where relatively flat-lying reflectors are present. Second, the
shape of the wings of the diffraction hyperbolas is velocity
dependent, and can be used to determine velocities where
point-like objects produce diffractions. Third, wells, outcrops,
or trenches can be used to directly correlate stratigraphy and
radar returns.
Finally, wave energy that travels through ground and air
directly from the transmitting antenna to the receiving an-
tenna must be interpreted differently than reflected arrivals.
This energy is invariably the first arrival, typically fairly uni-
form in appearance across a profile. An untrained analyst
might misinterpret these as shallow reflections parallel to the

Figure 7 Schematic illustration of GPR traces collected over 2-D targets with permittivities different from that background. Transmitting and
receiving antennas are typically moved in sync across the survey. Characteristic features of GPR profiles are illustrated. Steep surfaces are not
resolved because no energy is scattered back to the surface (orange dashed line). Sharp edges (green returns) and point-like objects (purple
returns) produce diffraction patterns. Energy traveling directly from transmitting antenna to receiving antenna (red arrows in inset, red pulses)
from the air and ground represent the first arrival. These direct waves are often not filtered from the record. Pulses typically have a tri-lobed
form so that a single contact can appear as a double-layer to the untrained interpreter.

surface. It is also important to note that these direct arrivals
render the earliest reflections undetectable, and, hence, GPR
profiles do not typically yield useful information at the shal-
lowest depths. Care is needed to extract information on
structure at depths less than B1.5 radar wavelengths, which
can be a few centimeters or as much as a few meters, de-
pending on antenna frequency. More recently, several authors
have demonstrated that the first arrival of the ground wave can
in fact yield useful information on the uppermost shallow
layer (Grote et al., 2003; van der Kruk et al., 2006, 2009). Such
surveys require different acquisition geometries than shown in
Figure 7.
3.5.5.1 Strengths
GPR permits rapid acquisition of stratigraphy in a range of
resistive environments. As Table 1 illustrates, GPR has proven
productive in various capacities for imaging most shallow
geomorphological targets. In clay-free sedimentary and volcanic
environments, stratigraphy can be recovered in the uppermost
few meters to, in exceptional cases, tens of meters. Because ice is
very resistive, penetration is excellent in ice and ice/rock mix-
tures, and glacial stratigraphy and ice thickness can be resolved
to depths of hundreds of meters. GPR systems can also be
floated on freshwater or run on lake floors to obtain excellent
sub-lake floor stratigraphy. Frozen-lake surfaces are similarly
good environments. It is important to note that there is no
penetration in saltwater or saltwater-saturated ground.
GPR data are relatively easy to acquire. Profiles can be
obtained at speeds ranging from hundreds of meters per day
in rough or wooded terrain to automobile speeds on open
roads. Airborne data, useful over glaciers, can be acquired even
more rapidly. With shielded antennas, data can be collected in
the presence of cultural noise, even inside buildings. Modern
systems are linked to global positioning system (GPS) or run
with odometers; therefore, positioning errors are significant
only in very-high-resolution 3-D surveys. Most commercial
systems are capable of operating over high and low tempera-
tures and in moderate rain. A final strength of the method is
that real-time data displays make it possible to rapidly deter-
mine whether the method is effective at a given site.
3.5.5.2 Limitations
GPR is not effective in conductive environments. Conductive
clay minerals and brackish–saline water reduce penetration
and render the method useless in many settings. Even worse,
when no real subsurface reflections are present, reflections
from aboveground objects and system noise can sometimes
appear to represent geologic features. Some training and care
must be taken to confirm a subsurface origin for interpreted
features. If a soil survey indicates high-conductivity soils, it is
unlikely that GPR will work well in that setting. In the United
States, Doolittle et al. (2007) have generated maps showing
GPR suitability. Even at locations where GPR penetration is
good, there is an inherent trade-off between depth of pene-
tration (better with low-frequency antennas) and spatial
resolution (better with high-frequency antennas). This
trade-off can be partially overcome by modern systems that
Near-Surface Geophysics in Geomorphology 115
permit simultaneous acquisition with multiple antenna
frequencies.
3.5.5.3 Case Studies
3.5.5.3.1 Faulting
Some of the most stunning examples of GPR imaging of
shallow (o10 m) fault structure come from 2-D and 3-D
surveys of normal faults within the Taupo Rift of New Zealand
(Tronicke et al., 2006; McClymont et al., 2008, 2009a, 2009b)
(Figure 8). By imaging overlapping fault strands, off-fault
folding, rotations of hanging-wall and footwall blocks, and
along-strike variations in vertical fault displacements, GPR
data provide information that significantly improves estimates
of fault slip rates based solely on topographic mapping and
trenching studies (McClymont et al., 2009a, 2009b). By cor-
relating GPR reflections with well-dated stratigraphic units
identified in a paleoseismic trench, it is possible to establish
short-term (o12.5 ka) variability in slip rates (McClymont
et al., 2009a). The high quality of the Taupo Rift surveys stems
from the presence of three strong and distinguishable sub-
horizontal marker horizons, high spatial resolution and ex-
tensive coverage (15–25-cm line spacing over hundreds of
meters), and from the use of advanced processing techniques.
These techniques include appropriate trace-binning methods,
3-D migration that incorporates topography (Lehmann and
Green, 2000; Heincke et al., 2005), and careful corrections for
day-to-day differences in subsurface velocities. The site’s po-
tential was recognized prior to acquisition of the large 3-D
GPR data set through reconnaissance work, including a series
of short 2-D profiles and a small 3-D GPR data set (Tronicke
et al., 2006).
Because GPR cannot directly resolve steeply dipping fea-
tures (Figure 7), faults must be recognized on the basis of
offsets of subhorizontal reflectors. Thus, GPR is not useful at
resolving fault features at sites that lack stratigraphic reflectors
or where relative slip is purely horizontal. Surveys are also
ineffective where weathering of surficial layers obscures
hanging wall/footwall contrasts, and where fault-scarp topo-
graphy is not properly accounted for in image processing.
3.5.5.3.2 Volcanic processes
GPR is exceptionally effective at illuminating stratigraphy in
volcanic deposits before the onset of significant weathering
(Russell and Stasiuk, 1997; Cagnoli and Ulrych, 2001). GPR
profiles illuminate the internal structure of pyroclastic flows
(Gomez et al., 2009; Cassidy et al., 2009) and surge and fall
deposits (Kruse et al., 2010). Ettinger and Kruse (2007)
documented the effectiveness of GPR in mapping stratigraphy
of lahar flows.
3.5.5.3.3 Topography and weathering
GPR is used extensively for the location and identification of
sinkholes, particularly in covered karst where sinkholes may
not have clear topographic expressions (e.g., Kruse et al., 2006;
Gutierrez et al., 2009). In many settings, chemical weathering
processes will increase the conductivity of the substrate
through the development of clays, and thereby reduce the
efficacy of GPR. Beauvais et al. (2004) showed, however, that

Table 1 Summary of geomorphic targets accessible with various geophysical methods
Feature/Process Gravity
Schematic Figure 1
Faulting Hanging wall/footwall
distinctions, Basin-
bounding faults, Figures
2 and 3
Volcanics Volcanic edifice
compensation mech-
anisms and erosion
patterns Identifying the
role of magma
movement in surface
deformation
Topography and Causes of large-scale
weathering surface dynamics;
subsidence processes
and causes; impact
structures
Slope processes Subsurface structures that
drive vertical collapse
Coastal processes and Causes of local observed
sea-level change sea-level changes
Aeolian
Fluvial
Glacial and periglacial Glacial isostatic
adjustment, subglacial
geology; permafrost
thickness
Rightmost column is a summary of comparative studies of various geophysical methods applied to geomorphology.
Magnetics
Figure 4
Hanging wall/footwall
distinctions, Basin-
bounding faults, Figures
2 and 3
Emplacement and
depositional processes
of volcanic units;
mapping debris
deposits; demagnet-
ization associated with
thermal activity
Soil formation, thickness,
and variability of
weathered horizons;
impact structures
Subglacial geology
116
Resistivity, EM GPR Seismic reflection, Mutiple methods;
refraction, and surface comparisons
Near-Surface Geophysics in Geomorphology
wave
Figure 5 Figure 7 Figure 10
Hanging wall/footwall Fault geometries, complex Fault geometries, complex Blakely et al. (2009);
distinctions, fault slip patterns, Figure 8 slip patterns, Figure 11 Lopes Cardozo et al.
attitudes, low-resistivity (2005); Peters et al.
fault zones (2005)
Hydrothermal zones in Internal structure of Volcanic edifice
volcanoes pyroclastic flows and compensation mech-
lahars anisms and erosion
patterns; internal
structure of offshore
volcanic deposits
Depth to bedrock; Quantifying erosion and McCall (2009); Schrott
sediment moisture; deposition, thickness and Sass (2008);
weathering patterns and variability of Ezersky et al. (2009)
Figure 6 weathered horizons
Landslide internal Granular flow Figure 9 Landslide structure, Bichler et al. (2004); Sass
structure, water content, talus thickness offshore slumping and (2006); Mauritsch et al.
and contact with landsliding (2000); Schrott and
substratum Sass (2008)
Coastal stratigraphy Sea-level change history, Offshore coastal Missiaen et al. (2008);
coastal evolution, barrier stratigraphy, coastal Gunn et al. (2006)
island architecture evolution, sea-level
change, tidal flat
stratigraphy, Figures 12
and 13
Dune architecture and
migration
Map fluvial architectural Fluvial architecture and Fluvial and estuarine Bowling et al. (2007)
elements; grain size evolution stratigraphy and
distributions sediment transport
Permafrost and rock Sub-glacial geology, Maurer and Hauck (2007);
glacier thickness, internal structure in Vonder Muehll et al.
distinguish ice-debris glaciers, glacier (2002); Schrott et al.
mixtures from massive thickness, stratigraphic (2003a); Hubbard and
ice records of ice-sheet Glasser (2005)
extents
 Near-Surface Geophysics in Geomorphology 117

175° E 180° E
Australian

ge
plate

c r id

e
gh

zon
rou
35° S

ade

tion
re t

duc
m
Maleme

Hav
Ker

sub
fault zone
0

ngi
47 mm yr−1

rift o

ura
p
Tau

Hik
5
40° S
Pacific
plate
10

15
250

200
N

150

100

50 m

50
S2

y (m) 50

0
0
−50

x (m)

4 5 6 7 8 9 10 11 12 13 14
Z (m)

Figure 8 Example of 3-D GPR surveying over an active fault. Perspective view of three cross sections extracted from a migrated 3-D ground-
penetrating radar volume. Subtle variations of fault slip could not be distinguished without 3-D acquisition and migration. Modified from
McClymont, A.F., Villamor, P., Green, A.G., 2009b. Assessing the contribution of off-fault deformation to slip-rate estimates within the Taupo Rift,
New Zealand, using 3-D ground-penetrating radar surveying and trenching. Terra Nova 21(6), 446–451.

in some places weathering horizons developed over crystalline
basement rocks can be mapped with both GPR and resistivity
methods.
3.5.5.3.4 Slope processes
Courtland et al. (2009) resolved grain avalanching packages
up to 3m thick and many tens of meters long in the highly
resistive deposits on Cerro Negro, Nicaragua, a small basaltic
cinder cone (Figure 9). Radar profiles exhibit clear evidence
of a cone built in large part by successive granular flows
(Figure 9(c)). During eruptions, growth of the cone becomes
oversteepened, resulting in grain avalanching, in turn pro-
ducing reversely graded deposits. The dramatic change in
particle size at the boundary between successive events is be-
lieved to cause strong radar reflections (Figure 9).
Sass (2006, 2007) showed that GPR can be used to derive
talus thickness on alpine slopes. He showed that where a re-
flection from the debris–bedrock contact is absent, the contact
may nevertheless be recognized in a change in the texture of
the radar returns, referred to as a radar facies change. Sass and
118 Near-Surface Geophysics in Geomorphology

20

40

60

80
Time (ns)

100

120

140

160

180

200
(a)
320 330 340 350 360 370 380 390 400 410 420
575
Caribbean
570 sea

565
Cerro negro
560
León
555
Pacific
Elev (m)

550 ocean
100 km
545

540

535

530
1 × vert exag
525

520
(b) 320 330 340 350 360 370 380 390 400 410 420
Distance (m)

(c)

Figure 9 Example of a GPR survey illustrating granular flow on the slope of the Cerro Negro cinder cone, Nicaragua (Courtland et al., 2009).
(a) GPR profile, no topography correction. (b) GPR profile, corrected for elevation. Horizon marked with red has geometry expected for base of a
granular flow package. (c) Schematic illustration of formation of package above red horizon. Data acquired with 100-MHz antennas.

Krautblatter (2007) used GPR-derived internal structure of
talus slopes to postulate models of talus development.
The more bouldery a deposit, the more chaotic the GPR
returns (e.g., Mills and Speece, 1997). It can be difficult to
make meaningful interpretations of GPR records from de-
posits with complex internal structure. For example, radar
surveys on pyroclastic flow deposits on Arenal Volcano, Costa
Rica, failed to resolve internal structure (Kruse, unpublished
data).
3.5.5.3.5 Coastal processes and sea-level change
For several decades, GPR has been a useful technique in
studies of coastal stratigraphy landward of the swash zone
(Bristow and Jol, 2003; Baker and Jol, 2007). The strong EM
response of small variations in porosity and lithology in
coastal sands makes this method sensitive to stratigraphic
variability that may not be detected even by visual inspection
of cores. Penetration depths range up to 10 m or more. In
south Florida, 50-MHz GPR resolves discontinuity surfaces
that record subaerial exposure during sea-level lowstands, re-
ferred to as Q1–Q5 in Florida stratigraphy (Kruse et al., 2000).
Studies on the beach system itself address beach ridge archi-
tecture (Clemmensen and Nielsen, 2010), barrier island and
barrier spit architecture (Jol et al., 2003; Nielsen et al., 2003),
barrier beach evolution (Bennett et al., 2009), coastal pro-
gradation (Moore et al., 2004; Bristow and Pucillo, 2006),
strand plain deposits (Tamura et al., 2008), relict beach scarps
(Buynevich et al., 2007), and overwash deposits (Wang and
Horwitz, 2007).
3.5.5.3.6 Aeolian processes
GPR’s utility in aeolian studies is comparable to that in coastal
settings (Bristow, 2009). This method offers unparalleled ac-
cess to the internal structure of a variety of types of dunes,
including linear (Bristow et al., 2000), crescentic (Derickson
et al., 2008), parabolic (Clemmensen et al., 2007), barchan
(Gomez-Ortiz et al., 2009), and reversing (Bristow et al.,
2010).
3.5.5.3.7 Fluvial processes
GPR is also used to identify the internal architecture of fluvial
deposits, including floodplain deposits (e.g., Leclerc and
Hickin, 1997). Pelpola and Hickin (2004) combined GPR
with aerial photography, bathymetry, and sediment sampling
to derive bed-load transport and fan-delta growth rates over a
52-year period on a steep drainage system. Grant et al. (2004)
used GPR to map buried fluvial channels and assess landscape
evolution.
3.5.5.3.8 Glacial and periglacial processes
GPR has been used for decades in glacial studies on scales
ranging from a few meters to hundreds of meters. It is widely
used as a tool to help calculate glacier volumes (Binder et al.,
2009; Fischer, 2009) and illuminate the internal structure of
glaciers (Fukui et al., 2008; Arcone and Kreutz, 2009; Degen-
hardt, 2009). These latter studies shed light on glacier-flow
dynamics and debris transport and depositional processes.
Studies relate the internal structure of glacial moraines
(Sadura et al., 2006; Burki et al., 2009) and eskers (Burke
et al., 2009) to their surface morphology. GPR studies have
Near-Surface Geophysics in Geomorphology 119
also helped increase understanding of ice-stream dynamics
(e.g., Stokes et al., 2008) and active-layer thickness (e.g.,
Brosten et al., 2009). Brandt et al. (2009) went beyond glacial
stratigraphic interpretation and examined with both theory
and field data the causes of radar returns in firn with ice lenses.
3.5.6 Seismic Methods
Seismic methods, like GPR, are based on generating and re-
cording waves that travel through the ground. As for GPR, any
subsurface heterogeneity associated with a change in the wave
velocity will scatter the waves. Seismic waves (elastic waves)
consist of oscillations of the ground material rather than the
GPR oscillations of EM fields and molecular charges. Hence
the seismic wave velocity is determined by completely differ-
ent physical parameters than the radar wave, namely the bulk
modulus and shear modulus (measures of compressibility and
resistance to shear), and the density. Of relevance to near-
surface geophysics are the facts that seismic wave velocities are
low in loose dry sediments, higher in saturated sediments,
higher in sedimentary rock, and highest in basement igneous
and metamorphic rock.
Thus, despite the fundamental physical differences, both
GPR and seismic methods are responsive to lithologic vari-
ations and the water table. Seismic waves have typically much
longer wavelengths (several meters or more), so the spatial
resolution of seismic studies is lower than for GPR, and it is
difficult (as for GPR) to extract information from depths of
less than a few wavelengths. In practice, the two methods often
provide complementary data, GPR imaging shallowest strata
(uppermost B10 m) and seismic methods imaging deeper
(Sloan et al., 2007).
On land, seismic pulses at the surface generate three dis-
tinct wave types of oscillations: P, S, and surface waves. Each
mode travels with a different velocity (P ¼ fastest, surface ¼
slowest) and can be analyzed and exploited, depending on the
survey targets. Unless otherwise stated, seismic surveys de-
scribed here are based on analysis of P waves. Shear (S) waves
do not propagate in fluids and are insensitive to water satur-
ation. S-wave studies can be useful where gases are present that
strongly attenuate P waves. Surface waves, traditionally a
source of noise in reflection surveys, have been the focus of
new method development in recent decades.
Seismic sources for marine use are designed to generate an
explosive or carefully modulated pulse while minimizing
secondary bubbles. Air guns, water guns, and boomer systems
create small explosions or implosions from/into cavities or
between plates. The dominant frequency of pulses used in
near-surface marine surveying (uppermost tens of meters)
ranges from tens of hertz to B4000 Hz. On land, practitioners
use hammer blows against plates or cylinders laid on the
surface, dropped or accelerated weights, or shotgun shells or
small explosives buried to shallow depths and detonated. The
sources offer peak frequencies in the range of 100–400 Hz.
S-wave sources require a horizontal motion, such as a hammer
blow against a block held in place on the ground.
Seismic wave energy from a blow or shot that returns to the
surface is recorded on an array of receivers called ‘hydrophones’
(in water) or ‘geophones’ (on land). The source and receiver
120 Near-Surface Geophysics in Geomorphology

array are then moved to the next shot location, and the process
repeated. This is relatively straightforward in marine surveying,
where source and receiver array are towed behind the boat,
but is labor intensive on land. Geophones have to make good
contact with the ground to record motion; most involve
pushing a probe some 10 cm into the ground. In high-
resolution surveys, many geophones are laid out, and the shot
location is progressively moved through or around the array.
Information on near-surface features can be derived from
waves refracted from subhorizontal contacts, reflected waves,
or spectral analysis of surface waves (Figure 10). Surveys ex-
ploiting refracted waves have long been a staple of engineering
geology. Survey geometries are designed so that refracted en-
ergy is the first to arrive at geophone arrays, making arrival
times of refracted waves easy to identify. Only contacts in
which a lower layer has a higher wave velocity than the
overlying layer will refract energy back to an array of surface
geophones (Figure 10). Such contacts typically include the
water table and sediment–basement boundary. Refraction
surveys suffer over sites with large lateral velocity variations,
with low-velocity layers, and with thin layers (Whiteley and
Eccleston, 2006). Increasingly sophisticated analysis techni-
ques include refraction tomography techniques, which at-
tempt to recover velocity structure throughout the strata above
the deepest refractor (Sheehan et al., 2005).
Reflection seismic surveys require analysis of all returns
recorded by geophones or hydrophones. These typically pro-
vide a much fuller picture of the subsurface, and to greater
depths than refraction surveys. However, many more shots
Time
and significantly more data processing are required than for
refraction surveying. Reflection seismics are the backbone of
petroleum exploration where target depths are many hundreds
of meters to kilometers. Processing and acquisition techniques
need to be modified for near-surface studies, which require
high frequencies for good spatial analysis, careful separation
of early (shallow) reflections from coincident refracted and
surface wave arrivals, and careful corrections of near-surface
velocity variations and topography.
Near-surface marine reflection seismic surveys are relatively
straightforward in that both a high-frequency source and
hydrophones are towed by boat, and easily moved into de-
sired survey positions. Reflections are easily distinguished in
marine environments, and surficial–lateral velocity variations
and topography are not problems. However, gassy sediments
and hard bottom reflections (such as exposed limestone) can
severely limit penetration. Reshef et al. (2007) reported good
correlations between high-resolution seismic penetration
depths and grain size distributions.
Recent advances in land-survey reflection processing
now permit resolution of strata as shallow as a few meters
(Baker et al., 2000). However, data processing for shallow
(o30 m) land-based seismic reflection surveys is extensive
and less standardized than for other geophysical methods.
One recent advance that can improve shallow resolution is
migration that incorporates topography (Lehmann and Green,
2000). Examples of effective data acquisition, filtering, and
topographic corrections are given in the case studies cited
below.

Figure 10 Schematic illustration of the first arrivals recorded at geophones in a survey designed for analysis of refracted energy. The seismic
source (hammer blow) generates a number of waves that will be recorded by the geophones: direct-body waves, direct-air waves, refracted-body
waves, reflected-body waves, and surface waves. Only the first wave to arrive at each geophone is shown. The nearest geophones record direct-
body waves as a first arrival (red line, red pulses). Farther geophones record refracted waves (head waves) as a first arrival (blue lines, blue
pulses). Wave energy is refracted (blue lines) when the velocity of the medium underlying the contact is higher than the overlying medium.
Velocities of the media above and below the contact and contact depth and dip can be reconstructed from arrival times from records of sources
(hammer blows) at both ends of the array of geophones. Information on a point-like object and steeply dipping contacts is unlikely to be
recovered from seismic-refraction surveys. Surveys designed for analysis of reflected energy will involve geometries similar to the GPR
schematic in Figure 7 and can capture more information, but require hammer blows (or other source) at regular intervals, close to the geophone
spacing, along the line.
 Near-Surface Geophysics in Geomorphology 121

Surface-wave studies can be conducted with geophones 3.5.6.2 Limitations

sensitive to low frequencies (B1 Hz) connected to con-
ventional near-surface seismic systems and sources. As lower
frequency waves permeate to greater depths, the dispersion of
the surface waves can be used to estimate the S-wave velocity
structure in the uppermost few tens of meters. The two pri-
mary methods of data analysis, spectral analysis of surface
waves (SASW) and multichannel analysis of surface waves
(MASW), are well summarized by Pelton (2005). These
methods are useful for mapping soil, imaging karst structure,
and identifying liquefaction potential.
3.5.6.1 Strengths
Reflection seismic methods can produce high-resolution im-
ages of subsurface stratigraphy over depths ranging from a few
meters to kilometers. The method is unparalleled in its reso-
lution over scales of hundreds of meters to kilometers (where
development has been fostered by the petroleum industry).
Efficient methods for rapid acquisition of near-surface data are
under development (e.g., Tsoflias et al., 2006).
Seismic methods are best suited to capture subhorizontal
strata. Land-based reflection methods tend to be labor inten-
sive in terms of data acquisition and processing. Processing
techniques needed to extract optimal images from reflection
data are sufficiently complex that they may not be accessible to
casual practitioners.
3.5.6.3 Case Studies
3.5.6.3.1 Faulting
Seismic reflection imaging of strata offset at faults has been a
standard method of recovering fault displacement and
geometry for many decades. Near-surface capabilities are
highlighted here with a high-quality case study of the trans-
pressive Alpine Fault Zone, New Zealand (Kaiser et al., 2011)
(Figure 11). The imaging of a Pleistocene erosional basement–
surface horizon provides the means to estimate late Quaternary
slip rates on a steeply dipping dominant fault strand (AF in
Figure 11). Although only a single fault scarp is visible at the
surface, deformation is inferred to extend over a 60-m-wide

165° E 170° E 175° E 180° E

Australian
plate
35° S

Late Pleistocene- SE
NW Holocene Moraine
e
40° S on AF postglacial sediments
lt z
e fau Schist
MFS
Alpin −1
m yr Marble
38 m
45° S
Pacific
plate Extent of seismic data

50° S
(a) (b)

ine )
ssl Inlin
Cro nce (m distan e
ta 50 100 ce (m
dis ) ine ) NW
200 ssl Inli
distan ne
100 Cro nce (m 50 100 ce (m
300 t a )
dis 200
400 100 SF1 300
400 SE

SF2

100
Depth (m)

100
Depth (m)

200
AF 200

(c) (d)

Figure 11 High-resolution 3-D land-based seismic reflection imaging of the extent of deformation structures within the shallow expression of
active faults in the transpressive Alpine Fault Zone, New Zealand. (a) Location map. (b) Local geology. The authors image a steeply dipping
dominant fault strand (AF) that significantly offsets the late Pleistocene erosional basement surface. (c) Seismic reflection image. (d) Interpreted
seismic reflection image. Although only a single fault scarp was visible at the surface, deformation was inferred to extend over a 60-m-wide zone
within the fault zone, encompassing two minor fault strands (SF1 and SF2) that were first imaged in shallow GPR studies (McClymont et al.,
2009). Modified from Kaiser, A.E., Horstmeyer, H., Green, A.G., Campbell, F.M., Langridge, R.M., McClymont, A.F., 2011. Detailed images of the
shallow Alpine Fault Zone, New Zealand, determined from narrow-azimuth 3D seismic reflection data. Geophysics 76, B19–B32, with permission
from AGU and A.E. Kaiser (personal communication).
122 Near-Surface Geophysics in Geomorphology
zone within the fault zone, encompassing two minor fault
strands (SF1 and SF2 in Figure 11) that were first imaged in
shallow GPR studies (McClymont et al., 2010). The detailed
seismic imaging of strata truncations, rotations, and tilting was
achieved through dense data coverage and a carefully tailored
processing scheme (Kaiser et al., 2011). Of particular import-
ance in processing were corrections for the severe static shifts
and strong source-generated noise that typically complicate
shallow seismic data. The imaging of complex dipping as well
as diffracted and crossing events required careful velocity ana-
lyses, dip-moveout corrections, and 3-D migration.
3.5.6.3.2 Volcanic processes
Seismic reflection profiling is useful in studying deposits of
near-surface volcanic structures when they lie under water, on
lakes (e.g., Johnson et al., 2003) or offshore (e.g., Morgan
et al., 2003a).
3.5.6.3.3 Topography and weathering
Over tectonic scales (kilometers to hundreds of kilometers), re-
flection seismic profiles can be used to reconstruct uplift and
regional topographic evolution (e.g., Regard et al., 2009). Re-
fraction surveys can be used to quantify rates of erosion and
deposition. For instance, Hunt and Wu (2004) used refraction
surveys in combination with soil sampling and field obser-
vations to estimate Holocene rates of soil erosion on a small hill.
Schrott et al. (2003b) established the volume of valley fill de-
posits in an alpine basin with an array of refraction surveys.
Refraction surveys with engineering targets yield data on litho-
logic parameters such as porosity that may be related to degrees
of weathering (e.g., Gay et al., 2006). In addition, both refraction
(e.g., Whiteley and Eccleston, 2006) and reflection methods (e.g.,
Roberts et al., 1992) can resolve depth to basement in a variety of
geologic settings. Surface-wave methods of mapping soil devel-
opment properties and voids have been, and are being, tested in
a variety of settings (e.g., Miller et al., 1999; Xu and Butt, 2006).
3.5.6.3.4 Slope processes
Marine seismic-reflection surveys can reveal the depositional
record of offshore slumping and landsliding (e.g., Morgan et al.,
2003a) and mass transport in general on passive and active
continental margins (e.g., Reis et al., 2010; Pedley et al., 2010).
In a study that combined both onshore and offshore seismic
surveys, Stucchi and Mazzotti (2009) mapped the extent of a
landslide across the coastline. Their work provides an excellent
illustration of the challenges involved in near-surface seismic
investigations in steep topography and noisy environments.
3.5.6.3.5 Coastal processes and sea-level change
Reflection seismic data are readily collected in the shallow
coastal marine environment using boomer seismic sources
with center frequencies of B1000–10 000 Hz, yielding vertical
resolution (as discussed for GPR) on the order of 20 cm to
2 m. Transitions between gravels, sands, muds, and limestone
tend to produce good seismic reflections. Unconformities,
fluvial valleys, and progradation and aggradation sequences are
commonly recognized in such data sets, permitting re-
constructions of coastal evolution. Such data are also valuable
for understanding the degree to which heterogeneities in the
geologic substrate control Quaternary coastal evolution.
Because coastal stratigraphy is typically subhorizontal, region-
ally extensive strata can be correlated between data gaps, and
data are commonly collected over grids. Mallinson et al. (2010)
combined seismic stratigraphy with amino acid racemization
and strontium-isotope age data from cores to determine the
Quaternary stratigraphic framework and evolution of the coast
near Cape Hatteras, USA (Figures 12 and 13). In such coastal
studies, the distances required between crossover points are
determined by geologic complexity and resources available.
Coring at sites close to seismic lines is critical for data inter-
pretation (Figure 12). In these environments, data processing
is typically limited to band-pass filtering and time-to-depth
conversions done with a constant velocity assumption. More
recently, surface-wave methods have been applied to non-
invasively map tidal-flat stratigraphy (Watabe and Sassa, 2008).
3.5.6.3.6 Fluvial processes
Stratigraphy mapped with shallow-marine seismic methods
can be used to trace morphological evolution of fluvial and
estuarine environments (e.g., Ginsberg et al., 2009; Wang
et al., 2007) and estimate fluvial sediment-transport volumes
(e.g., Liu et al., 2007).
3.5.6.3.7 Glacial and periglacial processes
Refraction and reflection seismics can be used to assess sub-
glacial geology (Ahmad et al., 2009) and infer variations in
frictional resistance to ice flow (Anandakrishnan et al., 1998).
Musil et al. (2002) showed that while seismic surveying in an
alpine rock glacier failed to yield information on reflecting
horizons, tomographic inversion of first-arrival times did yield
information on both the internal structure and the base of the
rock glacier. Baker et al. (2003) found strong seismic re-
flections from contacts between clean ice and underlying
debris-rich ice, as well as the glacier bottom. Wellner et al.
(2006) combined marine seismic lines with multibeam ba-
thymetry and coring to document the geomorphological fea-
tures and seaward extent of the Antarctic ice sheet during the
Last Glacial Maximum. Andreassen et al. (2008) similarly re-
constructed geomorphic features left by former ice sheets
using petroleum industry 3-D reflection seismic surveys. Re-
fraction methods, including refraction tomography, are useful
for mapping top of permafrost and heterogeneities within
permafrost (Musil et al., 2002; Maurer and Hauck, 2007).
3.5.7 Combining Geophysical Methods
Many of the aforementioned studies incorporated coincident
data acquisition using multiple geophysical techniques. The in-
cremental cost of acquiring a second type of geophysical data at
a survey site can be relatively small, because the initial survey
positioning, via GPS or taped grids, can be reused. There
can also be significant advantages associated with the analysis of
data acquired from different methods. For example, if a
resistivity survey shows strata of contrasting resistivity, these
differences could be due to variations in degree of saturation,
pore-water conductivity, or lithology. A seismic-refraction
survey could help distinguish between these possibilities. If a
refracting horizon is observed coincident with a resistivity
 Near-Surface Geophysics in Geomorphology 123

76°45′0″ W 76°30′0″ W 76°15′0″ W 76°0′0″ W 75°45′0″ W 75°30′0″ W

36°30′0″ N

Atlantic
36°15′0″ N Ocean

36°0′0″ N

35°45′0″ N

35°30′0″ N

35°15′0″ N

N
35°0′0″ N
W E
eastct N
S
Value W E 0 5 10 20 30 40
34°45′0″ N High: 0 S
km
0 5 10 20 30 40 50
Low: −20 km

(a) (b)

Figure 12 High-resolution shallow-marine seismic reflection imaging conducted as part of a study of the stratigraphy and geologic controls on
Quaternary evolution of the Cape Hatteras region, US. (a) Map showing bathymetry of the study area. Depth (gray shading) is meters below
mean sea level. (b) Map showing core locations (black dots) and seismic data transects (black lines) used in this investigation. Cores referred to
in this paper are labeled. Because strata are laterally continuous and subhorizontal, a loose grid of 2-D profiles is sufficient to reconstruct
stratigraphy. (Compare with Figures 8 and 11, where careful resolution of fault structure required full 3-D surveys.) Reproduced from Mallinson,
D.J., Culver, S.J., Riggs, S.R., et al., 2010. Regional seismic stratigraphy and controls on the Quaternary evolution of the Cape Hatteras region of
the Atlantic passive margin, USA. Marine Geology 268, 16–33.

gradient, then the resistivity gradient is better explained by a
water table or lithologic boundary than a pore-water conduct-
ivity gradient.
Multiple methods can add value to a survey in different
ways. Gravity and magnetic data offer distinct physical infor-
mation on similar spatial scales. The same is typically true for
resistivity and seismic refraction methods. On the other hand,
GPR and reflection-seismic surveys offer complementary
shallow and deep depths of investigations, respectively. A
combination of resistivity and EM methods offers different
spatial sensitivities of the same physical property, with more
labor-intensive resistivity profiles used to extract local resist-
ivity structure, whereas lateral mapping with more rapid EM
methods can cover areas between profiles. Studies that com-
pare and contrast individual methods at a given site are listed
in the rightmost column of Table 1.
3.5.8 Discussion and Conclusions
3.5.8.1 Practical Considerations
In geophysical studies, the importance of collecting in situ field
observations and reference data cannot be overstated. The best
way to define success of a geophysical method is by examining
results near outcrops or cores. If a geophysical method cap-
tures key parameters or features of interest, its utility lies in
being able to extrapolate the geometry of such conditions
away from calibration points. Where the collection of add-
itional reference data is simply impossible, comparisons of
multiple geophysical methods should clarify interpretation of
survey results, and any interpretations must be based on for-
ward or inverse modeling of the data and comparison of re-
covered geophysical parameters (velocities and resistivity)
with published values for expected lithologies.
Investigators commonly have the capability of acquiring
and processing 3-D data, but have to decide whether the gains
from the significant additional time expense are critical to the
study. For example, the McClymont et al. (2009a, 2009b,
2010) GPR example and the Kaiser et al. (2009, 2011) seismic
surveys resolved structural complexities that would have been
mis-estimated or mis-identified without fully 3-D acquisition
and processing. By contrast, the Braun et al. (2009) resistivity
study and the Mallinson et al. (2010) seismic-reflection study
achieved their research goals with a loose network of 2-D
profiles. Reconnaissance surveys often help clarify data density
needs, as well as giving a practitioner a realistic estimate of
data acquisition rates.
124 Near-Surface Geophysics in Geomorphology

North L1F2 South

0

Two-way travel time (s)

0.02
L1F2 L1F19

0.04

0.06 OBX-17

Cape
0.08 Hatteras
OBX-17
G S M
0

Pleistocene Holocene
Meters below sea level

Sed/water interface (Q100)

SSU VII
20 Q99
SSU V
40 Open shelf
Q30
Upper shoreface sand
SSU IV Lower shoreface sand
60
Q20

0 500 1000
1.4 km
(a) m

West L1F19 Eest

0.04
Two-way travel time (s)

0.05

0.06

0.07

0.08
1 km

30
Meters below sea level

Q30b Flooding surface TST

40 Q30
Subaerial unconformity LST
Deepening

FSST Outer shelf

50 Shoreface sands NR
TST
FR Conformity

60 Inner shelf and estuarine Inner shelf

Q20

70 1 km
(b)

Figure 13 Selected profiles from the Mallinson et al. (2010) study of the Cape Hatteras Quaternary coastal evolution (location shown in
Figure 12). (a) High-resolution single-channel seismic profile L1F2 (see inset map for location) shows prograding clinoforms characteristic of a
regressive shoreface. Two-way travel time (twtt) and meters below sea level (mbsl) based on an estimated seismic velocity of 1600 m s–1
(calculated by correlations to core data). (b) Seismic data with interpretation from L1F19_03 showing the interpretation of the falling stage
systems tract (FSST). Map labels represent normal regression (NR), forced regression (FR) and transgressive systems tract (TST). This study
was conducted with a 175 J boomer source with center frequency B1500-Hz, band-pass filtering 500–2000 Hz, 0.083-ms sampling frequency,
and a 0.5-s shot rate yielding B90-cm sampling interval. Reproduced from Mallinson, D.J., Culver, S.J., Riggs, S.R., et al., 2010. Regional
seismic stratigraphy and controls on the Quaternary evolution of the Cape Hatteras region of the Atlantic passive margin, USA. Marine Geology
268, 16–33.

3.5.8.2 Integrating Geophysics, Remote Sensing, and
Geographic Information Systems
The availability of high-resolution digital elevation models
(DEMs) has led to an emphasis on analysis of topography in
geomorphology studies. Yet, this chapter demonstrates that
geophysical investigations of the subsurface fundamentally
contributes to our understanding of geomorphological pro-
cesses. Clearly, better integration of subsurface and surface
data will advance the field. But better synthesis of geophysical,
remote sensing, and geographic information system GIS data
will require data acquisition at overlapping scales, and further
development of 3-D and temporal representational schemes.
Remotely sensed data are generally acquired over larger
spatial scales, with relatively low-to-moderate spatial reso-
lution, compared to higher-resolution near-surface geo-
physical studies. Such scale variations are being bridged by
advances in imaging sensor characteristics and in geophysical
methodologies. In particular, the development of terrestrial
laser scanning (TLS) brings the measurement scale of DEMs
down to the meter or centimeter scale of high-resolution near-
surface geophysical studies. At the same time, the spatial scale
of geophysical studies can be expanded through the use of
airborne instruments for gravity, magnetic, and EM methods.
Airborne combinations of remote sensing and geophysical
instrumentation such as thermal imagery and magnetic may
prove fruitful. Ground-based geophysical acquisition systems
are also becoming more efficient, making data sets over scales
of hundreds of meters to kilometers more affordable. Ad-
vances include capacitively coupled resistivity systems (e.g.,
Hickin et al., 2009), automated geophone installation systems
(Tsoflias et al., 2006), and side-by-side mounting of multiple
ground-penetrating radar antennas.
The temporal scale is very important in geomorphological
research, and studies should involve simultaneous surface-
imaging and geophysical surveying. For example, Carrivick et al.
(2010) combined hydrological measurements acquired over a
period of airborne LiDAR surveys of lahars to model lahar
transport dynamics; pre- and postevent geophysical surveys
would add another level of control on the transport models.
Time-lapse remote-sensing studies synchronized with repeated
geophysical surveys can be productive for understanding geo-
morphic processes in a variety of rapidly evolving environ-
ments. In coastal environments, airborne imagery coupled
with near-shore seismic or onshore ground-penetrating radar
profiling would illuminate sediment-transport processes.
Similarly, on landslides and debris flows, further studies
linking remote sensing to resistivity or ground-penetrating
radar profiles will help connect surficial and subsurface de-
formation mechanisms.
Better integration of geophysical, remote sensing, and GIS
data is at present largely limited by costs. In addition to the
acquisition issues described above, significant resources can be
required for data processing, visualization, and interpretation.
Many commonly used GIS software platforms do not readily
accommodate subsurface data. Software packages for 4-D
(space and time) visualization of data can be expensive, even
as computer-processing costs shrink. Large data volumes are
most easily understood with specialized visualization tools
such as stereo projectors and wall screens. However, as use of
Near-Surface Geophysics in Geomorphology 125
these software and hardware tools becomes more widespread,
their relative costs should fall.
References
Ahmad, J., Schmitt, D.R., Rokosh, C.D., Pawlowicz, J.G., 2009. High-resolution
seismic and resistivity profiling of a buried Quaternary subglacial valley;
northern Alberta, Canada. Geological Society of America Bulletin 121(11–12),
1570–1583.
Anandakrishnan, S., Blankenship, D.D., Alley, R.B., Stoffa, P.L., 1998. Influence of
subglacial geology on the position of a West Antarctic ice stream from seismic
observations. Nature 394, 62–65.
Andreassen, K., Laberg, J.S., Vorren, T.O., 2008. Sea-floor geomorphology of the
SW Barents Sea and its glaci-dynamic implications. Geomorphology 97(1–2),
157–177.
Arcone, S.A., Kreutz, K., 2009. GPR reflection profiles of Clark and Commonwealth
Glaciers, Dry Valleys, Antarctica. Annals of Glaciology 50(51), 121–129.
Baker, G.S., Jol, H.M., 2007. Stratigraphic Analyses Using GPR. The Geological
Society of America, Special Paper, 432 pp.
Baker, G.S., Steeples, D.W., Schmeissner, C., Spikes, K.T., 2000. Ultrashallow
seismic reflection monitoring of seasonal fluctuations in the water table.
Environmental and Engineering Geoscience 6(3), 271–277.
Baker, G.S., Strasser, J.C., Evenson, E.B., Lawson, D.E., Pyke, K., Bigl, R.A., 2003.
Near-surface seismic reflection profiling of the Matanuska Glacier; Alaska.
Geophysics 68(1), 147–156.
Barbiéro, L., Parate, H.R., Descloitres, M., et al., 2007. Using a structural approach
to identify relationships between soil and erosion in a non-anthropic forested
area, South India. Catena 70, 313–329.
Barde-Cabusson, S., Finizola, A., Revil, A., et al., 2009. New geological insights and
structural control on fluid circulation in La Fossa Cone (Vulcano, Aeolian Islands,
Italy). Journal of Volcanology and Geothermal Research 185(3), 231–245.
Battaglia, M., Gottsmann, J., Carbone, D., Fernández, J., 2008. 4D volcano
gravimetry. Geophysics 73(6), WA3–WA18.
Beauvais, A., Ritz, M., Parisot, J.-C., Bantsimba, C., Dukhan, M., 2004. Combined
ERT and GPR methods for investigating two-stepped lateritic weathering systems.
Geoderma 119(1–2), 121–132.
Bell, R.E., Blankenship, D.D., Finn, C.A., Morse, D.L., Scambos, T.A., Brozenak,
J.M., Hodge, S.M., 1998. Influence of subglacial geology on the onset of a
West Antarctic ice stream from aerogeophysical observations. Nature 394,
58–62.
Bennett, M.R., Cassidy, N.J., Pile, J., 2009. Internal structure of a barrier beach as
revealed by ground penetrating radar (GPR): Chesil beach, UK. Geomorphology
104, 218–229.
Bersezio, R., Giudici, M., Mele, M., 2007. Combining sedimentological and
geophysical data for high-resolution 3-D mapping of fluvial architectural
elements in the Quaternary Po Plain (Italy). Sedimentary Geology 202(1–2),
230–248.
Bichler, A., Bobrowsky, P., Best, M., Douma, M., Hunter, J., Calvert, T., Burns, R.,
2004. Three-dimensional mapping of a landslide using a multi-geophysical
approach: the Quesnel Forks landslide. Landslides 1, 29–40.
Binder, D., Brueckl, E., Roch, K.H., Behm, M., Schoener, W., Hynek, B., 2009.
Determination of total ice volume and ice-thickness distribution of two glaciers
in the Hohe Tauern region, Eastern Alps, from GPR data. Annals of Glaciology
50(51), 71–79.
Blakely, R.J., Sherrod, B.L., Hughes, J.F., Anderson, M.L., Wells, R.E., Weaver, C.S.,
2009. Saddle Mountain fault deformation zone, Olympic Peninsula, Washington:
western boundary of the Seattle uplift. Geosphere 5(2), 105–125.
Bowling, J.C., Harry, D.L., Rodriguez, A.B., Zheng, C., 2007. Integrated geophysical
and geological investigation of a heterogeneous fluvial aquifer in Columbus
Mississippi. Journal of Applied Geophysics 62(1), 58–73.
Brandt, O., Langley, K., Giannopoulos, A., Svein, E., Kohler, J., 2009. Radar
response of firn exposed to seasonal percolation, validation using cores and
FDTD modeling. IEEE Transactions on Geoscience and Remote Sensing 47(8),
2773–2786.
Branston, M.W., Styles, P., 2003. The application of time-lapse microgravity for
the investigation and monitoring of subsidence at Northwich, Cheshire.
Quarterly Journal of Engineering Geology and Hydrogeology 36(3),
231–244.
Braun, J.J., Descloitres, M., Riotte, J., et al., 2009. Regolith mass balance inferred
from combined mineralogical, geochemical and geophysical studies; Mule Hole
126 Near-Surface Geophysics in Geomorphology
gneissic watershed, south India. Geochimica et Cosmochimica Acta 73(4),
935–961.
Bristow, C., 2009. Ground penetrating radar in Aeolian dune sands. In: Jol, H.M.
(Ed.), Ground Penetrating Radar Theory and Applications Elsevier, Amsterdam,
The Netherlands 544 pp.
Bristow, C.S., Augustinus, P.C., Wallis, I.C., Jol, H.M., Rhodes, E.J., 2010.
Investigation of the age and migration of reversing dunes in Antarctica using
GPR and OSL, with implications for GPR on Mars. Earth and Planetary Science
Letters 289(1–2), 30–42.
Bristow, C.S., Bailey, S.D., Lancaster, N., 2000. The sedimentary structures of linear
dunes. Nature 406, 56–59.
Bristow, C.S., Jol, H.M. (Eds.), 2003. Special Publications, London.
Bristow, C.S., Pucillo, K., 2006. Quantifying rates of coastal progradation from
sediment volume using GPR and OSL, the Holocene fill of Guichen Bay, south-
east South Australia. Sedimentology 53, 769–788.
Brosten, T.R., Bradford, J.H., McNamara, J.P., Gooseff, M.N., Zarnetske, J.P.,
Bowden, W.B., Johnston, M.E., 2009. Estimating 3D variation in active-layer
thickness beneath Arctic streams using ground-penetrating radar. Journal of
Hydrology 373(3–4), 479–486.
Brown, G.C., Rymer, H., Thorpe, R.S., 1987. The evolution of andesite volcano
structures: new evidence from gravity studies in Costa Rica. Earth and Planetary
Science Letters 82, 323–334.
Budetta, G., Carbone, D., Greco, F., 1999. Subsurface mass redistributions at Mount
Etna (Italy) during the 1995–1996 explosive activity detected by microgravity
studies. Geophysical Journal International 138(1), 77–88.
Burke, M.J., Woodward, J., Russell, A.J., Fleisher, P.J., 2009. Structural controls on
englacial esker sedimentation; Skeidararjokull, Iceland. Annals of Glaciology
50(51), 85–92.
Burki, V., Larsen, E., Fredin, O., Margreth, A., 2009. The formation of sawtooth
moraine ridges in Bodalen, western Norway. Geomorphology 105(3–4),
182–192.
Buynevich, I.V., FitzGerald, D.M., Goble, R.J., 2007. A 1500 yr record of North
Atlantic storm activity based on optically dated relict beach scarps. Geology
(Boulder) 35(6), 543–546.
Cagnoli, B., Ulrych, T., 2001. Ground penetrating radar images of un-
exposed climbing dune-forms in the Ubahebe hydrovolcanic field (Death
Valley, California). Journal of Volcanology and Geothermal Research 109,
279–298.
Calhoun, K., Long, L.T., Schmitt, T.J., 1989. Sinkholes detected by microgravity
survey. Southeastern Geology 30(2), 109–119.
Carbone, D., Budetta, G., Greco, F., 2003. Bulk processes prior to the 2001 Mount
Etna eruption highlighted through microgravity studies. Journal Of Geophysical
Research 108(B12), 2556. http://dx.doi.org/10.1029/2003JB002542.
Carrivick, J.L., Manville, V., Graettinger, A., Cronin, S.J., 2010. Coupled fluid
dynamics-sediment transport modeling of a Crater Lake break-out lahar: Mt.
Raupehu, New Zealand. Journal of Hydrology 388, 399–413.
Cassidy, N.J., Calder, E.S., Pavez, A., Wooller, L., 2009. GPR-derived facies
architectures; a new perspective on mapping pyroclastic flow deposits.
In: Thordarson, T., Self, S., Larsen, G., Rowland, S.K., Hoskuldsson, A. (Eds.),
Studies in Volcanology; The Legacy of George, Walker. Special Publications of
the International Association of Volcanology and Chemistry of the Earth’s Interior
Geological Society for IAVCEI, London, vol. 2, pp. 181–210.
Clemmensen, L.B., Bjornsen, M., Murray, A., Pedersen, K., 2007. Formation of
aeolian dunes on Anholt, Denmark since AD 1560: a record of deforestation and
increased storminess. Sedimentary Geology 199, 171–187.
Clemmensen, L.B., Nielsen, L., 2010. Internal architecture of a raised beach ridge
system (Anholt, Denmark) resolved by ground-penetrating radar investigations.
Sedimentary Geology 223, 281–290.
Courtland, L.M., Kruse, S.E., Connor, C., 2009. Into the cone: a ground penetrating
radar investigation of near vent processes at Cerro Nego volcano. Eos
Transactions of AGU 90 (52). Fall Meeting Supplement, abstract V21C-2013,
Day-Lewis, F.D., White, E.A., Johnson, C.D., Lane, Jr. J.W., Belaval, M., 2006.
Continuous resistivity profiling to delineate submarine groundwater discharge:
examples and limitations. Leading Edge 25(6), 727–728.
Degenhardt, J.J., 2009. Development of tongue-shaped and multilobate rock
glaciers in alpine environments; interpretations from ground-penetrating radar
surveys. Geomorphology 109(3–4), 94–107.
Dentith, M.C., Bevan, A.W.R., Backhouse, J., Featherstone, W.E., Koeberl, C., 1999.
Yallalie: a buried structure of possible impact origin in the Perth Basin, Western
Australia. Geological Magazine 136, 619–632.
Derickson, D., Kocurek, G., Ewing, R.C., Bristow, C., 2008. Origin of a complex and
spatially diverse dune field pattern, Algodones, southeastern California.
Geomorphology 99(1–4), 186–204.
Diaferia, I., Barchi, M., Loddo, M., Schiavone, D., Siniscalchi, A., 2006.
Detailed imaging of tectonic structures by multiscale Earth resistivity
tomographies: the Colfiorito normal faults (central Italy). Geophysical Research
Letters 33, L09305.
Diez, M., Connor, C.B., Kruse, S.E., Connor, L., Savov, I.P., 2009. Evidence of
small-volume igneous diapirism in the shallow crust of the Colorado Plateau,
San Rafael desert, Utah. Lithosphere 1, 328–336.
Dimalanta, C.B., Ramos, E.G.L., Yumul, Jr. G.P., Bellon, H.N., 2009. New features
from the Romblon Island Group; key to understanding the arc-continent
collision in central Philippines. Tectonophysics 479(1–2), 120–129.
Doolittle, J.A., Minzenmayer, F.E., Waltman, S.W., Benham, E.C., Tuttle, J.W.,
Peaslee, S.D., 2007. Ground-penetrating radar soil suitability map of the
conterminous United States. Geoderma 141, 416–421.
Ettinger, S., Kruse, S., 2007. Using ground penetrating radar to help delineate lahar
hazard zones at Cotopaxi Volcano, Ecuador. EOS Transactions of AGU 88(52),
Fall Meeting Supplement, abstract V11D-0815.
Evans, M.E., Heller, F., 2003. Environmental Magnetism. Academic Press,
SanDiego, CA.
Ezersky, M., Legchenko, A., Camerlynck, C., Al-zoubi, A., 2009. Identification of
sinkhole development mechanism based on a combined geophysical study in
Nahal Hever South area (Dead Sea coast of Israel). Environmental Geology 58,
1123–1141.
Ferraccioli, F., Armadillo, E., Jordan, T., Bozzo, E., Corr, H.F.J., 2009. Aeromagnetic
exploration over the East Antarctic ice sheet; a new view of the Wilkes
subglacial basin. Tectonophysics 478(1–2), 62–77.
Finizola, A., Aubert, M., Revil, A., Schutze, C., Sortino, F., 2009. Importance
of structural history in the summit area of Stromboli during the 2002–
2003 eruptive crisis inferred from temperature, soil CO2, self-potential, and
electrical resistivity tomography. Journal of Volcanology and Geothermal
Research 183(3–4), 213–227.
Finn, C.A., Deszcz-Pan, M., Anderson, E.D., John, D.A., 2007. Three-dimensional
geophysical mapping of rock alteration and water content at Mount Adams,
Washington; implications for lahar hazards. Journal of Geophysical Research
112, B10204.
Fischer, A., 2009. Calculation of glacier volume from sparse ice-thickness data,
applied to Schaufelferner, Austria. Journal of Glaciology 55(191), 453–460.
Forte, A.M., Moucha, R., Simmons, N.A., Grand, S.P., Mitrovica, J.X., 2010. Deep
mantle contributions to the surface dynamics of the North American continent.
Tectonophysics 481(1–4), 3–15.
Froese, D.G., Smith, D.G., Clement, D.T., 2005. Characterizing large river history
with shallow geophysics; middle Yukon River, Yukon Territory and Alaska.
Geomorphology 67(3–4), 391–406.
Fukui, K., Sone, T., Strelin, J.A., Cesar, A., Mori, J., Fujii, Y., 2008. Dynamics and
GPR stratigraphy of a polar rock glacier on James Ross Island, Antarctic
Peninsula. Journal of Glaciology 54(186), 445–451.
Gay, D.A., Morgan, F.D., Sogade, J.A., Vichabian, Y., Reppert, P., Wharton, A.E.,
2006. Investigations of andesitic volcanic debris terrains; Part I, Geophysical.
Geophysics 71(1), B1–B8.
Ginsberg, S.S., Aliotta, S., Lizasoain, Guillermo, O., 2009. Morphodynamics and
seismostratigraphy of a deep hole at tidal channel confluence. Geomorphology
104(3–4), 253–261.
Gomez, C., Lavigne, F., Lespinasse, N., Hadmoko, D.S., Wassmer, P., 2009.
Longitudinal structure of pyroclastic-flow deposits, revealed by GPR survey, at
Merapi Volcano, Java, Indonesia. Journal of Volcanology and Geothermal
Research 174(4), 439–447.
Gomez-Ortiz, D., Martin-Crespo, T., Rodriguez, I., Sanchez, M.J., Montoya, L.,
2009. The internal structure of modern barchan dunes of the Ebro River delta
(Spain) from ground penetrating radar. Journal of Applied Geophysics 69(2),
159–170.
Gottsmann, J., Camacho, A.G., Marti, J., Wooller, L., Fernandez, J., Garcia, A.,
Rymer, H., 2008. Shallow structure beneath the Central Volcanic Complex of
Tenerife from new gravity data: implications for its evolution and recent
reactivation. Physics of the Earth and Planetary Interiors 168(3–4), 212–230.
Grant, J.A., Maxwell, T.A., Johnston, A.K., Kilani, A., Williams, K.K., 2004.
Documenting drainage evolution in Bir Kiseiba, southern Egypt: constraints from
ground-penetrating radar and implications for Mars. Journal of Geophysical
Research 109, EO9002.
Grote, K., Hubbard, S., Rubin, Y., 2003. Field-scale estimation of volumetric water
content using ground-penetrating radar ground wave techniques. Water
Resources Research 39, 1321–1333.
Guha, S., Kruse, S.E., Wright, E.E., Kruse, U.E., 2005. Spectral analysis of ground
penetrating radar response to thin sedimentary layers. Geophysical Research
Letters 32, L23304. http://dx.doi.org/10.1029/2005GL023933.

Gunn, D.A., Pearson, S.G., Chambers, J.E., et al., 2006. An evaluation of combined
geophysical and geotechnical methods to characterize beach thickness. Quarterly
Journal of Engineering Geology and Hydrogeology 39(4), 339–355.
Gutierrez, F., Galve, J.P., Lucha, P., Bonachea, J., Jorda, L., Jorda, R., 2009.
Investigation of a large collapse sinkhole affecting a multi-storey building by
means of geophysics and the trenching Technique (Zaragoza city, NE Spain).
Environmental Geology 58(5), 1107–1122.
Hauck, C., Vonder Mühll, D., Maurer, H., 2003. Using DC resistivity tomography to
detect and characterize mountain permafrost. Geophysical Prospecting 51,
273–284.
Heincke, B., Green, A.G., van der Kruk, J., Horstmeyer, H., 2005. Acquisition and
processing strategies for 3 D georadar surveying a region characterized by
rugged topography. Geophysics 70(6), K53–K61.
Herrero-Bervera, E., Canon-Tapia, E., Walker, G.P.L., Guerrero-Garcia, J.C., 2002.
The Nuuanu and Wailau giant landslides; insights from paleomagnetic and
anisotropy of magnetic susceptibility (AMS) studies. Physics of the Earth and
Planetary Interiors 129(1-2), 83–98.
Hickin, A.S., Kerr, B., Barchyn, T.E., Paulen, R.C., 2009. Using ground-penetrating
radar and capacitively coupled resistivity to investigate 3-D fluvial architecture
and grain-size distribution of a gravel floodplain in northeast British Columbia,
Canada. Journal of Sedimentary Research 79(6), 457–477.
Hubbard, B., Glasser, N., 2005. Field Techniques in Glaciology and Glacial
Geomorphology. Wiley, England, 400 pp.
Hsu, H.L., Yanites, B.J., Chen, C.C., Chen, Y.-G., 2010. Bedrock detection using 2D
electrical resistivity imaging along the Peikang River, central Taiwan.
Geomorphology 114(3), 406–414.
Hunt, A.G., Wu, J.Q., 2004. Climatic influences on Holocene variations in soil
erosion rates on a small hill in the Mojave Desert. Geomorphology 58(1–4),
263–289.
Hurst, A.W., Rickerby, P.C., Scott, B.J., Hashimoto, T., 2004. Magnetic field
changes on White Island, New Zealand, and the value of magnetic changes for
eruption forecasting. Journal of Volcanology and Geothermal Research 136,
53–70.
Irving, J.D., Knight, R.J., 2003. Removal of wavelet dispersion from ground-
penetrating radar data. Geophysics 63(3), 960–970.
Johnson, S.Y., Stephenson, W.J., Morgan, L.A., Shanks, III. W.C, Pierce, K.L., 2003.
Hydrothermal and tectonic activity in northern Yellowstone Lake, Wyoming.
Geological Society of America Bulletin 115(8), 954–971.
Jol, H.M., Lawton, D.C., Smith, D.G., 2003. Ground penetrating radar: 2-D and 3-D
subsurface imaging of a coastal barrier spit, Long Beach, WA, USA.
Geomorphology 53, 165–181.
Jomard, H., Lebourg, T., Guglielmi, Y., Tric, E., 2009. Electrical imaging of sliding
geometry and fluids associated with a deep seated landslide (La Clapière,
France). Earth Surface Processes and Landforms 35(5), 588–599.
Kaiser, A.E., Green, A.G., Campbell, F.M., et al., 2009. Ultrahigh-resolution seismic
reflection imaging of the Alpine Fault, New Zealand. Journal of Geophysical
Research 114, B11306. http://dx.doi.org/10.1029/2009JB006338.
Kaiser, A.E., Horstmeyer, H., Green, A.G., Campbell, F.M., Langridge, R.M.,
McClymont, A.F., 2011. Detailed images of the shallow Alpine Fault Zone
New Zealand, determined from narrow-azimuth 3D seismic reflection data.
Geophysics 76, B19–B32.
Kanda, W., Utsugi, M., Tanaka, Y., Hashimoto, T., Fujii, I., Hasenaka, T., Shigeno,
N., 2010. A heating process of Kuchi-erabu-jima Volcano, Japan, as inferred
from geomagnetic field variations and electrical structure. Journal of
Volcanology and Geothermal Research 189(1–2), 158–171.
Kristensen, L., Benn, D.I., Hormes, A., Ottesen, D., 2009. Mud aprons in front of
Svalbard surge moraines; evidence of subglacial deforming layers or proglacial
glaciotectonics? Geomorphology 111(3–4), 206–221.
Kruse, S., Grasmueck, M., Weiss, M., Viggiano, D., 2006. Sinkhole structure
imaging in covered karst terrain. Geophysical Research Letters 33, L16405.
Kruse, S., Mora, R., Ramirez, C., Alvarado, G., 2010. Ground penetrating radar
imaging of tephra stratigraphy on Poás and Irazú volcanoes, Costa Rica. Revista
Geológica de América Central 43, 119–136.
Kruse, S.E., Schneider, J.C., Campagna, D.J., Inman, J.A., Hickey, T.D., 2000.
Ground penetrating radar imaging of cap rock, caliche and carbonate strata.
Journal of Applied Geophysics 43, 239–249.
Kneisel, C., 2004. New insights into mountain permafrost occurrence and
characteristics in glacier forefields at high altitude through the application of 2D
resistivity imaging. Permafrost and Periglacial Processes 15, 221–227.
Kneisel, C., Kääb, A., 2007. Mountain permafrost dynamics within a recently
exposed glacier forefield inferred by a combined geomorphological, geophysical
and a photogrammetrical approach. Earth Surface Processes and Landform 32,
1797–1810.
Near-Surface Geophysics in Geomorphology 127
Konishi, N., 1998. Landslide surveys in Tertiary soft rock areas using HEM.
Exploration Geophysics (Melbourne) 29(1–2), 234–239.
Leclerc, R.F., Hickin, E.J., 1997. The internal structure of scrolled floodplain
deposits based on ground-penetrating radar, North Thompson River, British
Columbia. Geomorphology 21(1), 17–38.
Lee, C.-C., Tsai, L.L.-Y., Yang, C.-H., Wen, K.-L., Wang, Z.-B., Hsieh, Z.-H., Liu,
H.-C., 2009. The identified origin of a linear slope near chi–chi earthquake
rupture combining 2D, 3D resistivity image profiling and geological data.
Environmental Geology (Berlin) 58(7), 1397–1406.
Lehmann, F., Green, A.G., 2000. Topographic migration of georadar data:
implications for acquisition and processing. Geophysics 65, 836–848.
Liu, J.P., Xu, K.H., Li, A.C., Milliman, J.D., Velozzi, D.M., Xiao, S.B., Yang, Z.S.,
2007. Flux and fate of Yangtze River sediment delivered to the East China Sea.
Geomorphology 85(3–4), 208–224.
Lopes Cardozo, G., Gideon, G.O., Edel, J.B., Granet, M., 2005. Detection of active
crustal structures in the Upper Rhine Graben using local earthquake
tomography, gravimetry and reflection seismic. Quaternary Science Reviews
24(3–4), 337–344.
MacNeil, R.E., Sanford, W.E., Connor, C.B., Sandberg, S.K., Diez, M., 2007.
Investigation of the groundwater system at Masaya Caldera, Nicaragua, using
transient electromagnetics and numerical simulation. Journal of Volcanology and
Geothermal Research 166(3–4), 217–232.
Mallinson, D.J., Culver, S.J., Riggs, S.R., et al., 2010. Regional seismic stratigraphy
and controls on the Quaternary evolution of the Cape Hatteras region of the
Atlantic passive margin, USA. Marine Geology 268, 16–33.
Mamtani, M.A., Sengupta, A., 2009. Anisotropy of magnetic susceptibility analysis
of deformed kaolinite: implications for evaluating landslides. International
Journal of Earth Sciences 98(7), 1721–1725.
Massey, A.C., Taylor, G.K., 2007. Coastal evolution in south-west England, United
Kingdom; an enhanced reconstruction using geophysical surveys. Marine
Geology 245(1–4), 123–140.
Maurer, H., Hauck, C., 2007. Geophysical imaging of alpine rock glaciers. Journal
of Glaciology 53(180), 110–120.
Mauritsch, H.J., Seiberl, W., Arndt, R., Roemer, A., Schneiderbauer, K., Sendlhofer,
G.P., 2000. Geophysical investigations of large landslides in the Carnic region
of southern Austria. Engineering Geology 56(3–4), 373–388.
McCall, G.J.H., 2009. Half a century of progress in research on terrestrial impact
structures; a review. Earth-Science Reviews 92(3–4), 99–116.
McClymont, A.F., Green, A.G., Kaiser, A.E., Horstmeyer, H., Langridge, R.M., 2010.
Shallow fault segmentation of the Alpine fault zone, New Zealand revealed
from 2- and 3-D GPR surveying. Journal of Applied Geophysics 70,
343–354.
McClymont, A.F., Green, A.G., Villamor, P., Horstmeyer, H., Grass, C., Nobes, D.,
2008. Characterization of the shallow structures of active fault zones using 3-D
GPR data. Journal of Geophysical Research 113, B10315.
McClymont, A.F., Villamor, P., Green, A.G., 2009a. Fault displacement accumulation
and slip-rate variability within the Taupo Rift (New Zealand) based on 3-D
ground-penetrating radar data. Tectonics 28, TC4, 005.
McClymont, A.F., Villamor, P., Green, A.G., 2009b. Assessing the contribution of
off-fault deformation to slip-rate estimates within the Taupo Rift, New Zealand,
using 3-D ground-penetrating radar surveying and trenching. Terra Nova 21(6),
446–451.
Miller, R.D., Xia, J., Park, C.B., Ivanov, J.M., 1999. Multichannel analysis of surface
waves to map bedrock. Leading Edge (Tulsa, OK) 18(12), 1392–1396.
Mills, H.H., Speece, M.A., 1997. Ground-penetrating radar exploration of alluvial
fans in the southern Blue Ridge Province, North Carolina. Environmental and
Engineering Geoscience 3(4), 487–499.
Milne, G.A., Mitrovica, J.X., 2008. Searching for eustasy in deglacial sea-level
histories. Quaternary Science Reviews 27(25–26), 2292–2302.
Missiaen, T., Slob, E., Donselaar, M.E., 2008. Comparing different shallow
geophysical methods in a tidal estuary, Verdronken Land van Saeftinge, western
Scheldt, the Netherlands, Geologie en Mijnbouw. Netherlands Journal of
Geosciences 87(2), 151–164.
Mol, L., Viles, H.A., 2010. Geoelectric investigations into sandstone moisture
regimes: implications for rock weathering and the deterioration of San Rock Art
in the Golden Gate Reserve, South Africa. Geomorphology 118(3–4),
280–287.
Moore, L.J., Jol, H.M., Kruse, S., Vandenburgh, S., Kaminsky, G., 2004. Annual
layers revealed by GPR in the subsurface of a prograding coastal barrier,
southwest Washington, U.S.A. Journal of Sedimentary Research 74(5), 690–696.
Morgan, J.K., Moore, G.F., Clague, D.A., 2003a. Slope failure and volcanic
spreading along the submarine south flank of Kilauea volcano, Hawaii. Journal
of Geophysical Research 108, B9.
128 Near-Surface Geophysics in Geomorphology
Morgan, L.A., Shanks, III W.C., Lovalvo, D.A., et al., 2003b. Exploration and
discovery in Yellowstone Lake; results from high-resolution sonar imaging,
seismic reflection profiling, and submersible studies. Journal of Volcanology
and Geothermal Research 122(3–4), 221–242.
Musil, M., Maurer, H., Green, A.G., Horstmeyer, H., Nitsche, F.O., Vonder Muehll,
D., Springman, S., 2002. Shallow seismic surveying of an alpine rock glacier.
Geophysics 67(6), 1701–1710.
Nakatsuka, T., Utsugi, M., Okuma, S., Tanaka, Y., Hashimoto, T., 2009. Detection
of aeromagnetic anomaly change associated with volcanic activity; an
application of the generalized mis-tie control method. Tectonophysics 478(1–2),
3–18.
Naudet, V., Lazzari, M., Perrone, A., Loperte, A., Piscitelli, S., Lapenna, V., 2008.
Integrated geophysical and geomorphological approach to investigate the
snowmelt-triggered landslide of Bosco Piccolo village (Basilicata, southern Italy).
Engineering Geology 98, 156–167.
Nielsen, L., Moller, I., Nielsen, L.H., et al., 2003. Integrating ground-penetrating
radar and borehole data from a Wadden Sea barrier island. Journal of Applied
Geophysics 68(1), 47–59.
Okubo, A., Kanda, W., Tanaka, Y., et al., 2009. Apparent magnetization intensity
map on Sakura-jima Volcano, Kyushu, Japan, inferred from low altitude, high
density helicopter-borne aeromagnetic surveys. Tectonophysics 478(1–2),
34–42.
Okuma, S., Stotter, C., Supper, R., Nakatsuka, T., Furukawa, R., Motschka, K., 2009.
Aeromagnetic constraints on the subsurface structure of Stromboli Volcano,
Aeolian Islands, Italy. Tectonophysics 478(1–2), 19–33.
Pedley, K.L., Barnes, P.M., Pettinga, J.R., Lewis, K.B., 2010. Seafloor structural
geomorphic evolution of the accretionary frontal wedge in response to seamount
subduction, Poverty Indentation, New Zealand. Marine Geology 270(1–4),
119–138.
Pelpola, C.P., Hickin, E.J., 2004. Long-term bed load transport rate based on aerial-
photo and ground penetrating radar surveys of fan-delta growth, Coast
Mountains, British Columbia. Geomorphology 57(3–4), 169–181.
Pelton, J.R., 2005. Near-surface seismology: surface-based methods. In: Butler, D.K.
(Ed.), Near-Surface Geophysics. Society of Exploration Geophysicists, Tulsa, OK,
pp. 219–263.
Peters, G., Buchmann, T.J., Connolly, P., van Balen, R.T., Wenzel, F., Cloetingh,
S.A.P.L., 2005. Interplay between tectonic, fluvial and erosional processes along
the Western Border Fault of the northern Upper Rhine Graben, Germany.
Tectonophysics 406(1–2), 39–66.
Prezzi, C.B., Uba, C.E., Goetze, H.J., 2009. Flexural isostasy in the Bolivian Andes;
Chaco foreland basin development. Tectonophysics 474(3–4), 526–543.
Regard, V., Lagnous, R., Espurt, N., et al., 2009. Geomorphic evidence for recent
uplift of the Fitzcarrald Arch (Peru); a response to the Nazca Ridge subduction.
Geomorphology 107(3–4), 107–117.
Reshef, M., Ben-Avraham, Z., Tibor, G., Marco, S., 2007. The use of acoustic
imaging to reveal fossil fluvial systems; a case study from the southwestern Sea
of Galilee. Geomorphology 83(1–2), 58–66.
Ribolini, A., Chelli, A., Guglielmin, M., Pappalardo, M., 2007. Relationships
between glacier and rock glacier in the Maritime Alps, Schiantala Valley, Italy.
Quaternary Research 68(3), 353–363.
Ribolini, A., Guglielmin, M., Fabre, D., et al., 2010. The internal structure of rock
glaciers and recently deglaciated slopes as revealed by geoelectrical tomography:
insights on permafrost and recent glacial evolution in the Central and Western
Alps (Italy–France). Quaternary Science Reviews 29(3–4), 507–521.
Reis, A.T., Perovano, R., Silva, C.G., Vendeville, B.C., Araujo, E., Gorini, C.,
Oliveira, V., 2010. Two-scale gravitational collapse in the Amazon Fan; a
coupled system of gravity tectonics and mass-transport processes. Journal of
the Geological Society of London 167(3), 593–604.
Riva, R.E.M., Gunter, B.C., Urban, T.J., et al., 2009. Glacial isostatic adjustment
over Antarctica from combined ICESat and GRACE satellite data. Earth and
Planetary Science Letters 288(3–4), 516–523.
Roberts, M.C., Pullan, S.E., Hunter, J.A., 1992. Applications of land-based high
resolution seismic reflection analysis to Quaternary and geomorphic research.
Quaternary Science Reviews 11(5), 557–568.
Roberts, R.L., Daniels, J.J., 1996. Analysis of GPR polarization phenomena. Journal
of Environmental and Engineering Geophysics 1(2), 139–157.
Russell, J.K., Stasiuk, M.V., 1997. Characterization of volcanic deposits with
ground-penetrating radar. Bulletin of Volcanology 58, 515–527.
Rymer, H., 1994. Microgravity change as a precursor to volcanic activity. Journal of
Volcanology and Geothermal Research 61(3–4), 311–329.
Saballos, J.A., Malservisi, R., Connor, C., 2010. Gravity and geodetic studies at
Concepción volcano. American Geophysical Union, Nicaragua, Fall Meeting
2010, abstract G23C-0846.
Sadura, S., Martini, I.P., Endres, A.L., Wolf, K., 2006. Morphology and GPR
stratigraphy of a frontal part of an end moraine of the Laurentide ice
sheet; Paris Moraine near Guelph, ON, Canada. Geomorphology 75(1–2),
212–225.
Sass, O., 2006. Determination of the internal structure of alpine talus deposits
using different geophysical methods (Lechtaler Alps, Austria). Geomorphology
80(1–2), 45–58.
Sass, O., 2007. Bedrock detection and talus thickness assessment in the European
Alps using geophysical methods. Journal of Applied Geophysics 62(3),
254–269.
Sass, O., Bell, R., Glade, T., 2008. Comparison of GPR, 2D-resistivity and
traditional techniques for the subsurface exploration of the Öschingen landslide,
Swabian Alb (Germany). Geomorphology 93, 89–103.
Sass, O., Krautblatter, M., 2007. Debris flow-dominated and rockfall-dominated talus
slopes; genetic models derived from GPR measurements. Geomorphology
86(1–2), 176–192.
Schmutz, M., Guérin, R., Andrieux, P., Maquaire, O., 2009. Determination of the
3D structure of an Earthflow by geophysical methods: the case of Super
Sauze, in the French southern Alps. Journal of Applied Geophysics 68(4),
500–507.
Schneider, J.A., Kruse, S.E., 2005. Assessing natural and anthropogenic impacts on
freshwater lens morphology on small barrier islands: Dog Island and St. George
Island, Florida. Hydrogeology Journal 14, 131–145.
Schrott, L., Hördt, A., Dikau, R., 2003a. Geophysical applications in
geomorphology. Zeitschrift für Geomorphologie Supplementband, Stuttgart and
Berlin Gebrüder Borntraeger 132, 190.
Schrott, L., Hufschmidt, G., Hankammer, M., Hoffmann, T., Dikau, R., 2003b. Spatial
distribution of sediment storage types and quantification of valley fill deposits in
an alpine basin, Reintal, Bavarian Alps, Germany. Geomorphology 55(1–4),
45–63.
Schrott, L., Sass, O., 2008. Application of field geophysics in geomorphology;
advances and limitations exemplified by case studies. Geomorphology 93(1–2),
55–73.
Shah, A.K., Daniels, D.L., Kontny, A., Brozena, , John, M., 2009. Megablocks and
melt pockets in the Chesapeake Bay impact structure constrained by magnetic
field measurements and properties of the Eyreville and Cape Charles cores.
Special Paper – Geological Society of America 458, 195–208.
Sheehan, J.R., Doll, W.E., Mandell, W.A., 2005. An evaluation of methods and
available software for seismic refraction tomography analysis. Journal of
Environmental and Engineering Geophysics 10(1), 21–34.
Sloan, S.D., Tsoflias, G.S., Steeples, D.W., Vincent, P.D., 2007. High-resolution
ultra-shallow subsurface imaging by integrating near-surface seismic reflection
and ground-penetrating radar data in the depth domain. Journal of Applied
Geophysics 62(3), 281–286.
Steffen, H., Gitlein, O., Denker, H., Mueller, J., Timmen, L., 2009. Tectonophysics.
Present rate of uplift in Fennoscandia from GRACE and absolute gravimetry
474(1–2), 69–77.
Stokes, C.R., Lian, O.B., Tulaczyk, S., Clark, C.D., 2008. Superimposition of ribbed
moraines on a palaeo-ice-stream bed; implications for ice stream dynamics and
shutdown. Earth Surface Processes and Landforms 33(4), 593–609.
Srigutomo, W., Kagiyama, T., Kanda, W., et al., 2008. Resistivity structure of Unzen
Volcano derived from time domain electromagnetic (TDEM) survey. Journal of
Volcanology and Geothermal Research 175(1–2), 231–240.
Stucchi, E., Mazzotti, A., 2009. 2D seismic exploration of the Ancona landslide
(Adriatic Coast, Italy). Geophysics 74(5), B139–B151.
Stummer, P., Maurer, H., Green, A.G., 2004. Experimental design; electrical
resistivity data sets that provide optimum subsurface information. Geophysics
69(1), 120–139.
Suski, B., Brocard, G., Authemayou, C., Consenza Muralles, B., Teyssier, C.,
Holliger, K., 2010. Localization and characterization of an active fault in an
urbanized area in central Guatemala by means of geoelectrical imaging.
Tectonophysics 480, 88–98.
Tabor, R.W., Cady, W.M., 1978. The structure of the Olympic Mountains,
Washington – analysis of a subduction zone. U.S. Geological Survey
Professional Paper 1033, 25pp.
Tamisiea, M.E., Mitrovica, J.X., Davis, J.L., 2007. GRACE gravity data constrain
ancient ice geometries and continental dynamics over Laurentia. Science 316,
881–883.
Tamura, T., Murakami, F., Nanayama, F., Watanabe, K., Saito, Y., 2008. Ground-
penetrating radar profiles of Holocene raised-beach deposits in the Kujukuri
strand plain, Pacific coast of eastern Japan. Marine Geology 248(1–2), 11–27.
Telford, W.M., Geldart, L.P., Sheriff, R.E., 1990. Applied Geophysics, Second ed.
Cambridge University Press, New York, NY, 770 pp.

Tizzani, P., Battaglia, M., Zeni, G., Atzori, S., Berardino, P., Lanari, R., 2009. Uplift
and magma intrusion at Long Valley caldera from InSAR and gravity
measurements. Geology 37(1), 63–66.
Tronicke, J., Villamor, P., Green, A.G., 2006. Detailed shallow geometry and vertical
displacement estimates of the Maleme Fault Zone, New Zealand, using 2-D and
3-D georadar. Near Surface Geophysics 4, 155–161.
Tsoflias, G.P., Steeples, D.W., Czarnecki, G.P., Sloan, S.D., Eslick, R.C., 2006.
Automatic deployment of a 2-D geophone array for efficient ultra-shallow
seismic imaging. Geophysical Research Letters 33(9), 4.
Ueda, Y., 2007. A 3D magnetic structure of Izu-Oshima Volcano and their changes
after the eruption in 1986 as estimated from repeated airborne magnetic surveys.
Journal of Volcanology and Geothermal Research 164(3), 176–192.
van der Kruk, J., Streich, R., Green, A.G., 2006. Properties of surface waveguides
derived from separate and joint inversion of dispersive TE and TM GPR data.
Geophysics 71(1), K19–K29.
van der Kruk, J., Vereecken, H., Jacob, R.W., 2009. Identifying dispersive GPR
signals and inverting for surface wave-guide properties. Leading Edge 28(10),
1234–1239.
Vanneste, K., Verbeeck, K., Petermans, T., 2008. Pseudo-3D imaging of a low-slip-
rate, active normal fault using shallow geophysical methods: The Geleen fault in
the Belgian Maas River valley. Geophysics 73(1), B1–B9.
Voett, A., Brueckner, H., Brockmueller, S., et al., 2009. Traces of Holocene tsunamis
across the Sound of Lefkada, NW Greece. Global and Planetary Change
66(1–2), 112–128.
Vonder Muehll, D., Hauck, C., Gubler, H., 2002. Mapping of mountain permafrost
using geophysical methods. Progress in Physical Geography 26, 643–660.
Vonder Muehll, D.S., Hauck, C., Lehmann, F., 2000. Verification of geophysical
models in Alpine permafrost using borehole information. Annals of Glaciology
31(1), 300–306.
Wang, P., Horwitz, M.H., 2007. Erosional and depositional characteristics of
regional overwash deposits caused by multiple hurricanes. Sedimentology 54(3),
545–564.
Wang, Z., Chen, Z., Chen, J., Wei, Z., 2007. Seismic framework and the Holocene
morphological evolution of the Changjiang River mouth, China. Geomorphology
85(3–4), 237–248.
Biographical Sketch
Sarah Kruse is an associate professor in the Geology Department of the University of South Florida. She uses near-
surface geophysics to address hydrogeological, volcanological, and tectonic problems. She has worked on topics
ranging from the hydrogeology of small islands to volcano stratigraphy to the resolution limits of ground-
penetrating radar.
Near-Surface Geophysics in Geomorphology 129
Watabe, Y., Sassa, S., 2008. Application of MASW technology to identification of
tidal flat stratigraphy and its geoenvironmental interpretation. Marine Geology
252(3–4), 79–88.
Wellner, J.S., Heroy, D.C., Anderson, J.B., 2006. The death mask of the Antarctic
ice sheet; comparison of glacial geomorphic features across the continental
shelf. Geomorphology 75(1–2), 157–171.
Wetmore, P.H., Connor, C.B., Kruse, S.E., Callihan, S., Pignotta, G., Stremtan, C.,
Burke, A., 2009. Geometry of the Trachyte Mesa intrusion, Henry Mountains,
Utah: implications for the emplacement of small melt volumes into the upper
crust. Geochemistry, Geophysics, Geosystems 10, 8. http://dx.doi.org/10.1029/
2009GC002469.
Whiteley, R.J., Eccleston, P.J., 2006. Comparison of shallow seismic refraction
interpretation methods for regolith mapping. Exploration Geophysics
(Melbourne) 37(4), 340–347.
Wilkinson, P.B., Meldrum, P.I., Chambers, J.E., Kuras, O., Ogilvy, R.D., 2006.
Improved strategies for the automatic selection of optimized sets of electrical
resistivity tomography measurement configurations. Geophysical Journal
International 167(3), 1119–1126.
Williams-Jones, G., Rymer, H., Mauri, G., Gottsmann, J., Poland, M., Carbone, D.,
2008. Toward continuous 4D microgravity monitoring of volcanoes. Geophysics
73(6), WA19–WA28.
Xu, C., Butt, S.D., 2006. Evaluation of MASW techniques to image steeply dipping
cavities in laterally inhomogeneous terrain. Journal of Applied Geophysics 59,
106–116.
Yamaya, Y., Mogi, T., Hashimoto, T., Ichihara, H., 2009. Hydrothermal system
beneath the crater of Tarumai Volcano, Japan; 3-D resistivity structure revealed
using audio-magnetotellurics and induction vector. Journal of Volcanology and
Geothermal Research 187(3–4), 193–202.
de Zeeuw-van Dalfsen, E., Rymer, H., Sigmundsson, F., Sturkell, E., 2005. Net
gravity decrease at Askja Volcano, Iceland; constraints on processes responsible
for continuous caldera deflation, 1988–2003. Journal of Volcanology and
Geothermal Research 139(3–4), 227–239.
Zlotnicki, J., Sasai, Y., Toutain, J.P., et al., 2009. Combined electromagnetic,
geochemical and thermal surveys of Taal volcano (Philippines) during the period
2005–2006. Bulletin of Volcanology 71(1), 29–47.
 3.6 Digital Terrain Modeling
T Wasklewicz, East Carolina University, Greenville, NC, USA
DM Staley, US Geological Survey, Golden, CO, USA
K Reavis, East Carolina University, Greenville, NC, USA
T Oguchi, University of Tokyo, Kashiwa-shi, Japan
r 2013 Elsevier Inc. All rights reserved.

3.6.1 Introduction 131

3.6.2 Background 131
3.6.3 DTM Representation 133
3.6.3.1 Land Surface 133
3.6.3.2 Scales and Land Surface 134
3.6.3.3 Data Structures 138
3.6.3.3.1 Raster (grid) 138
3.6.3.3.2 Triangulated irregular network (TIN) 140
3.6.3.3.3 Contour lines 140
3.6.3.3.4 Breaklines 140
3.6.3.3.5 Mass points 140
3.6.4 Data Sources 141
3.6.4.1 Terrestrial Laser Scanning Techniques 141
3.6.4.2 Terrestrial Photogrammetry Techniques (TPT) 143
3.6.4.3 Airborne Laser Scanning 144
3.6.4.4 Interferometric Synthetic Aperture Radar 145
3.6.5 Preprocessing 146
3.6.5.1 Point-Cloud Reduction Techniques 147
3.6.5.1.1 Decimation 147
3.6.5.1.2 Filtering 147
3.6.5.1.3 Segmentation 148
3.6.5.2 Surface Interpolation 148
3.6.6 DTM Error Assessment 150
3.6.6.1 Error Issues 150
3.6.6.1.1 Global errors 150
3.6.6.1.2 Local errors 151
3.6.6.1.3 Systematic errors 151
3.6.7 Geomorphological Applications 151
3.6.7.1 High-Resolution DTMs 152
3.6.7.2 High-Resolution DTMs and Scale 154
3.6.7.3 Data Fusion 154
3.6.7.4 Temporal Data Acquisition 154
3.6.8 Conclusions 155
References 156

Glossary Digital terrain model A bare-earth digital elevation

Accuracy The degree of closeness of a measurement of a
quantity to its actual (true) value.
Airborne laser scanning Laser scanning of the Earth’s
surface (topographic and bathymetric) conducted from
either a fixed wing aircraft or a helicopter.
Digital elevation model A file or database of
numerical values that approximates topographic/
bathymetric variability at given x, y, and z;
coordinates.
model that is devoid of vegetation and human-built
structures.
Interferometric synthetic aperture radar Interferometric
synthetic aperture radar refers to a rapidly evolving form
of radar data processing that allows for the
characterization of topography using airborne or
spacecraft-mounted radar systems.
Point clouds A large number of points representing
topographic/bathymetric variations that are recorded as x, y,

Wasklewicz, T., Reavis, K., Staley, D.M., Oguchi, T., 2013. Digital terrain
modeling. In: Shroder, J. (Editor in Chief), Bishop, M.P. (Ed.), Treatise on
Geomorphology. Academic Press, San Diego, CA, vol. 3, Remote Sensing
and GIScience in Geomorphology, pp. 130–161.

130 Treatise on Geomorphology, Volume 3 http://dx.doi.org/10.1016/B978-0-12-374739-6.00048-8

 Digital Terrain Modeling 131

and z coordinates most commonly by laser scanners or
photogrammetric approaches.
Terrestrial laser scanning Laser scanning of the Earth’s
surface conducted with a tripod-mounted high-definition
surveying device.
Terrestrial photogrammetry techniques The
development of high-resolution digital terrain models for
geomorphic analysis using digital cameras and specialized
photogrammetric software. Conducted properly, this
complex process allows high-resolution DTMs to be
created from vertical or oblique images of the Earth’s
surface.

Abstract

Geomorphologists require quantitative information about the land surface. New sensors can now measure elevation
changes at a variety of scales and these data are used to generate digital terrain model (DTM) that accurately characterize
topography. A variety of DTM analytics are used to support a multitude of geomorphological studies. However, there are
numerous issues involving representation, sampling, interpolation, and error assessment and correction, which must be
addressed before using the elevation data. Data reduction, filtering, and accuracy are key aspects to consider. Knowledge of
these issues is critical for terrain analysis and the communication of information derived from a DTM.

3.6.1 Introduction determine the magnitude of process rates. In some instances,

Quantitative characterization of topography is required to ex-
plain the formation, diversity, and dynamics of surface processes
and landforms, as well as unraveling the complexity of anthro-
pogenic forcing on landscape evolution (Tucker and Hancock,
2010). Historically, the generation and use of topographic maps
greatly facilitated terrain analysis and geomorphological map-
ping. Modern-day terrain analysis, or geomorphometry, requires
digital terrain modeling and the production of digital terrain
models (DTMs) that accurately characterize the topography. The
rapid proliferation of geospatial technologies greatly facilitates
digital terrain modeling and terrain analysis. Scientific and ap-
plied research in geomorphology is increasingly making use of
DTMs. Digital terrain modeling and terrain analysis are now
fundamental to geomorphology, and applications span the
spectrum of surface-process domains.
Advances in remote-sensing technology permit multiple
sensors on numerous platforms to acquire digital data that
can be used to generate DTMs. What seems to be an ever-
increasing capacity to acquire detailed landscape information,
greatly facilitates digital terrain modeling and terrain analysis.
Consequently, new spatio-temporal data permit many new
advances in geomorphometry, involving the development and
evaluation of new algorithms and analysis procedures. Tech-
nically, the cost effectiveness (monetary expenses and human
labor), increased spatial extent, accuracy, measurement scales,
and an ability to rapidly mobilize before and after a geo-
morphic event represent important developments that are
unprecedented in geomorphological research.
The availability of new and improved data, analytical cap-
abilities, and progress in theoretical/conceptual treatments of
geomorphological systems now permit new opportunities to
study process-form relationships. Numerous geomorphometric
parameters have been developed to characterize the multiscale
morphological attributes of the topography. Many of these
same parameters also play a role in specific process mechanics,
as topographic parameters govern many surface processes, and
the rates of erosion and deposition can be accurately estimated
by using multitemporal DTMs. Research reveals, however, that
high-resolution data with increased spatial coverage is required
to account for the operational scale-dependencies of many
surface processes and their linkage to multiscale terrain par-
ameters. Such process-form linkages have already resulted in
new insights into short-term and long-term landscape evo-
lution (e.g., Hancock et al., 2010; Stark et al., 2010). Neverthe-
less, there are many issues associated with digital terrain
modeling that require knowledge of the process and limitations
associated with analysis and information extraction.
Consequently, the objective of this chapter is to cover the
many developments related to digital terrain modeling and the
use of digital elevation models (DEMs). The focus is on im-
portant background information and key issues. The authors
first address terrain representation concepts and discuss scale
and data structure issues. Recent advances in DTM generation are
highlighted with particular emphasis placed on terrestrial laser
scanning, terrestrial photogrammetry techniques, airborne laser
scanning, and interferometric synthetic-aperture radar. The
aforementioned data sources characteristically produce large
volumes of data, and it is important for geomorphologists to
note that not all DTMs represent bare-earth information.
The authors then discuss the numerous types of preprocessing
that are required to address missing data and error via various
algorithms and spatial interpolation procedures. Postprocessing
is also covered and related to DTM error and uncertainty. Finally,
the chapter concludes by considering the importance of DTMs to
current and future studies in geomorphology, with consideration
of undeveloped opportunities and research directions.
3.6.2 Background
A fledgling group of geomorphologists began to present re-
search on landforms in a more systematic and quantitative
fashion during the late 1940s and early 1950s at the onset of
132 Digital Terrain Modeling
the quantitative revolution (Horton, 1945; Mackin, 1948;
Strahler, 1952). These works led to a proliferation of research
that examined the operation of process-mechanics within
geomorphological systems, and became a dominant approach
for almost 50 years of applied and scientific studies in geo-
morphology (e.g., Dury, 1983; Morisawa, 1988; Strahler,
1992; Rhoads, 2006; Haggett, 2008). This is not to suggest that
research during this period did not consider process-form re-
lationships. Observations of process-oriented information
were linked with two-dimensional (2D) landform shapes
(Horton, 1945; Schumm, 1956; Strahler, 1958; Melton,
1957). Studies linking process and form remained spatially,
temporally, and morphometrically limited and continued to
be underrepresented in the literature until the 1980s. The
limited scope of process-form studies (beyond 2D measure-
ments of landform shape) can be directly related to an in-
ability to produce measures/metrics that describe various
aspects of process and form. Detailed information on particle
entrainment and movement, wind speeds, and fluid dis-
charges were generally not in accordance with 2D measure-
ments of topography and landforms. Early emphases on
quantitative process mechanics had rapidly eclipsed the geo-
morphologist’s ability to capture topographic information in a
manner capable of producing comparable measures of form
(Pike, 1995). This was in part a function of rapid advances in
sampling instruments for measuring flows of energy (wind,
water, and waves) and matter (sediment), as well as equations
to approximate the dynamics of energy and matter (e.g., unit
stream power and critical threshold shear velocity), and lack of
geospatial technologies critical to the emergence and evo-
lution of digital terrain modeling.
Attempts were made to quantify terrain and landforms
before the availability of DTMs. A variety of measures and
indices from topographic maps were employed to characterize
basic morphometric characteristics such as length, area, and
relief (Schumm, 1956; Strahler, 1958; Mark, 1974). Many
studies, however, did not follow consistent rules of collecting
or analyzing morphometric data, which further inhibited the
broader application of the research and brought into question
the representativeness of the measures and indices for char-
acterizing specific aspects of terrain and landforms (Evans,
1972, 1980). This work was also hindered by a general lack of
spatial analyses of topographic data during the quantitative
revolution (Gregory, 2000). Much of the work of this period
relied on statistical and later mathematical approaches to
testing hypotheses, where more recent work has focused on
integrating statistics, mathematics, and spatial analyses
(Hayakawa and Oguchi, 2006; Lin and Oguchi, 2006;
Dunning et al., 2009).
Early morphometric and process-mechanic oriented tech-
niques devised during the quantitative revolution led to ad-
vances in geomorphology. They also established a foundation
and a need for digital data that could represent the continuous
nature of the land surface and permit topographic change
detection. The emergence of computers in the early- to mid-
1950s led to the first digital analysis of topography. Although
not a DTM in the sense of representing a continuous surface,
initial models were numerical points used in the development
of topographic profiles (Pike et al., 2009) and used to assess
microrelief (Walls et al., 1954) and elevation spectra
(Jaeger and Schuring, 1966). It is evident from these and other
examples that progress was occurring in geomorphology and
engineering. These approaches would later be adopted to
initiate analysis of the topography using DTMs.
Pike et al. (2009) provided a detailed history of DTM use in
the field of geomorphometry, which is now fundamental to
geomorphological research. Herein, the authors synthesize their
historical interpretation to provide a short-history of DTM in-
corporation into geomorphology. Digital terrain modeling as a
concept had its early origins in the late 1950s (Miller and
Leflamme, 1958). Practical applications for DTMs were not
appreciated until the development of 63 m gridded data de-
rived from 1:250 000 scale maps of the US during 1963–72
(Noma and Misulia, 1959; U.S. Army Map Service, 1963), or in
a very few cases through laborious efforts by individuals who
generated DTMs from topographic maps, field surveys, or
photogrammetry techniques (Tobler and Davis, 1968; Saka-
guchi, 1969). The visually interpreted DTMs were first used to
formulate mathematical and statistical means for measuring
elevation and its surface derivatives in geomorphology (Tobler,
1969; Evans, 1972). The geomorphological relevance of this
work was expounded on in the context of hillslope geo-
morphology (Carson and Kirkby, 1972), which provided a
subsequent framework to advance quantitative DTM ap-
proaches into other subfields of geomorphology.
Computer processing, visualization, and software with the
capability to handle DTMs remained an issue throughout
much of the 1970s and 1980s. Statistical software packages
produced descriptive, univariate, and multivariate statistics
from tabular elevation and surface derivative data sets during
this time frame. It was not until the development of personal
computers, however, and the advent of software packages with
the capability to process and analyze raster data such as
GRASS (http://grass.osgeo.org/, 1985), MicroDEM (Guth
et al., 1987), IDRISI (Eastman, 1989), and ArcView (http://
www.esri.com/about-esri/index.html, 1991) that there was
broad-scale use of DTMs. Data availability from the 1980s to
the present was also a driving factor in the advancement of
terrain-analysis techniques in geomorphology. The develop-
ment of DTMs by a variety of governmental agencies across
the globe provided users with free access to data (e.g., USGS
7.5-min DEM, 11 DEM and 2 asec DEM). Users were also able
to purchase instruments capable of producing digital elevation
data (e.g., total stations, Global positioning systems (GPS)
surveying equipment, and terrestrial laser scanners (TLS)), or
had access to a variety of remote-sensing data (e.g., Shuttle
Radar Topography Mission, digital photogrammetry, InSAR/
IfSAR, ASTER, SPOT, and Airborne laser scanning (ALS;
commonly referred to as LiDAR)) that could be used to pro-
vide products more tailored to the user’s needs in terms of
accuracy and acquisition time and frequency.
The combination of advances in sensors, computer tech-
nology, and analysis capabilities is clearly documented in the
peer-reviewed literature. A Web-of-Science topics search using
the terms ‘digital and elevation,’ revealed 33 publications
during 1975–89, 995 during 1990–99, and 3602 during
2000–09. Similar increases, but smaller overall numbers of
articles were noted for a title search using the same terms, the
addition of ‘geomorphology’ with the previous terms as a
topic and title search, and a topic search with the terms

‘digital’ and ‘terrain.’ It is obvious from these numbers that the
use of DTMs increased exponentially, and the likelihood for
continued use in research and education is very high, given the
current training and job opportunities afforded to students
and scientists in multiple sectors of society.
3.6.3 DTM Representation
Information representation is an important topic in geo-
morphology and geographic information science (Bishop and
Shroder, 2004). Quantitative characterization of the topo-
graphy is required if we want to formalize our understanding
of surface processes and landscape evolution. This raises the
complex question of how do we best characterize various as-
pects of the topography? Do we want to have formal repre-
sentations of surface processes and landforms, or do we want
to simply represent the altitude field? Furthermore, do we re-
quire 3D and 4D representations for modeling and mapping,
or are static cartographic representations suitable? These and
many more representation questions (e.g., indeterminate
boundaries) are related to formalizing our knowledge and
setting the stage for developing and testing new concepts in
geomorphology.
Historically, many concepts have evolved from detailed
fieldwork. However, field investigations sometimes do not
result in a definitive or quantitative statement about the
complexities of the landscape. Many observers do not possess
the ability to recognize subtle features or changes critical to
interpreting landscape or landform dynamics. DTM develop-
ment is based on a particular representational scheme (e.g.,
entity or field model) that permits a detailed means of map-
ping, quantifying, and analyzing the topography. Detailed
maps and measurements support existing field observations
and offer new virtual observations about geomorphic change.
Nevertheless, capabilities and the quality of the information
extracted from the DTM is only as good as the representational
scheme.
Similarly, geomorphological research has brought about the
need for quantitative parameters that characterize various con-
cepts of topographic structure and process-form relationships.
This information is commonly combined with process-oriented
data (i.e., rates of erosion) to provide a more definitive view of
landscape or landform dynamics. However, our ability to ac-
curately characterize specific parameters is ultimately related to
representation (parameterization schemes).
Finally, scale is an important topic in geomorphology and
GIScience, and geomorphologists require information at a
variety of spatial and temporal scales. Geospatial technologies
permit the capture and production of information at a
variety of measurement scales, spatial extents, and temporal
frequencies. Nevertheless, in addition to geographic, carto-
graphic, and measurements scales, the operational scale-
dependencies of surface processes and the hierarchical
scale-dependencies of topographic structure/organization must
also be accounted for. Currently, several aspects of scale can be
addressed via spatial resolution and spatial analysis software
(e.g., Bishop et al., 2003), however, standard cartographic
representational schemes do not formally account for the 2D
Digital Terrain Modeling 133
or 3D spatial structure of the topography that is related to
process domains, landform evolution, and tectonics.
The aforementioned discussion, however, involves ac-
counting for numerous conceptual treatments of what the
landscape actually is. It can be viewed as a complex entity or a
simple entity depending on ontological considerations. Based
on existing conceptual treatments and geospatial technologies,
a DTM is a representation of the land surface. Geomorph-
ologists must be mindful that DTMs are not exact replicas of
the Earth’s surface, and do not formally characterize surface
processes and landforms. The term representation reflects an
inability to formalize many parameters and linkages that in-
tegrate the total complexity associated with the concepts of
topography and land surface. Therefore, DTMs are surface
models developed to best represent the spatial continuity of
the altitude field. These caveats require us to formally define
what the land-surface represents and address scale issues, as
background to DTMs and their associated data structures.
3.6.3.1 Land Surface
The field of geomorphology is concerned with the geometry,
structure (internal and external), coexistence with other forms
(biota and humans), and dynamics (mode of evolution and
processes integral to their existence and evolution) of the land
surface (Bauer, 1996). The land surface is defined here as the
entire solid Earth surface (oceanic and continental crust). Al-
though geomorphologists are conducting studies at the scale
of continents or ocean basins, far more research has been and
continues to be focused on the dynamics of landscapes and
landforms. In general, a landscape is frequently thought of as
a collection of systems that produce a unique landscape
character including climate, geology, topography and land-
forms, soils, and vegetation. A more restricted interpretation
can be envisaged as a unique composite of landforms. For
example, the landscape of western North Carolina is com-
posed of the Appalachian Mountains (Figure 1). Watersheds
are a major structural landform governing the collection and
drainage of water. Each watershed is composed of some
combination of geomorphic features such as hillslopes, cliffs,
talus, colluvial hollows, rills, gullies, channels, and alluvial
Figure 1 A photograph of a section of the Appalachian Mountains
to the west of Mount Mitchell showing the drainage basins, which
are a common landform on the landscape.
134 Digital Terrain Modeling
fans. Mountainous topography of western North Carolina is
also punctuated by other landforms such as peaks, ridges, and
passes. Each of these landforms possesses their own charac-
teristic morphology and surface expression that are tied to a
complex response to recent and historical processes. The
morphometric and topographic expressions of the landforms
and landscapes are generally what are sought after in digital
terrain modeling and terrain analyses.
Terrain analysis at the landscape-scale has traditionally
focused on the Earth’s surface at a scale of 101–102 m or
greater, dependent on application and the area of interest
(Wilson and Gallant, 2000; Hengl and Evans, 2009). Terrain
analysis at the landscape-scale typically produces information
that captures patterns attributed to spatial and temporal
variations in the topography, which may or may not be related
to specific landforms or processes (Minar and Evans, 2008).
Landforms represent smaller distinct components of a land-
scape that may be characteristic of a particular process or the
result of polygenetic evolution. They exhibit internal facies,
surface roughness, and 2D and 3D morphological character-
istics (Smith and Mark, 2003). Geomorphological analyses at
the landform-scale ranges from 101 to 102 m. Character-
ization of landforms involves assessing spatial and temporal
variations in topography through a variety of morphometric
measurements, process-based analyses, statistical and math-
ematical approaches, various modeling approaches, or some
combination of these.
Space, time, and scale variations in the land surface have
been of particular interest to geomorphologists and are gen-
erally considered in the conceptual context of dynamism
within geomorphology (Rhoads, 2006). Throughout much of
the history of geomorphology, these items have generally been
investigated conceptually and analytically as separate entities.
Advances in geomorphology beyond the ‘quantitative era’
have begun to place more emphasis on integrating these items
in the context of land-surface dynamics (Walker, 2010). Land-
surface dynamics involve process-form interactions during an
event that promotes a net spatial and temporal change in the
landscape or landform topography. Operational scales of
processes, feedback mechanisms, and systems vary from event
to event. Thus, landscapes and landforms represent an amal-
gamation of events that can produce different topographic
signatures (Volker et al., 2007) or might lead to emergence
(Phillips, 1999b). Regardless of the outcome, DTMs are critical
data sets for capturing the topographic changes associated
with land-surface dynamism. Topography reflects both the
external and internal dynamism affecting change of the land
surface, as well as represents a critical forcing factor interacting
with processes during subsequent events.
Land-surface dynamism is not completely driven by ‘nat-
ural’ processes and events. An ever increasing population has
also intensified the role of humans as agents of landscape and
landform change. Impacts of early populations were more
localized as population density was lower and mechanization
was limited (Hooke, 2000). The development of large urban
areas and the use of mechanized equipment brought a shift in
impacts from the local-scale to the regional- and global-scales.
Land-use change associated with human development and
agriculture has in general played a significant role in altering
the processes associated with landscape and landform
evolution (Brierley, 2010). Ground-cover alteration or removal
can initially accelerate the magnitude and timing of sediment
and water flux. The initial increase may diminish through time
as the source material is reduced, urbanization covers surfaces,
or exotic species establish (Houben, 2008). Increasing tem-
perature and in some locales increased rates of precipitation,
associated with human-induced global climate change, are
major factors in changing weathering rates and patterns
(Gislason et al., 2008). Modification of weathering rates and
patterns can lead to significantly different rates of soil pro-
duction, soil erosion, bedrock erosion, and saprolite pro-
duction (Dixon et al., 2009). The combination of the internal
and external energy and matter within systems ultimately
alters the magnitude and form of landscape and landform
evolution. Although a consensus exists that humans play a
significant role as agents of land-surface change, major efforts
are required to accurately determine the spatial and temporal
scales at which humans modify processes and forms.
3.6.3.2 Scales and Land Surface
Identifying an objective geographical scale of unique land-
scapes and landforms through time has been an elusive goal
of science in general, and of geomorphology in particular
(Atkinson and Tate, 2000). Many studies only address the
concept of scale from a computational or analysis point-of-
view (Dikau, 1990; Evans, 2003; Bishop, 2013), or only ac-
count for a very narrow range of geographic scales (Schmidt
and Andrew, 2005; Shary et al., 2005; Staley et al., 2006),
despite scale being a major issue in geomorphology. Further-
more, these treatments generally exclude micro- to local-scale
surface roughness (Reuter et al., 2009). Surface roughness
represents local topographic variability. Topographic vari-
ations result from various objects (e.g., boulder- or cobble-
sized material on the surface of an alluvial fan) that are part of
the land surface. An adequate assessment of surface roughness
requires an initial understanding of the measurement scale
required to collect the data. The operational and geographic
scales of the processes and structures under investigation, and
the representational resolution of the DTM are important
pieces of information needed to accurately capture landform
and landscape change. Yet, the appropriate levels of scale re-
quired to characterize the topography have rarely been
examined or synthesized in a systematic way for studying
landform and landscape evolution.
From a digital terrain modeling perspective, geographical
scale is defined as the spatial extent of a land-surface feature.
Our ability to characterize and map landforms at various scales
is largely controlled by the horizontal and vertical resolution of
the DTM. In general, as the DTM resolution (cell size, triangle
size, or point spacing) becomes coarser, there is a corres-
ponding decrease in the captured land-surface information
(terrain details are lost). A finer resolution DTM possesses more
data and generally more precisely represents the topographic
variability indicative of a landscape (Figure 2). A logical infer-
ence from the previous statements is that users want to actively
seek the highest resolution data possible. There are, however,
potential pitfalls with this approach, as higher resolution data
require a greater computational expense (and might be cost
 Digital Terrain Modeling 135

Location
of Fig. 2D

NED 10 m
High : 4345.43

0 0.15 0.3 km 0 0.15 0.3 km

Low : 2396.18
(a) (c)

Elevation (m) TLS

High: 1002.81

Low: 950.627
0 3 6m

ALS Data (1m)

High: 4595.84

0 0.15 0.3 km
Low: 2494.64
(b) (d)

Figure 2 Three different DTM resolutions from the USGS Natural Debris Flow Laboratory at Chalk Cliffs, CO that show the differences in detail
ranging from medium to fine resolution. (a) 10 m DTM from the National Elevation Dataset. (b) 1 m resolution DTM generated with ALS
techniques. (c) Hillshade map from ALS data showing the location of the DTM in Figure 2(d). (d) A 2 cm DTM from a section of the channel in the
intensively monitored section of the Natural Laboratory (work funded by the National Science Foundation-Grant No. 0934131 and EAR-0643240).

prohibitive to produce), and many algorithms are not designed
to address the added variability in the data. Additionally, with
higher resolution data there is a tendency to produce data for a
smaller spatial footprint of the landscape. In this case, the
characteristics of the feature under investigation may be
overwhelmed by roughness elements existing at a finer spatial
scale. Important coarse-scale patterns will be lost when small
footprints of the land surface are examined, and this ultimately
limits a researcher’s ability to make broader inferences and
connections about landscape or landform evolution.
136 Digital Terrain Modeling
A goal then is to come up with some general guiding
principles with regard to spatial-scale and DTM resolution
before conducting an analysis. An initial step to investigating
the spatial scale of a landform or landscape is through field
observation and field measurement/mapping. A DTM user
must know the spatial scale of the land-surface object(s) of
interest, as well as the operational scales of processes that are
altering and are modified by the object(s). The goal of this
endeavor is to develop a framework for understanding the
DTM resolution required to investigate the feature(s) or pro-
cess(es). Once the observations have been made and meas-
urements obtained, a good rule of thumb in resolution
selection is that, at a minimum, the user should select a DTM
whose resolution is half of the size of the smallest feature
under investigation (Maune et al., 2007). For example, a sci-
entist concerned with quantifying gullies within a colluvial
hollow would need to have knowledge of the cross-sectional
dimensions of the gullies. If the gullies had an average width
of 1 m, then the user would want to use a square tessellation
with a cell resolution of 0.5 m or less in order to adequately
produce measures that will capture the form of the gully sys-
tems. Hengl (2006) presented a more quantitative approach to
determining a suitable grid resolution based on the com-
plexity of the terrain. For sampled elevation data (e.g., con-
tours and elevation), the density of inflection points on a land
surface can be used where the cell size (p) is at least half the
average spacing between the inflection points or finer:
l
pr ½1
2  nðdzÞ
where l is the length of the transect and n(dz) is the number of
inflection points observed. If the DTM is based on digitized
contours, a suitable raster resolution can be estimated from
the total length of the contours. Inflection points are not
actually mapped; instead they are approximated from con-
tours selected to present planform changes in the surface.
A suitable cell is defined as:
A 2003). A key to unraveling the dynamic conditions occurring
p¼ ½2
2  Sl
P
where A is the total size of the study area and l is the total
cumulative length of all the digitized contours. Hengl (2006)
also identified a more precise approach than equation [2] that
produces a 5% probability of the smallest spacing between
contours determined from the density of contours in a
given area.
The previous examples highlight how to determine the
representational scale and the horizontal cell resolution, but
one may also want to consider the vertical resolution of the
DTM. In the previous example of measuring the cross-sections
of gully systems, a depth measurement would need to be de-
termined, and this value may be used independently or in
conjunction with the horizontal measures. Determining the
vertical resolution also requires a priori information about the
depth of the gully feature. Initial work in examining the ver-
tical resolution requirements of DTMs began in fluvial
geomorphology with the analysis or determination of hydro-
logically accurate drainage networks. Gyasi-Agyei et al. (1995)
discovered that accurate hydrological networks could be
extracted from a DTM only if the ratio of average elevation
change per pixel (pixel relief or drop) to vertical resolution
(accuracy of elevation) of the DTM was greater than unity (a
ratio below unity was unacceptable for the development and
analysis of drainage networks). Average pixel drop is calcu-
lated from average slope and grid size, and vertical resolution
is measured by the error between topographic data and
benchmark elevation points. The main inference from this
work was that areas of lower relief required greater vertical
accuracy than areas in higher relief. McMaster (2002) designed
a new method to calculate hillslope length based largely on
the ratio test of Gyasi-Agyei et al. (1995) for DTM suitability,
and confirmed earlier qualitative work by Quinn et al. (1991)
and Zhang and Montgomery (1994) that the DTM resolution
must be greater than the average spatial scale of the hillslope
for accurate analyses.
Vertical resolution consideration has also been addressed
in the context of slope measurements. The ratio of the vertical
and horizontal resolution determines the minimum (non-
zero) gradient measurement that can be determined from a
DTM (Gruber and Peckham, 2009). Results from this simple
ratio can be used to determine the lower bounds of surface
gradient on hillsides that can be computed with minimal
error. Vertical accuracy, independent of elevation and gradient,
was also shown to be an important measure in governing
horizontal resolution (Walker and Willgoose, 1999). The
vertical accuracy, however, must be consistent throughout
the DTM. If vertical accuracy is consistent, then topography in
the flattest locales of a watershed controls the horizontal
resolution (Walker and Willgoose, 1999).
Determining the spatial scale necessary for examining a
particular form or landscape is not always clear cut, despite the
above guidelines. This endeavor is further confounded by the
fact that each landform or landscape undergoes processes that
operate at different spatial and temporal scales. This complex
space–time coupling of process(es) and form(s) is critical to
capturing information on landscape evolution (Lane and
Richards, 1997; Phillips, 1999a, b, 2006; Thoms and Parsons,
in many landscapes is to capture process imprints that are
recorded at different spatial scales within individual land-
forms (Schmidt and Andrew, 2005). In the previous example
of gully systems, only one spatial scale was considered.
Therefore, it may be necessary to consider using a finer reso-
lution DTM to examine topographic changes at a variety of
spatial scales. This approach permits an assessment of differ-
ences that arise within and between landforms. One approach
might be to resample the DTM to the different resolutions to
produce these measures, but there is a concern of decreasing
the accuracy of the original high-resolution data by producing
coarser resolution land-surface measurements (Liu, 2008). An
alternative is to develop an approach that samples scalar dif-
ferences to the same DTM resolution by measuring standard-
ized values with an increasing window size (Volker et al.,
2007). Another option would be to consider measures from a
variety of different DTM sources with varying scales from fine-
to coarse-resolution. In fact, given the constraints of current
high-resolution DTMs (discussed more in the following
paragraph), integrating data from multiple sources is required
to capture the landscape and landform complexity associated
 Digital Terrain Modeling 137

with hierarchical scales of the land-surface structure. There is
no simple approach to fusing data sets with varying scales.
Many concerns exist, but a fundamental concern stems from
the varying data quality associated with the different data sets.
At issue is how to assimilate the varying error budgets to
produce reliable results that are comparable from location to
location.
The aforementioned discussion highlights the inherent
space–time hierarchical connections in landscapes, and the
need for terrain analyses to consider multiple space–time
scales and multiple-resolution characterization. Space–
time hierarchical connections are by no means novel to geo-
morphology. Kugler (1964) and Dikau (1989, 1990)
conceptualized distinct space–time hierarchies for surfi-
cial features. These scales remain relevant to current geo-
morphological research (MacMillan and Shary, 2009). The
distinct space–time hierarchical scales can also serve as a
means for considering the wide-range of DTM data sources
(Table 1). Horizontal and vertical resolutions have significant
impacts on how the landforms are portrayed (Shary et al.,
2002). Table 1 was developed with the general idea of putting
commonly used DTM data sources from geomorphological
studies into the context of fine, intermediate, and coarse
resolution. Fine, intermediate, and coarse scale classifications
were used to categorize the DTM data sources and, in turn,
capture the relevant hierarchical landform levels, as con-
ceptualized by Dikau (1990), applicable for each of the DTM
sources. Table 1 demonstrates that many high-resolution data
sources are limited in terms of their applicability across all
scales, and highlights the previous point that to capture the
full range of scales across broad landscapes, there is a need to
consider integrating DTMs from multiple sources.
The ability to integrate multiple DTM sources is tempered
by a limited amount of research that has accounted for scale
dependencies within geomorphology. Phillips (2012) has
highlighted both operational problems (e.g., determining
conditions of scale dependence, what to feature or process to
represent at a given scale, and developing tools necessary to
extract and integrate the information and analyses) and the-
oretical issues (e.g., bottom–up vs. top–down influences, and

Table 1 DTM data sources in the context of Dikau’s (1990) range of scales associated with the land surface

Resolution class Data source DTM resolution Accuracy Landform

Fine-resolution Close-range oMillimeter No theoretical limitation Pico-microrelief

photogrammetry
Terrestrial laser scanning Millimeter Millimeter–centimeter Pico-microrelief
(TLS)
GPS-RTK surveying Centimeter Centimeter Pico-microrelief
Surveying Centimeter Millimeter–centimeter Pico-microrelief
Intermediate-resolution Soft-copy Meter Decimeter Micromegarelief
photogrammetry
Orthophotograpy Meter Centimeter Micromegarelief
Airborne laser swath Meter Decimeter Micromegarelief
mapping (LiDAR)
Interferometric synthetic Meter X-band Mesomegarelief
aperture radar (IfSAR H–1.25–2.5 m
or InSAR) V–0.3–3.0 m
P-band
H–2.0–4.0 m
V–1.0–5.0 m
National elevation 3.3 m, 10 m, and 30 m 1/9 arcs Mesomegarelief
dataset (NED) V–1.0 m
1/3 arcs
V–7.0 m
1 arcs
V–7.0–15.0 m
Advanced space-borne 30 m H–7.0–50.0 m Mesomegarelief
termal emission and V–7.0–50.0 m
reflection radiometer
(ASTER)
SPOT 20–30 m H–15.0 m Mesomegarelief
V–10.0 m
SRTM X-Band 30 m H–6.0 m Mesomegarelief
V–10.0 m
SRTM C-Band 90 m H–20.0 m Mesomegarelief
V–16.0 m
Coarse-resolution Global 30 arcs elevation 1 km Variable by location Macromegarelief
data ser (GTOPO30)
The Global land 1 km 1 km Variable by location Macromegarelief
base elevation project
(GLOBE)
138 Digital Terrain Modeling
multiple scale causality (see Phillips (2005) for more details)).
Future advancement of scale dependence will likely require
further advancements in data structures that account for
multiscale representation and analyses.
3.6.3.3 Data Structures
The land surface and its inherent complexity can be repre-
sented by a variety of data models (Bishop, 2012). As many
DTM are developed or analyzed in a GIS, some of the more
common data models include field (layer), entity (object),
and network data models (Goodchild, 1992). Each of the
aforementioned data models can also be linked with a re-
lational data model. Continuous topographic data are com-
monly represented by a variety of different types of field
models. Traditional field models include sampled points,
contours, polygons, tessellations, and triangular nets. If dis-
crete boundaries are present between landforms or features
within the landscape, vectors can be used to map objects.
Concerns pervade in the literature, however, with regard to
how field and object models can be effectively used when
processes and dynamism enter into the equation (Raper and
Livingstone, 1995). Despite these concerns, the afore-
mentioned data models represent the most widely used and
currently accepted approaches to DTM representation.
The terminology related to digital-elevation data has be-
come somewhat convoluted, as terrain data are used for a
variety of applications, and advances in terrain modeling and
analysis have evolved in a variety of fields. In the broadest
sense, all digital-elevation data can be considered as a digital
elevation model (DEM). A DEM is a collection of numerical
values that can be represented in a file or database to ap-
proximate topographic/bathymetric variability at given x- and
y-coordinates (Kidner and Smith, 1993; Kimerling et al.,
2009). Although the importance of bathymetric data to a
variety of research areas is recognized (including geomorph-
ology), subaerial topography is the focus throughout this
chapter. The x- and y-coordinates may be expressed as geo-
graphic coordinates (e.g., latitude and longitude, UTM co-
ordinates or other local coordinate system), or they might
simply represent a user-defined Cartesian coordinate system.
Topography is recorded as the z-value and is represented using
the function z¼ f(x,y). A height value z is mapped to a 2D
location (x,y). Elevation represents a height above a vertical
datum (see Maune (2007) for definitions and explanations of
data and various height measures). Throughout this chapter,
elevation refers to orthometric height, which is the height
measured along a plumbline from the surface of a geoid
model to the Earth’s surface (Maune, 2007), unless otherwise
specified. A bare-earth DEM is devoid of vegetation and
human-built structures and herein is referred to as a DTM.
DTMs commonly contain a combination of ground and
water-surface information, and these two measures can also be
incorporated with bathymetric data. A digital surface model
(DSM) is a DEM that contains vegetation and human-built
structures, and this term is also adopted herein.
Each of the aforementioned terms contains the word
model. The term model evolved from an ability to use com-
puters to automatically map and analyze topographic features
from a database containing x, y, and z values. Current sam-
pling techniques used to capture topographic data are unable
to record all of the heights for a given locale. An inability to
record all spatial information regarding topography means
that the elevation data and any subsequent media produced
within a computer environment represent an approximation
of the land surface, hence the term model. DTMs are com-
monly represented as raster or triangulated irregular network
(TIN) data structures (Figure 3). More recent work has focused
on developing hybrid data structures that are a combination of
raster and TIN formats (Maune et al., 2007). Other data
structures can also be used to enhance information contained
within DTMs and these include contour lines, breaklines, and
mass points.
3.6.3.3.1 Raster (grid)
Raster files are the most common data structure used by
geomorphologists for terrain analysis. This simple data struc-
ture is based on a 2D matrix containing x and y coordinates,
which are gathered at regular intervals to produce a grid
pattern of square-shaped cells (Kidner and Smith, 1993;
El-Sheimy et al., 2005; Maune et al., 2007). Raster data are
commonly used in geomorphology because they require less
physical storage space. Many current software packages possess
algorithms to rapidly produce surface measures from raster
datasets. DTM mapping software stores data in the form of a
‘hierarchical tile-block structure’ that uses the grid cell as its
foundation (El-Sheimy et al., 2005). The process of repre-
senting data in this structure is called a tessellation and is
commonly represented as a square with similar x and y di-
mensions. Tessellation shape can vary, however, and the
choice of shape is dependent on a variety of other factors (e.g.,
distance calculations). Many other shapes have been used in
place of squares. The grid is then divided up into smaller
squares called tiles, which are divided up into blocks, and then
cells, creating a grid pattern. Breaking the surface down into
tessellations permits information about a specific cell or area
to be retrieved more efficiently, thereby reducing the process-
ing time (El-Sheimy et al., 2005).
Each grid cell contains its own specific value and the value
only changes at the borders with an adjacent cell (Maune
et al., 2007). A smaller cell size represents a higher resolution,
creating a more detailed depiction of the surface area, com-
pared to a larger cell size (El-Sheimy et al., 2005). A smaller
cell size, however, generates larger file sizes (Maune et al.,
2007). A smaller storage size is ideal as it saves time and
money to store and analyze the data (El-Shrimy et al., 2005).
Although compression techniques can be applied to help
lower the storage volume (Maune et al., 2007), there are no
generally accepted compression methods.
A main disadvantage of the raster data structure is that it
does not conform to the ever-changing characteristics of the
surface topography that reflect surface changes as well as in-
ternal structure of the landform or feature. Raster DTMs are
static cartographic representations, which would require a
large number of maps (data sets) to capture even the most
basic changes to surface topography. Even in the case of high-
resolution (small cell sizes) data sources, the data structure
does not represent numerous attributes of the topography.
This partially results from the fact that a raster file is only
 Digital Terrain Modeling 139

Elevation (m)
14.376 − 16.342
12.41 − 14.376
10.444 − 12.41
8.478 − 10.444
6.512 − 8.478
4.546 − 6.512
2.58 − 4.546
N

0.613 − 2.58
−1.353 − 0.613 0 2.5 5 10 meters
(a)

N
2 meters

(b)

Figure 3 Example of a raster file (top) and a TIN file (bottom). The images are from a stream reach that was scanned with a terrestrial laser
scanner. The raster file contains 1  1 cm cells. The TIN is a mesh constructed in Cyclone that has been decimated to enhance the triangles of
the TIN points (work funded by the National Science Foundation-Grant No. 0934131).

representative of individual cells or points that have been
captured at one particular time given a data collection tech-
nique (Kidner and Smith, 1993). In essence, not all of the
variations in the land surface are represented in the raster file
because of the inability to collect all of the surface topographic
expressions at a location given the limitations in the sensors
and current data collection approaches. Therefore, there is a
need to capture a large number of raster files to characterize
the temporal variation of the topography.
Raster files are best at representing the actual bare surface
of the Earth, because the data structure does not adequately
represent specific features like vegetation or buildings as well
as other data structures. This is also a limiting factor because
these other land-surface features are important for character-
izing human alteration and land-use for better representation
and understanding of geomorphic change. Recent work using
voxels (3D pixels), however, has shown better results in
quantifying vegetation (Popescu and Zhao, 2008), as well as
140 Digital Terrain Modeling
quantifying and visualizing true 3D (surface and subsurface)
data (Hickin et al., 2009).
3.6.3.3.2 Triangulated irregular network (TIN)
Triangulation of mass points (a point cloud of x, y, z data) or
grids generally serves as the basis for generating TINs (El-
Sheimy et al., 2005). If triangulation is performed on a raster
file, a Triangular Regular Network (TRN) is generated, as op-
posed to irregularly distributed mass points, from which tri-
angulation would generate a TIN (a series of contiguous
triangles of irregular size and shape). For TIN models mass
points are connected by nonoverlapping 3D lines, creating
planes of neighboring triangles (Figure 3). Triangles are
formed from any one of a wide variety of triangulation tech-
niques. Delaunay triangulation (Kidner and Smith, 1993) has
been most commonly used in the geosciences because it meets
three basic requirements for TIN formation (Li et al., 2005):
(1) the resulting TIN from any set of points should be iden-
tical if the same algorithm is used, regardless of the starting
point of the algorithm; (2) each triangle should be nearly
equilateral; and (3) nearest-neighbor points must be used to
generate each triangle. Furthermore, the circumscribing circle
of each triangle would not include any other surrounding
points than the three nearest neighbors, hence no other vertex
exists (Kidner and Smith, 1993; Li et al., 2005).
Unlike grids, TINs are adaptable to topographic changes
that are generally associated with the heterogeneity of many
land surfaces. Therefore, TINs are a more reliable source when
it comes to specific surface details (Kidner and Smith, 1993).
For example, the triangles of a TIN define topography by
representing features (vectors) as points, lines, or polygons
(Kidner and Smith, 1993). Points representing the angle of
each triangle are called nodes, and an edge is a line connecting
two nodes (El-Sheimy et al., 2005). As a result, a TIN repre-
sents a more complex data structure. This complexity does
have advantages, as TINs can be used to characterize linear
features such as coast-lines and rivers more accurately than
grids. TINs also require fewer points than raster files to rep-
resent the land surface (El-Shrimy et al., 2005). Several
problems, however, arise when using the TIN data structure.
One problem is that the TIN size (data volume) increases
exponentially with the complexity of the terrain it represents,
which results from the data storage approach required to
maintain triangles in a TIN (Maune et al., 2007). TIN data-
bases store the topological relations of the triangular data.
Information is stored in a clockwise list of all the neighboring
edges that produces an efficient means of accessing data. A
single triangle uses the equivalent of 10 raster cells of storage
space (Kidner and Smith, 1993).
3.6.3.3.3 Contour lines
One of the most common techniques to represent topography
on a 2D map is the use of contour lines (Figure 4). Contour
lines are one of several forms of isoline maps that record equal
lines of elevation at a specified interval (National Digital
Elevation Program (NDEP), 2004; Kimerling et al., 2009).
Contours were initially derived directly from aerial photog-
raphy using a photogrammetric stereoplotter, but can also be
interpolated from point data. Spatial interpolation in this
manner results in a less accurate representation of the land
surface (El-Sheimy et al., 2005). Contour lines are commonly
used on topographic maps to reveal features in the landscape
such as mountains, ridges, and valleys. They represent a rea-
sonable approach for representing and visualizing the land
surface, because consistent rules are used to produce contours.
For example, contours will never intersect one another, which
produce a contiguous visual surface to the reader. The contour
interval (vertical distance between contours) also aids in rapid
visual connections to the mapped surface, as closely spaced
contours represent steeper terrain. Contour maps have also
played a critical role in past geomorphologic research where
profiles and cross-sections were used to quantify form (Hack,
1957; Rapp and Nyberg, 1981).
3.6.3.3.4 Breaklines
Land-surface complexity can be represented as variations in
the local relief of a landscape or landform. This might exist as
a break-in-slope, as the landscape transitions from a gradual
slope to a steep cliff. Breaklines are used to assure that dra-
matic discontinuities are maintained, as terrain conditions can
dramatically change (National Digital Elevation Program
(NDEP), 2004). Two types of breaklines are commonly util-
ized and include hard and soft breaklines. Hard breaklines
define abrupt disruptions in the continuity and evenness of
surface topography (e.g., rivers, shorelines, dams, and build-
ings). They are commonly illustrated in 3D, but can also be
illustrated in 2D. This is because there are a number of fea-
tures that are only defined by x and y points. Soft breaklines
are used to make certain that elevations are accurately
represented along linear features such as roads or pipelines
(National Digital Elevation Program (NDEP), 2004). Break-
lines have many important roles in the development of DTMs,
as they force a data structure to adhere to the heterogeneity of
the land surface (Maune et al., 2007). This would include
transitions to flatter surfaces (e.g., lake surface) or sharp
breaks-in-slope such as from the flooplain to a channel. If
breaklines were not present, many of these features would be
lost or generalized in the final DTM.
3.6.3.3.5 Mass points
Mass points represent topography as x, y, and z values at
random locations over a land surface (National Digital Ele-
vation Program (NDEP), 2004). Each point represents a slight
change in the morphology of the terrain. The use of mass
points, however, is not a completely accurate representation of
the surface, because the spaces in between the points are not
taken into account (El-Shrimy et al., 2005). Recent develop-
ments in ALS and TLS provide dense point clouds (often
oversampling a site) whereby over small areas mass points can
be used to accurately record the morphometry of a feature and
the morphometric changes through time (Wasklewicz and
Hattanji, 2009). Dense mass points also provide information
on the location of vegetation and other nontopographic ob-
jects. The removal of these objects can be accomplished by
knowing a specific location and extracting the feature, or by
applying any one of a variety of data reduction approaches to
the mass points.

5 m contours
0 12.5 25 50 meters
Figure 4 A hillshade map overlain by contours. The area is a small, recently burned basin near Goleta, CA. The hillshade was developed from a
1 cm DTM and the contours represent relative elevation in meters (work funded by the National Geographic Society-CRE Grant #8590-08).
3.6.4 Data Sources
A wide variety of DTM data sources exists, and many of these
have been covered in detail in a variety of books and review
articles. Here, the authors focus on four DTM data sources that
have been recently adopted in geomorphology. In particular,
the authors focus on TLS and terrestrial photogrammetry
techniques (TPT), ALS, and interferometric synthetic aperture
radar (InSAR). The basic principles associated with each are
presented.
3.6.4.1 Terrestrial Laser Scanning Techniques
At present there are two types of TLS instruments in use: static
and dynamic. Static scanners are stationary and are generally
mounted on a surveying tripod or mast (Figure 5). Dynamic
scanners are vehicle-mounted and motion-compensated
through the use of an inertial measuring unit (controls roll,
pitch, and yaw) and a GPS to record the location of the
scanner during the sampling campaign. A majority of the TLS
review will focus on the static scanners, as they are more fre-
quently used in geomorphology, although usage of motion-
compensated scanners is growing in coastal geomorphology.
For more detailed information see Petrie and Toth (2009a, b).
Digital Terrain Modeling 141
N
Static TLS can be classified in a number of ways, but the
authors focus on a general spatial categorization of short-,
medium-, and long-range scanners. Short-range scanners are
generally limited to less than 100 m maximum scanning
range, but commonly the practical limits are 40–60 m. A vast
majority of the short-range scanners are suitable for indoor
and outdoor use. Short-range scanners frequently employ
phase-measuring methods with a Class 3R continuous wave
laser that operates in the near-infrared part of the spectrum at
780–790 nm. A few manufacturers employ pulse-ranging
methods using Class R lasers that emit the pulsed light at
658 nm. Short-range scanners with phase-measuring capabil-
ities can possess a manufacturer’s specified accuracy of
3–6 mm and can record up to 100 000–200 000 points
per second. Many of these scanners record point-cloud data to
internal hard-drives or to a computer via Ethernet or FireWire
connections. Medium-range scanners employ pulsed-ranging
methods and can measure objects at maximum distances
ranging from 150 to 350 m dependent on the object’s surface
reflectivity (Figure 5). These scanners also use a Class 3R at
350 nm. This style of scanner collects data at the 5000–50 000
points per second range at a manufacturer’s specified accuracy
of 3–6 mm at 50 m. Long-range scanners also use pulse-
ranging techniques and can measure up to distances of
142 Digital Terrain Modeling
Figure 5 Terrestrial-laser scanner set-up in the Front Range of the
Rocky Mountains, CO (work funded by the National Science
Foundation Grant No. 0502343 and 0239749).
500 m–1.5 km. Although these scanners are considered long-
range, they do possess the capability to measure objects as
close as 3 m. A caveat with long-range scanners is that in order
to attain long-distance measurements, the object must be
highly reflective. A majority of the long-range scanners use a
Class 1 laser rangefinder emitting its infrared radiation at
905–1550 nm. However, some also use a Class R laser at
905 nm. In general, this range of TLS records 4000–12 000
points per second with accuracies in the range of 5–7 mm at
100 m.
Recording topographic data with TLS is conducted in a
similar manner regardless of the scanner’s range. All of these
instruments are using laser-based ranging to accurately record
distance. This is accomplished by measuring the time-of-flight
(TOF) of the laser light or the phase variation between the
transmitted and received signal. A TOF instrument emits a
series of short bursts of light from the laser ranger to the object
and back to the instrument. The distance is resolved by ac-
curately knowing the speed of light through the following
equation:
R ¼ v  t=2, ½3
where R is the slant range, v is the speed of the electromagnetic
radiation, and t is the measured time interval. The second
ranging option is referred to as phase-measuring, where the
laser transmits a continuous beam of radiation instead of a
pulse. Range is measured via the phase difference of the
sinusoidal wave pattern of the transmitted and received light
signals. Phase measurements are generally performed using a
digital pulse-counting technique, which records a fractional
part of the total distance (Dl). The amplitude of the laser
radiation is modulated with a sinusoidal signal and by chan-
ging the modulation pattern; the integer number of the
wavelength can be determined and added to the fractional
distance to provide a slant range, such that
R ¼ ðMl þ DlÞ=2, ½4
where M is the integer number of wavelengths, l is the known
value of the wavelength, Dl is the fractional part of the
wavelength ¼ (j/2p)  l, and j is the phase angle.
In addition to distance measurements, each instrument
also possesses a scanning mechanism (e.g., rotating mirror or
prism) that permits vertical measurements and a motor drive
or robotic servo to capture horizontal (azimuthal) direction.
In essence, the scanner measures a series of profiles around the
scanner.
A majority of the TLS instruments are also single return.
That is, once an object reflects radiation from the pulsed or
phase-based laser light, any object behind that object is not
recorded. This can produce ‘shadowing’ in the point cloud
(Figure 6). Shadowing can be reduced or removed by moving
the TLS to a position that will fill-in the data void. This ap-
proach, however, produces multiple point clouds that must be
merged into a single point cloud.
Development of a single point cloud, or what is termed as
registration of the point cloud, commonly requires a min-
imum of three common points between each cloud. The three
common points can be selected manually within a manu-
facturer’s registration program. This approach is often used in
built environments where corners of structures and other
known points are quite easily picked. This is often difficult to
apply, however, in natural topographic settings that are most
frequently used in DTM analyses. Another alternative is to use
a series of control points established before scanning a lo-
cation (Figure 7). These control points can be referenced to
real-world coordinates via any number of surveying techni-
ques, or they can just represent fixed points in 3D Cartesian
coordinate space, which has an origin at the point where the
laser is emitted from the scanner (points are recorded as
above, below, left, and right of this point of origin). Control
points are used to set-up targets or objects of known dimen-
sions or volumes. The target and objects of known dimensions
are highly reflective, so they produce quality laser returns
during the scanning process. The targets or objects of known
dimensions serve as a surface from which a series of vertices
are established in the manufacturer’s software or after-market
software. Each vertex represents a common point that can be
used to accurately link together multiple point clouds
(Figure 7). The number of targets or objects in the study site is
generally dependent on the complexity of the terrain. In
general, the more complex the terrain, the more targets or
objects will be required. Target numbers also increase with
increasing area, especially when short- and medium-range
scanners are employed in the data acquisition process. Both
manual and control-point registration approaches produce a
registration-error budget. Registration-error budgets must also

be considered in the context of ground measurements, objects
placed in the field-of-view, or spot elevation checks from other
accurately defined positions. Targets or objects of known di-
mension have been proven to provide the most accurate and
rapid way to register the data (Hetherington et al., 2005;
Heritage and Hetherington, 2007; Wasklewicz and Hattanji,
2009).
Figure 6 An image of a TLS point cloud collected from one of the
2009 USGS Landslide Hazards Program’s intensive research site in
the Station Fire located in the San Gabriel Mountains. Note the
shadowing (data void) in the gullies from a single scan location (xyz
indicator in the upper right-hand corner). The shadowing on the
right-hand side of the gullies is a result of topographic complexity,
scanner location, and the single-return nature of the scanner. The
shadows were filled by scanning from other locations within the
basin (work funded by the National Science Foundation-Grant No.
0934131).
Figure 7 Terrestrial-laser scanner targets, as seen from a digital photograph on the left. The right image is the same targets (t7, t8) as seen in
a point cloud with colors from scanner gathered with a terrestrial laser scanner (work funded by the National Science Foundation-Grant No.
0934131).
Digital Terrain Modeling 143
3.6.4.2 Terrestrial Photogrammetry Techniques (TPT)
Photogrammetry, the extraction of quantitative 3D data from
stereo-imagery, is an important tool for the collection of DTM
data (Lane et al., 2000). Traditionally, photogrammetric data
of the Earth’s surface have been derived from airborne imagery
sources, where the camera is oriented vertically in relation to
the ground surface, and there is sufficient overlap (typically
60%) between sequential photographs to replicate the optical
parallax needed to view images in three dimensions (Mat-
thews, 2008). More recently, advancements in camera tech-
nology and photogrammetric software have increased the
popularity of TPT (Chandler, 1999; Chandler et al., 2005;
Matthews, 2008). TPT permits the development of high-
resolution DTMs for geomorphic analysis using digital cam-
eras and specialized photogrammetric software. Conducted
properly, this complex process allows high-resolution DTMs to
be created from vertical or oblique images of the ground
surface (Clarke, 1995; Lane et al., 2001; Redmondino and
El-Hakim, 2006; Matthews, 2008). DTM derivation from TPT
requires careful consideration of three phases of the process
including: (1) image acquisition; (2) stereo-matching of
photographs; and (3) manual or automated DTM generation.
Image acquisition refers to the process of camera selection,
calibration, and the recording of the images. Traditional
photogrammetry techniques relied on metric cameras for the
acquisition of images. Metric cameras are more easily imple-
mented in photogrammetric work, as the geometry of the
camera is stable, the coordinate system is included in the
images as fiducial marks, and the lens distortion is well de-
fined (Peipe and Stephani, 2003). Although these highly
144 Digital Terrain Modeling
accurate cameras are ideal for this type of data collection, the
camera costs are prohibitive to most researchers. In recent
years, geoscientists have begun to rely on consumer-grade
nonmetric digital cameras to acquire images for photogram-
metry and DTM creation (Chandler et al., 2005).
Since the mid-1990s, the cost of consumer-grade digital
cameras has decreased dramatically whereas sensor and lens
quality have increased. Chandler et al. (2005) compared the
photogrammetric capabilities of low-cost digital cameras. They
found that these types of cameras are suitable for the pro-
duction of high-quality DTMs provided that errors associated
with lens imperfections are considered (Chandler et al., 2005;
Peipe and Tecklenburg, 2006). The process of determining
camera parameters and lens distortions is known as camera
calibration (Fraser, 1997; Chandler et al., 2005; Remondino
and Fraser, 2006; Peipe and Tecklenburg, 2006). Stand-alone
software is available that allows users to calibrate their cam-
eras and lenses. These self-calibration bundle adjustments
(SCBA) provide users with distortion and camera parameters
for digital photogrammetry systems (Fraser, 1997; Peipe and
Tecklenburg, 2006). In addition, many photogrammetric
software packages now include SCBAs (Chandler et al., 2005;
Matthews, 2008; Sturzenegger and Stead, 2009). Once cam-
eras have been properly calibrated, images of the research area
need to be acquired. Although there is some flexibility in the
methods used to take pictures, extracting the best possible
DTM from photogrammetric techniques requires careful at-
tention to internal and external control, camera position, and
image processing.
Internal control of a TPT study area refers to the insertion
of objects of known length into the acquired images. These
objects are used, in combination with the lens parameters
obtained from camera calibration, to produce a highly ac-
curate and precise local coordinate system (Matthews, 2008).
Local coordinate systems can be projected to geographic co-
ordinates by surveying control objects using a total station or
GPS survey, though these methods may degrade the relative
accuracy of the photogrammetrically derived DTM.
TPT images can be obtained from strategic vantage points
at ground level, at elevated positions (e.g., from ladders or
platforms), or from low-altitude airborne positions, such as
unmanned aerial vehicles (UAVs). Regardless of the platform
from which images are acquired, the derivation of highly ac-
curate DTMs requires a minimum overlap of 60% between
images and a camera angle as close to nadir (perpendicular to
the object of interest) as possible (Matthews, 2008). Where
possible, the use of a tripod or some other stable base is rec-
ommended (Haneberg, 2008). In addition, the axes for each
camera image must not cross (i.e., converge) between con-
secutive images or the software may not be able to obtain a
stereographic model.
Once the study site is photographed, photogrammetric
software is used to produce DTMs. This process requires three
steps: pixel matching, feature matching, and surface creation
(Fraser and Cronk, 2009). Pixel matching refers to the process
by which similar points are identified in multiple images.
These points may be obtained from both control objects in the
scene, or from points with other distinct characteristics. Fea-
ture matching follows the same process, but matches similar
areas between images. Both the pixel- and feature-matching
processes are automated in most current generation photo-
grammetric software (Haneberg, 2008; Matthews, 2008). Once
pixels and features have been matched, DTMs are automatic-
ally created by triangulation (for matched points and areas)
and interpolation (unmatched points and areas) (Brasington
and Smart, 2003).
3.6.4.3 Airborne Laser Scanning
ALS is conducted from either a fixed-wing aircraft or a heli-
copter. A basic ALS system is composed of a laser-ranging unit
(mostly pulsed time-of-flight measurement systems) that in-
cludes a transmitter and receiver optics. This is mounted over a
hole in the fuselage of an aircraft. There is also an optical-
scanning mechanism, such as a rotating mirror (with angular
encoder) that directs the pulsed-laser light in a cross-track
direction at known angles. It should be mentioned that there
are other types of optical-scanning mechanisms that use dual
mirrors, a spinning optical polygon, a nutating (rocking,
swaying, or nodding motion) mirror producing an elliptical-
scan pattern (called a Palmer scan), or a pair of linear fiber-
optics arrays. The forward motion of the plane allows a series
of ranges to be gathered along the flight line. The combination
of a series of ranges along successive lines and angular meas-
urements permit large swaths of the land-surface elevation to
be recorded. When overlapping swaths are combined, the
measurement densities are often in the range 0.2 to 50 points
per meter square. The range of values depends on the aircraft
height and velocity (Beraldin et al., 2010).
An airborne GPS antenna is mounted on top of the aircraft
and data are differentially corrected with a ground-based ref-
erence GPS. Ground-based reference information can be
provided from a GPS receiver operating at a base station, a
wide-area augmentation system, or satellite-based augmen-
tation system (e.g., the Japanese Multifunctional Satellite
Augmentation System or commercial systems such as StarFire
or OmniSTAR), continuously operation reference stations
(CORS), or by employing precise point positioning (PPP)
technology. An inertial measurement unit (IMU) is used to
record acceleration data and orientation of the aircraft (Petrie
and Toth, 2009c). A typical sampling rate for the IMU is
B200 Hz. Data from the GPS and IMU are combined to re-
construct the air trajectory of the aircraft to an accuracy of
o10 cm. The scanner, GPS, and IMU are synchronized by an
onboard control and data-recording unit. The synchronization
is keyed by the GPS pulse-per-second signal. This signal trig-
gers the internal clocks in the IMU and scanner to provide a
time stamp for each range measurement and trajectory pos-
itions (Beraldin et al., 2010).
Many commercial ALS systems operate at wavelengths of
800–1550 nm. The wavelength is important because different
wavelengths will perform differently depending on the object
being recorded on the land surface. For example, in alpine
settings where snow or glacial ice is present, it would be
preferable to use an ALS system that makes use of near infra-
red wavelengths (800–1300 nm), as snow and ice have low
reflectivity at 1550 nm (Beraldin et al., 2010). Laser-pulse
repetition rates can be measured at up to 150 Khz, but many
ALS systems have a scan rate in the 30–50 Khz range.

ALS systems have the ability to record multiple echos from
a single pulse. For example, in a forested ecosystem, the beam
is wide enough that it could hit the edge of a leaf or branch in
the upper canopy and the remaining portion of the beam
would continue downward either striking the ground or
additional branches or leaves. The elevation at the top of the
canopy is called the ‘first return’, whereas the ‘last return’ could
potentially be the ground. The terms terrain (ground) and off-
terrain (above the ground) are also commonly used terms to
classify the multiple echoes. Thus, a coarse (in the sense that it
is not recording every branch and leave) DSM could be created
incorporating the multiple echoes (in earlier ALS systems this
ranged from 3–5 echoes). Alternatively, if a data-reduction
method was employed to only consider the last returns, a
DTM could be developed from this same dataset.
Recent developments (B2004 to present) in ALS technol-
ogy have allowed the adoption of small footprint (recording a
diameter of up to 100 cm) ALS systems that record the full
waveform (FW) of a laser pulse. FW-scanners permit the
complete waveform of each backscattered pulse to be digit-
ized. A major benefit of this approach to ALS is that it provides
the user with the ability to improve the classification of terrain
and off-terrain objects. The major advantages of FW ALS sen-
sors to geomorphological research are the production of
more accurate DTMs, an ability to consider vegetation as a
roughness component, and the ability to measure the spatial
extent or spread of vegetation (Devereux and Amable, 2009).
Classification of terrain and off-terrain points, however,
is reliant on proper segmentation or filtering of the FW data.
There has not been a great deal of application of FW to
date within geomorphology because of the novelty of the
technology.
After each ALS surveying mission significant data process-
ing is required to reach a final product. At the heart of this
processing is the merging of the three datasets, GPS ground
data, navigation data (GPS/IMU), and the ranging data, from
each mission. A variety of other information also includes logs
of aircraft behavior, instrument behavior, and recording errors
that may be included at some point in the processing step
(Devereux and Amable, 2009). These data streams are inte-
grated into a database.
Systematic errors also are commonly associated with ALS
instrumentation. Several conditions produced during oper-
ation of a light aircraft lead to systematic errors and they in-
clude: high vibration; variable pressure and temperature;
impacts from rough landings; changing orientation; and in-
consistent electrical supply. Each of the aforementioned con-
ditions is a major factor leading to varying degrees of
systematic error (Shrestha et al., 2007). Furthermore, the
natural aging process of the ALS system can also cause vari-
ations in measurements over time, and this is particularly
important to space–time measurements in geomorphology.
The NSF funded National Center for Airborne Laser Mapping
(NCALM) has presented a number of scenarios by which the
calibration processes varies from mission to mission (Shrestha
et al., 2007), and Katzenbeisser (2003) has also provided de-
tailed examples of calibration issues and techniques. One
common approach, used by NCALM has been point-by-point
comparisons of kinematic GPS data at a permanent test lo-
cation (1000–10 000 points). Before each extended mission or
Digital Terrain Modeling 145
when the ALS instrument is deployed, field crews establish
kinematic GPS lines on hard surfaces such as roads and land
strips. These point-to-point comparisons are used to accurately
identify the zero point and scale of the scanner to within a few
centimeters (Shrestha et al., 2007). Vertical accuracies of ALS
derived data range from 710 to 715 cm, but this is
dependent on careful calibration and processing. Horizontal
accuracies are typically worse than the vertical accuracies and
range from 15 to 100 cm (Baltsavias, 1999; Maas, 2002;
Maune, 2007).
3.6.4.4 Interferometric Synthetic Aperture Radar
Synthetic aperture radar (SAR) refers to a rapidly evolving
form of radar data processing that allows for the character-
ization of topography using airborne or spacecraft-mounted
radar systems (Gabriel et al., 1989; Gens and Van Genderen,
1996; Massonnet and Feigl, 1998; Bürgmann et al., 2000;
Rosen et al., 2000; Smith, 2002; Zhou et al., 2009). SAR sys-
tems use an antenna to produce a narrow band of electro-
magnetic energy directed at the surface. SAR systems both
transmit and receive electromagnetic waves at wavelengths
commonly ranging from 3 cm (X band) to 24 cm (C band)
(Massonnet and Feigl, 1998), although, shorter wavelength
sensors in the Ku band (roughly 1.7 cm) have been recently
used (Okada et al., 2007). At these wavelengths, the energy is
able to travel through the atmosphere with minimal degrad-
ation in all weather conditions (Okada et al., 2007). The phase
(time delay) and amplitude (energy intensity) of the radar
echoes are then used to derive both topographic and pos-
itional data from the SAR image, as well as information about
the surface (Smith, 2002; Zhou et al., 2009).
Interferometric SAR (InSAR) refers to the process of using
information from two or more SAR images of the same lo-
cation to improve the accuracy of elevation data or calculate
changes in surface elevation over the study area. The product
of the two images is known as an interferogram. The inter-
ferogram can be produced using: (1) two (or more) SAR an-
tennae mounted perpendicular to the flight path or orbit on
the same platform (single pass InSAR); or (2) the same SAR
antenna during sequential flight paths or orbits (multipass
InSAR) (Gabriel et al., 1989; Smith, 2002; Zhou et al., 2009).
The length of time between images used to create the inter-
ferogram, referred to as the temporal baseline, may range from
o1 s (for single pass InSAR) to days, months or years (for
multipass InSAR) (Zhou et al., 2009). The temporal baseline
for multipass is generally dictated by the orbital frequency
of the satellite or the time-scale of interest for the particular
application. Readers seeking a more thorough review of InSAR
theory and methods are referred to Gabriel et al. (1989),
Gens and Van Genderen (1996), Massonnet and Feigl (1998),
Rosen et al. (2000), Smith (2002), and Zhou et al. (2009).
Depending on aircraft height, sensor size and atmospheric
conditions, InSAR is able to produce DTMs ranging in reso-
lution from B30 m (Roering et al., 2009) to less than 1 m
(Okada et al., 2007). The ground resolution of the pixel is
dependent on numerous factors, including the wavelength of
radiation emitted by the sensor, the distance of the aircraft or
satellite from the surface, and the dielectric properties of the
146 Digital Terrain Modeling
surface (e.g., the amount of scatter and absorption) (Zhou
et al., 2009). DEMs produced from InSAR are generally of
inferior quality when compared to those derived from ALSM
or TLS methods. The higher level of inaccuracy associated with
a single set of topographic measurements from InSAR is re-
lated to the nature of satellite navigation and tracking, errors
in the estimation of interferometric phase and atmospheric
conditions (e.g., water vapor content) (Smith, 2002). These
factors combine to produce elevation measurements that may
contain uncertainties on the order of meters (Smith, 2002).
The strength of InSAR data lies in its ability to detect fine-
scale changes in surface elevation or characteristics associated
with both natural and anthropogenic processes (Gabriel et al.,
1989; Smith, 2002; Zhou et al., 2009). A temporal sequence of
data collection allows for the differencing of interferograms
and calculation of phase changes in the spatial or dialectic
properties of the surface (Zhou et al., 2009). The accuracy of
such an analysis is highly dependent on the proper selection
of a control area, where surface deformation and dielectric
changes have not occurred between image acquisitions.
A perfect control area would be topographically homogenous,
low gradient and stable. In addition, optimum quantification
of the differences in phase between images would be achieved
when the sensor is in the same orbital configuration and ob-
servational direction, and all postprocessing was conducted
using the same procedures (Zhou et al., 2009). Differences in
phase recorded at the control area are then used to correct for
the interferogram so that differences in dielectric properties
and surface deformation may be calculated (Gabriel et al.,
1989; Zhou et al., 2009).
InSAR data have been used in a wide variety of applications
in geomorphology. Although these data were originally used
to produce models of surface topography at a single point in
time (Graham, 1974; Zebker and Goldstein, 1986), the recent
trend has been to analyze phase differences in the inter-
ferograms to quantify landscape change. Dielectric differences
related to changes in soil moisture content (Nolan et al.,
2009), depth to bedrock (Elsherbini and Sarabandi, 2010),
Figure 8 Ground (brown) and off-terrain (highlighted in green) data collected with a TLS. The image is from a point cloud with a picture
superimposed over the points (work funded by the National Science Foundation-Grant No. 0934131).
wetland analysis (Hong et al., 2010), and forest canopy
structure (Kellndorfer et al., 2010; Solberg et al., 2010) are all
recent examples of analysis of dielectric differences in the
interferogram. Phase differences have also been used to
quantify surface deformations (changes in elevation) associ-
ated with both natural and anthropogenic processes. Recent
examples include deformation associated landsliding and
sediment transport (Catani et al., 2005; Roering et al., 2009;
Lauknes et al., 2010), tectonic activity (Jo et al., 2010; Poland,
2010; Wen and Ma, 2010; Xu et al., 2010), soil consolidation
(Kim et al., 2010), glacial dynamics (Capps et al., 2010; Shugar
et al., 2010) and mine collapse (Plattner et al., 2010).
3.6.5 Preprocessing
Ground and off-terrain data (e.g., vegetation and components
of the built environment) are simultaneously gathered as x, y,
z data in a point cloud (Figure 8). This may not be a problem
in studies employing a DSM, but most geomorphological
studies require bare-earth models. The issue of classifying
ground and off-terrain points must be considered at the onset
of preprocessing the DTM source data. Of particular import-
ance is the removal of off-terrain points from the point cloud
before DTM production. Removal of off-terrain points is a
form of data reduction and is a critical step to producing a
hydrologically sound DTM. Even after removal of the off-
terrain points, the point cloud may remain exceedingly large,
and further reduction of ground points might be required to
permit data analysis or representation to occur in an efficient
manner. Larger point clouds might also be reduced regardless
of vegetation removal. Extracting ground data from and re-
ducing data from point clouds is an operational activity, but it
is particularly important for geomorphologists to understand
these processes to either critically evaluate the already pre-
processed data or to determine a starting point for pre-
processing on her/his own.
Y X
0.5

3.6.5.1 Point-Cloud Reduction Techniques
Reducing the complexity of point clouds is not a trivial matter.
Pfeifer and Mandlburger (2009) conveyed a number of rea-
sons why this is the case: (1) the geometrical structure of
objects located only slightly above the ground surface can be
very similar to topographic features, making it quite difficult
to differentiate between the two; (2) other attribute infor-
mation is commonly recorded with the locational data (e.g.,
echo number, intensity, rgb values from digital imagery, or
echo-width extracted from full-waveform) and cannot be used
to classify off-ground points because items such as last echoes
from multiple echoes (in ALS these represent multiple return
data classifiers) can contain multipath errors that would place
the point well below the land surface; (3) it is difficult to code
and incorporate an algorithm that accounts for contextual
knowledge (a common form of human interpretation) into a
feature-extraction program; and (4) several data-reduction
techniques are highly influenced by systematic and random
measurement errors (Mandlburger et al., 2008). At a min-
imum, data quality must be determined from both field
measurements and from information gathered from the
digital data set. Once this is determined, errors must be re-
moved or minimized through statistical or modeling ap-
proaches to proceed forward with the data reduction
component. Each of these steps increases the potential for
more data storage as well as increasing the labor and com-
putational time involved in the data reduction process. Three
types of data reduction techniques are highlighted and include
decimation, filtering, and segmentation.
3.6.5.1.1 Decimation
A variety of fields such as geodesy, computer vision, computer
graphics, computer science, and applied engineers are de-
veloping techniques that ‘thin’ or reduce the number of points
used in the production of DTMs. The thinning of data is
commonly termed decimation. Decimation can be accom-
plished at a variety of different levels, and involves the re-
duction of the original point cloud via point removal,
refinement, or cloud segmentation. Point-based simplification
is more rapid and less demanding of computer memory be-
cause unlike meshes (another term for a type of TIN) topo-
logical and geometric features are not preserved (Dey et al.,
2001; Moenning and Dodgson, 2003). The greater efficiency
afforded point-based approaches indicates a potential for
further advances in the near future. A more detailed review can
be found in Pauly et al. (2002).
Three main types of point-based simplification occur: (1)
clustering; (2) particle simulation; and (3) iterative simplifi-
cation methods (Pauly et al., 2002). Clustering is an approach
that splits the point clouds into subsets and replaces them
with representative points (Rossignac and Borrel, 1993;
Brodsky and Watson, 2000; Shaffer and Garland, 2001; Pauly
et al., 2002). Particle simulation uses interparticle repeling
forces to compute sampling positions based on a desired
number of points, but have not been widely applied (Turk,
1992). The iterative simplification method iteratively reduces
the number of points along a local plane or surface by re-
moving the least important points at each of the iterations
(Garland and Heckbert, 1997; Alexa et al., 2001; Linsen, 2001;
Digital Terrain Modeling 147
Fleishman et al., 2003). Iterative simplification is the most
promising approach for geomorphology because it relies on
point extraction based on curvature, distance between a point
and a specified target, and locality within a neighborhood as a
means of defining the points to remove. This approach should
only be used in the case where all of the systematic errors in
the point cloud have been removed through sensor cali-
bration. Random errors should also be removed by an inter-
polation strategy with measurement noise filtering.
Despite the promise of this technique, concern exists at
present with the current algorithms used to decimate point
clouds. Mandlburger et al. (2008) suggested not using deci-
mation methods for large high-resolution DTM data because
they work from a fine-to-coarse, which involves triangulation
of the entire point cloud. This is computationally complex and
some GISs would not be able to handle the number of tri-
angles required to conduct this approach. Therefore, deci-
mation is currently feasible for high-resolution DTMs with
small spatial coverage, as has been applied by Dupuis et al.
(2006). Decimation was used to develop small, but accurate
TIN data for use with autonomous long-range rover navi-
gation. Dupuis et al. (2006) particular application required
only a narrow DTM (small area), measuring slightly larger
than the width of the rover and was therefore, computation-
ally feasible.
Refinement methods represent multipass algorithms that
resample from coarse-to-fine resolution starting with a min-
imal initial approximation (Heckbert and Garland, 1997;
Pfeifer and Mandlburger, 2009). Each subsequent pass in-
corporates one or more additional points as vertices for tri-
angulation until the desired tolerance is achieved (Pfeifer and
Mandlburger, 2009). Mandlburger et al. (2008) presented a
modified refinement approach using a hybrid DTM (regular
grid and breaklines, structure lines, and spot heights). A
maximum height tolerance and a maximum planimetric point
distance were used as the basic parameters for this data thin-
ning approach. An initial approximation of the DEM is gen-
erated using Constrained Delaunay Triangulation, and each
cell is subsequently refined by iteratively inserting additional
points until the height tolerance is achieved. Mandlburger
et al. (2008) reduced the original data set to B8% of its ori-
ginal size. A comparison of this simplified data set with low-
resolution hydrological modeling found similarities in terms
of water height, but drastic differences in a variety of other
hydrological measurements (e.g., flooded area and flow vel-
ocity). Refinement and feature approaches, in conjunction
with calibrated and corrected high-resolution DTMs, are im-
portant to geomorphological research where data reduction is
needed. Furthermore, Mandlburger et al. (2008) made a
strong case for the need of high-resolution DTMs in hydrology
research.
3.6.5.1.2 Filtering
Filtering is a process designed to extract terrain points from
off-terrain points that commonly occur in many DTM data
sources. A vast majority of the ground-filtering techniques for
point clouds assume gradual variations in topography (e.g.,
smooth variation in slope from point to point). Therefore,
terrain points are detected through a combination of slope,
elevation difference, and local elevations (Vosselman, 2000;
148 Digital Terrain Modeling
Zhang and Whitman, 2005). The more complex the topo-
graphy, however, the more difficult this process becomes, and
is further complicated by other natural and human-made
objects on the surface. Silván-Cárdenas and Wang (2006),
subsequently reviewed by Liu (2008) and Pfeifer and Man-
dlburger (2009), have identified three generally accepted
classes of filtering processes: interpolation-based (Kraus and
Pfeifer, 1998), slope-based (Vosselman, 2000; Roggero, 2001;
Sithole, 2001; Shan and Sampath, 2005), and morphological
(Zhang et al., 2003; Chen et al., 2007).
Interpolation-based filters use weighted linear least-squares
interpolation to iteratively estimate the topography under
consideration (Chen et al., 2007). Residuals from the ap-
proximated surface are assigned to the elevation values
(negative for terrain points and positive for off-terrain points).
Negative residuals are more heavily weighted, whereas positive
residuals are weighted less. Residuals with higher weights are
kept in the iterative process until the surface gets closer and
closer to the ground (Crosilla et al., 2004). This technique
does not work well in steep topography, but Lee and Younan
(2003) developed a combined modified linear-prediction fil-
ter and an adaptive filter to produce better results.
Slope-based point-cloud filters assume distinct differences
between the slope of the topography and the slope of an off-
terrain object (Zhang et al., 2003). If the maximum value of
the slope between a point and any other point for a given
neighborhood is less than a threshold value, it is classified as a
terrain point (Vosselman, 2000). Threshold development is
highly dependent on knowledge of the topographic variability
in a study area. Slope-based filters perform well in gentle
topography, but accuracy declines as the slope of the terrain
increases (Sithole and Vosselman, 2004). Sithole (2001) im-
proved the ability to produce accurate data in steep topo-
graphy by developing a varying threshold with slope.
Morphological filters are derived from mathematical
morphology (Harlick and Shapiro, 1992). These point-cloud
filters are based on an admissible height difference that is a
function of planimetric distance (Pfeifer and Mandlburger,
2009). Points are identified as off-terrain if one or more height
differences between it and its surrounding points are above
the admissible height difference. The neighborhood contrib-
uting to the height difference is limited to a maximum dis-
tance. Other variants of this style of filter have expanded its
capabilities. For information on these variants, the reader is
referred to Zhang et al. (2003) and Kobler et al. (2007).
3.6.5.1.3 Segmentation
Point-cloud segmentation is a type of data reduction whereby
point clouds are aggregated into homogeneous segments (e.g.,
roads, buildings, motor vehicles, and topography) and sub-
sequent analyses are performed within the context of the
segmented cloud. This process can be accomplished in variety
of ways. Typically, smooth surfaces are segmented using
region-growing techniques. A normal vector or its change is
used to produce a homogeneity criterion to segment out pla-
nar surfaces and topographically varying surfaces (Pfeifer and
Mandlburger, 2009). A cloud segment is generated by using a
threshold value. If the threshold value of the angle between
the surface normal and normal at the nearest surface point to
be expanded is below the threshold, the point is accepted into
the segment (Vosselman et al., 2004). Other criteria might be
employed such as proximity of points (considering only
points close to one another) and locally planar (a candidate
point is accepted based on its orthogonal distance to a plane).
Cloud segmentation may be most advantageous to geo-
morphologists for filtering and extracting objects. Large
buildings and other objects might be removed to limit data for
analyses. The potential also exists to use segmentation to ex-
tract surficial features of varying topographic complexity, but
to date, this has not been fully applied in geomorphology.
3.6.5.2 Surface Interpolation
In general, interpolation is required to generate a higher
resolution DTM, or to utilize spot elevations to generate a
continuous gridded surface that is typically required for
morphometric analyses or modeling efforts (El-Sheimy et al.,
2005). The need for interpolation techniques stems from a
variety of reasons. Shadows or data voids might result from
radar shadowing. Radar illumination is absent during sur-
veying missions because the topography, either concave or
convex relief features, reflects the energy, which does not
permit objects behind or below the terrain to be recorded. It is
also the case that some objects absorb the energy associated
with the radar illumination to generate a data void. Shadows
are also associated with TLS techniques (Figure 6). Another
reason to interpolate data stems from the data source not
being at the desired sampling interval. A DTM user might have
two data sets that are of different sampling intervals and to
make these data sets comparable, one or both might be in-
terpolated to the same resolution. Sampling intervals might
also be discontinuous or at a scale that does not meet the
demands of the desired representation or analysis. Inter-
polation could be used in both cases to improve the resolution
of a DTM. Off-terrain data might produce another scenario
where interpolation is needed. For example, filtering the
vegetation from the point cloud would leave discontinuous
data (larger intervals between terrain points) or data voids.
Interpolation algorithms could be used to minimize the effects
of these voids in the final DTM.
Point heights are commonly used to generate DTMs with the
aid of any number of interpolation techniques. Interpolation is
a process whereby an unknown elevation is estimated from
known elevations of neighboring points. Interpolation methods
are categorized into three dominant approaches (Table 2).
Local neighborhood methods use surrounding points to esti-
mate unknown elevation values. Two algorithmic approaches
include fitted functions and weighted averages. The fitted-
function algorithms examine small patches of data to dete-
rmine the coefficients of selected analytical bivariate functions
based on sample points inside the window. This is frequently
accomplished using least-squared criterion. Weighted average
algorithms use a weighted summation of elevation values from
a running subset of surrounding elevation values, without
the aid of an intermediate parametric surface. Geostatistical
interpolation approaches involve understanding the inherent
spatial structure of the data (local, regional, and global spatial
structures) and then predicting missing values based on the
spatial structure. Kriging techniques are a common geostatistical

Table 2 Examples of methods employed for DTM generation from each of the three major categories of spatial interpolation techniques (Mitas
and Mitasova, 1999)
Local Geostatistical
Univariate
Nearest neighbors Simple kriging
Natural neighbors Ordinary kriging
Inverse distance weighting Block kriging
Triangular irregular network related interpolations Factorial kriging
approach. Splines are another common gridding or inter-
polation technique that have also been referred to as variational
approaches (Mitas and Mitasova, 1999). Spatial interpolation
via this approach assumes missing values should be closely
related to, or pass through existing data points and should be as
smooth as possible (Hutchinson, 1989; Mitasova and Hofierka,
1993). It is clear from this very brief overview of the different
interpolation algorithms and approaches, that each technique is
not equivalent and does not produce the same result. Therefore,
an analyst/scientist must have a clear understanding of the
mathematical underpinnings of interpolation algorithms that
are used to generate a DTM, or what is needed to produce a
high-quality DTM from the point cloud. Individuals requiring a
mathematical and statistical treatment of spatial interpolation
should refer to Li et al. (2005), El-Sheimy et al. (2005), and
Maune (2007).
Interpolation techniques estimate unknown elevations and
have different levels of accuracy. Several comparative studies
have examined how different interpolation techniques per-
form under a variety of test scenarios (Carrara et al., 1997;
Robeson, 1997; Burrough and McDonnell, 1998; Zimmerman
et al., 1999; Wilson and Gallant, 2000). In general, geosta-
tistical approaches have performed better than other inter-
polation approaches in terms of interpolating elevation values
accurately (Zimmerman et al., 1999; Wilson and Gallant,
2000, Erdogan, 2010). In some situations, however, neigh-
borhood approaches such as the inverse distance weighted
(IDW) algorithm and radial-basis functions can match or
outperform results from geostatistical approaches (Weber and
Englund, 1992; Brus et al., 1996; Declercq, 1996; Aguilar
et al., 2005).
These disparate conclusions have more recently been rec-
onciled in studies conducted by Chaplot et al. (2006) and
Heritage et al. (2009). Chaplot et al. (2006) used a nested
scales approach (DTMs at the microplot, hillslope, and
catchment scales) to evaluate the accuracy of several inter-
polation techniques (IDW; ordinary kriging (OK); universal
kriging (UK); multiquadratic radial basis function (MRBF);
and regularized spline with tension (RST)). Heritage et al.
(2009) tested five different interpolation algorithms under five
different sampling strategies that included cross section, bar
outline only, bar and chute outline, bar and chute outline with
spot heights, and aerial ALS equivalent derived from degraded
TLS data. Key findings from both studies indicated data
sources with greater point densities and those that contained a
greater number of survey point positions relative to the
breaks-in-slope of the surficial feature(s) of interest signifi-
cantly reduced the error differences between interpolation
Digital Terrain Modeling 149
Spline
Multivariate
Universal kriging Regular thin plate spline
Kriging with an external drift Regular spline with tension
Simple cokriging Thin plate spline
Ordinary cokriging Thin plate spline þ tension
techniques. The basic premise from this work is that with
higher resolution data, the differences between the various
interpolation approaches are reduced. Chaplot et al. (2006)
also identified where particular interpolation techniques
would be most effectively used: (1) kriging worked particularly
well where sample data densities were low; (2) IDW and RST
performed better in environments where structure of the
height was weak (microplots); (3) OK was better suited in
topographically smooth conditions and interpolation across
multiple scales; (4) MRBF provided better results in moun-
tainous topography; and (5) RST was found to perform well at
hillslope and catchment scales.
Beyond existing interpolation techniques, there is a need to
advance and test new interpolation algorithms that will im-
prove our ability to address topographic complexity and per-
mit the generation of high-resolution DTMs. A variety of new
approaches have been proposed over the past decade. Almansa
et al. (2002) attempted to apply an image interpolation ap-
proach called the absolutely minimizing Lipschitz extension
(AMLE) model to DTM development. The main advantage of
this approach was that it did not produce artifacts in the in-
terpolated surface, which satisfies a maximum principle
(Almansa et al., 2002). A major disadvantage of AMLE to
geomorphologists is that it cannot interpolate slopes and
other land-surface parameters. Furthermore, AMLE requires
use with another interpolation technique, which further
complicates the use of this promising approach. Almansa et al.
(2002) suggested combing AMLE with Kriging or thin-plate
spline to overcome this issue. A relatively recent novel advance
was a hybrid interpolation method, which integrates linear
(bilinear) and nonlinear (bicubic) interpolation methods (Shi
and Tian, 2006). Root mean square error (RMSE) values from
a hybrid interpolation method were found to be significantly
lower (greater accuracy) than the other bilinear or bicubic
methods. This approach has promise for areas with hetero-
geneous topography. A final example of advances in new in-
terpolation approaches is an approach specifically designed to
handle high-resolution data sets. Isenburg et al. (2006) com-
bined finalized point streams and streaming meshes to link
several streaming software modules, which collectively used
large data sets of ALS points, produced giant TINs, and gen-
erated a high-resolution raster DTM. This approach generated
a 50 394  30 500 raster DTM with 6 m post spacing in 16 bit
binary BIL format (3 GB), using less than 100 MB of main
memory and 300 MB of temporary disk space in just over an
hour on a standard laptop (Isenburg et al., 2006). This ap-
proach provides an opportunity to convert large mass point
data structures to a DTM with minimal computer processing
150 Digital Terrain Modeling

power and less time than traditional techniques. This type of DTM under investigation (Reuter et al., 2009). The RMSE is
interpolation technique has been under developed to this often expressed as follows:
point. We believe that further advances in this type of inter- sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Pn 2
polation technique would greatly benefit geomorphologists i ¼ 1 ½zðsi Þ  zREF ðsi Þ
interested in high-resolution DTMs. RMSE ¼ ½5
n

3.6.6 DTM Error Assessment
Error assessment is used to evaluate the accuracy of a DTM.
Geomorphologists rely on the accurate representation of land-
surface heights to generate land-surface parameters, assess
topographic patterns, and analyze terrain changes (Reuter
et al., 2009). An initial assessment of DTM errors is an
evaluation of elevation values compared to corresponding
field measurements that are more accurate. This is not always
practical, however, because field measurements also exhibit
error (Florinsky, 1998), and field data may be difficult to
obtain. Unfortunately, acceptable guidelines for assessing
DTM accuracy for geomorphological applications do not exist
(Reuter et al., 2009), although numerous approaches can be
used to assess the quality of a DTM. These include visual,
manual, mathematical, statistical, or modeling approaches to
error assessment.
3.6.6.1 Error Issues
where z(si) is the DTM elevation at the spot location (si), and
zREF(si) is the reference elevation at the same location.
The RMSE metric is the most common metric used to
characterize global error. For example, we used it to examine
the global error associated with TLS measurements of the
topography in a recently burned watershed (Figure 9). Two
TLS surveying missions were conducted with seven control
points (Leica Geosystems HDS targets), which produced a
total of 187 reference points. The RMSE for the first survey was
1.38 mm and 2.94 mm for the second survey. These errors are
below the centimeter-scale change of measurement used to
detect changes in sediment transport after a rainfall event.
They also highlight the importance of targets in reducing error
associated with TLS techniques. Inferences from error
TLS survey parameter summary (09/27−29/08)
n 192
RMSE (m) 0.001
Quantiles (m)

Wise (2000) grouped errors into three categories that include 90 0.002
artifacts, systematic errors, and random errors. Artifacts are 75 0.001
obvious spurious data that are commonly visually detected
50 0.000
with the aid of viewing hillshade maps and 3D perspectives.
They are generally the result of timing errors, which ultimately 25 −0.001
produces erroneous positioning of data, or may be holes in 10 −0.002
the DTM data that are the result of improperly coded data
(Reuter et al., 2009). Systematic errors result from inherent
bias in the instrumentation that causes the data to be distorted
in the x, y, or z direction, physical causes (i.e., temperature,
clouds, and air quality), or result from human-observer limi-
tations (Li et al., 2005). Systematic errors are not as visually
evident as artifacts and require the use of mathematical or
statistical techniques to ascertain the level to which these
impact a particular DTM. Random errors are commonly ran-
dom noise and may be normally distributed in the data. A
major goal of digital terrain modeling is to identify, reduce,
remove, or at least quantify these errors (Reuter et al., 2009).
Other sources of error can arise from issues such as data
density, data distribution, and interpolation algorithms (Gong
et al., 2000).

3.6.6.1.1 Global errors

A standard way to assess the accuracy of a DTM is analysis of Control point location
the entire data set to produce a global summary statistic that
TLS instrument location
characterizes the average error in the altitude field (e.g., 0 10 20 40
RMSE). The RMSE represents a direct comparison of DTM meters
elevations to reference elevations obtained from GPS, maps, Figure 9 A control network of high-definition targets from the 2008
and field survey (e.g., GPS real-time kinematic surveying). USGS Landslide Hazards Program’s intensive research site located
Reference points should be widely and systematically distrib- outside of Santa Barbara, CA. The targets were used to calculate
uted across the surface to capture the complexity of the terrain global measures of error for the DTM (work funded by the National
and be measured at a much higher degree of accuracy than the Geographic Society - CRE Grant #8590-08).

assessment indicated DTM errors are spatially variable and are
linked to topographic complexity (Chaplot et al., 2006).
Many studies also utilize the mean error (ME) and the
standard deviation (SD) as highlighted by Fisher and Tate
(2006): Pn
½zðsi Þ  zREF ðsi Þ2
ME ¼ i ¼ 1 , ½6
n quired to identify and assess these types of error (Reuter et al.,
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Pn
i ¼ 1 ½zðsi Þ  zREF ðsi Þ  ME
SD ¼
n1
Equations [5]–[7] represent global measures of DTM ele-
vation accuracy. They do not account for localized patterns of
errors that are commonly tied to the heterogeneous nature of
the topography in many geomophic settings.
3.6.6.1.2 Local errors
Recent studies have begun to focus on assessing accuracy at
local scales (Kraus et al., 2006; Erdogan, 2010). Kraus et al.
(2006) presented an approach that considers the accuracy of
every grid cell in a DTM with particular emphasis on the
number, alignment, and distance of the neighboring original
points, the terrain curvature of the neighborhood of each grid
point, and the height accuracy of the original points. The
approach provides a good starting point to assess local ac-
curacy in geomorphological studies, but should not be the
only approach considered in the assessment of local accuracy
issues. Kraus et al. (2006) highlighted the need to properly
qualify the DTM errors by visually depicting points that are
either unusable or contain larger errors. Erdogan (2010) used
an error mapping approach to assess the spatial patterns of
DTM errors. A combination of local spatial statistics and
geographically weighted regression was used to reveal that
topographic complexity and point spacing were linked to lo-
calized DTM errors.
The importance of local versus global error measures to
accurately assess DTMs requires additional research. Other
approaches have also been presented, and surely new ap-
proaches will emerge. There is no consensus on the validity of
accuracy measures, as they characterize different aspects of
error. An example of a new research direction was presented by
the work of Hohle and Hohle (2009) who assessed errors in
laser scanning and photogrammetry DTM source data to de-
termine if outliers would significantly degrade the effective-
ness of the RMSE and other approaches that assume that error
is normally distributed. Their findings led to the conclusion
that if the DTM error distribution departs from normality,
other approaches such as sample quantiles of error, median,
and the normalized absolute deviation should be used to as-
sess DTM error. Aguilar et al. (2007, 2010), Aguilar and Mills
(2008), and Habib (2009) also provided important reviews of
new techniques for assessing the accuracy of ALS-derived DTM
data sets.
3.6.6.1.3 Systematic errors
Systematic errors are commonly manifested as systematic
biases associated with sensor systems and procedures used in
the DTM generation process (Albani and Klinkenberg, 2003).
Reuter et al. (2009) also showed systematic error could be
Digital Terrain Modeling 151
introduced or enhanced by geomorphometric algorithms. This
is significant and indicates that error can be introduced and
amplified in terrain analysis. Systematic errors are not easily
detectable and generally not recognized via the visualization
process, as biases generally represent lower frequency variation
(Wechsler, 1999). Statistical or mathematical analyses are re-
2009; Blanchard et al., 2010). If the systematic error can be
identified, it can be modeled and significantly reduced to
2
improve the quality of the DTM. At a minimum, there should
: ½7
be qualifying statements about the quality of the DTM that
accurately quantifies the systematic error (Daniel and Tennant,
2007).
There are a variety of ways to assess or reduce systematic
errors. The NCALM’s calibration procedure is an example of
how careful preparation of the data collection mission and
independent survey data can be used to reduce the potential
for systematic error. Field conditions, however, may not be
optimal for calibration or the field conditions may change
during the calibration process, thereby making the collected
data unreliable in terms accuracy assessment. In these scen-
arios, the lack of calibrated elevation data or the unreliable
calibrated elevation data should be replaced with reliable data
from other sources (i.e., benchmarks, prior surveys, or other
high-quality DTMs). For example, Rignot et al. (2001) re-
ported systematic errors in SRTM data due to microwave
penetration into snow cover on alpine glaciers, thereby im-
pacting the accuracy of glacial volumetric estimates based on
analysis. Knowing the location of such systematic biases per-
mits the use of other data sources to be used in those locations
for assessing elevation or volumetric changes.
Although the term ‘reliable data’ is used here, each data set
should be independently evaluated for systematic biases. This
is particularly important to studies that examine land-surface
changes with data from different sensors. Systematic errors
associated with different sensors are generally hard to detect,
which makes it difficult to draw conclusive inferences from the
results. Mills et al. (2005) identified systematic over- or
underestimation of the actual land-surface altitudes, and in
other locations a significant increase in noise. A lack of
knowledge of the existence of these types of errors would have
led to misinterpretation of the changes in the coastal land-
scape investigated by these researchers. Furthermore, it is
critical to consider systematic error when conducting DTM
differencing, as systematic error propagation between data sets
can occur. The systematic error can be either linear or non-
linear in nature.
3.6.7 Geomorphological Applications
The use of DTMs in geomorphological research has risen sig-
nificantly over the last three decades, and applications extend
over all process domains. Numerous advances have resulted
from the rapid proliferation of geospatial technologies, in-
cluding remote sensing, GPS, ALS, TLS, and TPT. Recent ad-
vances in geospatial technologies have provided new
opportunities to map landscapes and landforms without
issues traditionally associated with field mapping (Jones et al.,
2007). Issues commonly associated with field mapping
152 Digital Terrain Modeling
include: (1) access to locales (e.g., private property, road
closures, or hazardous trekking to the site); (2) time to con-
duct the survey and cost; and (3) an inability to capture subtle
features in the topography (topographic expressions missed by
the human eye). Furthermore, multitemporal data offer the
ability to capture topographic changes through repeat surveys
and provide critical information regarding land-surface
evolution.
The ability to capture repeat high-resolution data and gen-
erate multitemporal DTMs provides much better spatial and
temporal information regarding surface roughness and topo-
graphic change. Therefore, the use of geospatial technologies
have directly led to new and innovative findings that could not
have been generated using traditional techniques. For example,
Volker et al. (2007) used local relief (a proxy of surface
roughness) to identify scalar differences in alluvial fan mor-
phometry from fans with disparate formative processes. The
results revealed a distinct topographic signature associated with
debris flow and fluvial dominated fans. Frankel and Dolan
(2007) also used surface roughness to differentiate the ages of
fan units, as they were found to exhibit statistically unique
topography. They found a smoothing trend in the topography
of fan units up to B70 ka, with older units exhibiting greater
surface roughness, as secondary erosional processes begin to
dominate the process regime on older units.
Advances are also being made by integrating surface
roughness with other landscape ecology metrics to unravel
ecosystem function and structure (Hoechstetter et al., 2008).
In particular, Hoechstetter et al. (2008) found that area and
distance landscape metrics were sensitive to surface roughness.
Forest-island distributions associated with fire in the central
Andes Mountains were found to be significantly correlated to
locations that exhibit higher topographic roughness (Coblentz
and Keating, 2008). The integration of surface roughness and
other land-surface and ecological parameters provided greater
insight into landscape and landform evolution.
Despite many new insights into how geomorphic systems
operate, more terrain-analysis research is required. Oppor-
tunities exist for incorporating high-resolution DTMs with
spatially and temporally continuous process-oriented data
(e.g., rates of weathering, sediment flux, surface and sub-
surface hydrology, ablation, etc.). The potential to fuse data
from a variety of remote-sensing techniques also exists to en-
hance geomorphologist’s ability to extract more information
about features and processes. This will become more import-
ant as existing and emerging instruments provide the cap-
ability to conduct repeat surveys in a timely and cost efficient
manner. This provides geomorphologists with the ability to
push-the-envelope in terms of accurately estimating process-
rates from assessment of morphological changes. The in-
creased frequency in the temporal resolution of acquiring
digital elevation data has the potential to advance our know-
ledge of shorter term landscape and landform changes. Such
change detection capabilities are also relevant from resource
management and policy perspectives. Nevertheless, more
progress is required to effectively acquire, transfer, process, and
analyze large data sets that represent computationally inten-
sive endeavors. Technically, there is also a need for improved
storage capabilities and data/processing quality control. Such
developments will further advance geomorphological
research. With these technical or methodological issues in
mind, we take the liberty to speculate about the possible re-
search pathways involving the use of elevation data and terrain
analysis in geomorphology. These are based on our field ex-
periences and an extensive literature review.
3.6.7.1 High-Resolution DTMs
One of the important aspects of high-resolution DTMs is the
ability to extract information at the length-scale of particle
roughness in many environmental settings (Heritage and
Milan, 2009; Hodge et al., 2009a,b). This level of detail is
critical, as it provides an accurate depiction of the boundary
conditions that commonly influence, and in turn, are modi-
fied by process-mechanics. For example, particle roughness
has a direct impact on fluid mechanics, creating frictional drag
that generally reduces velocity or generates turbulent flow.
Flow over particles exerts drag that is important to sediment
entrainment and subsequent sediment transport. Therefore,
particle dimension variability is critical for understanding
fluid-particle interactions. TLS and TPT can be used to char-
acterize these conditions, but with some limitations. TLS work
has commonly been constrained to studies of specific events.
There are limitations to capturing within-event temporal data
because scanning rates do not permit rapid scans of an area
(generally only very small areas can be scanned rapidly
enough and therefore a large portion of the landscape is not
assessed), the laser does not penetrate water or solid materials
in the case of thick granular flows, and inclement weather
conditions can hinder scanning or at least make it dangerous
to the user. TPT can be used to obtain higher temporal reso-
lution data, but there are limitations with it in terms of
lighting conditions, particularly for events that occur during
darkness, unless a significant light source were added to illu-
minate the site. TPT, much like TLS, is limited in its ability to
obtain data during an event, as the sediment movement im-
pedes accurately capturing temporal variability of the surface.
It is also the case that TPT is commonly hampered by in-
clement weather conditions thereby further limiting the ability
to capture all of the temporally relevant topographic changes.
The novelty of obtaining high-resolution topographic data
is also tempered by an inability to measure process-related
data at similar spatial extents. For example, TLS surveying can
be used to accurately measure topographic changes in a
channel as the direct result of sediment transport associated
with runoff-generated debris flow (McCoy et al., 2010). Even
in the previously mentioned study, however, the process-
related data (gathered with pressure transducers, a load cell,
rainfall gauges, laser and sonic stage devices, soil moisture
probes, digital videography, etc.) were limited to cross-
sectional process-oriented information (stage, pore water
pressure, velocity, force, etc.) or point-based process-oriented
data (rain fall, soil moisture, etc.). The ability to correlate
energy and matter dynamics with an accurate depiction of the
topographic change (obtained with the TLS) is hindered by
limitations in the spatial extent of process-oriented data and
the temporal extent of the TLS data. The TLS was used to only
sample before and after each event. Topographic changes be-
tween individual debris-flow surges that are captured in the

digital videography are not obtained in our TLS sampling
scheme. Therefore, our temporal resolution is inconsistent
with the permanent field instruments that are sampling during
the rainfall event on the order of 10s to 100s of Hz. Despite
these high sampling rates from the permanent field instru-
ments, the measurements occurred at two cross-sections and
two hillslope locations along the B150 m extent of channel
scanned with repeat TLS surveys. Topographic changes from
2 cm DTM data gathered along the B150 m channel segment
were spatially more contiguous than the cross-section and
point information gathered by the permanent field
instruments.
Point and cross-sectional data are generally not sufficient
to estimate local-scale erosion or for considering the local and
nonlocal interactions that speak to connectivity within a var-
iety of environmental settings. For example, studies of ero-
sional and depositional processes have been hindered by the
uncertainties associated with spatial and temporal lumping of
data to attain results on sediment delivery, yield, transport,
and causal mechanisms (Walling, 1983; de Vente et al., 2007).
This lack of information has led to the development of spa-
tially distributed sediment transport models (Merritt et al.,
2003; de Vente and Poesen, 2005), and transport models
based on the conservation of mass and energy (see Dietrich
et al., 2003). Wainwright et al. (2008) have shown that many
of the spatially distributed models are significantly over sim-
plified, whereas other researchers (Tucker and Bras, 1998;
Roering et al., 1999) have found these models to be based on
misinterpretations of the literature. Studies examining geo-
morphic transport laws have the potential to characterize
causal mechanisms for long-term landscape evolution, but
assume local transport. A reasonable assumption for some
processes (e.g., grain detachment by raindrop impact) can be
expressed as an at-a-point measurement (i.e., local surface
gradient). However, transport laws do not significantly inte-
grate process with form parameters across the landscape or
landform. This limitation has led to a recent proliferation of
papers focused on local and nonlocal sediment flux (e.g.,
Tucker and Bradley, 2010; Foufoula-Georgiou et al., 2010;
Foufoula-Georgiou and Stark, 2010).
Spatially and temporally continuous process-related data
are required to make connections between local and nonlocal
sediment fluxes in the field. Ultimately, spatially and tempo-
rally continuous process-related data can be made available
from numerical models. The models, however, need to be
based on and continuously calibrated with field measure-
ments, but not field measurements as they are currently ob-
tained. Denser sampling networks are required to capture the
variability within geomorphic systems. In the best-
case scenario, increased spatially and temporally continuous
data sets could be developed in conjunction with high-
resolution digital elevation measurements. This would allow
geomorphologists to gain a better appreciation the connect-
ivity of local and nonlocal sediment fluxes as they relate to
landscape and landform change. At a minimum, this add-
itional data might be used to validate or extrapolate modeling
approaches to bring a new perspective on the issue of local
and nonlocal fluxes as they govern the magnitude of landscape
and landform change. Alternatively, process-related data might
also provide a greater understanding of diffusion processes,
Digital Terrain Modeling 153
that when linked with high-resolution DTM data, may provide
additional information and insight required to unravel the
complexities of landscape and landform evolution. This
would seemingly make the local and nonlocal sources
issue moot.
Several new field sampling approaches and instruments are
likely necessary to increase the amount of process data that
need to be collected in order to obtain spatial coverage asso-
ciated with DTM coverage. These data must also be integrated
with data collected at-a-point (i.e., cross-sectional or point-
based measurements) and with data and information generated
from various modeling approaches (Tucker and Bradley, 2010;
Hancock et al., 2010) to provide spatially and temporally con-
tinuous measures of hydrologic and sedimentologic fluxes. The
lack of spatial integration has continuously plagued sediment
and hydrological transport modeling by reducing the ability to
validate model results with field data. Therefore, a logical goal
might be to develop new instrumentation that can be used in
conjunction with measurements from high-resolution DTM
data to obtain field measurements. This approach, although
likely expensive, would produce the best results moving for-
ward and likely become less cost prohibitive over time.
Another fundamental geomorphological research topic is
landscape evolution or change. High-resolution DTM data
from TLS and TPT offer the ability to obtain a time-series of
topographic conditions that can be compared, such that land-
surface change can be characterized, as opposed to relying
solely on numerical modeling approaches. Wheaton et al.
(2010a) have formalized this approach producing DEM of
Difference (DoD) maps, based on the fundamental work of
Lane et al. (1994). The same approach can be used with
coarser resolution DTMs. However, ALS, TLS, and TPT provide
users with the ability to conduct repeat surveys with signifi-
cantly greater temporal frequency. Indeed, TLS and TPT can be
automated to acquire continuous data at a locale. These new
advances in repeat high-resolution topographical mapping
techniques have improved morphological measurements of
sediment budgets that were traditionally done with repeat
surveys at cross-sections and longitudinal profiles (Lane, 1998;
Stott, 2002; Bezzi et al., 2009; Martin-Vide et al., 2010). High-
resolution DoD approaches can increase our understanding of
space–time relationships by providing accurate measures of
sediment loading (i.e., dry raveling and rock fall; Lim et al.,
2010), sediment entrainment, and event-by-event, sediment-
transport rates (Oppikofer et al., 2009). This information can
be derived at a variety of scales from the original data to
provide information on important concepts in geomorph-
ology. Examples include: (1) local and nonlocal variations in
sediment flux (Gabet, 2003; Tucker and Bradley, 2010); (2)
connectivity within the landscape (Brierley et al., 2006;
Bracken and Croke, 2007); and (3) multiple-scale causality
(Phillips, 2005).
High-resolution geomorphological maps that capture the
space–time details of morphological features can be used to
estimate rates of slow-moving geomorphic events (Prokop and
Panholzer, 2009). Rates of movement on an event-by-event
basis can also improve estimates of the extent to which a
specific feature spatially evolves on the landscape given vari-
ations in erosion and deposition. For example, a dune crest
(Rubin et al., 2008) or sediment waves (James, 2006) might
154 Digital Terrain Modeling
be mapped before and after a wind or flood event to deter-
mine the extent that these features have migrated during that
event. Rates of movement from geomorphological mapping
are not only important to understanding process mechanisms,
but they also have significant implications for habitat changes
and management (Wheaton et al., 2010b).
Surface roughness can provide topographic boundary
conditions that alter a variety of flow properties and in turn
influence subsequent flow events (Frankel and Dolan, 2007).
High-resolution topographic data can be used to represent
initial boundary conditions for process-form interactions
(Heritage and Milan, 2009). This source of information can
establish baseline topographic conditions for model initiation
as well as an end form to calibrate results from model simu-
lations. The integration of models with high-resolution
topographic data might serve as potential solutions to con-
cerns with the lack of spatially continuous process-related data
and model validation. A practical issue in moving forward,
however, is that high-resolution DTM data require substantial
computational resources such as memory and fast processors
given the data volume. Although this is a definite short-
coming, the use of high-resolution data in numerical process
models offers significant potential for improved para-
meterization schemes and validation of model results.
3.6.7.2 High-Resolution DTMs and Scale
High-resolution DTMs also show promise for formalizing
some of the previously mentioned shortcomings in addressing
issues of scale, such as the scale at which processes and
landforms interact, and hierarchical connections that may
exist between scales. The use of TLS and TPT offer the potential
to develop millimeter to centimeter DTMs, a scale that ap-
proximates particle roughness and the operational scale of
many processes. Variations in the computational scale of an-
alysis can then be used to determine what scales are best for
assessing topographic variability and landscape change. Such
scale-dependent analysis is of interest because multiple pro-
cesses operate on landforms, and each process may have a
unique space–time signature that could potentially be char-
acterized (Chorley et al., 1984). As Phillips (2012) has out-
lined, analyses of varying spatial scales in geomorphology
should identify: (1) characteristic scales (Schmidt and
Andrews, 2005; Perron et al., 2008); (2) conditions of scale
independence (Pelletier, 1999; Southgate and Möller, 2000);
and (3) tools for visualization and data exchange between
scales (Murzyn et al., 2006). Data exchanges and analyses
between scales are critical for interpreting hierarchical con-
nections between scales. Although a great deal of literature in
geomorphology has highlighted scale invariance (Rodrı́guez-
Iturbe and Rinaldo, 2001), scale linkages have been identified
in the landscape (Phillips, 2012). High-resolution DTMs can
aid in the assessment of fine-scale fluxes in mass and energy
(as represented by spatio-temporal morphologic variations
that govern those fluxes) and for identifying and mapping
process-pattern relationships that are manifested at these
scales. In turn, these results can be investigated in the context
of geomorphic responses at coarser scales within a system that
produces different patterns at different spatial and temporal
scales. This information could ultimately serve as an
important starting point for formalizing our interpretations
and predictions of landscape evolution.
A limiting factor with high-resolution DTMs is the coarsest
scale that can be analyzed given their limited spatial extent.
Techniques used to generate high-resolution topographic data,
such as TLS and TPT, record small areas (100 s to 1000 s of m2).
Topographic wavelengths and complexity of larger landforms
are generally not represented in the data, and this limits the use
of the data for selected geomorphological applications. Op-
portunities exist, however, to use coarser resolution DTMs from
other imaging sensors to provide additional information to
address the spatial structure and hierarchical organization of
the topography at regional and global scales. New conceptual
and technical approaches need to be developed that integrate a
wide variety of information from numerous DTM sources. This
will permit a better assessment and mapping of the inherent
hierarchical organization of the land surface.
3.6.7.3 Data Fusion
Very few attempts have been made to fuse other remotely sensed
data with DTMs within the constructs of geomorphological
methodologies. As an example, ALS and TLS both capture in-
tensity data. Intensity data are underutilized for geomorphic
research and could be used in conjunction with elevation to aid
in mapping surficial features. Intensity offers a means of cap-
turing information about the reflectivity of an object and phe-
nomena that are both visible and invisible, with some modest
fine-tuning of the energy returns (Wasklewicz et al., 2005,
2007). Intensity values provide geomorphology with an in-
creased ability to map objects, place topographic variability in
context of the mapped objects, and put the discipline in a
position to better understand human impacts on the natural
geomorphic system. The later is possible because human-
modified landscapes generally have different spectral signatures
than the materials that make up geomorphic features.
Additional information can be obtained from imaging
sensors that include spectrometers and radar systems. For ex-
ample, high-resolution multispectral imagery has been com-
bined with ALS data to map the distribution of marsh features
across the PIE-LTER site in northeastern Massachusetts (Mill-
ette et al., 2010). The majority of the work involves the inte-
gration of hyperspectral data, multispectral imagery, and
thermal imagery with a focus on classification and thematic
mapping (e.g., Gilvear et al., 2004). Further developments are
needed that consider landscape and landform dynamics. Work
conducted in a coastal setting to examine sediment dynamics
provides a snapshot of what could be attempted to advance
geomorphological research (e.g., Deronde et al., 2008). They
used hyperspectral data to classify and map different natural
and artificial berm and dune deposits. Spectral information
was combined with ALS data from multiple surveys to in-
vestigate sediment transport amounts and directions. This
fusion of data was found to be particularly useful to capturing
berm dynamics associated with beach renourishment projects.
3.6.7.4 Temporal Data Acquisition
The advances in instruments such as TLS, ALS, and TPT have the
potential to increase the temporal frequency of data acquisition

and increase DTM inventories. As instrument weight and cost
continue to decrease, it has become more feasible to mobilize
equipment rapidly or set-up these instruments at various lo-
cations for extended periods of time. Instruments such as TLS
are presently capable of being set-up and programmed at fixed
locations to sample periodically throughout the day and night.
This particular application has scarcely been applied in geo-
morphology, but is used in the mining industry to monitor rock
fall and volumes of mined material. A similar approach should
be used to increase our temporal understanding of landscape
evolution and supplement existing information gathered at
long-term sampling sites (e.g., stream gauges, weather stations,
coastal-wave monitoring sites). Marzolff and Poesen (2009)
demonstrated how using multiple DTMs can improve our
space-time understanding of morpho-dynamics within gully
systems. They examined changes to two different bank gully
types using TPT. A major finding from their research was the
large degree of storage of sediment within the gully network
after headward erosion of the bank gullies. This was not an-
ticipated, as most hypotheses would suggest that the gullies
erode material into other parts of the drainage network with
little deposition taking place within the gully system. Further-
more, TPT has been used to quantify the rate of parabolic dune
migration (e.g., Arteaga et al., 2008).
A decrease in instrument weight and cost may also make it
feasible to mount instruments on UAVs. This would greatly
enhance the geomorphologists’ ability to capture more data
and provide greater flexibility in planning field-sampling cam-
paigns. Missions (ALS) commonly take a great deal of time to
mobilize because equipment is not located near relevant field
sites. A UAV is much easier to mobilize and would permit
measurement of time-sensitive data within a matter of hours.
The ability to respond rapidly would permit the user to obtain
digital elevation data immediately before and after an event.
This flexibility greatly facilitates fieldwork and the assessment of
geomorphological hazards. For example, multitemporal DTMs
can be used to calibrate debris-flow sediment yield and pre-
diction models after wildfires. This information is important for
hazard mitigation schemes (Gartner et al., 2009; Cannon et al.,
2010a,b). Furthermore, the ability to rapidly respond to events
or forcing factors is critical, and DTMs can be combined with
rainfall intensity data to study the timing, location, and run-out
distances of debris flow-prone areas (Cannon et al., 2010a,b).
3.6.8 Conclusions
Advances in sensors, computer technology, and infor-
mation technology have provided many new insights into
geomorphological systems. This is clearly demonstrated by the
ever-increasing usage of DTMs and geomorphometric research
over the last four decades. This phenomenon has arisen from a
shift in how geomorphologists conduct research, whereby
greater emphasis has been placed on enumerating topographic
complexity to capture spatial, temporal, and scalar differences
in the land surface. There are, however, growing pains with
implementing geospatial technologies, as many technical
issues must be addressed. The objective of this chapter has
been to reduce some of these difficulties by providing a de-
finitive introduction to digital terrain modeling and examples
Digital Terrain Modeling 155
of the incorporation of DTMs in geomorphological research.
Topics covered include the basic history of digital terrain
modeling, recent data sources, preprocessing issues, DTM
production, DTM error analysis, and the use of DTMs in the
current context of geomorphology.
Recent advances in sensors and methods include light de-
tection and ranging, radio detection and ranging systems, and
progress in terrestrial photogrammetry techniques. These in-
struments are capable of producing moderate to high-
resolution elevation data over a large range of spatial extents,
with LiDAR and terrestrial methods producing data with much
greater temporal frequency than previously possible. A con-
cern with instruments that capture high-resolution data is the
nondiscrete nature of obtaining elevation data. The point
clouds captured by many instruments contain both terrain
and off-terrain points. A variety of methods can be used to
remove vegetation and other off-terrain objects from the point
cloud, and geomorphologists must be cognizant of land-
surface concepts, sensor characteristics, preprocessing, error
analysis and postprocessing, and the limitations associated
with utilizing any DTM for geomorphological research. These
stages in digital terrain modeling are critical to understand, as
the quality of DTMs is a major issue, and high quality bare-
earth models are most often sought after in geomorphology.
The generation of DTMs generally requires selecting a
spatial interpolation algorithm. Spatial interpolation of the
cleaned point clouds is an essential step in preprocessing. Each
interpolation technique has advantages and disadvantages.
The advantages and limitations associated with various inter-
polation approaches were examined in the context of en-
vironments with different topographic complexity.
Another key aspect of digital terrain modeling involves
error assessment, although this is not always accounted for in
many studies. It is important to note that inherent error is due
to sensor systems, the data acquisition approach, the repre-
sentational scheme, and issues related to preprocessing ap-
proaches. In general, the error can be classified as random or
systematic, although global, local, and systematic errors
should be evaluated. Global measures are well established and
are commonly used in engineering, statistics, and cartography.
Systematic errors tend to be associated with field methods,
instrumentation, and choice of algorithms in the production
of the DTM. Local error analysis represents a relatively new
research area that assesses variations in the local error that
greatly influences terrain analysis results. Change detection
using multitemporal DTMs requires the evaluation of all error
in or to assess topographic changes. Further research is re-
quired to determine the extent to which these types of error
influence geomorphological research, although numerous
studies have demonstrated that error propagation does occur,
stemming from DTM error that influences the accuracy of
geomorphometric parameters and modeling results, as many
models make use of specific parameters.
Digital terrain modeling is now critical in geomorpho-
logical research. The availability of high-resolution DTMs has
greatly improved our ability to assess geomorphological sys-
tems and maps landforms at a variety of scales. These data also
greatly facilitate change detection and the generation of pro-
cess-related data to investigate the linkages of process and
form, and process and patterns. Such data and the derived
156 Digital Terrain Modeling
information are important for process modeling, as we at-
tempt to improve our process parameterization schemes,
calibrate numerical models and validate results from simu-
lations. Integrating process-oriented data from the field with
terrain analysis and simulation results enable new investi-
gations of land-surface evolution, evaluation of the impact of
scalar variations on landscape change, and promote better
understandings of specific landform evolution. Digital terrain
modeling and geomorphometry have the potential to provide
us with many more new understandings of complex geo-
morphological systems and landscape evolution.
References
Aguilar, F.J., Agüera, F., Aguilar, M.A., Carvajal, F., 2005. Effects of terrain
morphology, sampling density, and interpolation methods on grid DEM
accuracy. Photogrammetric Engineering and Remote Sensing 71, 805–816.
Aguilar, F.J., Aguilar, M.A., Aguera, F., 2007. Accuracy assessment of digital
elevation models using a non-parametric approach. International Journal of
Geographical Information Science 21, 667–686.
Aguilar, F.J., Mills, J.P., 2008. Accuracy assessment of LiDAR-derived digital
elevation models. Photogrammetric Record 23, 148–169.
Aguilar, F.J., Mills, J.P., Delgado, J., Aguilar, M.A., Negreiros, J.G., Pérez, J.L.,
2010. Modelling vertical error in LiDAR-derived digital elevation models. ISPRS
Journal of Photogrammetry and Remote Sensing 65, 103–110.
Albani, M., Klinkenberg, B., 2003. A spatial filter for the removal of striping artifacts
in digital elevation models. Photogrammetric Engineering and Remote Sensing
69, 755–765.
Alexa, M., Behr, J., Cohen-Or, D., Fleishman, S., Levin, D., Silva, T., 2001. Point
set surfaces. In Proceeding of the 12th IEEE Visualization Conference, San
Diego, USA, pp. 21–28.
Almansa, A., Cao, F., Gousseau, Y., Rouge, B., 2002. Interpolation of digital
elevation models using AMLE and related methods. IEEE Transactions on
Geoscience and Remote Sensing 40, 314–325.
Arteaga, C., Juan de Sanjosé, J., Serrano, E., 2008. Terrestrial photogrammetric
techniques applied to the control of a parabolic dune in the Liencres dune system,
Cantabria (Spain). Earth Surface Processes and Landforms 33, 2201–2210.
Atkinson, P.M., Tate, N.J., 2000. Spatial scale problems and geostatistical solutions:
a review. Professional Geographer 54, 607–623.
Baltsavias, E.P., 1999. A comparison between photogrammetry and laser scanning.
ISPRS Journal of Photogrammetry and Remote Sensing 54, 83–94.
Bauer, B., 1996. Geomorphology, geography and science. In: Rhoads, B.L., Thorn,
C.E. (Eds.), The Scientific Nature of Geomorphology: Proceedings of the 27th
Binghamton Symposium in Geomorphology. Wiley, New York, pp. 381–413.
Beraldin, J.-A., Blais, F., Lohr, U., 2010. Laser scanning technology. In: Vosselman,
G., Maas, H-G. (Eds.), Airborne and Terrestrial Laser Scanning. Whittles
Publishing, Dunbeath, Scotland, UK, pp. 1–39.
Bezzi, A., Fontolan, G., Nordstrom, K.F., Carrer, D., Jackson, N.L., 2009. Beach
nourishment and foredune restoration: practices and constraints along the
Venetian shoreline, Italy. Journal of Coastal Research 56, 287–291.
Bishop, M.P., 2013. Remote sensing and GIScience in geomorphology: introduction
and overview. In: Shroder, J. (Editor in Chief), Bishop, M.P. (Eds.), Treatise on
Geomorphology. Academic Press, San Diego, CA, vol. 3, pp. 1–24.
Bishop, Jr. M.P., Shroder, Jr. J.F. (Eds.), 2004. Geographic Information Science and
Mountain Geomorphology. Springer-Praxis, Chichester.
Bishop, M.P., Shroder, Jr. J.F., Colby, J.D., 2003. Remote sensing and
geomorphometry for studying relief production in high mountains.
Geomorphology 55, 345–361.
Blanchard, S.D., Rogan, J., Woodcock, D.W., 2010. Geomorphic change analysis
using ASTER and SRTM digital elevation models in Central Massachusetts,
USA. GIScience and Remote Sensing 47, 1–24.
Bracken, L.J., Croke, J., 2007. The concept of hydrological connectivity and its
contribution to understanding runoff-dominated geomorphic systems.
Hydrological Processes 21, 1749–1763.
Brasington, J., Smart, R.M.A., 2003. Close range digital photogrammetric analysis
of experimental drainage basin evolution. Earth Surface Processes and
Landforms 28, 231–247.
Brierley, G.J., 2010. Landscape memory: the imprint of the past on contemporary
landscape forms and processes. Area 42, 76–85.
Brierley, G.J., Fryirs, K., Jain, V., 2006. Landscape connectivity: the geographic
basis of geomorphic applications. Area 38.2, 165–174.
Brodsky, D., Watson, B., 2000. Model simplification through refinement.
In: MacKenzie, S., Stewart, J. (Eds.), Proceedings of Graphics Interface 2000.
Morgan Kaufmann Publishers, Montreal, Canada, pp. 221–228.
Brus, D.J., de Gruijter, J.J., Marsman, B.A., Visschers, R., Bregt, A.K., Breeuwsma,
A., 1996. The performance of spatial interpolation methods and choropleth maps
to estimate properties at points: a soil survey case study. Environmetrics 7,
1–16.
Burrough, P.A., McDonnell, R.A., 1998. Principles of Geographical Information
Systems. Oxford University Press, New York, 333 pp.
Bürgmann, R., Rosen, P.A., Fielding, E.J., 2000. Synthetic aperture radar
interferometry to measure Earth’s surface topography and its deformation. Annual
Review of Earth and Planetary Sciences 28, 169–209.
Cannon, S.H., Boldt, E.M., Kean, J.W., Laber, J., Staley, D.M., 2010a. Relations
between rainfall and postfire debris-flow and flood magnitudes for emergency-
response planning, San Gabriel Mountains, Southern U.S. U.S. Geological
Survey Open-File Report #2010–1039: 31 pp.
Cannon, S.H., Gartner, J.E., Rupert, M.G., Michael, J.A., Rea, A.H., Parrett, C.,
2010b. Predicting the probability and volume of postwildfire debris flows in the
intermountain western United States. Geological Society of America Bulletin 122,
127–144.
Capps, D.M., Rabus, B., Clague, J.J., Shugar, D.H., 2010. Identification and
characterization of alpine subglacial lakes using interferometric synthetic aperture
radar (InSAR): Brady Glacier, Alaska, USA. Journal of Glaciology 56, 861–870.
Carrara, A., Bitelli, G., Carla, R., 1997. Comparison of techniques for generating
digital terrain models from contour lines. International Journal of Geographical
Information Science 11, 451–473.
Carson, M.A., Kirkby, M.J., 1972. Hillslope Form and Process. Cambridge
University Press, Cambridge.
Catani, F., Farina, P., Moretti, S., Nico, G., Strozzi, T., 2005. On the application of
SAR interferometry to geomorphological studies: estimation of landform
attributes and mass movements. Geomorphology 66, 119–131.
Chandler, J.H., 1999. Effective application of automated digital photogrammetry for
geomorphological research. Earth Surface Processes and Landforms 24, 51–63.
Chandler, J.H., Fryer, J.G., Jack, A., 2005. Metric capabilities of low cost digital
cameras for close range surface measurement. The Photogrammetric Record 20,
12–26.
Chaplot, V., Darboux, F., Bourennane, H., Leguédois, S., Silvera, N., Phachomphon,
K., 2006. Accuracy of interpolation techniques for the derivation of digital
elevation models in relation to landform types and data density. Geomorphology
77, 126–141.
Chen, Q., Gong, P., Baldocchi, D., Xin, G., 2007. Filtering airborne laser scanning
data with morphological methods. Photogrammetric Engineering and Remote
Sensing 73, 175–185.
Chorley, R.J., Schumm, S.A., Sugden, D.E., 1984. Geomorphology. Methuen,
London, 605 pp.
Clarke, T.A., 1995. An analysis of the prospects for digital close-range
photogrammetry. ISPRS Journal of Photogrammetry and Remote Sensing 50,
4–7.
Coblentz, D., Keating, P.L., 2008. Topographic controls on the distribution of tree
islands in the high Andes of south-western Ecuador. Journal of Biogeography
35, 2026–2038.
Crosilla, F., Visintini, D., Prearo, G., 2004. A robust method for filtering nonground
measurements from airborne LiDAR data. International Archives of
Photogrammetry, Remote Sensing and Spatial Information Sciences 35,
196–201.
Daniel, C., Tennant, K., 2007. DEM quality assessment. In: Maune, D.F. (Ed.),
Digital Elevation Model Technologies and Applications: The Dem Users Manual.
American Society for Photogrammetry and Remote Sensing, Bethesda, MD,
pp. 395–440.
Declercq, F.A.N., 1996. Interpolation methods for scattered sample data: accuracy,
spatial patterns, processing time. Cartography and Geographic Information
Systems 23, 128–144.
Deronde, B., Houthuys, R., Henriet, J.P., van Lancker, V., 2008. Monitoring of the
sediment dynamics along a sandy shoreline by means of airborne hyperspectral
remote sensing and LIDAR: a case study in Belgium. Earth Surface Processes
and Landforms 33, 280–294.
Devereux, B., Amable, G., 2009. Aiborne LiDAR: instrumentation, data acquisition
and handling. In: Heritage, G.L., Large, A.R.G. (Eds.), Laser Scanning for the
Environmental Sciences. Blackwell Publishing Ltd., Chicester, UK, pp. 49–66.

Dey, T.K., Giesen, J., Hudson, J., 2001. Decimating samples for mesh
simplification. Proceedings of the 13th Canadial Conference on Computational
Geometry, Waterloo, Canada, pp. 85–88.
Dietrich, W.E., Bellugi, D., Heimsath, A.M., Roering, J.J., Sklar, L., Stock, J.D.,
2003. Geomorphic transport laws for predicting landscape form and dynamics.
In: Wilcock, P., Iverson, R. (Eds.), Prediction in Geomorphology, Geophysical
Monograph Series. AGU, Washington, D.C., vol. 136, pp. 103–123.
Dikau, R., 1989. Application of a digital relief model to landform analysis in
geomorphology. In: Raper, J. (Ed.), Three Dimensional Application in
Geographic Information Systems. Taylor and Francis, London, pp. 51–77.
Dikau, R. 1990. Geomorphic landform modelling based on Hierarchy Theory. In:
Brassel K., Kishimoto H. (Eds.). Proceedings 4th International Symposium on
Spatial Data Handling, 23rd–27th July, Zürich, pp. 230–239.
Dixon, J.L., Heimsath, A.M., Kaste, J.M., Amundson, R., 2009. Climate driven
processes of hillslope weathering. Geology 37, 975–978.
Dunning, S.A., Massey, C.I., Rosser, N.K., 2009. Structural and geomorphological
features of landslides in the Bhutan Himalaya derived from terrestrial laser
scanning. Geomorphology 108, 17–29.
Dupuis, E., Bakambu, J.N., Rekleitis, I., Bedwani, J., Gemme, S., Rivest-Caissy, J.,
2006. Autonomous long-range rover navigation – experimental results.
Proceedings of the 9th ESA Workshop on Advanced Space Technologies for
Robotics and Automation, Noordwijk, The Netherlands, 8 pp.
Dury, G.H., 1983. Geography and geomorphology: the last fifty years. Transactions,
Institute of British Geographers, NS 8, 90–99.
Eastman, J.R., 1989. IDRISI: A Grid-based Geographical Information System. Clark
University, Worcester, MA.
El-Sheimy, N., Valeo, C., Habib, A., 2005. Digital Terrain Modeling: Acquisition,
Manipulation and Applications. Artech House Publishers, Boston, MA, 257 pp.
Elsherbini, A., Sarabandi, K., 2010. Mapping of sand layer thickness in deserts
using SAR Interferometry. IEEE Transactions on Geoscience and Remote Sensing
48, 3550–3559.
Erdogan, S., 2010. Modelling the spatial distribution of DEM error with
geographically weighted regression: an experimental study. Computers and
Geosciences 36, 34–43.
Evans, I.S., 1972. General geomorphometry, derivatives of altitude, and description
statistics. In: Chorley, R.J. (Ed.), Spatial Analysis in Geomorphology. Methuen &
Co., London, pp. 17–90.
Evans, I.S., 1980. An integrated system of terrain analysis and slope mapping.
Zeitschrift für Geomorphologie N.F. Supplementband 36, 274–295.
Evans, I.S., 2003. Scale-specific landforms and aspects of the land surface.
In: Evans, I.S., Dikau, R., Tokunaga, E., Ohmori, H., Hirano, M. (Eds.), Concepts
and Modelling in Geomorphology: international Perspectives. Tokyo, Terrapub,
pp. 61–84.
Fisher, P.F., Tate, N.J., 2006. Causes and consequences of error in digital elevation
models. Progress in Physical Geography 30, 467–489.
Fleishman, S., Cohen-Or, D., Alexa, M., Silva, C.T., 2003. Progressive point set
surfaces. ACM Transactions on Graphics 22, 997–1011.
Florinsky, I.V., 1998. Accuracy of local topographic variables derived from digital elevation
models. International Journal of Geographical Information Science 12, 47–61.
Foufoula-Georgiou, E., Stark, C., 2010. Rethinking geomorphic transport: stochastic
theories, broad scales of motion and non-locality. Journal of Geophysical
Research – Earth Surface 115, F00A01.
Foufoula-Georgiou, E., Ganti, V., Dietrich, W.E., 2010. A non-local theory for
sediment transport on hillslopes. Journal of Geophysical Research - Earth
Surface 115, F00A16.
Frankel, K.L., Dolan, J.F., 2007. Characterizing arid region alluvial fan surface
roughness with airborne laser swath mapping digital topographic data. Journal
of Geophysical Research 112, F02025.
Fraser, C.S., 1997. Digital camera self-calibration. ISPRS Journal of Photogrammetry
and Remote Sensing 52, 149–159.
Fraser, C.S., Cronk, S., 2009. A hybrid measurement approach for close-range
photogrammetry. ISPRS Journal of Photogrammety and Remote Sensing 64,
328–333.
Gabet, E.J., 2003. Sediment transport by dry ravel. Journal of Geophysical Research
108, 2049–2057.
Gabriel, A.K., Goldstein, R.M., Zebker, H.A., 1989. Mapping small elevation changes
over large areas; differential radar interferometry. Journal of Geophysical
Research 94, 9183–9191.
Garland, M., Heckbert, P.S., 1997. Surface simplification using quadratic error
metrics. SIGGRAPH ’97 97, 209–216.
Gartner, J.E., Cannon, S.H., Helsel, D.R., Bandurraga M., 2009. Multivariate
Statistical models for predicting sediment yields from Southern California
watersheds. US Geological Survey Open-File Report #2009-1200, 42 pp.
Digital Terrain Modeling 157
Gens, R., Van Genderen, J.L., 1996. Review Article SAR interferometry – issues,
techniques, applications. International Journal of Remote Sensing 17,
1803–1835.
Gilvear, D., Tyler, A., Davids, C., 2004. Detection of estuarine and tidal river
hydromorphology using hyper-spectral and LiDAR data: Forth Estuary, Scotland.
Estuarine Coastal and Shelf Science 61, 379–392.
Gislason, S.R., Oelkers, E.H., Eiriksdottir, E.S., et al., 2008. Direct evidence of the
feedback between climate and weathering. Earth and Planetary Science Letters
277, 213–222.
Gong, J., Zhilin, L., Zhu, Q., Sui, H.G., Zhou, Y., 2000. Effect of various factors on
the accuracy of DEMs: an intensive experimental investigation. Photogrammetric
Engineering and Remote Sensing 66, 1113–1117.
Goodchild, M., 1992. Geographical information science. International Journal of
Geographical Information Systems 6, 31–45.
Graham, L.C., 1974. Synthetic interferometer radar for topographic mapping.
Proceedings of the IEEE 62, 763–788.
Gregory, K.J., 2000. The Changing Nature of Physical Geography. Arnold, New
York, 368 pp.
Gruber, S., Peckham, S., 2009. Land surface parameters and objects specfic to
hydrology. In: Hengl, T., Reuter, H.I. (Eds.), Geomorphometry: Geomorphometry:
Concepts, Software, Applications. Developments in Soil Science. Elsevier,
Amsterdam, vol. 33, pp. 171–194.
Guth, P.L., Ressler, E.K., Bacastow, T.S., 1987. Microcomputer program for
manipulating large digital terrain models. Computers and Geosciences 13,
209–213.
Gyasi-Agyei, Y., Willgoose, G.R., De Troch, F.P., 1995. Effects of vertical resolution
and map scale of digital elevation models on geomorphological parameters used
in hydrology. Hydrological Processes 9, 363–382.
Habib, A., 2009. Accuracy, quality assurance, and quality control of LiDAR data.
In: Shan, J., Toth, C.K. (Eds.), Topograpic Laser Ranging and Scanning. CRC
Press, Boca Raton, pp. 269–294.
Hack, J.,1957. Studies of Longitudinal Stream Profiles in Virginia and Maryland.
United States Geological Survey Professional Paper 294-B: 45-97.
Haggett, P., 2008. The local shape of revolution: reflections on quantitative
geography at Cambridge in the ‘fifties and ’sixties. Geographical Analysis 40,
336–352.
Hancock, G.R., Lowry, J.B.C., Coulthard, T.J., Evans, K.G., Moliere, D.R., 2010. A
catchment scale evaluation of the SIBERIA and CAESAR landscape evolution
models. Earth Surface Processes and Landforms 35, 863–876.
Haneberg, W.C., 2008. Using close-range digital photogrammetry for 3-D rock slope
modeling and discontinuity mapping in the United States. Bulletin of
Engineering Geology and the Environment 67, 457–469.
Harlick, R.M., Shapiro, L.G., 1992. Computer and Robot Vision: vol. 1 and vol. 2.
Addison-Wesley, Reading, MA, 1302 pp.
Hayakawa, Y.S., Oguchi, T., 2006. DEM-based identification of fluvial knickzones
and its application to Japanese mountain rivers. Geomorphology 78, 90–106.
Heckbert, P. and Garland, M., 1997. Survey of polygonal surface simplification
algorithms. multi-resolution surface modeling course, ACM Siggraph Course notes.
Hengl, T., 2006. Finding the right pixel size. Computers and Geosciences 32,
1283–1298.
Hengl, T., Evans, I.S., 2009. Mathematical and digital models of the land surface.
In: Hengl, T., Reuter, H.I. (Eds.), Geomorphometry – Concepts, Software,
Applications: Developments in Soil Science. Elsevier, Amsterdam, vol. 33, pp.
31–63.
Heritage, G.L., Hetherington, D., 2007. Toward a protocol for laser scanning in
fluvial geomorphology. Earth Surface Processes and Landforms 32, 66–74.
Heritage, G.L., Milan, D.J., 2009. Terrestrial laser scanning of grain roughness in a
gravel-bed river. Geomorphology 113, 4–11.
Heritage, G.L., Milan, D.J., Large, A.R.G., Fuller, I., 2009. Influence of survey strategy
and interpolation model upon DEM quality. Geomorphology 112, 334–344.
Hetherington, D., Heritage, G.L., Milan, D.J., 2005. Reach scale sub-bar dynamics
elucidated through oblique LiDAR survey. International Association of
Hydrological Scientists Red Book Publication, IAHS Publication 291, 278–284.
Hickin, A.S., Kerr, B., Barchyn, T.E., Paulen, R.C., 2009. Using ground-penetrating
radar and capacitively coupled resistivity to investigate 3-D fluvial architecture
and grain-size distribution of a gravel floodplain in Northeast British Columbia,
Canada. Journal of Sedimentary Research 79, 457–477.
Hodge, R., Brasington, J., Richards, K., 2009a. In situ characterization of grain-
scale fluvial morphology using Terrestrial Laser Scanning. Earth Surface
Processes and Landforms 34, 954–968.
Hodge, R., Brasington, J., Richards, K., 2009b. Analysing laser-scanned digital
terrain models of gravel bed surfaces: linking morphology to sediment transport
processes and hydraulics. Sedimentology 56, 2024–2044.
158 Digital Terrain Modeling
Hoechstetter, S., Walz, U., Dang, L.H., Thinh, N.X., 2008. Effects of topography and
surface roughness in analyses of landscape structure–a proposal to modify the
existing set of landscape metrics. Landscape Online 1, 1–14.
Hohle, J., Hohle, M., 2009. Accuracy assessment of digital elevation models by
means of robust statistical methods. ISPRS Journal of Photogrammetry and
Remote Sensing 64, 398–406.
Hong, S.-H., Wdowinski, S., Kim, S.-W., 2010. Evaluation of TerraSAR-X
observations for wetland InSAR application. IEEE Transactions on Geoscience
and Remote Sensing 48, 864–873.
Hooke, R., Le, B., 2000. On the history of humans as geomorphic agents. Geology
28, 843–846.
Horton, R.E., 1945. Erosional development of streams and their drainage basins:
Hydrophysical approach to quantitative morphology. Geological Society of
America Bulletin 56, 275–370.
Houben, P., 2008. Scale linkage and contingency effects of field-scale and hillslope-
scale controls of long-term soil erosion: anthropogeomorphic sediment flux in
agricultural loess watersheds of southern Germany. Geomorphology 101,
172–191.
Hutchinson, M.F., 1989. A new procedure for gridding elevation and stream line
data with automatic removal of spurious pits. Journal of Hydrology 106,
211–232.
Isenburg, M., Liu, Y., Shewchuk, J., Snoeyink, J., Thirion, T., 2006, Generating
raster DEM from mass points via TIN streaming. In Raubal, H.J. Miller, A.U.
Frank, M.F. Goodchild (Eds.). Proceedings of the Geographic Information
Science, 4th International Conference, GIScience, Munster, Germany, September
2006M. 4197 of Lecture Notes on Computer Science, Springer, Berlin,
pp. 186–198.
Jaeger, R.M., Schuring, D.J., 1966. Spectrum analysis of terrain of Mare Cognitum
(Lunar surface roughness in Mare Cognitum analyzed spectrally and compared
with two terrestrial areas). Journal of Geophysical Research 71, 2023–2028.
James, L.A., 2006. Bed waves at the basin scale: implications for river management
and restoration. Earth Surface Processes and Landforms 31, 1692–1706.
Jo, M.-J., Won, J.-S., Kim, S.-W., Jung, H.-S., 2010. A time-series SAR
observation of surface deformation at the southern end of the San Andreas Fault
Zone. Geosciences Journal 14, 277–287.
Jones, A.F., Brewer, P.A., Johnstone, E., Macklin, M.G., 2007. High-resolution
interpretative geomorphological mapping of river valley environments using
airborne LiDAR data. Earth Surface Processes and Landforms 32, 1574–1592.
Katzenbeisser, R., 2003. About the calibration of LIDAR sensors. Proceedings of
ISPRS Workshop ‘‘3D Reconstruction from Airborne Laser-Scanner and InSAR
data’’, Dresden, Germany. http://www.isprs.org/publications/archives.html
Kellndorfer, J.M., Walker, W.S., LaPoint, E., Kirsch, K., Bishop, J., Fiske, G., 2010.
Statistical fusion of lidar, InSAR, and optical remote sensing data for forest
stand height characterization: a regional-scale method based on LVIS, SRTM,
Landsat ETM þ , and ancillary data sets. Journal of Geophysical Research. 115,
G00E08.
Kidner, D.B., Smith, D.H., 1993. Terrain modelling and ground cover data for
propagation studies, IEE Colloquium on Terrain Modelling and Ground Cover
Data for Propagation Studies. The Institution of Electrical Engineers, London,
7/1-7/9.
Kim, S.-W., Wdowinski, S., Dixon, T.H., Amelung, F., Kim, J.W., Won, J.-S., 2010.
Measurements and predictions of subsidence induced by soil consolidation
using persistent scatterer InSAR and a hyperbolic model. Geophysical Research
Letters 37, L05304.
Kimerling, A.J., Buckley, A.R., Muehrcke, P.C., Muehrcke, J.O., 2009. Map Use:
Reading and Analysis. ESRI Press, Redlands, CA, 493 pp.
Kobler, A., Pfeifer, N., Ogrinc, P., Todorovski, L., Oštir, K., Džeroski, S., 2007.
Repetitive interpolation: a robust algorithm for DTM generation from aerial laser
scanner data in forested terrain. Remote Sensing of Environment 108, 9–23.
Kraus, K., Pfeifer, N., 1998. Determination of terrain models in wooded areas with
airborne laser scanner data. ISPRS Journal of Photogrammetry and Remote
Sensing 53, 193–203.
Kraus, K., Karel, W., Briese, C., Mandlburger, G., 2006. Local accuracy measures
for digital terrain models. The Photogrammetric Record 21, 342–354.
Kugler, H., 1964. Die geomorphologische Reiefanalyse als Grundlage
grossmassstabiger geomorphologischer Kartierung. Wissenschaft
Veroffentlichungen der Deustsche Institut fur. Landerkunde N.F. 21/22, 541–655.
Lane, SN., 1998. The use of digital terrain modelling in the understanding of
dynamic river channel systems. In: Lane, SN, Richards, K, Chandler, J (Eds.),
Landform Monitoring, Modelling and Analysis. Wiley, Chichester, pp. 311–342.
Lane, S.N., Chandler, J.H., Richards, K.S., 1994. Developments in monitoring and
modeling small-scale river bed topography. Earth Surface Processes and
Landforms 19, 349–368.
Lane, S.N., Chandler, J.H., Porfiri, K., 2001. Monitoring river channel and flume
surfaces with digital photogrammetry. Journal of Hydraulic Engineering 127,
871–877.
Lane, S.N., James, T.D., Crowell, M.D., 2000. Application of digital photogrammetry
to complex topography for geomorphological research. Photogrammetric Record
16, 793–821.
Lane, S.N., Richards, K.S., 1997. Linking river channel form and process: time,
space and causality revisited. Earth Surface Processes and Landforms 22,
249–260.
Lauknes, T.R., Piyush Shanker, A., Dehls, J.F., Zebker, H.A., Henderson, I.H.C.,
Larsen, Y., 2010. Detailed rockslide mapping in northern Norway with small
baseline and persistent scatterer interferometric SAR time series methods.
Remote Sensing of Environment 114, 2097–2109.
Lee, H.S., Younan, N.H., 2003. DTM extraction of LiDAR returns via adaptive
processing. IEEE Transactions on Geoscience and Remote Sensing 41,
2063–2069.
Li, Z., Zhu, Q., Gold, C., 2005. Digital Terrain Modeling: Principles and
Methodology. CRC Press, Boca Raton, FL, 323 pp.
Lim, M., Rosser, N.J., Allison, R.J., Petley, D.N., 2010. Erosional processes in the
hard rock coastal cliffs at Staithes, North Yorkshire. Geomorphology 114,
12–21.
Lin, Z., Oguchi, T., 2006. DEM analysis on longitudinal and transverse profiles of
steep mountainous watersheds. Geomorphology 78, 77–89.
Linsen, L., 2001. Point cloud representation. Technical report 2001-3, Science, FOC
University of Karlsruhe, Germany, 18 pp.
Liu, X., 2008. Airborne LiDAR for DEM generation: some critical issues. Progress
in Physical Geography 32, 31–49.
Maas, H., 2002. Methods for measuring height and planimetry discrepancies in
airborne laser scanner data. Phogogrammetric Engineering and Remote Sensing
68, 933–940.
MacMillan, R.A., Shary, P.A., 2009. Landforms and landform elements in
geomorphometry. In: Hengl, T., Reuter, H.I. (Eds.), Geomorphometry: Concepts,
Software, and Applications. Elsevier, Oxford, U.K., pp. 227–254.
Mackin, J.H., 1948. Concept of the graded river. Geological Society of America
Bulletin 59, 463–512.
Mandlburger, G., Hauer, C., Hofle, B., Habersack, H., Pfeifer, N, 2008. Optimisation
of LiDAR derived terrain models for river flow modelling, Hydrology and Earth
System Sciences Discussions 5, 3605–3638.
Mark, D.M., 1974. Line intersection method for estimating drainage density.
Geology 2, 235–236.
Martin-Vide, J.P., Ferrer-Boix, C., Ollero, A., 2010. Incision due to gravel mining:
modeling a case study from the Gallego River, Spain. Geomorphology 117,
261–271.
Marzolff, I., Poesen, J., 2009. The potential of 3D gully monitoring with GIS using
high-resolution aerial photography and a digital photogrammetry system.
Geomorphology 111, 48–60.
Massonnet, D., Feigl, K.L., 1998. Radar interferometry and its application to
changes in the Earth’s surface. Reviews in Geophysics 36, 441–500.
Matthews, N.A., 2008. Aerial and close-range photogrammetric technology:
providing resource documentation, interpretation, and preservation. Technical
Note 428. US Department of the Interior, Bureau of Land Management, National
Operations Center, Denver, Colorado, 42 pp.
Maune, D.F., 2007. Digital Elevation Model Technologies and Applications: The
Dem Users Manual. American Society for Photogrammetry and Remote Sensing,
Bethesda, MD.
Maune, D.F., Kopp, S.M., Crawford, C.A., Zervas, C.E., 2007. Introduction.
In: Maune, D.F. (Ed.), Digital Elevation Model Technologies and Applications:
The Dem Users Manual. American Society for Photogrammetry and Remote
Sensing, Bethesda, MD, pp. 1–35.
McCoy, S.W., Kean, J.W., Coe, J.A., Staley, D.M., Wasklewicz, T.A., Tucker, G.E.,
2010. Evolution of a natural debris flow: in situ measurements of flow
dynamics, video imagery, and terrestrial laser scanning. Geology 38, 735–738.
McMaster, K.J., 2002. Effects of digital elevation model resolution on derived
stream network positions. Water Resources Research 38, 13-1–13-8.
Melton, M.A., 1957. An analysis of the relations among elements of climate, surface
properties and geomorphology. Columbia University, Office of Naval Research.
Project NR 389–042, Technical Report 11.
Merritt, W.S., Letcher, R.A., Jakeman, A.J., 2003. A review of erosion and sediment
transport models. Environmental Modelling and Software 18, 761–799.
Miller, C.L., Leflamme, R.A., 1958. The digital terrain model - Theory and
application. Photogrammetric Engineering 24, 433–442.
Millette, T.L., Argow, B.A., Marcano, E., Hayward, C., Hopkinson, C.S., Valentine, V.,
2010. Salt marsh geomorphological analyses via integration of multitemporal

multispectral remote sensing with LIDAR and GIS. Journal of Coastal Research
26, 809–816.
Mills, J.P., Buckley, S.J., Mitchell, H.L., Clarke, P.J., Edwards, S.J., 2005. A
geomatics data integration technique for coastal change monitoring. Earth
Surface Processes and Landforms 30, 651–664.
Minar, J., Evans, I.S., 2008. Elementary forms for land surface segmentation: the
theoretical basis of terrain analysis and geomorphological mapping.
Geomorphology 95, 236–259.
Mitas, L., Mitasova, H., 1999. Spatial interpolation. In: Longley, P.A., Goodchild,
M.F., Maguire, D.J., Rhind, D.W. (Eds.), Geographical Information Systems.
Wiley, New York, pp. 481–492.
Mitasova, H., Hofierka, J., 1993. Interpolation by regularized spline with tension: II.
Application to terrain modeling and surface geometry analysis. Mathematical
Geology 25, 657–669.
Moenning, C., Dodgson, N., 2003. A new point cloud simplification algorithm.
Proceedings of the 3rd IASTED International Conference on Visualization,
Imaging and Image Processing. Benalmadena, Spain, pp. 1027–1033.
Morisawa, M., 1988. The Geological Society of America Bulletin and the
development of quantitative geomorphology. Geological Society of America
Bulletin 100, 1016–1022.
Murzyn, F., Mouaze., D., Chaplin, J.R., 2006. Flow visualization and free surface
length scales measurements in a horizontal jet beneath a free surface.
Experimental Thermal and Fluid Science 30, 703–710.
Nolan, M., Fatland, D.R., Hinzman, L., 2009. DInSAR measurement of soil moisture.
IEEE Transactions on Geoscience and Remote Sensing 41, 2802–2813.
Noma, A.A., Misulia, M.G., 1959. Programming topographic maps for automatic
terrain model construction. Surveying Mapping 19, 355–366.
National Digital Elevation Program (NDEP), 2004. Guidelines for Digital Elevation
Data v.1. www.ndep.gov/NDEP_Elevation_Guidelines_Ver1_10May2004.pdf
(visited 6-2-10), 93 pp.
Okada, Y., Hirao, C., Horiuchi, T., Hara, Y., Yedidia, J.S., Azarbayejani, A., Oishi, N.,
2007. Highly accurate DSM reconstruction using Ku-band Airborne InSAR, IEEE
International Geoscience and Remote Sensing Society Symposium (IGARSS),
July 2007, pp. 5049–5052.
Oppikofer, T., Jaboyedoff, M., Blikra, L., Derron, M.-H., Metzger, R., 2009.
Characterization and monitoring of the Åknes rockslide using terrestrial laser
scanning. Natural Hazards and Earth System Science 9, 1003–1019.
Pauly, M., Gross, M., Kobbelt, L.P., 2002. Efficient simplification of point-sampled
surfaces. Proceedings of the 13th IEEE Visualization Conference, Boston, MA,
USA, pp. 163–170.
Peipe, J., Stephani, M., 2003. Performance evaluation of a 5 megapixel digital
metric camera for use in architectural photogrammetry. The International
Archives of the Photogrammetry, Remote Sensing and Spatial Information
Sciences 34, 259–261.
Peipe, J. and Tecklenburg, W., 2006. Photogrammetric camera calibration software -
a comparison. Proceedings of the ISPRS Commission V Symposium ‘Image
Engineering and Vision Metrology’, Dresden, Germany, 4 pp.
Pelletier, J.D., 1999. Self-organization and scaling relationships of evolving river
networks. Journal of Geophysical Research 104, 7359–7375.
Perron, J.T., Kirchner, J.W., Dietrich, W.E., 2008. Spectral evidence of characteristic
spatial scales and nonfractal structure in landscapes. Journal of Geophysical
Research 113, F04003.
Petrie, G., Toth, C.K., 2009a. Introduction to laser ranging, profiling and
scanning. In: Shan, J., Toth, C.K. (Eds.), Topographic Laser Ranging
and Scanning: Principles and Processing. CRC Press, Boca Raton, FL,
pp. 1–27.
Petrie, G., Toth, C.K., 2009b. Terrestrial laser scanners. In: Shan, J., Toth, C.K.
(Eds.), Topographic Laser Ranging and Scanning: Principles and Processing.
CRC Press, Boca Raton, FL, pp. 87–127.
Petrie, G., Toth, C.K., 2009c. Airborne and spaceborne laser profilers and scanners.
In: Shan, J., Toth, C.K. (Eds.), Topographic Laser Ranging and Scanning:
Principles and Processing. CRC Press, Boca Raton, FL, pp. 29–85.
Pfeifer, N., Mandlburger, G., 2009. LiDAR data filtering and DTM generation.
In: Shan, J., Toth, C.K. (Eds.), Topographic Laser Ranging and Scanning. CRC
Press, Boca Raton, FL, USA, pp. 308–333.
Phillips, J.D., 1999a. Earth surface systems. Complexity, order and scale. Basil
Blackwell, Oxford.
Phillips, J.D., 1999b. Divergence, convergence, and self-organization in landscapes.
Annals of the Association of American Geographers 89, 466–488.
Phillips, J.D., 2005. Entropy analysis of multiple scale causality and qualitative
causal shifts in spatial systems. The Professional Geographer 57, 83–93.
Phillips, J.D., 2006. Deterministic chaos and historical geomorphology: a review
and look forward. Geomorphology 76, 109–121.
Digital Terrain Modeling 159
Phillips, J.D., 2012. Synchronization and scale in geomorphic systems.
Geomorphology 137, 150–158.
Pike, R.J., 1995. Geomorphometry – progress, practice, and prospect. Zeitschrift fur
Geomorphologie, Supplement band 101, 221–238.
Pike, R.J., Evans, I.S., Hengl, T., 2009. Geomorphometry: a brief guide. In: Hengl,
T., Reuter, H.I. (Eds.), Geomorphometry: Concepts, Software, and Applications.
Elsevier, Oxford, U.K., pp. 3–30.
Plattner, C., Wdowinski, S., Dixon, T.H., Biggs, J., 2010. Surface subsidence
induced by the Crandall Canyon Mine (Utah) collapse: InSAR observations and
elasto-plastic modelling. Geophysical Journal International 183, 1089–1096.
Poland, M., 2010. Localized surface disruptions observed by InSAR during strong
earthquakes in Java and Hawai’i. Bulletin of the Seismological Society of
America 100, 532–540.
Popescu, S.C., Zhao, K., 2008. A voxel-based LiDAR method for estimating crown
base height for deciduous and pine trees. Remote Sensing of Environment 112,
761–781.
Prokop, A., Panholzer, H., 2009. Assessing the capability of terrestrial laser
scanning for monitoring slow moving landslides. Natural Hazards and Earth
System Science 9, 1921–1928.
Quinn, P., Beven, K., Chevallier, P., Planchon, O., 1991. The prediction of hillslope
flow paths for distributed hydrological modelling using digital terrain models.
Hydrological Processes 5, 59–79.
Raper, J., Livingstone, D., 1995. Development of a geomorphological spatial model
using object-oriented design. International Journal of Geographical Information
Systems 9, 359–383.
Rapp, A., Nyberg, R., 1981. Alpine debris flows in northern Scandinavia:
morphology and dating by lichenometry. Geografiska Annaler Series A, Physical
Geography 63, 183–196.
Redmondino, F., El-Hakim, S., 2006. Image-based 3D modelling: a review. The
Photogrammetric Record 21, 269–291.
Remondino, F., Fraser, C., 2006. Digital camera calibration methods: consideration
and comparisons. The International Archives of the Photogrammetry and Remote
Sensing Commission V Symposium ‘Image Engineering and Vision Metrology’,
Dresden, Germany, vol. 36, pp. 266–272.
Reuter, H.I., Hengl, T., Gessler, P., Soille, P., 2009. Preparation of DEMs for
geomorphometric analyses. In: Hengl, T., Reuter, H.I. (Eds.), Geomorphometry:
Geomorphometry: Concepts, Software, Applications. Developments in Soil
Science. Elsevier, Amsterdam, vol. 33, pp. 87–120.
Rhoads, B.L., 2006. The Dynamic Basis of Geomorphology Reenvisioned. Annals of
the Association of American Geographers 96, 14–30.
Rignot, E., Echelmeyer, K., Krabill, W., 2001. Penetration depth of interferometric
synthetic-aperture radar signals in snow and ice. Geophysical Research Letters
28, 3501–3504.
Robeson, S.M., 1997. Spherical methods for spatial interpolation: review and
evaluation. Cartography and Geographic Information Systems 24, 3–20.
Rodrı́guez-Iturbe, I., Rinaldo, A., 2001. Fractal River Basins: Chance and Self-
Organization. Cambridge University Press, Cambridge, U.K., 564 pp.
Roering, J., Kirchner, J., Dietrich, W.E., 1999. Evidence for nonlinear, diffusive
sediment transport on hillslopes and implications for landscape morphology.
Water Resources Research 36, 853–870.
Roering, J.J., Stimely, L.L., Mackey, B.H., Schmidt, D.A., 2009. Using DInSAR,
airborne LiDAR, and archival air photos to quantify landsliding and sediment
transport. Geophysical Research Letters 36, L19402.
Roggero, M., 2001. Airborne laser scanning: clustering in raw data. International
Archives of Photogrammetry, Remote Sensing and Spatial Information Sciences
34, 227–232.
Rosen, P.A., Hensley, S., Joughin, I.R., Li, F.K., Madsen, S.N., Rodriguez, E.,
Goldstein, R.M., 2000. Synthetic aperture radar interferometry. Proceedings of
the IEEE 88, 333–382.
Rossignac, J., Borrel, P., 1993. Multiresolution 3D approximations for rendering
complex scenes. Modeling in Computer Graphics: Methods and Applications.
Springer, New York, 455-465.
Rubin, D.M., Tsoar, H., Blumberg, D.G., 2008. A second look at western Sinai seif
dunes and their lateral migration. Geomorphology 93, 335–342.
Sakaguchi, Y., 1969. A theory of relief forming. Bulletin Department of Geography
University of Tokyo 1, 33–66.
Schmidt, J., Andrew, R., 2005. Multi-scale landform characterization. Area 37,
341–350.
Schumm, S.A., 1956. Evolution of drainage basins and slopes in badlands at Perth
Amboy, New Jersey. Bulletin of Geological Society of America 67, 597–646.
Shaffer, E., Garland, M., 2001. Efficient adaptive simplification of massive meshes.
Proceedings of the Conference on Visualization ’01, Washington DC.,
pp. 127–134.
160 Digital Terrain Modeling
Shan, J., Sampath, A., 2005. Urban DEM generation from raw LiDAR data: a
labeling algorthm and its performance. Photogrammetric Engineering and
Remote Sensing 71, 217–226.
Shary, P.A., Sharaya, L.S., Mitusov, A.V., 2002. Fundamental quantitative methods of
land surface analysis. Geoderma 107, 1–32.
Shary, P.A., Sharaya, L.S., Mitusov, A.V., 2005. The problem of scale-specific and
scale-free approaches in geomorphometry. Geografia Fisica e Dinamica
Quaternaria 28, 81–101.
Shi, W.Z., Tian, Y., 2006. A hybrid interpolation method for the refinement of a
regular grid digital elevation model. International Journal of Geographical
Information Science 20, 53–67.
Shrestha, R., Cather, W., Slatton, C., Dietrich, W., 2007. ‘‘Research-Quality’’ Airborne
Laser Swath Mapping: The Defining Factors. National Center for Airborne Laser
Mapping White Paper (v. 1.2), http://www.ncalm.cive.uh.edu/assets/
publication_pdf/NCALM_WhitePaper_v1.2.pdf.
Shugar, D.H., Rabus, B., Clague, J.J., 2010. Elevation changes (1949–1995) of
Black Rapids Glacier, Alaska, derived from a multi-baseline InSAR DEM and
historical maps. Journal of Glaciology 56, 625–634.
Silván-Cárdenas, J.L., Wang, L., 2006. A multiresolution approach for filtering
LiDAR altimetry data. ISPRS Journal of Photogrammetry and Remote Sensing
61, 11–22.
Sithole, G., 2001. Filtering of laser altimetry data using a slope adaptive filter.
International Archives of Photogrammetry, Remote Sensing and Spatial
Information Sciences 34, 203–210.
Sithole, G., Vosselman, G., 2004. Experiomental comparison of filter algorithms for
for bare-Earth extraction from airborne laser scanning point clouds. ISPRS
Journal of Photogrammetry and Remote Sensing 59, 85–101.
Smith, B., Mark, D.M., 2003. Do mountains exist? Toward an ontology of
landforms. Environment and Planning B: Planning and Design 30, 411–427.
Smith, L.C., 2002. Emerging applications of interferometric synthetic aperture radar
(InSAR) in geomorphology and hydrology. Annals of the Association of
American Geographers 92, 385–398.
Solberg, S., Astrup, R., Gobakken, T., Næsset, E., Weydahl, D.J., 2010. Estimating
spruce and pine biomass with interferometric X-band SAR. Remote Sensing of
Environment 114, 2353–2360.
Southgate, H.N., Möller, N.K., 2000. Fractal properties of coastal profile evolution at
Duck, North Carolina. Journal of Geophysical Research 105, 11489–11507.
Staley, D.M., Wasklewicz, T.A., Blaszczynksi, J.S., 2006. Surficial patterns of debris
flow deposition on alluvial fans in Death Valley, CA using airborne laser swath
mapping data. Geomorphology 74, 152–163.
Stark, C.P., Barbour, J.R., Hayakawa, Y., et al., 2010. The climatic signature of
incised river meanders. Science 327, 1497–1501.
Stott, T., 2002. Bedload transport and channel bed changes in the proglacial Skeldal
River, Northeast Greenland. Arctic, Antarctic, and Alpine Research 34, 334–344.
Strahler, A.N., 1952. Dynamic basis of geomorphology. Bulletin Geological Society
of America 63, 923–938.
Strahler, A.N., 1958. Dimensional analysis applied to fluvially eroded landforms.
Geological Society of America Bulletin 69, 279–300.
Strahler, A.N., 1992. Quantitative/dynamic geomorphology at Columbia 1945-60: a
retrospective. Progress in Physical Geography 16, 65–84.
Sturzenegger, M., Stead, D., 2009. Close-range terrestrial digital photogrammetry
and terrestrial laser scanning for discontinuity characterization on rock cuts.
Engineering Geology 106, 163–182.
Thoms, M.C., Parsons, M., 2003. Identifying spatial and temporal patterns in the
hydrological character of the Condamine-Balonne River, Australia, using
multivariate statistics. River Research and Applications 19, 443–458.
Tobler, W.R., 1969. An analysis of a digitized surface. In: Davis, C.M. (Ed.), A Study
of the Land Type, Final Report on Contract No. DA-31-124-ARO-D-456 to the
U.S. Army Research Office, Durham, NC, Department of Geography, University of
Michigan, Ann Arbor, pp. 59–83.
Tobler, W.R., Davis, C.M., 1968. A digital terrain library. Technical Report ORA
08055, Department of Geography, University of Michigan, Ann Arbor, 28 pp.
Tucker, G.E., Bradley, D.N., 2010. Trouble with diffusion: reassessing hillslope
erosion laws with a particle based model. Journal of Geophysical Research 115,
F00A10.
Tucker, G.E., Bras, R.L., 1998. Hillslope processes, drainage density, and landscape
morphology. Water Resources Research 34, 2751–2764.
Tucker, G.E., Hancock, G.R., 2010. Modelling landscape evolution. Earth Surface
Processes and Landforms 35, 28–50.
Turk, G., 1992. Re-tiling polygonal surfaces. SIGGRAPH 92 Conference Proceedings
26, 55–64.
U.S. Army Map Service, 1963. Computer production of terrain models.
Communication of the Association for Computing Machinery 6, 190–191.
de Vente, J., Poesen, J., 2005. Predicting soil erosion and sediment yield at the
basin scale: scale issues and semi-quantitative models. Earth-Science Reviews
71, 95–125.
de Vente, J., Poesen, J., Arabkhedri, M., Verstraeten, G., 2007. The sediment
delivery problem revisited. Progress in Physical Geography 31, 155–178.
Volker, H.X., Wasklewicz, T.A., Ellis, M.A., 2007. A topographic fingerprint to
distinguish alluvial fan formative processes. Geomorphology 88, 34–45.
Vosselman, G., 2000. Slope based filtering of laser altimetry data. International
Archives of Photogrammetry, Remote Sensing and Spatial Information Sciences
33, 958–964.
Vosselman, G., Gorte, B., Sithole, G., Rabbani, T., 2004. Recognising structure in
laser scanner point clouds. International Archives of Photogrammetry, Remote
Sensing and Spatial Information Sciences 46, 33–38.
Wainwright, J., Parsons, A.J., Müller, E.N., Brazier, R.E., Powell, M., Fenti, B., 2008.
A transport-distance approach to scaling erosion rates: 1. Background and
model development. Earth Surface Processes and Landforms 33, 813–826.
Walker, I.J., 2010. Changing views in Canadian Geomorphology: are we seeing the
landscape for the processes? The Canadian Geographer 54, 261–276.
Walker, J.P., Willgoose, G.R., 1999. On the effect of digital elevation model
accuracy on hydrology and geomorphology. Water Resources Research 35,
2259–2268.
Walling, D.E., 1983. The sediment delivery problem. Journal of Hydrology 65,
209–237.
Walls, J.H., Houbolt, J.V., Press, H., 1954. Some measurements and power spectra
of runway roughness. NACA Technical Note 3305, Washington, DC.
Wasklewicz, T., Whitley, D., Volker, H., Staley, D., 2005. 3D Laser Scanning and
Intensity Mapping: A New Methodology for Mapping and Monitoring Rock Art
Sites. International Newsletter on Rock Art 41, 16–25.
Wasklewicz, T.A., Hattanji, T., 2009. High-resolution Analysis of Channel Change in
a Headwater Stream after a Debris Flow, Ashio Mountains, Japan. Professional
Geographer 61, 231–249.
Wasklewicz, T.A., Staley, D.M., Mihir, M., Seruntine, L., 2007. Virtual recording of
lichen species: integrating terrestrial laser scanning and GIS techniques.
Physical Geography 28, 183–192.
Weber, D., Englund, E., 1992. Evaluation and comparison of spatial interpolators.
Mathematical Geology 24, 381–391.
Wechsler, S.P., 1999. Digital Elevation Model (DEM) uncertainty: evaluation and
effect on topographic parameters. In Proceedings of the 1999 ESRI User
Conference, San Diego, California, http://gis.esri.com/library/userconf/proc99/
proceed/papers/pap262/p262.htm.
Wen, Y.-Y., Ma, K.-F., 2010. Fault geometry and distribution of asperities
of the 1997 Manyi, China (Mw ¼ 7.5), earthquake: integrated analysis
from seismological and InSAR data. Geophysical Research Letters 37, L05303.
Wheaton, J.M., Brasington, J., Darby, S.E., Merz, J.E., Pasternack, G.B., Sear, D.A.,
Vericat, D., 2010b. Linking Geomorphic Changes to Salmonid Habitat at a Scale
Relevant to Fish. River Research and Applications 26, 469–486.
Wheaton, J.M., Brasington, J., Darby, S.E., Sear, D., 2010a. Accounting for
Uncertainty in DEMs from Repeat Topographic Surveys: Improved Sediment
Budgets. Earth Surface Processes and Landforms 35, 136–156.
Wilson, J.P., Gallant, J.C., 2000. Terrain Analysis. Principles and Applications.
Wiley, New York, 479 pp.
Wise, S., 2000. Assessing the quality for hydrological applications of digital elevation
models derived from contours. Hydrological Processes 14, 1909–1929.
Xu, C., Liu, Y., Wen, Y., Wang, R., 2010. Coseismic slip distribution of the 2008
Mw 7.9 Wenchuan Earthquake from joint inversion of GPS and InSAR data.
Bulletin of the Seismological Society of America 100, 2736–2749.
Zebker, H.A., Goldstein, R.M., 1986. Topographic mapping from Interferometric
Synthetic Aperture Radar observations. Journal of Geophysical Research 91,
4993–4999.
Zhang, K., Whitman, D., 2005. Comparison of three algorithms for filtering airborne
LiDAR data. Photogrammetric Engineering and Remote Sensing 71, 313–324.
Zhang, K.Q., Chen, S.C., Whitman, D., Shyu, M.L., Yan, J.H., Zhang, C.C., 2003.
A progressive morphological filter for removing nonground measurements from
airborne LiDAR data. IEEE Transactions on Geoscience and Remote Sensing 41,
872–882.
Zhang, W., Montgomery, D.R., 1994. Digital elevation model grid size, landscape
representation, and hydrologic simulations. Water Resources Research 30,
1019–1028.
Zhou, X., Chang, N.-B., Li, S., 2009. Applications of SAR interferometry in Earth
and environmental science research. Sensors 9, 1876–1912.
Zimmerman, D., Pavlik, C., Ruggles, A., Armstrong, M., 1999. An experimental
comparison of ordinary and universal kriging and inverse distance weighting.
Mathematical Geology 31, 375–390.
 Digital Terrain Modeling 161

Biographical Sketch

Thad Wasklewicz is an Associate Professor in the Department of Geography at East Carolina University. Wask-
lewicz is a geomorphologist interested in debris-flow processes as they relate to the evolution of hillslopes,
channels, and alluvial fans. Research and teaching experiences include the application of GIS, ALS, TLS, TPT, and
GPS to geomorphic studies.

Dennis Staley (PhD) is a research scientist with the United States Geological Survey, Central Region Geologic
Hazards Team in Denver, CO, USA. Staley has research experience in geomorphology with special interest on mass
wasting and expertize in GIS, ALS, TLS, TPT, and GPS. His current research involves postwildfire debris flow
susceptibility, hazard assessments, and warning system development.

Takashi Oguchi is the Vice-Director and a Prof. at the Center for Spatial Information Science at the University of
Tokyo. Prof. Oguchi has been working on the application of GIS to various fields of earth science, especially
geomorphology, hydrology, and Quaternary science. Oguchi is also Co-Editor-in-Chief of the journal
Geomorphology.

Kathryn Reavis is an undergraduate student in the Department of Geography at East Carolina University. Reavis’s
research-educational experiences focus on geomorphology with applications of GIS, TLS, TPT, and GPS.
 3.7 Geomorphometry
JP Wilson, University of Southern California, Los Angeles, CA, USA
MP Bishop, Texas A&M University College Station, TX, USA
r 2013 Elsevier Inc. All rights reserved.

3.7.1 Introduction 163

3.7.2 Digital Terrain Modeling 164
3.7.2.1 Representation 164
3.7.2.2 Data Capture 165
3.7.2.3 Data Preprocessing and DEM Construction 167
3.7.2.4 Error and Artifacts 168
3.7.3 Land-Surface Parameters 169
3.7.3.1 Primary Parameters 169
3.7.3.2 Secondary Land-Surface Parameters 173
3.7.3.2.1 Hydrology 173
3.7.3.2.2 Climatology 175
3.7.4 Land-Surface Objects and Landforms 178
3.7.5 Conclusions 180
References 181

Abstract

The study of surface processes and landforms requires quantitative characterization of the topography. New theoretical/
conceptual and practical advances in understanding and mapping various aspects of geomorphological systems have
emerged from new geospatial data and analysis of the topography. This chapter describes geomorphometry, or the science
of quantitative land-surface characterization, and how it can be used to represent and sample the land surface, generate
digital elevation models (DEMs), correct errors and artifacts from surface models, compute land-surface parameters and
objects, and use various forms of quantitative information in different application domains to address or solve problems.

Glossary analysis. Such errors are propagated in terrain analysis in

Digital elevation model A digital elevation model (DEM)
is generally a land-surface model that attempts to accurately
portray the altitude field of the topography. In
geomorphology, it commonly takes the form of a raster data
layer representing a field of square tessellations. The
resolution of the grid cells is usually determined based on
the source data utilized and the desired scale for
representing the topography.
Digital terrain modeling Digital terrain modeling (DTM)
refers to a workflow process of acquiring data that samples
the altitude field, preprocessing the data to generate a digital
elevation model, and error and uncertainty analysis to
identify and remove systematic and random errors.
Error propagation Digital elevation models exhibit an
inherent error due to digital terrain modeling that includes
choice of sensor in data capture, algorithm and methods
selection in DEM generation, and effective implementation
of DEM error correction, given DEM error and uncertainty
the computation of land-surface parameters and objects
and in modeling efforts that utilize parameters and objects.
Geomorphometry The discipline that is concerned with
the science of quantitative land-surface characterization.
Land-surface objects Land-surface objects are spatial
entities that represent a meaningful segmentation of the
Earth’s surface. They are generated from land-surface
parameters using a variety of algorithms and methods, and
generally relate to the morphology of the topography in
terms of landform elements, features, or functional units.
Terrain segmentation can also relate to topographic
position and structure, surface material, or process
domains, if properly defined and delineated.
Land-surface parameters Land-surface parameters are
also called geomorphometric parameters, and they attempt
to quantitatively characterize various aspects of the
topography. They can be defined and classified based on
geometry, scale, and by surface-process characterization.

Wilson, J.P., Bishop, M.P., 2013. Geomorphometry. In: Shroder, J. (Editor in

Chief), Bishop, M.P. (Ed.), Treatise on Geomorphology. Academic Press,
San Diego, CA, vol. 3, Remote Sensing and GIScience in Geomorphology,
pp. 162–186.

162 Treatise on Geomorphology, Volume 3 http://dx.doi.org/10.1016/B978-0-12-374739-6.00049-X


They are used to generate land-surface objects and
characterize process mechanics in surface-process modeling.
A variety of parameters such as slope, slope azimuth,
curvature, surface roughness, and relief are used for
studying geomorphological systems and for
geomorphological mapping.
Segmentation Segmentation refers to the partitioning of
the land surface into meaningful spatial entities.
3.7.1 Introduction
The topography plays a fundamental role in modulating several
components of Earth’s dynamic systems including atmospheric,
geomorphic, hydrologic, ecologic, and geological processes.
The topography constrains the operational scale of surface
processes, and partially governs both climate and tectonic for-
cing (Molnar and England, 1990; Bishop et al., 2010; Koons
et al., 2012). The strength of the linkage between form and
process can range from weak to strong, and may or may not be
inherently visible on the landscape depending on the com-
plexity of the topography. Nevertheless, moderate to strong
linkages have been observed, such that an understanding of the
nature of the land surface can provide insights and under-
standing of the nature and magnitude of several processes
(Hutchinson and Gallant, 2000). Consequently, it is necessary
to quantitatively characterize the land surface and segment the
topography into fundamental spatial units, as the topography
inherently represents the results of the interplay between vari-
ous systems, and records an imprint of landscape dynamics
(over a limited time). Therefore, the utility of digital elevation
models (DEMs) and the analysis of topography (geomorpho-
metry) have and will continue to revolutionize the field of
geomorphology, as critical information regarding process
mechanics, process domains, feedback mechanisms, poly-
genetic evolution, and landforms continues to be investigated
and generated. Furthermore, with the rapid proliferation of
geographic information technologies, new data, algorithms,
and analysis/modeling techniques allow new capabilities. These
capabilities represent the evolution of the field of geomor-
phometry, which, in its broadest sense, refers to the science of
quantitative land-surface characterization (Pike, 1995, 2000) or
digital terrain analysis. For more details regarding the definition
and terminologies used in geomorphometry, see Wilson and
Gallant (2000a), Li et al. (2005), Zhou et al. (2008), Pike et al.
(2009), Hengl and Reuter (2009), and Wilson (2012).
Modern geomorphometry focuses on the extraction of
land-surface parameters and the segmentation of the land-
scape into spatial entities/features (land-surface objects) from
digital topography. This characterization relies on the so-called
specific and general modes of geomorphometric analysis that
were first defined by Evans (1972). The specific mode of an-
alysis attempts to describe discrete surface features (i.e.,
landforms), whereas the general mode attempts to describe
the continuous land surface. Pike et al. (2009) have since
updated these definitions, such that a land-surface parameter
is a descriptive measure of surface form (e.g., slope, slope
azimuth, curvature) and a land-surface object is a discrete
Geomorphometry 163
Segmentation is required for object-oriented analysis of
land-surface objects.
Surface-process modeling Surface-process modeling
refers to physics-based deterministic modeling that
attempts to characterize process mechanics and how
topography governs various process rates, and the
influence of various processes on topographic landscape
evolution.
surface feature (e.g., watershed, cirque, alluvial fan, drainage
network). Although this definition represents an improve-
ment, it is worth noting that this is a somewhat arbitrary
distinction, and there are already research examples that
demonstrate that these two perspectives are closely linked
to one another (e.g., Gallant and Dowling, 2003; Deng and
Wilson, 2008). Furthermore, it is clear that formalizing such
linkages is necessary in geomorphological research to address
concepts such as surface-process overprinting and polygenetic
evolution and to address multiple perspectives in geo-
morphological mapping (Bishop et al., 2012).
Collectively, geomorphometry is a rapidly evolving and
complicated field. This is in part due to its multidisciplinary
nature and the inclusion of information technology. Similar to
the field of geographic information science, it is based on de-
velopments in a variety of fields including source and end-user
disciplines. It not only attempts to deal with theoretical/con-
ceptual issues involving representation and spatio-temporal
variation, but also includes issues of data collection and an-
alysis, numerical modeling, and the utilization of other domain
knowledge for conceptual and practical problem solving.
Consequently, the field of geomorphometry is based on a sci-
entific treatment of ‘land surface’ and its characterization that
accounts for surface processes and morphology. Rapid
evolution has been facilitated by geographic information
technology and the widespread availability of DEMs. Geo-
morphologists now have many new capabilities to manipulate
and extract information from a variety of data sources. Never-
theless, it is important to recognize the empirical nature of
many forms of spatial analysis and modeling, and many issues
raise important questions about the assumptions and validity
of various approaches (Bishop et al., 2001).
Many questions still remain, and geomorphologists must
be aware of the advantages and limitations associated with
various representations and data structures, metrics/indices,
spatial modeling approaches, and their utility for geo-
morphological investigations. Furthermore, investigators must
be familiar with the mathematical underpinnings of geo-
morphometric analysis in order to adequately use information
and to interpret the results (Bishop and Shroder, 2004a).
Collectively, many issues point to a series of key questions that
in general include: (1) How should the land surface be rep-
resented? (2) What preprocessing is required to produce a
useable DEM? (3) What approach to error and uncertainty
analysis is required? (4) What algorithms are best for pro-
ducing land-surface parameters? (5) What methods are best
for producing land-surface objects? (6) Is there a need to
develop new parameters and objects to address a particular
164 Geomorphometry
problem? (7) What algorithms and approaches are best suited
for a particular mapping application or do methods even
exist? (8) Does an adequate model exist or is there a need to
develop or modify one?
Many of these questions relate to data sources, issues, and
capabilities, although it is important to note that considerable
research is still required to address a whole host of issues in
geomorphology. In many cases, the answers to these questions
may not be clear, as software-tool development has focused on
the tool-box approach, foregoing formal scientific treatments
to mapping and geographic information system (GIS)-based
analysis/modeling. These challenges, however, can also be
attributed to the steady growth in the development and sheer
number of parameters and algorithms for processing DEMs.
Consequently, the values of parameters depend on a variety of
factors, including the parameterization scheme, measurement
scale of data, computational scale of analysis, and many em-
pirical parameters that are used to address several conceptual
issues. This represents a serious issue in geomorphology, as the
spatial patterns associated with many metrics/indices may not
represent real-world phenomena, and the use of empirical
parameters reduces the ability to formalize important aspects
of the gemorphological system (Bishop et al., 2012).
Notwithstanding these challenges, land-surface parameters
and objects have been adopted in a variety of applications and
environmental settings. Geomorphometry has been used to
predict the distribution of soil properties (e.g., Zhu et al.,
1997; Bishop and Minasny, 2005), model soil redistribution
(i.e., erosion and deposition) processes (e.g., Mitášová et al.,
1995), assess the likelihood of slope hazards (e.g., Guzzetti
et al., 2005; Kheir et al., 2007), model solar-radiation poten-
tial (e.g., Reuter et al., 2005), improve vegetation mapping
(e.g., Bolstad and Lillesand, 1992; Antonić et al., 2003),
analyze wildfire propagation (e.g., Hernández Encinas et al.,
2007), assess the role of surface processes in mountain topo-
graphic evolution (Burbank et al., 1996; Bishop and Shroder,
2000; Bishop et al., 2003), and to predict water flow, drainage,
and flooding in many hydrological applications. More
applications are rapidly emerging, and geomorphologists play
an important role in the development and evaluation of ap-
proaches that are based on geomorphological concepts.
Collectively, this chapter examines the data sources and
methods used to generate DEMs, and how land-surface par-
ameters and objects can be used in geomorphology. A typical
digital terrain analysis workflow is presented that includes basic
data capture, data preprocessing, DEM generation, compu-
tation of land-surface parameters and objects, landform classi-
fication, and surface-process modeling. Throughout, examples
are provided of how parameters and geomorphometric ana-
lyses can provide new opportunities for geomorphological re-
search. The focus is on presenting current capabilities and
issues associated with the quantitative characterization of the
topography.
3.7.2 Digital Terrain Modeling
The generation of DEMs involves data modeling or represen-
tation choices, sampling the land surface, representing
and creating a surface model from the sampled heights, and
correcting the errors and artifacts in the surface model (Hengl
and Reuter, 2009). Each phase in the production of a DEM is
critical for determining its utility, and in assessing the amount
of error that will propagate through the analysis phase. The
general workflow is depicted in Figure 1.
3.7.2.1 Representation
Many topics in geomorphology and geomorphometry are in-
herently related to the space–time representation of topo-
graphy (Bishop and Shroder, 2004b). This topic is complex,
and a variety of philosophical, cognitive, and natural-science
perspectives exist. The current use of representation is dom-
inated by static cartographic representations. Although this
approach provides many advantages in terms of spatial over-
lay, management of data, basic spatial analysis, and infor-
mation distribution, it does not address many issues related to
surface processes and landforms.
Topographic variation can be represented in a variety of
ways using data models. The common data models are the
field (layer), entity (object), and network data models, which
can be linked to a relational data model (Goodchild, 1992).
These data models are represented in a computer using data
structures (i.e., raster and vector). Consequently, topography
can be represented by many field models (sampled points,
contours, polygons, tessellations, triangular nets) to charac-
terize the continuous spatial variation in altitude. Object
models are used to define well-defined features, assuming that
discrete boundaries actually exist, whereas indeterminant
boundaries have been recognized to pose a unique challenge,
as environmental gradients or zones of homogeneous and
Applications
Interpretations Visualization
DEM
DEM generation
Data capture
Reality
Figure 1 The main tasks associated with digital terrain modeling.
Reproduced from Hutchinson, M.F., Gallant, J.C., 2000. Digital
elevation models and representation of terrain shape. In: Wilson, J.P.,
Gallant, J.C. (Eds.), Terrain Analysis: Principles and Applications.
John Wiley and Sons, New York, pp. 29–50.

heterogeneous surface properties can effectively represent
boundaries or limits to the spatial distribution of phenomena
(Burrough, 1996; Usery, 1996; Lagacherie et al., 1996). Earth
scientists have noted the advantages and disadvantages of such
data models and have recognized that these representations do
not effectively address process mechanics or dynamics (Raper
and Livingstone, 1995). It is also important to note that field
and object models do not formally represent the complex na-
ture of landforms, as issues of scale, organization, composition,
and age must be taken into consideration. Furthermore, the
degree to which qualitative and quantitative analysis should be
used to characterize the land surface needs to be determined, as
qualitative analysis is supported by the entity view, whereas the
science of studying process mechanics, feedback mechanisms,
geodynamics and landscape evolution tends to focus on con-
tinuous space (Raper and Livingstone, 1995).
Clearly, determining how to best represent the land surface
is a complex issue. Dikau (1989) indicated that a digital relief
model involving the parameterization of relief units could be
used to represent topography that is hierarchically organized.
Relief is scale dependent, and the concept of homogeneous
relief can be defined based on distance and direction. Never-
theless, relief and many other land-surface parameters are
scale dependent. Does this mean that the scale dependence
and the anisotropic nature of the topography need to be
represented? Should the hierarchical spatial structure of the
topography be represented? Furthermore, how should process
mechanics, process–form relationships, and temporal dy-
namics be characterized?
An intriguing proposition for geomorphometry has been
presented by Cova and Goodchild (2002) that involves the
extension of spatial representation to include fields of spatial
objects. This effectively represents the linking of continuous
space with object representation. It also allows considerable
flexibility in terms of representing the complexity associated
with landforms, as the issues of homogeneity, heterogeneity,
complexity, and other concepts can be addressed, as a tessel-
lation can have more than one object and the objects can have
discrete or fuzzy boundaries. In addition, an object hierarchy
can be developed to address issues associated with scale.
Furthermore, it also allows the representation of process via
‘process objects,’ wherein a multitude of process objects can
simultaneously alter the topography at fundamentally different
scales. This allows the integration of process modeling and
mapping in a seamless way, and raises the important issue
of parameterization schemes for characterizing the process
mechanics and specific process–form relationships. Such for-
mal representations of the topography in geomorphology are
required to validate the results obtained via empirical analysis
using geospatial technologies. In addition, such a represen-
tation can handle temporally changing spatial patterns by using
a dynamic representational scheme that results from the pro-
cess dynamics. Consequently, spatio-temporal relationships are
inherently represented. Several complexities associated with
3-D and temporal representation, however, remain.
3.7.2.2 Data Capture
The data sources and processing methods for generating DEMs
have evolved rapidly over the past 20–30 years. Data collection
Geomorphometry 165
approaches (Table 1) have been categorized by Nelson
et al. (2009) and include: (1) ground-survey techniques
(electronic theodolites, total stations, electronic distance
measuring (EDM), and global positioning system (GPS)
units); (2) existing topographic maps (derivation of contours,
streams, lakes, and spot heights from existing hardcopy
topographic maps); and (3) remote sensing (both air and
space-borne optical, radar, and Light Detection and Ranging
(LiDAR) sensors).
Modern-day data collection is based on remote sensing for
the production of DEMs, as rapid progress has occurred over
the past two decades. Global DEM data products from the
Shuttle Radar Topographic Mission (SRTM) and the Advanced
Spaceborne Thermal Emission and Reflectance Radiometer
Global Digital Elevation Model (ASTER GDEM) were released
in 2000 and 2009, respectively. The 3 arc-second SRTM DEM
covers a large fraction of the globe (from 601 N to 581 S) and
has already emerged as one of the most consistent, complete,
and popular environmental datasets in the world (Zandber-
gen, 2008; Nelson et al., 2009). The 3 arc-second (B90 m)
grid spacing is much better than the 1 km spacing of the
worldwide GTOPO30 DEM, and an accuracy assessment using
kinematic GPS data showed good absolute height accuracy,
with 90% of the errors o5 m (Rodriguez et al., 2006). This
product must nevertheless be used carefully because: (1) it
represents a digital surface model (DSM) (not a bare-earth
model); (2) surface characteristics may affect accuracy; (3)
voids generally occur at land–water margins; (4) problems
may occur in desert and mountain areas due to foreshadowing
and shadowing effects (Rodriguez et al., 2005); and (5) the
current 90-m resolution provided by the global SRTM DEM is
not fine enough for accurate surface characterization and the
mapping of soils, vegetation, and many landforms (Gessler
et al., 2009). It represents, however, an excellent dataset for
geomorphometric analysis of mountain environments to
study the influence of glaciations and tectonics on landscape
evolution (Figure 2).
Some, but probably not all of the aforementioned prob-
lems, may be addressed by the ASTER GDEM. This relatively
new product offers better resolution (1 vs. 3 arc-seconds),
improved spatial coverage (831 N to 831 S), as well as com-
parable vertical and horizontal accuracy (Hiranoa et al., 2003;
Nelson et al., 2009; Slater et al., 2009). The release of this
product is so recent that there are few published reports
documenting the strengths and weaknesses of this new data
source for specific locations and applications, although the
improvement in the measurement scale should greatly facili-
tate geomorphological investigations (Figure 3).
Recently, LiDAR surveys have been conducted, resulting in
the generation of DEMs with improved resolution (Figure 4).
Consequently, the reliance on LiDAR surveys has increased
quickly, and this source now dominates local and regional
projects everywhere. Belgium and the Netherlands, for ex-
ample, have produced national LiDAR DSMs at resolutions of
2–5 m, and much finer-resolution DEMs have been produced
for many smaller areas as well (Nelson et al., 2009). The ad-
vantages of using LiDAR include the high density of sampling,
high vertical accuracy, and the opportunity to derive a set
of surface models, given that some laser-scanning systems
can already provide at least two versions of the surface: the
166 Geomorphometry

Table 1 Key characteristics of data sources

Source Resolution (m) Accuracy Footprint (km2) Postprocessing Elevation/

requirements surface

Ground survey Variable but usually Very high vertical and Variable, but usually small Low Elevation
o5 m horizontal
GPS Variable but usually Medium vertical and Variable, but usually small Low Elevation
o5 m horizontal
Table digitizing Depends on map scale Medium vertical and Depends on map footprint Medium Elevation
and contour interval horizontal
On-screen digitizing Depends on map scale Medium vertical and Depends on map footprint Medium Elevation
and contour interval horizontal
Scanned topo-map Depends on map scale Medium vertical and Depends on map footprint High Elevation
and contour interval horizontal
Ortho photography o1 Very high vertical and – High Surface
horizontal
LiDAR 1–3 0.15–0.11 m vertical, 1 m 30–50 h  1 High Surface
horizontal
InSAR/IfSAR 2.5–5 1–2 m vertical, 2.5–10 m Depends on method of High Surface
horizontal acquisition

SRTM, Band C 90 (30) 16 m vertical, 20 m Almost global 601 N–581 S Potentially high Surface
horizontal
SRTM, Band X 30 16 m vertical, 6 m Similar to Band C, but only Potentially high Surface
horizontal every second path is
available
ASTER 30 7–50 m vertical, 7–50 m 3600 Medium Surface
horizontal
SPOT 30 10 m vertical, 15 m 72 000 per swath Medium Surface
horizontal

Source: Modified from Nelson, A., Reuter, H.I., Gessler, P., 2009. DEM production methods and sources. In: Hengl, T., Reuter, H.I. (Eds.), Geomorphometry: Concepts, Software, and
Applications. Elsevier, Amsterdam, pp. 65–85.

Figure 2 Shuttle Radar Topographic Mission (SRTM) 3 arc-second

DEM for the Shimshal Valley in northern Pakistan. The 90 m
resolution allows a relatively accurate geomorphometric
characterization of the region for studying surface processes and
tectonics.

Figure 3 Advanced Spaceborne Thermal Emission and Reflectance

vegetation canopy (first returns) and ground surface (last
returns), which should help with the modeling of erosion
and water budgets. The smaller footprint and measurement
challenges encountered in areas with tall buildings, dense-
vegetation canopies, and water surfaces make this a relatively
Radiometer Global Digital Elevation Model (ASTER GDEM) for the Mt.
Everest region in Nepal. The 30 m resolution allows a more detailed
assessment of the mountains, although several errors and artifacts
are found in such high-mountain environments. ASTER GDEMs in a
less complex topography are usually of higher quality. The displayed
x and y dimensions are 216 km, respectively.

Figure 4 Five-meter resolution digital elevation model (DEM)
generated from bare-earth LiDAR data from the North Carolina
Floodplain Mapping Program. The area is located in the Blue-Ridge
Province of western North Carolina, and is approximately 232 km2 in
size. The region exhibits an elevation range of 867 m (1454–587 m)
and has a mean slope of 171. The incised terrain in the southeast
portion of the DEM represents the beginning of the Blue-Ridge
Escarpment. DEM courtesy of Jeffrey Colby, Appalachian State
University.
expensive option, and several studies have documented how
the elevation accuracy of LiDAR data varies with both the
sensor system and the land-cover characteristics (e.g., Dow-
man, 2004; Hodgson et al., 2005).
3.7.2.3 Data Preprocessing and DEM Construction
Regardless of the representational scheme, preprocessing of
the elevation data for analysis can be difficult, as altitude is but
one of the land-surface characteristics that are required for
various applications. The morphological accuracy can only be
assessed by measuring surface parameters and objects such as
the slope angle, slope azimuth, curvature, drainage lines, and
landforms in the field, and then comparing their magnitudes,
shapes, distributions, and locations with the values obtained
by geomorphometric analysis (e.g., Fisher, 1998; Wilson et al.,
2008). Reuter et al. (2009) suggested that the true utility of
DEMs for geomorphometric analysis can only be assessed by
determining the following: (1) accuracy of surface-roughness
representation; (2) accuracy of land-surface morphology; (3)
accurate detection of actual ridge- and stream-lines; and (4)
spatial consistency of altitude measurements. Such issues are
interrelated, and the frequency and magnitude of errors will
depend on the technologies and methods used to collect the
source data, the preprocessing algorithms that are applied, and
the complexity of the land surface itself.
Not surprisingly, the horizontal and vertical resolution of the
elevation data used to portray a terrain surface will have a sig-
nificant influence on the level of detail and the accuracy of the
Geomorphometry 167
DEM, and on the analysis results (MacMillan and Shary, 2009).
Several authors have documented the effects of grid spacing on
the value and accuracy of land-surface parameters and landform
objects (e.g., Zhang and Montgomery, 1994; Florinsky, 1998;
Jones, 1998; Wilson et al., 2000; Thompson et al., 2001; Shary
et al., 2002; Tang et al., 2002; Kienzle, 2004; Warren et al., 2004;
Zhou and Liu, 2004; Raaflaub and Collins, 2006). The increasing
interest in various forms of multiscale analysis (e.g., Bishop et al.,
2003; Gallant and Dowling, 2003; Sulebak and Hjelle, 2003;
Deng and Wilson, 2008; Bishop et al., 2012) and the need to be
able to move seamlessly across scales indicate that more research
on scale and its effects is required.
Other decisions made about unwanted depressions (i.e.,
spurious pits or sinks) will also have an impact on subsequent
geomorphometric analysis and interpretation of the results. Two
approaches have been utilized and include: (1) progressively
filling the sinks by increasing their elevation values until the
elevation of their lowest outflow point is reached (e.g., Jenson
and Domingue, 1988; Martz and de Jong, 1988; Soille and
Gratin, 1994; Planchon and Darboux, 2001; Wang and Liu,
2006) and (2) creating a descending path from the bottom of the
sink by carving the terrain along this path until the nearest point
is reached, which has an elevation lower than the bottom of the
sink (e.g., Reiger, 1992; Morris and Heerdegen, 1988; Martz and
Garbrecht, 1999; Soille et al., 2003; Soille, 2004). Reuter et al.
(2009) recently used both of these approaches along with one
that combined sink filling and carving, such that the sum of the
differences in elevation between the input and the output DEMs
that did not have sinks was minimized. Lindsay and Creed
(2005a, b, 2006) have also combined elements of the afore-
mentioned approaches and used them to distinguish artifact and
real depressions in digital elevation data, and propose a min-
imum-impact approach for removing artifact depressions in
relatively flat landscapes. Grimaldi et al. (2007) have proposed
an alternative physically-based approach to remove spurious pits
as well.
At least two other related challenges may need to be ad-
dressed. The first concerns the problem of unresolved flow
directions on flat terrain, because the assignment of flow dir-
ections relies on the presence of elevation differences between
adjacent cells to drive the flow. The presence of lakes and
reservoirs, and reliance on the first of the aforementioned
approaches for filling sinks, may exacerbate this challenge by
creating artificial flat regions. Whatever the cause, one of two
approaches is typically used to remove or minimize these
kinds of problems. The first relies on an iterative procedure to
assign a single flow direction to a neighboring cell without
alteration of the elevation values (e.g., Jenson and Domingue,
1988), whereas the second method makes small alterations to
the elevation of the flat cell(s) in order to create a small arti-
ficial gradient (e.g., Garbrecht and Martz, 1997). The solutions
obtained with these approaches will vary slightly from one to
the other, and an in-depth knowledge of field conditions will
generally be required to know whether one approach pro-
duces superior results in most landscape settings.
The second challenge is the need to reconcile the DEM and
drainage lines acquired from some other dataset (Lindsay
et al., 2008). One approach relies on ‘stream burning,’ where
the local topography is altered to provide consistency with
some existing vector hydrography dataset (Saunders and
168 Geomorphometry
Maidment, 1996) and the second method utilizes the stream
network as a part of the surface-fitting approach used to
generate square-grid DEMs (Hutchinson, 1989). The latter has
been used here almost exclusively in the authors’ hydrological
work, in part because of existing software tools that tackle the
past three problems (i.e., unwanted depressions, unresolved
flow directions on flat terrain, and reconciliation of elevation
and hydrography datasets) simultaneously.
The rapid growth in sources of mass-produced, remote-
sensing-derived DEMs during the past two decades has de-
manded new forms of DEM preprocessing. For example, Reuter
et al. (2009) and Webster and Dias (2006) described various
approaches and opportunities for ortho-rectifying DEMs, re-
ducing local outliers and noise, filtering water surfaces, filtering
pure noise, filtering forests in SRTM DEMs, reducing padi terraces
(i.e., areas with closed contours where all the surrounding pixels
show the same value), filling voids and sinks, mosaicking ad-
jacent DEMs, and filtering LiDAR DEMs. Some problems are
more difficult to fix than others, and several authors have noted
the presence of systematic and random errors that are not so easy
to detect and correct in LiDAR datasets (e.g., Filin, 2003; Kat-
zenbeisser, 2003; LaLonde et al., 2010). The detection of varying
deflection errors and of time delays between measurements is
especially difficult and specific to the composition of the LiDAR
sensor system, and the large number of parameters that were
assigned when the individual sensor systems were manufactured
(Dowman, 2004). Reuter et al. (2009) also noted that two
additional trends have emerged in recent years. The first is the
integration of topographic and auxiliary information (such that
the location of lakes, streams, ridges, and breaks will be identi-
fied from satellites and incorporated into the DEM processing
chain) and the second is the increased use of fully data-driven
simulation methods that reduce some or all of the afore-
mentioned errors by calculating the average value of the land-
surface parameter from multiple equi-probable realizations of
the DEM (e.g., Burrough et al., 2000b; Hengl et al., 2004; Raa-
flaub and Collins, 2006).
3.7.2.4 Error and Artifacts
With new data sources and information technologies, DEMs
can now be produced in a variety of ways. Many subtleties are
embedded in the various data sources and methods, and
errors can be introduced at many stages of the production
process. Some can be attributed to the fact that these errors
may vary with the choice of the sensor and a specific appli-
cation (i.e., method of deployment). Consequently, DEM
differencing cannot be accurately used to detect altitude vari-
ations for assessing change, erosion, deposition, etc. (Burns
et al., 2010). A second set of challenges concerns the propa-
gation of the elevation errors in primary and secondary
land-surface parameters, and the considerable effort that is
generally required to identify them. The general approach
for propagating errors incorporates statistical modeling of the
error in the DEM (which is generally only partially known)
and running a Monte Carlo analysis (Temme et al., 2009).
Digital terrain modeling will utilize various techniques to
detect and remove some or all of these errors. Data source
errors, however, cannot always be eliminated, and those
interested in using land-surface parameters calculated from
DEMs must be cognizant of these errors and how they may
affect the analysis and interpretation of results.
It is worth reviewing what is known about the accuracy of
DEM elevation values and the land-surface parameters calcu-
lated from these elevations. Several approaches have been
proposed to assess the accuracy of DEM elevation values (e.g.,
Hutchinson, 2008; Temme et al., 2009). Many researchers
have compared DEM-derived altitudes with elevation values
taken from a more accurate source of topographic data,
computing the root-mean-square error (Wise, 2000). One
problem with this approach is that it ignores both the pres-
ence of systematic bias and the spatial pattern of errors, which
is critical for those land-surface parameters that are heavily
influenced by the shape of the land surface (Hutchinson and
Gallant, 2000; Deng et al., 2008). Carara et al. (1997) sug-
gested simple criteria to evaluate DEM quality when the DEM
is constructed from contours: (1) the DEM should have the
same values as contours close to the contour lines; (2) the
DEM values must be in the range given by the bounding
contour lines; (3) the DEM values should vary almost linearly
between the values of the bounding contour lines; (4) the
DEM patterns must reflect realistic shapes in flat areas; and (5)
the artifacts must be limited to a small proportion of the data
set. Hutchinson and Gallant (2000) have suggested a larger
and more diverse list of simple metrics for measuring quality
for DEMs constructed from surface-specific point elevation
and contour- and stream-line data that incorporate some of
the same ideas, and a rapidly growing literature is docu-
menting the quality of DEMs constructed from remotely
sensed sources (e.g., Carabajal and Harding, 2006; Hoften
et al., 2006; Rodriguez et al., 2006; Shortridge, 2006; Berry
et al., 2007; Bhang et al., 2007).
Whatever the source of the elevation data, a high-resolution
DEM may still have greater uncertainty than a low-resolution
DEM. Furthermore, the errors in DEMs may propagate to the
land-surface parameters and modeling results in ways that are
not easily predicted. See Bolstad and Stowe (1994), Band et al.
(1995), Desmet (1997), Hunter and Goodchild (1997), Wood
et al. (1997), Wise (1998), Holmes et al. (2000), Endreny and
Wood (2001), Aerts et al. (2003), Van Niel et al. (2004),
Lindsay and Creed (2005b), Fisher and Tate (2006), Lindsay
(2006), Lindsay and Evans (2006), and Chow and Hodgson
(2009), for examples spanning multiple DEM data sources and
land-surface parameters.
In one particularly impressive study, Temme et al. (2009)
examined the propagation of errors from DEMs in the
computation of the slope (a local parameter), the topo-
graphic-wetness index (a regional parameter), and the soil
redistribution resulting from water erosion (a complex model
output) in the Baranja Hill watershed in Croatia. The DEM
errors propagated strongly to slope (the mean coefficient
of variation across 100 Monte Carlo simulations was 42%
for unfilled DEMs and 49% for filled DEMs), but only mod-
erately for the wetness index (the mean coefficient of variation
was 10% for unfilled and 16% for filled DEMs), although
the coefficient of variation for the index varied more spatially
than that of slope. These results show that the wetness index
values were less sensitive than slope to the input DEM, but
this may have been influenced by the flow-routing algorithm

(Holmgren, 1994) that was used to calculate upslope con-
tributing areas.
Temme et al. (2009) next used the water-erosion module
of the LAPSUS landscape-evolution model (Schoorl et al.,
2000) to simulate erosion and deposition in the Baranja Hill
study area for 10 years. The model utilizes water flow and slope
to calculate a sediment-transport capacity, and calculates ero-
sion and deposition by comparing this transport capacity with
the predicted amount of sediment in transport. The latter was
simulated with the same multiple flow-direction algorithm
used to calculate the wetness index, and the approach
of Temme et al. (2006) was used to handle the flows of water
and sediment into sinks. The latter capability was very im-
portant because it meant that the model could simulate erosion
and deposition using both unfilled and filled DEMs. The
results, at first glance, might be viewed as reassuring because the
general erosion and deposition patterns were similar for un-
filled and filled DEMs, with erosion occurring in the upper
valleys and deposition occurring in flat areas. The mean
soil-redistribution maps of the 100 simulations on unfilled
DEMs, however, showed considerably more deposition and less
erosion than the filled DEMs (in part because the depressions
were filled before the model runs in the latter case), and the
results in both sets of model runs were very sensitive to errors in
the DEM (the mean coefficient of variation of soil redistri-
bution was 4600% for unfilled and 1000% for filled DEMs).
Hence, the coefficients of variation were larger and more spa-
tially variable for soil redistribution than they were for the
wetness index and slope, because the LAPSUS model results
were sensitive to three forms of error in the input DEM –
those associated with the error in the DEM and those that were
introduced into the slope and topographic wetness index terms
by the same source error. These kinds of dependencies are likely
to be embedded in many of the analysis and modeling appli-
cations that incorporate one or more of the aforementioned
primary and secondary terrain parameters. Consequently, error
analysis is critical in geomorphometry, and preprocessing error-
removal techniques are usually required.
3.7.3 Land-Surface Parameters
Computations of land-surface parameters attempt to charac-
terize various multiscale properties of the terrain, and are used
to extract land-surface objects. Consequently, many par-
ameters can be classified based on geometry considerations,
scale, or use in process mechanics and numerical modeling.
Wilson and Gallant (2000a) classified them as primary and
secondary, given that they constitute the basic building blocks
for landform classification and other forms of more sophisti-
cated analysis and modeling.
3.7.3.1 Primary Parameters
Many local parameters are calculated by moving a three-
by-three window across a grid and computing land-surface
parameters for the target cell (i.e., the central cell in the three-
by-three window). There are special rules on how to handle
the edges, and this approach produces a new grid or GIS layer
with the same dimensions as the DEM for each parameter. The
Geomorphometry 169
most frequently used parameters represent the first and second
derivatives of the altitude field. The two main geometric
properties are the average slope gradient, S, and the slope
azimuth, f. The average slope gradient accounts for orth-
ogonal directions such that
 
qz qz
S¼ , ½1
qx qy
where z represents the altitude, and x and y represent the
directions. Several algorithms or modeling approaches may
require the use of S, or the slope angle b, which is defined as
b ¼ arctanðSÞ ½2
Slope information can also be expressed as slope percent-
age and surface area, As, which can be approximated by
Acell
As ¼ ½3
cos b
where Acell is the area of a grid cell. Slope is routinely used in
sediment-transport modeling, landform mapping, surface-
energy-budget studies, and for characterizing various aspects
of process mechanics related to fluvial, mass movement, and
glacier-erosion dynamics (Figure 5).
The direction of the slope, or slope azimuth, is another
critical geometric property that governs water and sediment
flows, while also reflecting the orientation structure of the
topography as governed by lithology and structural influences.
Gallant and Wilson (1996) defined it as
 
q p
f ¼ 180  arctan þ 90 ½4
p 9p9
Figure 5 Slope-angle map generated from 5 m LiDAR DEM
displayed in Figure 4. Slope variations are significant in mountain
environments and can reflect variations in erosion, lithology, and
tectonics. Slope information is also critical for evaluating natural-
hazard potential and for planning purposes.
170 Geomorphometry

where

qz qz
p¼ , q¼ ½5
qx qy

Slope azimuth is a circular parameter and is frequently

transformed using cos f and sin f to examine linear trends
in northerness and easterness, respectively. Slope azimuth is also
important is solar-radiation modeling. As with slope, various
algorithms can be used for computation, and each will produce
slightly different estimates across a range of land-surface con-
ditions. Details about the performance of various equations can
be found in Skidmore (1989) and Hengl and Evans (2009).
Terrain curvature is also frequently used to estimate the
magnitude of concavity and convexity of the land surface. The
convention followed in the Earth sciences is for a positive
curvature value to represent a convex surface shape, whereas a
negative value represents a concave surface shape (Olaya, 2009).
The profile (or vertical) curvature, Cprofile, and tangential (hori-
zontal) curvature, Ctangential, are generally used to distinguish
locally convex and concave shapes. They are defined as

p2 ðr þ 2Þpqrs þ q2 t Figure 6 Profile-curvature map generated from the 5 m LiDAR DEM

Cprofile ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ½6 displayed in Figure 4.
ðp2 þ q2 Þ ð1 þ p2 þ q2 Þ3

q2 ðr  2Þpqs þ p2 t
Ctangential ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ½7
ðp2 þ q2 Þ 1 þ p2 þ q2

where r, s, and t represent

q 2z q 2z q 2z
r¼ , s¼ , t¼ ½8
q x2 q xq y q y2

Convex profile curvature indicates acceleration of flows

and a local increase in potential energy, whereas concave
profile curvature indicates a flattening of the slope and
therefore a decline in potential energy (Figure 6). Concave
tangential curvature indicates convergence and convex tan-
gential curvature indicates divergence of flow lines, which may
in turn influence overland flow paths, soil moisture distri-
bution, and soil redistribution (Figure 7). Planform curvature
is sometimes used to describe the curvature of contour
lines and should yield results similar to tangential curvature,
as long as the contour lines describe the shape of the land
surface (Gallant and Wilson, 2000). Other forms of curvature
can also be utilized, and these include mean curvature,
unsphericity curvature (Figure 8), Gaussian curvature, and
Figure 7 Tangential-curvature map generated from the 5 m LiDAR
curvature of flow lines. See Olaya (2009) for the details on DEM displayed in Figure 4.
their computation and their potential significance in the Earth
sciences.
Other local-statistical parameters can also be used to values over the spatial extent of the computational window.
characterize key aspects of the surface. Relief and surface It is important to note that relief is scale dependent, and
roughness represent two important parameters valuable examining relief variation at different scales highlights differ-
in geomorphology. Local relief is highly correlated to slope, ent aspects of the geomorphological system (i.e., tectonics at
although a nonlinear relationship has been found in extreme larger scales). Similarly, surface roughness can be estimated in
mountain environments that can be useful for differentiating a variety of ways. An interesting characterization makes use of
process–form relationships, and the nature of the relationship the vector approach such that
may be related to the magnitude of erosion and the rate of
relief production. It simply represents the range in the altitude Xi ¼ sinðbÞ cosðfÞ, Yi ¼ sinðbÞ sinðfÞ, Zi ¼ cosðbÞ ½9

Figure 8 Unsphericity-curvature map generated from the 5 m LiDAR
DEM displayed in Figure 4. The unsphericity parameter highlights
land-surface conditions that are planar in nature.
where Xi, Yi, and Zi are the components of the unit vector
normal to the land surface. The surface-roughness factor (SRF)
can then be computed as
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
P P
ð ni Xi Þ2 þ ð ni Yi Þ2 þ ð ni Zi Þ2
SRF ¼
n Last but not the least, the complex issue of scale as it relates
where n is the number of cells in a window. Other approaches
to characterizing surface roughness include semivariogram
analysis and the use of the fractal dimension to characterize
terrain complexity. Terrain roughness has utility related to
weathering studies, estimating aerodynamic drag, and other
surface applications.
The regional (i.e., nonlocal) land-surface parameters are
mainly concerned with the climatic, geomorphic, hydrologic,
or visual properties of landscapes. The first category relies
on the accurate delineation of the shadowing, sky-view, and
terrain-view nature of the surrounding topography as it in-
fluences irradiance, temperature, and precipitation. The geo-
morphic and hydrological parameters focus on the movement
of water and sediment and, as such, rely on the accurate de-
lineation of flow paths, watersheds, and scale-dependent slope
and relief. The most common parameters are sediment flux,
upslope contributing area, flow-path length, and a variety
of statistical measures. For example, positive and negative
openness (Yokoyama et al., 2002) can be extremely valuable
to geomorphologists, as positive openness highlights areas of
extreme relief and has been used to discover active erosion
zones in the Himalaya (Bishop et al., 2010). Negative open-
ness (Figure 9) highlights the scale-dependent hydrological
network, and has potential for use in mapping valley bottoms
and hydrological modeling. For the final category, the visi-
bility (i.e., from what other points can a single point be seen
Geomorphometry 171
Figure 9 Negative-openness map generated from the 5 m LiDAR
DEM displayed in Figure 4. A window search radius of 1 km was
used to highlight the scale-dependent hydrological network. This
parameter, when combined with other geomorphometric parameters,
can greatly facilitate the assessment of local and regional
hydrological conditions.
or the reverse, what other points can be seen from a single
point) can be calculated by drawing the line of sight from the
point of interest to all other points, and checking whether or
not the relief forms that occur between them block visibility.
From here, various measures of visual exposure, such as the
number of cells that can be seen from each cell, can be cal-
culated. See Fisher (1991, 1992, 1993, 1995, 1996) and Ruiz
P ffi
(1997) for examples of these types of applications and some
½10 of the pitfalls that should be avoided.
to representing and characterizing the topography must be
discussed. Unfortunately, key issues associated with conceptual
and practical treatments of scale have not been appropriately
addressed (Bishop et al., 2012). These issues are related to
measurement, cartographic, geographical, computational, and
operational scales, coupled with hierarchical organization, and
the anisotropic nature of the topography. First, the local terrain
shape, which is generally thought of as the continuous vari-
ation of the altitude field from point to point, has an enor-
mous impact on local and regional terrain parameters, but this
role is primarily influenced by cartographic and computational
scale. Florinsky (1998) suggested that local parameters, such as
slope gradient, slope azimuth, and curvatures, are mathemat-
ical variables rather than real-world characteristics. This state-
ment may be extended to all local terrain parameters for two
reasons. First, local terrain shape may rely on different math-
ematical descriptions, so that the local parameters calculated
depend on algorithm selection. Furthermore, the terrain shape
portrayed by DEMs is a function of cartographic scale, com-
bining the complexity of the terrain (geographic scale of fea-
tures), computational scale, and measurement scale at which
the terrain surface was sampled (e.g., Deng et al., 2008). Thus,
it is possible to use the same local parameter to describe terrain
shape at different scales (resolutions, distances, and dir-
ections). The special feature of nonlocal primary attributes is
that they rely on the terrain shape of a larger, non-neighbor
area (computational scale) and need to be defined with
172 Geomorphometry

reference to other nonlocal points. Therefore, calculation of
nonlocal attributes is more difficult because it incurs additional
efforts in constructing point to-point connections over the
landscape and involves more complex algorithms and scale
considerations (e.g., Desmet and Govers, 1996; Gallant and
Wilson, 2000).
Finally, the computation of geomorphometric parameters
does not usually account for a meaningful computational
scale corresponding to the geographic scale of landforms,
to the operational scale of surface processes, or to coupled
systems dynamics. This disconnect significantly affects the
magnitude and interpretation of a land-surface parameter,
such that it does not accurately characterize the topography, as
governed by theory and practical utility. Such spatial con-
straint problems of computational scale are related to surface
materials and the hierarchical organization of the topography,
where hierarchically organized landforms and features effect-
ively represent complex topographic patterns. Such scale
dependencies and organizational structure have yet to be
formally addressed, and are rarely accounted for by many
land-surface parameters. The partial exception to this is basic
semivariogram analysis, as it effectively accounts for spatial
complexity, although a spatial limit based on hierarchical
structure has yet to be rigorously evaluated. Furthermore, the
anisotropic nature of the topography at a multitude of dis-
tances must also be accounted for. This is demonstrated at the
local level using slope computed from the eight directions
within a 3  3 window. Principal component analysis of
the entire slope data set reveals fundamentally important
anisotropic information that can be used to highlight topo-
graphic structure related to surface processes, lithology, and
tectonics (Figure 10). The second principal component image
highlights the local basin structure, such that many spatial
characteristics can be used for geomorphological mapping and
system characterization. This aspect of scale is very important
in geomorphology, as it may improve the ability to assess
new aspects of geomorphological systems. Another example
at a larger computation scale is presented by Koons et al.
(2012), as they reveal that rock strength may be related to the
anisotropic nature of scale-dependent relief. Land-surface
parameters and objects that characterize operational scales
and structural constraints are urgently needed, and may aid in
establishing new theories about landscape evolution. Table 2
lists the most commonly used primary land-surface par-
ameters and their significance.
Figure 10 Second principal-component image representing
anisotropic slope information extracted from local slope data over the
Shimshal Valley in northern Pakistan. The slope data set was
generated from a Shuttle Radar Topographic Mission (SRTM) 3 arc-
second DEM. Spatial variation patterns are related to lithological units
and tectonics. Less resistant metasedimetary rocks in the
northeastern region and less precipitation generate a higher spatial-
frequency pattern, compared with a lower spatial-frequency pattern in
the southwestern region associated with more resistant lithologic
units, active uplift, and greater relief.

Table 2 Select list of primary and secondary land-surface parameters and their significance

Parameters Type Significance

Elevation Local Climate, vegetation, potential energy

Slope Local Precipitation, overland and subsurface flow velocity and runoff rate, soil-water content
Slope azimuth Local Flow direction, solar insolation, evapotranspiration, flora and fauna distribution and abundance
Profile curvature Local Flow acceleration and deceleration, soil erosion and deposition rates
Tangential curvature Local Local flow convergence and divergence
Roughness Local Terrain complexity
Elevation percentile Local Relative landscape position, flora and fauna distribution and abundance
Flow width Local Flow velocity, runoff rate, and sediment load
Upslope contributing area Regional Runoff volume, soil water content, soil redistribution
Flow-path length Regional Runoff volume, soil water content, soil redistribution
Upslope height, elevation- Regional Distribution of height values, potential energy, flow characteristics
relief ratio, hypsometric
curve, etc.
Mean slope of upslope area Regional Runoff velocity and possibly other flow characteristics
Mean slope of dispersal area Regional Rate of soil drainage
Visual exposure Regional Exposure, solar insolation, wind patterns
Topographic wetness index Regional Spatial distribution and extent of zones of saturation (i.e., variable source areas) for runoff generation as
a function of upslope contributing area, soil transmissivity, and slope
Stream-power index Regional Erosive power of flowing water (based on the assumption that discharge is proportional to the specific
catchment area)

Source: Modified from Wilson, J.P., Gallant, J.C., 2000a. Digital terrain analysis. In: Wilson, J.P., Gallant, J.C. (Eds.), Terrain Analysis: Principles and Applications. John Wiley and
Sons, New York, pp. 1–27, and Olaya, V., 2009. Basic land-surface parameters. In: Hengl, T., Reuter, H.I. (Eds.), Geomorphometry: Concepts, Software, Applications. Elsevier,
Amsterdam, pp. 141–169.

3.7.3.2 Secondary Land-Surface Parameters
In general, there are two basic sets of secondary land-surface
parameters. The first is for hydrologic characterization related to
quantifying water flow and related surface processes, and the
second is a series of climatology parameters that are related to
multiscale topographic influences on radiation, temperature, and
precipitation. Collectively, these parameters attempt to quantify
the interactions between the atmosphere and surface processes.
The underlying theory for both is well established, and the
parameterization schemes and computational methods have
evolved continuously over the past 20 years.
3.7.3.2.1 Hydrology
The movement of water is primarily driven by gravity and, to
some degree, modified by the properties of the material it
flows through or over (Gruber and Peckham, 2009). The effect
of gravity can be approximated easily with a DEM, but the
surface and subsurface properties and conditions are cum-
bersome to describe and treat. There are steadily improving
regional and national databases describing the spatial vari-
ability of selected land-surface and soil characteristics (e.g.,
Miller and White, 1998; Feng et al., 2009), but these have a
much coarser resolution than the authors’ DEMs, and seldom
include the various properties needed for specific applications.
The typical approach relies on a series of parameter estimation
equations (e.g., Rawls, 1983; Saxton et al., 1986; Abdulla and
Lettenmier, 1997; Homann et al., 1998; Waltman et al., 2003;
Saxton and Rawls, 2006), although these will undoubtedly
introduce some additional uncertainty and error into the
analytical workflow or modeling application at hand (e.g.,
Band, 1993; Wilson et al., 1996; Zhu and Mackay, 2001;
Quinn et al., 2005). Given this state of affairs, it can be as-
sumed that the DEM-based parameters will do better where
the relative importance of gravity is greatest (i.e., in headwater
areas and on steep slopes).
The initial development and use of flow-based land-surface
parameters can be traced to the introduction of the D8 algo-
rithm (O’Callaghan and Mark, 1984). This is now but one of
more than a dozen flow-routing algorithms, however, and
a distinction is usually drawn between single- and multiple-
flow direction algorithms. The single flow-routing algorithms,
which direct flow to just one downslope or neighboring cell,
include the Rho8 (Fairfield and Leymarie, 1991) and aspect-
driven kinematic routing (Lea, 1992) algorithms in addition
to the D8 algorithm. The multiple flow-routing algorithms,
which are capable of directing flow to two or more downslope
or neighboring cells, include the FD8 (Freeman, 1991), TOP-
MODEL (Quinn et al., 1991, 1995), DEMON (Costa-Cabral
and Burges, 1994), DN (Tarboton, 1997), and Mass-Flux
algorithms (Gruber and Peckham, 2009). These flow-routing
algorithms will generally generate very different results
(Figures 11 and 12).
The performance of many of these flow-routing algorithms
has been compared across a variety of landscapes (e.g., Wolock
and McCabe, 1995; Desmet and Govers, 1996; Zhou and Liu,
2002; Endreny and Wood, 2003; Chirico et al., 2005; Wilson
et al., 2000, 2007). Wilson et al. (2008) recently evaluated
the performance of several algorithms including ANSWERS
(Beasley and Huggins, 1978), D8 (O’Callaghan and
Geomorphometry 173
Contributing
area (ha)
200
20
2
0.2
0.02
N
0 300 600
(m)
Figure 11 Gridded contributing area (ha) map derived from the
Cottonwood Creek, Montana DEM, using the D8 single flow direction
algorithm, with the catchment boundary overlaid. Reproduced from
Wilson, J.P., Gallant, J.C., 2000b. Secondary topographic attributes.
In: Wilson, J.P., Gallant, J.C. (Eds.), Terrain Analysis: Principles and
Applications. John Wiley and Sons, New York, pp. 51–85.
Mark, 1984), Rho8 (Fairfield and Leymarie, 1991), FD8/
TOPMODEL (Freeman, 1991; Quinn et al., 1991, 1995), an
aspect-driven kinematic routing algorithm (Lea, 1992),
DEMON (Costa-Cabral and Burges, 1994), a flow de-
composition algorithm (Desmet and Govers, 1996), DN
(Tarboton, 1997), and MFD-md (Qin et al., 2007). They
showed that various algorithms can be expected to generate
different patterns of flow (based on upslope contributing
areas) on different parts of a hill slope or a watershed. The
gridded contributing area map reproduced in Figure 11 shows
how the D8 algorithm often generates many parallel flow
lines, which of course do not match the dendritic patterns that
characterize most surface and channel flow systems, whereas
the DEMON algorithm (Figure 12) produces more realistic
patterns, but is computationally slow when used with large
DEMs and prone to failure in flat areas. The final choice of the
flow-routing algorithm should aim to minimize the most
important of these tradeoffs for the particular study area and
application at hand.
Notwithstanding the presence of these kinds of tradeoffs, the
multiple flow-routing algorithms have grown in popularity over
time. The fundamental goal with this class of algorithms is to
find a method or a sequence of methods that move water into
one or more downslope cells, and Gruber and Peckham (2009)
174 Geomorphometry
Contributing
area (ha)
200
20
2 areas for runoff generation) as a function of upslope contributing
0.2
0.02
N
0 300
(m)
Figure 12 Gridded contributing area (ha) map derived from the
Cottonwood Creek, Montana DEM, using the DEMON stream-tube
algorithm, with the catchment boundary overlaid. Reproduced from
Wilson, J.P., Gallant, J.C., 2000b. Secondary topographic attributes.
In: Wilson, J.P., Gallant, J.C. (Eds.), Terrain Analysis: Principles and
Applications. John Wiley and Sons, New York, pp. 51–85.
have noted how this approach might be justified by actual di-
vergence (i.e., the need to treat flows across convergent and
divergent land surfaces) and/or the attempt to overcome the
limits of having only eight adjacent cells (i.e., methods to
overcome the limitations generated when a continuous flow
field is represented with a regular grid that has only eight pos-
sible directions in multiples of 451).
This pair of explanations helps to explain why the final
choice of a flow-routing method for a specific application will be
a compromise. The single flow-direction algorithms can-
not represent divergent flow, but for the same reason, have no
problem with overdispersal (i.e., the dispersal of the available
flow over too many cells or too large an area). The multiple
flow-direction algorithms, however, can represent divergent flow,
but usually also suffer from some overdispersal. In addition, the
subtleties and outcomes of the methods are concerned with the
need to: (1) treat ambiguous flow directions (as for example
occurs along ridgelines or saddles and across flat plains or valley
bottoms); and (2) reconcile the DEM-delineated flow lines and
the drainage lines acquired from some other source that are also
likely to influence the results generated with these different flow-
routing algorithms.
Therefore, the flow directions are generally computed to cal-
culate upslope contributing areas (i.e., flow accumulation areas)
and delineate the drainage networks along with the basin
boundaries. See Band (1986, 1989), Montgomery and Dietrich
(1989, 1992) and Peckham (1998) for examples of methods for
delineating drainage networks with single flow-direction algo-
rithms. The topographic wetness and stream-power indices are
among the most popular of various stream attributes, and unlike
the catchment (basin) boundaries and some other attributes, this
pair of attributes can be calculated with both the single and the
multiple flow-direction algorithms.
The typical form of the topographic wetness index (TWI)
assumes steady-state conditions and describes the spatial distri-
bution and extent of zones of saturation (i.e., variable source
area Ac, slope, and occasionally soil transmissivity (this last term
is often excluded because the transmissivity is assumed to be
constant throughout the catchment), such that
 
Ac
TWI ¼ ln ½11
tanb
The steady-state form of the TWI predicts zones of saturation
where the specific catchment area is large (which typically occurs
in converging areas of the landscape), the slope is small (which
600 typically occurs at the base of concave slopes), and soil trans-
missivity is low (which is often characteristic of areas with
shallow soils). This index has been used successfully in a variety
of hydrological applications because the aforementioned con-
ditions are commonly encountered along drainage paths and in
zones of water concentration in many landscapes (e.g., Beven
and Kirkby, 1979; Burt and Butcher, 1985; Moore and Burch,
1986; O’Loughlin, 1986; Sivapalan et al., 1987; Moore et al.,
1988; Phillips, 1990; Montgomery and Dietrich, 1994; Moore
and Wilson, 1992, 1994; Fried et al., 2000; Kheir et al., 2007).
These types of static indices, however, must be used care-
fully to predict the distribution of dynamic phenomena like
soil-water content because surface saturation is a threshold
process, the presence of hysteric effects, and the reliance
on one or more assumptions. The two most important as-
sumptions in this case are that: (1) the gradient of the
piezometric head, which governs the direction of subsurface
flow, is parallel to the land surface; and (2) there is
sufficient time between rainstorms for the subsurface flow to
achieve a steady state (Moore et al., 1993a). Several authors
have described the pitfalls of using these kinds of indices in
inappropriate ways. For example, Jones (1986, 1987) docu-
mented some of the advantages and limitations of using
wetness indices to describe the spatial patterns of soil-water
content and drainage, and Quinn et al. (1995) summarized
the various problems and described how the steady-state TWI
can be calculated and used effectively as part of the TOP-
MODEL hydrologic-modeling framework. Several variants
of the original equation have also been proposed. Barling
(1992) proposed a quasi-dynamic topographic wetness index
(QD-TWI) to overcome the limitations of the steady-state
assumption and used it to show how the topographic
hollows, and not the drainage channels themselves, deter-
mined the response of a semiarid catchment in New South
Wales, Australia (see Barling et al. (1994) for additional de-
tails). Wood et al. (1997) later proposed an alternative index
to predict the saturated-zone thickness that incorporated both

spatial and temporal variation in recharge. Both Fried et al.
(2000) and Nguyen and Wilson (2010) calculated QD-TWI
using a variety of flow-routing algorithms and showed how
the results varied depending on the flow-routing algorithm
that was utilized.
3.7.3.2.2 Climatology
The topography also governs microclimate, and surface-energy
conditions influence the magnitude of various surface pro-
cesses. Böhner and Antonić (2009) reviewed topoclimatology
parameters, as the land surface controls the spatial variability
of near-ground atmospheric processes and meso-scale climatic
variations. The regionalization approaches that use kriging,
universal kriging, and splines to map the climate variables
measured at climate stations are ignored, and the focus is on
the land-surface parameters that are used to assess the vari-
ability of the short- and long-wave radiation fluxes, as these
influence surface temperature, evapotranspiration, air move-
ment, and other surface processes.
Understanding and predicting the magnitude of solar and
surface irradiance is of primary concern. Solar irradiance varies
as a function of: (1) the changing distance from the Sun to the
Earth; (2) intrinsic variation in the output of the Sun; and (3)
changes in the radiation field from the Sun toward the Earth.
Annual changes in irradiance can be B6%, independent
of wavelength. A nonuniform distribution of active regions
also occurs on the solar disk that are modulated by a 27-day
rotation period, which results in irradiance variations that are
wavelength (l) dependent. Irradiance variations are also
caused by solar magnetic activity (22-year cycle). These vari-
ations are caused by eruptive phenomena such as flares, and
range in temporal scale from minutes, to months, to years,
and include the 11-year sunspot cycle.
The exo-atmospheric irradiance, E0, is therefore a function
of orbital parameters that determine its magnitude. Most GIS-
based solar-radiation models do not account for orbital-
parameter variation of obliquity and eccentricity; therefore,
they cannot be used effectively for paleo-climate and future-
scenario studies too far into the past or future. For example,
such models cannot be used to study radiative forcing in the
Holocene. Rather, such models make use of standard irradi-
ance spectra and account for the annual variation in the
Earth–Sun distance to portray seasonal and diurnal variations
in E0.
Atmospheric conditions then determine the amount
of atmospheric attenuation, and atmospheric-transmission
functions are wavelength dependent. Atmospheric conditions
are generally prescribed based on modeled atmospheric con-
ditions. For solar-radiation modeling and estimation of sur-
face-process rates, there is a need to accurately estimate the
surface spectral irradiance (E) which is a composite of three
downward irradiance components
EðlÞ ¼ Eb ðlÞ þ Ed ðlÞ þ Et ðlÞ ½12
The direct/beam irradiance (Eb) is typically dominant,
followed by the diffuse-skylight irradiance (Ed) and the
adjacent terrain irradiance (Et). Variations in atmospheric,
topographic, and land-cover conditions determine the se-
quential dominance of irradiance partitioning.
Geomorphometry 175
Under cloudless skies, Eb is the dominant term in eqn [12].
Consequently, considerable research has focused on modeling
the direct atmospheric-transmittance functions accurately.
The atmosphere attenuates the direct irradiance primarily by
gaseous absorption and molecular and aerosol scattering
(Chavez, 1996). These atmospheric processes are wavelength
dependent, and spatially and temporally controlled by chan-
ging atmospheric and landscape conditions. The total down-
ward atmospheric transmission (Tk) is a function of the total
optical depth of the atmosphere, which varies with solar-
zenith angle and altitude, and can be represented as
T k ðlÞ ¼ Tr ðlÞTa ðlÞTO3 ðlÞTgas ðlÞTH2 O ðlÞ ½13
where Tr is the Rayleigh transmittance, Ta is the aerosol
transmittance, TO3 is the ozone transmittance, Tgas is the
transmittance for miscellaneous well-mixed gases, and TH2 O is
the water-vapor transmittance. Atmospheric attenuation is
highly variable with wavelength, with Rayleigh and aerosol
scattering dominating at shorter wavelengths and water vapor
dominating at longer wavelengths.
The direct irradiance is also governed by multiscale topo-
graphic parameters. Local or microscale topographic variation
is represented by the incidence angle of illumination between
the Sun and the vector normal to the ground, such that
cos i ¼ cos yi cos b þ sinyi sin b cosðf  fi Þ ½14
where yi is the incident solar-zenith angle and fi is the inci-
dent solar-azimuth angle.
It is possible to estimate cos i using a DEM, and uncertainty
in the estimate is related to the resolution, as subpixel-scale
topographic variation is not accounted for. Values of cos i can
be r0.0, indicating no direct irradiance due to the orientation
of the topography. It is important to note that the incident
solar geometry varies across the landscape, although this is
usually assumed to be constant when working with image
scenes (i.e., small-angle approximation). In addition, the
meso-scale topographic relief in the direction of fi determines
whether a pixel is in shadow (Sc). This can be accounted for by
ray tracing, shadow detection, and shadow interpolation al-
gorithms that alter cos i values appropriately (Dozier et al.,
1981; Rossi et al., 1994; Giles, 2001). The local and meso-scale
topographic influences on the direct irradiance are significant
over annual and diurnal time scales (Figure 13). Con-
sequently, direct irradiance exhibits a high degree of spatio-
temporal variability. The direct irradiance component can be
estimated as
Eb ðlÞ ¼ E0 ðlÞT k ðlÞcos iSc ½15
Atmospheric scattering also generates a hemispherical
source of irradiance that should be calculated as an integration
of the total sky irradiance. This source can be simplistically
represented as a composite of a Rayleigh-scattered component
(Er), an aerosol-scattered component (Ea), and the ground-
backscattered component (Eg) that represent inter-reflections
between the land surface and the atmosphere, where
Ed ðlÞ ¼ Er ðlÞ þ Ea ðlÞ þ Eg ðlÞ ½16
176 Geomorphometry

(a) (b)

(c) (d)

Figure 13 Simulated topographic influences on the direct irradiance (cos iSc) over the Mt. Everest region in Nepal. Darker tones represent less
direct irradiance whereas light tones represent greater direct irradiance. Simulations based on an Advanced Spaceborne Thermal Emission and
Reflectance Radiometer Global Digital Elevation Model of the region. Simulations account for local topographic conditions including slope and slope
azimuth, as well as meso-scale topographic shielding that casts shadows. The solar azimuth was held constant at 135.01: (a) Simulation with a
solar-zenith angle of 0.01; (b) Simulation with a solar-zenith angle of 45.01; (c) Simulation with a solar-zenith angle of 70.01; and (d) Simulation with
a solar-zenith angle of 85.01. Assuming relatively constant atmospheric conditions, direct irradiance in this region exhibits a high degree of
spatiotemporal variability.

Its accurate estimation is complicated by the fact that an and direction. In general, the solid angle will increase with
anisotropic parameterization scheme is required. In general, altitude. It is frequently referred to as the sky-view factor (Vf )
the irradiance decreases with angular distance from the Sun. in the remote-sensing and energy-balance literature, and can
In addition, this irradiance component is also influenced be estimated using a DEM (Figure 14) such that
by meso-scale hemispherical shielding of the topography. X
360
Df
Consequently, only a solid angle of the sky will contribute Vf ¼ cos2 ymax ðf,dÞ ½17
to Ed, and this angle will change as a function of pixel location f¼0
360

Figure 14 Sky-view factor over northern Pakistan. This parameter
was computed from a Shuttle Radar Topographic Mission (SRTM)
3 arc-second DEM. Darker tones represent those areas that do not
receive as much diffuse-skylight irradiance, whereas lighter tone
areas receive more. The spatial patterns reveal that meso-scale relief
conditions are highly variable in the Himalaya. Examination of the
magnitude and spatial patterns reveals that some glaciers are
receiving more diffuse-skylight irradiance than others, whereas
different glaciers exhibit different trends in their sky-view-factor
altitude profile. Such topographic variation regulates ablation rates
and governs the sensitivity of glacier response to climate forcing.
This is turn regulates melt-water production and glacial and fluvial
erosion.
where ymax is the maximum local horizon angle at a given
azimuth, f, over a radial distance of d.
In mountain environments exhibiting extreme relief and
deep valleys, topographic shielding of the skylight-diffuse ir-
radiance can be significant (Proy et al., 1989). Furthermore,
difficulties in accurately predicting the bidirectional reflect-
ance distribution function (BRDF) of land-cover character-
istics surrounding individual pixels generate uncertainty in the
estimation of the ground-backscattered component.
The irradiance components Eb and Ed interact with the
terrain and land-cover biophysical characteristics to generate
an adjacent-terrain irradiance component. This irradiance
component is not generally considered in remote-sensing,
GIS-based radiation modeling, and energy-balance studies be-
cause it is assumed that its magnitude is relatively minor, and it
is a difficult parameter to estimate accurately. A first-order ap-
proximation has been formulated by Proy et al. (1989) and
assumes that surface radiance is Lambertian. It is then possible
to estimate the radiance received at any pixel, by accounting for
the geometry between two pixels (p1 and p2) such that
  
Ap
L12 ¼ cosy1 L2 cosy2 2 ½18
d regional relief structure, deformation orientation patterns, and
where L12 represents the radiance received at p1 from the lu-
minance of p2 (L2), y1 and y2 are the angles between the
normal to the terrain and the direct line of sight from p1–p2,
Ap is the pixel area (p2), and d represents the distance between
p1 and p2.
This equation can be used to estimate Et for any pixel by
integrating over all of the pixels whose slopes are oriented
toward a pixel of interest and where the line of sight is not
blocked by the topography. High-altitude and extreme local
Geomorphometry 177
relief areas can exhibit a strong adjacent-terrain irradiance
component due to highly reflective features such as snow and
vegetation.
The magnitude of the reflected and emitted radiance at the
surface is determined by the conservation of energy, such that
rðlÞ þ aa ðlÞ þ TðlÞ ¼ 1:0 ½19
where, r, aa, and T represent the reflectance, absorption, and
transmission, respectively. For opaque objects, T¼ 0.0. It is
commonly assumed that reflectance is isotropic (surface re-
flects radiation equally in all directions) and the surface
spectral radiance (L) can be computed as follows
 
EðlÞ
LðlÞ ¼ rðlÞ ½20
p
The Lambert assumption, however, is not an accurate
characterization of the anisotropic nature of surface reflection.
The BRDF describes such reflectance variations and can be
used to estimate the surface albedo, a key parameter in surface
energy-budget modeling needed to compute the net short-
wave radiant flux. Several parameterization schemes exist for
the long-wave irradiance and net long-wave radiant flux, and
other surface energy-budget components are also dependent
on topographic parameters including surface roughness.
Several GIS-based solar-radiation models and surface en-
ergy-budget models can be used to produce maps of various
surface irradiance and energy parameters. Such models can be
used over user-specified periods ranging from 1 day to a year
in length. SRAD (Moore et al., 1993b; Wilson and Gallant,
2000b) is but one of a number of models that have been
proposed for calculating the radiation fluxes and it incorpor-
ates the effects of cloudiness into the calculations. Others in-
clude r.sun (Hofierka, 1997), Solar Analyst (Fu and Rich,
2000), Solar Flux (Dubyah and Rich, 1995; Hetrick et al.,
1993a, b), and Solei (Mészároš, 1998; Miklánek, 1993). All
the aforementioned models document how spatial variability
in elevation, slope, slope azimuth, cast shadows, sky-view
factor, and other variables can create very strong local gradi-
ents in solar radiation and surface temperature, and thereby
exert a large influence on surface processes including photo-
synthesis, evapotranspiration, ablation, weathering, as well as
influencing vegetation diversity and biomass production at
specific locations on the land surface.
Analysis of the topography is also required for the char-
acterization of precipitation, air flow, and wind speed, as
rainfall is governed by local slopes and meso-scale altitude
variations, and air-flow direction and speed is governed by
local and meso-scale surface roughness. For more details, see
Böhner and Antonić (2009).
Finally, it is important to note that it is difficult to verify
some parameter estimates because it is not easy to accurately
measure the spatio-temporal variability of key variables (e.g.,
albedo, BRDF, Et). Furthermore, climate stations are not gen-
erally established in complex terrain, but occur in low-altitude
areas in flat terrain, and may not collect data representative
of higher-altitude areas or more complex topography. One
possible way around this problem is to use satellite data for
178 Geomorphometry
estimating incoming solar radiation and precipitation. Con-
sequently, there is a need to develop and validate new para-
meterization schemes that address process mechanics and the
space–time issues connected to data, analysis, and modeling.
Table 2 also lists the most frequently used secondary land-
surface parameters and their significance.
3.7.4 Land-Surface Objects and Landforms
The use of land-surface parameters to segment the landscape
into terrain features or landform classes (i.e., objects) can be
traced to the pioneering work of Speight (1968) and Dikau
(1989). Recent developments have included the use of
automated fuzzy-classification algorithms to detect landform
elements (e.g., Burrough et al., 2000a; Schmidt and Hewitt,
2004). The focus of these kinds of applications may range
from the identification of specific landforms (e.g., mountains,
valleys, glaciers, alluvial fans) to landform elements (i.e.,
geometric shapes that constitute part or all of a specific
landform) and repeating landforms types (e.g., a series of
rolling hills and valleys). Here, the focus is on the extraction
and classification of landform elements, since these constitute
the basic building blocks for segmenting the landscape into
landform classes.
The early landform classification approaches relied on
various representations of landscape position and the shape of
the land surface itself. For example, Dikau (1989) divided the
landscape into combinations of concave, straight, and convex
planform curvatures, on the one hand, and concave, straight,
and convex profile curvatures on the other. This approach (like
many subsequent ones) relies on the inferred relationship
between surface shape (i.e., local curvature) and the accumu-
lation of surface flow and consequently that of surface de-
position through two accumulation mechanisms. The first
mechanism reflects the divergence and convergence of flow
across a hill slope, whereas the second reflects the relative
deceleration of flow in the downslope direction, as influenced
by changes in profile curvature (Moore et al., 1991; MacMillan
and Shary, 2009).
Shary and his colleagues have criticized Dikaus’ (1989)
original approach on two levels, and proposed a more robust
and predictable classification based on curvatures. Their first
suggestion was to use tangential curvature in place of plan-
form curvature in the classification of basic form elements
(Figure 15), because both tangential and profile curvatures are
curvatures of normal sections and both exhibit similar stat-
istical distributions, unlike planform curvature (MacMillan
and Shary, 2009). The second criticism concerned the con-
tradiction in Dikau’s (1989) original premise that this ap-
proach differentiated form elements with a homogeneous
plan and profile curvature, because these facets invariably
contain even more homogeneous form facets with similar
gradients, aspects, and curvatures. Shary (1995) and Shary
et al. (2005), in turn, have proposed an objective, local, scale-
specific classification of elemental landform features based
only on the consideration of the signs of the tangential, pro-
file, mean, difference, and total Guassian curvatures, as a way
of avoiding both of these sets of problems (Figure 16).
Profile curvature
Convex Profile-straight Concave
Convex
X/X SF/X V/X
Tangential curvature
Tangential-
straight
X/SL SF/SL V/SL
Concave
X/V SF/V V/V
Figure 15 The modified Dikau (1989) classification of form
elements based on the profile and tangential (i.e., across slope)
curvatures. The elements have been further classified as positive or
negative based on the radius of curvatures (4600 or o600 m) and
the planform curvature in the original classification was replaced by
tangential curvature based on Shary and Stepanov (1991).
Reproduced from MacMillan, R.A., Shary, P.A., 2009. Landforms and
landform elements in geomorphometry. In: Hengl, T., Reuter, H.I.
(Eds.), Geomorphometry: Concepts, Software, Applications. Elsevier,
Amsterdam, pp. 227–254.
None of the aforementioned approaches, however, in-
corporates the contextual position of specific facets that make
up the landscape. Most of the automated classifications that
incorporate context build on the conceptual classifications of
hillslopes like those of Ruhe (1960), which divided hill slopes
into five units (summits, shoulders, back slopes, foot slopes,
and toe slopes). Many other such classifications have been
proposed, and Conacher and Dalrymple (1977) and Speight
(1990) divided hill slopes into 9 and 10 units, respectively.
This was accomplished by delineating finer-resolution hill
slope facets or by extending the hill slope to include the
channel or various parts of the channel at the bottom of the
hill slope. Taken as a whole, these conceptual classes consider
slope gradient and relative-slope position along a topo-
sequence from divide to channel. In addition to curvatures
and automated classification, various approaches have in-
cluded absolute and relative horizontal and vertical distance
to ridge lines or channels (e.g., Skidmore et al., 1991) and
position in the landscape relative to the order of the nearest
stream channel below a hillslope (e.g., Schmidt et al., 1998;
Schmidt and Dikau, 1999).
The adoption and use of fuzzy-classification algorithms to
detect landform elements marked an important step forward.
The importance of this innovation can be traced to the fact
that each of the aforementioned approaches will work in
some instances and not others, and they will seldom produce
satisfactory answers to questions linked to the locations of
a specific mountain or valley (Fisher et al., 2004). Indeed,
there are many phenomena that are difficult to locate or de-
lineate because their meaning is not well defined or because of
the subjectivity, vagueness, and ambiguity that have often
characterized the ways in which the world is described

Difference
curvature
Total gaussian
curvature
1 3
7 8
9
Tangential
curvature
Figure 16 Shary’s complete system of classification of landform elements based on signs of tangential, profile, mean, difference, and total
Guassian curvatures. Reproduced from Shary, P.A., Sharaya, L.S., Mitusov, A.V., 2005. The problem of scale-specific and scale-free approaches
to geomorphometry. Geografia Fisica e Dinamica Quaternaria 28, 81–101.
(Burrough, 1996; Wilson and Burrough, 1999). Fuzzy-set
theory represents an alternative approach to classic set theory
(Burrough and McDonnell, 1998; Robinson, 2003) and has
been used in many environmental domains to solve these
kinds of problems. Hence, one or more forms of fuzzy clas-
sification have been used to describe soil variability (e.g.,
McBratney and Odeh, 1997; Zhu, 1997a, b, 1999; Ahn et al.,
1999), land cover (e.g., Fisher and Pathirana, 1994; Foody,
1996, Brown, 1998; DeBruin, 2000), site selection and mul-
ticriteria evaluation (Charnpratheep et al., 1997; Jiang and
Eastman, 2000), and the parameterization of land-surface
models (e.g., Mackay et al., 2003).
The above mentioned examples indicate how many
authors have used fuzzy sets and fuzzy-logic operators with
land-surface parameters to generate partial and multiple
memberships of spatial objects of various kinds during the past
15 years. The various membership functions that have been
described in considerable detail by Robinson (2003) represent
the core of the method because they allow the expression
of irreducible observation and measurement uncertainties in
their various manifestations, and make these uncertainties in-
trinsic to the classification (i.e., using grades of membership).
In this way, the fuzzy-logic approach will associate a fuzzy
likelihood of each output class with each value or class on each
input map (Figure 17). This means that when fuzzy data are
processed, their intrinsic uncertainties are processed as well,
and their results are more meaningful than their counterparts
obtained by processing the usual crisp data (Klir and Yuan,
1995; Robinson, 2003). The advantages of using such an
approach are evident in the landform elements specified by
MacMillan et al. (2000) on the basis of a combination of
measures of local-surface shape (convexity/concavity) plus
the relative-slope position for a 64 ha site in Alberta, Canada.
Geomorphometry 179
Total gaussian
curvature
4
5 6
11 Mean
12
curvature
10
Profile
curvature
Two basic fuzzy-set approaches exist. The first relies on
expert knowledge and is often referred to as the Semantic-
Import model, whereas the second approach relies on the
identification and characterization of feature-space clusters
(e.g., Bezdek et al., 1984). The fuzzy sets generated in both of
these instances will often provide tremendous expressive
power (Robinson, 2003). For example, Zhu (1997a, b, 1999)
work clearly demonstrates the utility of the fuzzy paradigm for
capturing and representing spatially explicit soils knowledge
from human experts. Zhu and his colleagues used knowledge
of observable environmental inputs including several land-
surface parameters, and targeted outputs (soil-map units), and
adopted a limiting-factor approach for computing the overall
similarity. The value for the overall similarity between an
unclassified entity and a reference entity was obtained by
simply selecting the smallest similarity value from among all
similarity values computed for all attributes for an unclassified
entity in this application.
Both the expert- and data-driven fuzzy-classification ap-
proaches will result in multiple maps of membership, where
each class will be represented on a separate map. The advan-
tages of using memberships are threefold since it is possible
to: (1) determine which classes are connected with which
land-surface parameters; (2) assess the confusion between
the classes; and thereby (3) detect the areas where the con-
fusion between two or more classes is high (Burrough and
McDonnell, 1998; Hengl et al., 2004; Shi et al., 2005; Evans
et al., 2009).
There are many subtleties, however, connected to these
fuzzy-classification methods. The knowledge in the afore-
mentioned approach could have been acquired via several
methods, and each would have varied in the degree to which it
is theoretically, empirically, or statistically valid. See Qi and
180 Geomorphometry
1.0
Membership
A
0.5
not A
0.0
Threshold value
(a) Land-surface parameter
Figure 17 Schematic showing the derivation of fuzzy memberships using (a) the definition of thresholds and (b) the definition of class centers.
Reproduced from Hengl, T., Reuter, H.I. (Eds.), 2009. Geomorphometry: Concepts, Software, and Applications. Developments in Soil Science.
Elsevier, Amsterdam.
Zhu (2003), Qi (2004), and Qi et al. (2006) for an extended
discussion of these issues. Similarly, the results can be ex-
pected to vary with the method used to compute the overall
similarity scores. MacMillan et al. (2000), in their imple-
mentation of the Semantic-Import model, relied on a weigh-
ted-average method to calculate the overall similarity of an
unclassified site to a reference entity, based on the assumption
that all input variables should be included in calculating the
similarity of a site to a reference entity. However, a case can
also be made, that some inputs may deserve to be afforded
a greater importance or weight than others (Hengl and
MacMillan, 2009). Indeed, the sensitivity of the fuzzy c-means
approach to the choice of input variables and weights assigned
to them when calculating the overall similarity values has
been explored extensively (e.g., Deng et al., 2006; Deng and
Wilson, 2006).
There is also the need to choose an appropriate scale for
both the Semantic-Import model and the fuzzy c-means fuzzy-
classification approaches. The fact that there is no single true
or fixed value for local land-surface parameters such as slope
or curvature at a point, but rather a whole range of values that
are dependent on the horizontal and vertical resolution has
already been noted. Not unexpectedly, no best resolution
can be singled out at which to compute local land-surface
parameters to portray and classify terrain (Hengl, 2006; Smith
et al., 2006; Deng et al., 2007), and the final scale that is
chosen should be appropriate for capturing and describing the
surface features of interest for a particular application (Deng
et al., 2008). The size or the extent of the study area needs to
be added to this list of sensitive variables since some land-
surface parameters will vary in systematic ways across the
landscape, and may generate locally specific results when the
fuzzy classification is implemented for a limited area (Evans
et al., 2009).
The aforementioned discussion gives some sense of the
great progress that has been made with automated landform
classification during the past quarter century. The successful
deployment of these techniques, however, requires consider-
able knowledge and experience with the techniques them-
selves, and of the study area to which they are to be applied.
Furthermore, the generation of land-surface objects requires
better formalization that links process and form. Addressing
Land-surface parameter #2
A
0.8
6
0.
0.6
0.8 8
0. C
6
0.
Class centre
B
Land-surface parameter #1
(b)
various issues associated with the formal characterization of
topographic structure and the use of object-oriented technol-
ogy also has the potential to lead to significant progress. Un-
fortunately, new representational schemes and the use of
Earth-science concepts in analysis and modeling have not kept
pace with empirical exploration (Bishop et al., 2012). Both are
required for formalizing the generation of land-surface objects
that will facilitate diagnostic geomorphological mapping
efforts.
3.7.5 Conclusions
There have been tremendous advances in DEM data sources,
digital terrain modeling techniques, new algorithms for land-
surface parameterization, and new geomorphological appli-
cations driven by geomorphometry. Nevertheless, new theo-
retical/conceptual and information technology advances must
also occur that formalize the understanding of geomorpho-
logical systems and topographic complexity. There are at least
four research paths that can be expected to yield substantial
benefits.
1. Knowledge of the presence of, and propagation of, errors in
both the current and the new remote-sensing data sources
that emerge needs to be improved. This is a challenging
task because many of the systematic and random errors in
the current data streams are specific to the sensor used and
the specific protocols and methods that have been used in
individual projects (Dowman, 2004). This suggests that
ways to clarify and publish information related to data
quality need to be established, since much of this is pro-
prietary information of firms that have built and deployed
the aforementioned technologies.
2. Field observations and the development and testing of new
analytical methods are required. Taking the modeling of
flow directions and upslope contributing areas as ex-
amples, there is an urgent need to learn more about the
ways in which the land surface and the interactions with
the underlying soil and regolith influence rainfall–runoff
relationships and the growth and contraction of flow
networks in specific environments. Lindsay and colleagues

at the University of Guelph in Ontario, Canada, for
example, are exploring the spatial pattern and timing of
ephemeral flows in headwater channels that may provide
some important new insights. These kinds of projects are
time-consuming, but are vital for the development of DEM
datasets and analytical methods that support the represen-
tation of the key hydrologic and geomorphic processes
(i.e., those influencing nonpoint source pollution) oper-
ating in specific landscapes (e.g., Mitášová et al., 1995).
3. The critical evaluation and adoption of key algorithms and
best approaches for solving specific problems is required. This
is similar to the second path, although it is likely to yield
faster returns. The goal here would be to combine and inte-
grate ‘best’ practices, as exemplified by the following example.
The QD-TWI model proposed by Barling (1992) would ap-
pear to have considerable merit, given what is known about
the distribution of soil moisture along with both the surface
and subsurface flow patterns in a variety of semiarid and arid
landscapes. The original QD-TWI model, however, in-
corporated the D8 flow-routing algorithm, and several studies
have demonstrated that D8 generates many undesirable ar-
tifacts. The DEMON flow-routing algorithm, however, offers
many advantages but sometimes fails in areas with flat ter-
rain, and shows slow performance when applied to relatively
fine-resolution DEMs covering large areas. Nguyen (2011) has
exploited this opportunity by building a faster and more ro-
bust version of DEMON along with a flexible version of the
QD-TWI model that allows the user to choose from a variety
of flow-routing algorithms. Consequently, a terrain analyst
can work with both approaches simultaneously. There are
many opportunities like this that can be exploited in the
immediate future.
4. The final research path concerns issues of scale. The rapid
advent and adoption of fine-resolution remote-sensing
data sources, and the need to characterize coupled geo-
morphological systems, means that there is an urgent need
to address multiple issues of scale that affect the ability to:
(1) collect, represent, and integrate data and information
across multiple scales; (2) characterize land-surface par-
ameters; (3) generate meaningful land-surface objects that
are based on scientific principles and concepts; and (4)
develop and refine techniques that allow multiscale char-
acterization and visualization to address a variety of
problems. These topics have been recognized by a variety of
researchers as key agendas (Gallant et al., 2000; Gallant
and Dowling, 2003; Sulebak and Hjelle, 2003; Fisher et al.,
2005; Deng, 2007; Deng and Wilson, 2008; Bishop et al.,
2012).
Finally, various equations have been included for those
interested in calculating one or more of the aforementioned
land-surface parameters, as these can be used for segmenting
the topography into land-surface objects that are useful for
geomorphological mapping. Furthermore, many of the sec-
ondary parameters can be used in geomorphological research
related to assessment of hydrological, glaciological, and geo-
logical conditions. The current state of the art suggests that the
present-day terrain analyst will need to choose wisely among
the extreme multitude of options (i.e., data, algorithms, an-
alysis approaches, and models), while paying special attention
Geomorphometry 181
to their own project goals, the advantages and disadvantages
of different data sources and digital terrain modeling techni-
ques, the characteristics of their study area(s), and how errors
might have been introduced and propagated, and the likely
significance of these errors, given the results that are produced.
References
Abdulla, F.A., Lettenmier, D.P., 1997. Development of regional parameter estimation
equations for a macroscale hydrologic model. Journal of Hydrology 197,
230–257.
Aerts, C.J.H., Heuvelink, G.B.M., Goodchild, M.F., 2003. Accounting for spatial
uncertainty in optimization with spatial decision support systems. Transactions in
GIS 7, 211–230.
Ahn, C.W., Baumgartner, M.F., Biehl, I.I., 1999. Delineation of soil variability using
geostatistics and fuzzy clustering analysis of hypospectral data. Soil Science
Society of America Journal 63, 142–150.
Antonić, O., Pernar, N., Jelaska, S.D., 2003. Spatial distribution of main forest soil
groups in Croatia as a function of basic pedogenetic factors. Ecological
Modeling 170, 363–371.
Band, L.E., 1986. Topographic partition of watersheds with digital elevation models.
Water Resources Research 22, 15–24.
Band, L.E., 1989. A terrain based, watershed information system. Hydrological
Processes 3, 151–162.
Band, L.E., 1993. Effect of land surface representation on forest water and carbon
budgets. Journal of Hydrology 150, 749–772.
Band, L.E., Vertessey, R., Lammers, R.B., 1995. The effect of different terrain
representation schemes and resolution on simulated watershed processes.
Zeitschrift fur Geomorphologie, Suppl-Bd 101, 187–199.
Barling, R.D., 1992. Saturation zones and ephemeral gullies on arable land in
Southeastern Australia. University of Melbourne, Unpublished PhD dissertation.
Barling, R.D., Moore, I.D., Grayson, R.B., 1994. A quasi-dynamic wetness index for
characterizing the spatial distribution of zones of surface saturation and soil
water content. Water Resources Research 30, 1029–1044.
Beasley, D.G., Huggins, L.F., 1978. ANSWERS: a model for watershed planning.
Proceedings of the Tenth Conference on Water Simulation Volume 2, Miami
Beach, FL. 503–515
Berry, P.A.M., Garlick, J.D., Smith, R.G., 2007. Near-global validation of the SRTM
DEM using satellite radar altimetry. Remote Sensing of Environment 106,
17–27.
Beven, K.J., Kirkby, M.J., 1979. A physically-based, variable contributing area
model of basin hydrology. Hydrological Sciences Bulletin 24, 43–69.
Bezdek, J.C., Ehrlick, R., Full, W., 1984. FCM: the fuzzy c-means clustering
algorithm. Computers and Geosciences 10, 191–203.
Bhang, K.J., Schwartz, F.W., Braun, A., 2007. Verification of the vertical error in C-
Band SRTM DEM using ICESat and Landsat-7, Otter Tail County, MN. IEEE
Transactions on Geoscience and Remote Sensing 45, 36–44.
Bishop, M.P., Bonk, R., Kamp, Jr. U., Shroder, Jr. J.F., 2001. Terrain analysis and
data modeling for alpine glacier mapping. Polar Geography 24, 257–276.
Bishop, M.P., Bush, A.B.G., Copland, L., Kamp, U., Owen, L.A., Seong, Y.B.,
Shroder, J.F., 2010. Climate change and mountain topographic evolution in the
Central Karakoram, Pakistan. Annals of the Association of American Geographers
100(4), 772–793.
Bishop, M.P., James, L.A., Shroder, J.F., Walsh, S.J., 2012. Geospatial technologies
and digital geomorphological mapping: Concepts, issues and research.
Geomorphology 137, 5–26.
Bishop, M.P., Shroder, Jr. J.F., 2000. Remote sensing and geomorphometric
assessment of topographic complexity and erosion dynamics in the Nanga
Parbat massif. In: Khan, M.A., Treloar, P.J., Searle, M.P., Jan, M.Q. (Eds.),
Tectonics of the Nanga Parbat Syntaxis and the Western Himalaya. Geological
Society, London, No. 170 in Special Publication, pp. 181–200.
Bishop, M.P., Shroder, Jr. J.F. (Eds.), 2004a. Geographic Information Science and
Mountain Geomorphology. Springer-Praxis, Chichester.
Bishop, M.P., Shroder, Jr. J.F., 2004b. GIScience and mountain geomorphology:
Overview, feedbacks, and research directions. In: Bishop, M.P., Shroder, Jr. J.F.
(Eds.), Geographic Information Science and Mountain Geomorphology. Springer-
Praxis, Chichester, pp. 1–31.
Bishop, M.P., Shroder, Jr. J.F., Colby, J.D., 2003. Remote sensing and
geomorphometry for studying relief production in high mountains.
Geomorphology 55, 345–361.
182 Geomorphometry
Bishop, T.F.A., Minasny, B., 2005. Environmental soil-terrain modeling: the
predictive potential and uncertainty. In: Grunwald, S. (Ed.), Environmental Soil-
Landscape Modeling: Geographic Information Technologies and Pedometrics.
CRC Press, Boca Raton, FL, pp. 185–213.
Böhner, J., Antonić, O., 2009. Land-surface parameters specific to topo-climatology.
In: Hengl, T., Reuter, H.I. (Eds.), Geomorphometry: Concepts, Software,
Applications. Elsevier, Amsterdam, pp. 195–226.
Bolstad, P.V., Lillesand, T.M., 1992. Improved classification of forest vegetation in
northern Wisconsin through a rule-based combination of soils, terrain, and
Landsat TM data. Forest Science 38, 5–20.
Bolstad, P.V., Stowe, T., 1994. An evaluation of DEM accuracy: elevation, slope and
aspect. Photogrammetric Engineering and Remote Sensing 60, 1327–1332.
Brown, D.G., 1998. Classification and boundary vagueness in mapping pre-
settlement forest types. International Journal of Geographical Information Science
12, 105–129.
Burbank, D., Leland, J., Fielding, E., Anderson, R.S., Brozovik, N., Reid, M.R.,
Duncan, C., 1996. Bedrock incision, rock uplift and threshold hillslopes in the
northwestern Himalaya. Nature 379, 505–510.
Burns, W.J., Coe, J.A., Kay, B.S., Ma, L., 2010. Analysis of elevation changes
detected from multi-temporal LiDAR surveys in forested landslide terrain in
western Oregon. Environmental and Engineering Geoscience 16, 315–341.
Burrough, P., 1996. Natural objects with indeterminate boundaries. In: Burrough,
P.A., Frank, A.U. (Eds.)., Geographic Objects with Indeterminate Boundaries.
Taylor and Francis, London, pp. 3–28.
Burrough, P.A., McDonnell, R., 1998. Principles of Geographical Information
Science. Oxford University Press, New York.
Burrough, P.A., van Gaans, P.F.M., MacMillan, R.A., 2000a. High-resolution
landform classification using fuzzy k-means. Fuzzy Sets and Systems 113,
37–52.
Burrough, P.A., Wilson, J.P., van Gaans, P.F.M., Hansen, A.J., 2000b. Fuzzy k-
means classification of topo-climatic data as an aid to forest mapping in the
Greater Yellowstone Area, USA. Landscape Ecology 16, 523–546.
Burt, T.P., Butcher, D.P., 1985. Topographic controls of soil moisture distribution.
Journal of Soil Science 36, 469–486.
Carabajal, C., Harding, D., 2006. SRTM C-Band and ICESat laser altimetry elevation
comparisons as a function of tree cover and relief. Photogrammetric Engineering
and Remote Sensing 72, 287–298.
Carara, A., Bitelli, G., Carla, R., 1997. Comparison of techniques for generating
digital terrain models from contour lines. International Journal of Geographical
Information Science 11, 451–473.
Charnpratheep, K., Zhou, Q., Garner, B., 1997. Preliminary landfill site screening
using fuzzy geographical information systems. Waste Management and Research
15, 197–215.
Chavez, Jr. P.S., 1996. Image-based atmospheric corrections: Revisited and improved.
Photogrammetric Engineering and Remote Sensing 62, 1025–1036.
Chirico, G.B., Western, A.W., Grayson, R.B., Günter, B., 2005. On the definition of
the flow width for calculating specific catchment area patterns from gridded
elevation data. Hydrological Processes 19, 2539–2556.
Chow, T.E., Hodgson, M.E., 2009. Effects of LiDAR post-spacing and DEM
resolution to mean slope estimation. International Journal of Geographic
Information Science 23, 1277–1295.
Conacher, A.J., Dalrymple, J.B., 1977. The nine-unit land surface model: an
approach to pedogeomorphic research. Geoderma 18, 1–154.
Costa-Cabral, M., Burges, S.J., 1994. Digital Elevation Model Networks (DEMON), a
model of flow over hillslopes for computation of contributing and dispersal
areas. Water Resources Research 30, 1681–1692.
Cova, T.J., Goodchild, M.F., 2002. Extending geographical representation to include
fields of spatial objects. International Journal of Geographical Information
Science 16, 509–532.
DeBruin, S., 2000. Querying probabilistic land cover data using fuzzy set theory.
International Journal of Geographical Information Science 14, 359–372.
Deng, Y.X., 2007. New trends in digital terrain analysis: landform definition,
representation, and classification. Progress in Physical Geography 31,
405–419.
Deng, Y.X., Wilson, J.P., 2006. The role of attribute selection in GIS representations
of the biophysical environment. Annals of the Association of American
Geographers 96, 47–63.
Deng, Y.X., Wilson, J.P., 2008. Multi-scale and multi-criteria mapping of mountain
peaks as fuzzy entities. International Journal of Geographical Information Science
22, 205–218.
Deng, Y.X., Wilson, J.P., Bauer, B.O., 2007. DEM resolution dependencies of terrain
attributes across a landscape. International Journal of Geographical Information
Science 21, 187–213.
Deng, Y.X., Wilson, J.P., Gallant, J.C., 2008. Terrain analysis. In: Wilson, J.P.,
Fotheringham, A.S. (Eds.), Handbook of Geographic Information Science.
Blackwell Publishers, Oxford, pp. 417–435.
Deng, Y.X., Wilson, J.P., Sheng, J., 2006. The sensitivity of fuzzy landform
classification to variable attribute weights. Earth Surface Processes and
Landforms 31, 1452–1462.
Desmet, P.J.J., 1997. Effects of interpolation errors on the analysis of DEMs. Earth
Surface Processes and Landforms 22, 563–580.
Desmet, P.J.J., Govers, G., 1996. Comparison of routing algorithms for
digital elevation models and their implications for predicting ephemeral
gullies. International Journal of Geographical Information Systems 10,
311–331.
Dikau, R., 1989. The application of a digital relief model to landform analysis in
geomorphology. In: Raper, J. (Ed.), Three Dimensional Applications in
Geographical Information Systems. Taylor and Francis, London, pp. 51–77.
Dowman, I.J., 2004. Integration of LIDAR and IFSAR for mapping. International
Archives of the Photogrammetry, Remote Sensing and Spatial Information
Sciences 35, 90–100.
Dozier, J., Bruno, J., Downey, P., 1981. A faster solution to the horizon problem.
Computers and Geosciences 7, 145–151.
Dubyah, R., Rich, P.M., 1995. Topographic solar radiation models for GIS.
International Journal of Geographical Information Systems 9, 405–419.
Endreny, T.A., Wood, E.F., 2001. Representing elevation uncertainty in runoff
modeling and flowpath mapping. Hydrological Processes 15, 2223–2236.
Endreny, T.A., Wood, E.F., 2003. Maximizing spatial congruence of observed and
DEM-delineated overland flow networks. International Journal of Geographical
Information Science 17, 699–713.
Evans, I.S., 1972. General geomorphometry, derivatives of altitude, and descriptive
statistics. In: Chorley, R.J. (Ed.), Spatial Analysis in Geomorphology. Harper and
Row, London, pp. 17–90.
Evans, I.S., Hengl, T., Gorsevski, P., 2009. Applications in geomorphology.
In: Hengl, T., Reuter, H.I. (Eds.), Geomorphometry: Concepts, Software,
Applications. Elsevier, Amsterdam, pp. 497–525.
Fairfield, J., Leymarie, P., 1991. Drainage networks from grid digital elevation
models. Water Resources Research 27, 709–717.
Feng, G., Sharratt, B., Vaughan, J., Lamb, B., 2009. A multiscale database of soil
properties for regional environmental quality modeling in the western United
States. Journal of Soil and Water Conservation 64, 363–373.
Filin, S., 2003. Recovery of systematic biases in laser altimetry data using natural
surfaces. Photogrammetric Engineering and Remote Sensing 57, 1235–1242.
Fisher, P.F., 1991. First experiments in viewshed uncertainty: the accuracy of the
viewshed area. Photogrammetric Engineering and Remote Sensing 57,
1321–1327.
Fisher, P.F., 1992. First experiments in viewshed uncertainty: simulating fuzzy
viewsheds. Photogrammetric Engineering and Remote Sensing 58, 345–352.
Fisher, P.F., 1993. Algorithm and implementation uncertainty in viewshed analysis.
International Journal of Geographical Information Systems 7, 331–347.
Fisher, P.F., 1995. An exploration of probable viewsheds in landscape planning.
Environment and Planning B 22, 527–546.
Fisher, P.F., 1996. Reconsideration of the viewshed function in terrain modeling.
Geographical Systems 3, 33–58.
Fisher, P.F., 1998. Improved modeling of elevation error with geostatistics.
Geoinformatica 2, 215–233.
Fisher, P.F., Pathirana, S., 1994. The evaluation of fuzzy membership of land cover
classes in the suburban zone. Remote Sensing of the Environment 34, 121–132.
Fisher, P.F., Tate, N.J., 2006. Causes and consequences of error in digital elevation
models. Progress in Physical Geography 30, 467–489.
Fisher, P.F., Wood, J., Cheng, T., 2004. Where is Helvellyn? Fuzziness of multi-
scale landscape morphometry. Transactions of the Institute of British
Geographers 29, 106–128.
Fisher, P.F., Wood, J., Cheng, T., 2005. Fuzziness and ambiguity in multi-scale
analysis of landscape morphometry. In: Petry, F.E., Robinson, V.B., Cobb, M.A.
(Eds.), Fuzzy Modeling with Spatial Information for Geographic Problems.
Springer-Verlag, New York, pp. 209–232.
Florinsky, I.V., 1998. Accuracy of local topographic variables derived from digital
elevation models. International Journal of Geographical Information Science 12,
47–62.
Foody, G.M., 1996. Fuzzy modeling of vegetation from remotely sensed imagery.
Ecological Modeling 85, 3–12.
Freeman, G.T., 1991. Calculating catchment area with divergent flow based on a
regular grid. Computers and Geosciences 17, 413–422.
Fried, J.S., Brown, D.G., Zweifler, M.O., Gold, M.A., 2000. Mapping contributing
areas for stormwater discharge to streams using terrain analysis. In: Wilson,

J.P., Gallant, J.C. (Eds.), Terrain Analysis: Principles and Applications. John
Wiley and Sons, New York, pp. 183–203.
Fu, P., Rich, P.M., 2000. A geometric solar radiation model with applications
in agriculture and forestry. Computers and Electronics in Agriculture 37,
25–35.
Gallant, J.C., Dowling, T.I., 2003. A multi-resolution index of valley bottom flatness
for mapping depositional areas. Water Resources Research 39, 1347–1360.
Gallant, J.C., Hutchinson, M.F., Wilson., J.P., 2000. Future directions for terrain
analysis. In: Wilson, J.P., Gallant, J.C. (Eds.), Terrain Analysis: Principles and
Applications. John Wiley and Sons, New York, pp. 423–427.
Gallant, J.C., Wilson, J.P., 1996. TAPES-G: a grid-based terrain analysis program
for the environmental sciences. Computers and Geosciences 22, 713–722.
Gallant, J.C., Wilson, J.P., 2000. Primary topographic attributes. In: Wilson, J.P.,
Gallant, J.C. (Eds.), Terrain Analysis: Principles and Applications. John Wiley
and Sons, New York, pp. 51–85.
Garbrecht., J., Martz, L.W., 1997. The assignment of drainage direction over
flat surfaces in raster digital elevation models. Journal of Hydrology 193,
204–213.
Gessler, P., Pike, R., MacMillan, R.A., Hengl, T., Reuter, H.I., 2009. The future of
geomorphometry. In: Hengl, T., Reuter, H.I. (Eds.), Geomorphometry: Concepts,
Software, Applications. Elsevier, Amsterdam, pp. 31–63.
Giles, P.T., 2001. Remote sensing and cast shadows in mountainous terrain.
Photogrammetric Engineering and Remote Sensing 67, 833–839.
Grimaldi, S., Nardi, F., Di Benedetto, F., Istanbulluoglu, E., Bras, R.L., 2007. A
physically-based method for removing pits in digital elevation models. Advances
in Water Resources 30, 2151–2158.
Goodchild, M.F., 1992. Geographical data modeling. Computers and Geosciences
18, 401–408.
Gruber, S., Peckham, S., 2009. Land-surface parameters and objects in hydrology.
In: Hengl, T., Reuter, H.I. (Eds.), Geomorphometry: Concepts, Software,
Applications. Elsevier, Amsterdam, pp. 171–194.
Guzzetti, F., Reichenbach, P., Cardinali, M., Galli, M., Ardizzone, F., 2005.
Probabilistic landslide hazard assessment at the basin scale. Geomorphology 72,
272–299.
Hengl, T., 2006. Finding the right pixel size. Computers and Geosciences 32,
1283–1298.
Hengl, T., Evans, I.S., 2009. Mathematical and digital models of the land surface.
In: Hengl, T., Reuter, H.I. (Eds.), Geomorphometry: Concepts, Software,
Applications. Elsevier, Amsterdam, pp. 31–63.
Hengl, T., Gruber, S., Shrestha, D.P., 2004. Reduction of errors in digital terrain
parameters used in soil-landscape modeling. International Journal of Applied
Earth Observation and Geoinformation 5, 97–112.
Hengl, T., MacMillan, R.A., 2009. Geomorphometry: a key to landscape mapping
and modeling. In: Hengl, T., Reuter, H.I. (Eds.), Geomorphometry: Concepts,
Software, Applications. Elsevier, Amsterdam, pp. 433–460.
Hengl, T., Reuter, H. I. (Eds.), 2009. Geomorphometry: Concepts, Software, and
Applications. Developments in Soil Science. Elsevier, Amsterdam.
Hernández Encinas, A., Hernández Encinas, L., Hoya White, S., Martin del Rey, A.,
Rodriquez Sanchez, G., 2007. Simulation of forest fire fronts using cellular
automata. Advances in Engineering Software 38, 372–378.
Hetrick, W.A., Rich, P.M., Barnes, F.J., Weiss, S.B., 1993a. GIS–based solar
radiation flux models. In: Proceedings of the ASPRS-ACSM Annual Convention,
Volume 3. American Society for Photogrammetry and Remote Sensing,
Bethesda, Maryland, pp. 132–143.
Hetrick, W.A., Rich, P.M., Weiss, S.B., 1993b. Modeling insolation on complex
surfaces. Proceedings of the Thirteenth Annual ESRI International User
Conference, Volume 2. Esri, Redlands, CA, pp. 447–458.
Hiranoa, A., Welch, R., Lang, H., 2003. Mapping from ASTER stereo image data:
DEM validation and accuracy assessment. ISPRS Journal of Photogrammetry
and Remote Sensing 57, 356–370.
Hodgson, M.E., Jensen, J.R., Raber, G., Tullis, J., Davis, B., Schuckman, K.,
Thompson, G., 2005. An evaluation of LiDAR-derived elevation and terrain slope
in leaf-off conditions. Photogrammetric Engineering and Remote Sensing 71,
817–823.
Hofierka, J., 1997. Direct solar radiation within an open GIS environment.
Proceedings of the 1997 Joint European GI Conference, Vienna, Austria,
pp. 575–584.
Hofton, M., Dubayah, R., Blair, J.B., Rabine, D., 2006. Validation of SRTM
elevations over vegetated and non-vegetated terrain using medium footprinting
LiDAR. Photogrammetric Engineering and Remote Sensing 72, 279–285.
Holmes, K.W., Chadwick, O.A., Kyriakidis, P.C., 2000. Error in a USGS 30 m digital
elevation model and its impact on digital terrain modeling. Journal of Hydrology
233, 154–173.
Geomorphometry 183
Holmgren, P., 1994. Multiple flow direction algorithms for runoff modeling in grid
based elevation models: an empirical evaluation. Hydrological Processes 8,
327–334.
Homann, P.S., Sollins, P., Fiorella, M., Thorson, T., Kern, J.S., 1998. Regional soil
organic carbon storage estimates for western Oregon by multiple approaches.
Soil Science Society of America Journal 62, 789–796.
Hunter, G.J., Goodchild, M.F., 1997. Modeling the uncertainty of slope and
aspect estimates derived from spatial databases. Geographical Analysis 29,
35–49.
Hutchinson, M.F., 1989. A new procedure for gridding elevation and stream line
data with automatic removal of spurious pits. Journal of Hydrology 106,
211–232.
Hutchinson, M.F., 2008. Adding the Z-dimension. In: Wilson, J.P., Fotheringham,
A.S. (Eds.), Handbook of Geographic Information Science. Blackwell Publishers,
Oxford, pp. 144–168.
Hutchinson, M.F., Gallant, J.C., 2000. Digital elevation models and representation of
terrain shape. In: Wilson, J.P., Gallant, J.C. (Eds.), Terrain Analysis: Principles
and Applications. John Wiley and Sons, New York, pp. 29–50.
Jenson, S.K., Domingue, J.O., 1988. Extracting topographic attributes from digital
elevation data for geographical information system analysis. Photogrammetric
Engineering and Remote Sensing 54, 1593–1600.
Jiang, H., Eastman, J.R., 2000. Application of fuzzy measures in multi-criteria
evaluation in GIS. International Journal of Geographical Information Science 14,
173–184.
Jones, J.A., 1986. Some limitations to the a/s index for predicting basin-wide
patterns of soil water drainage. Zeitschrift für Geomorphologie 60, 7–20.
Jones, J.A., 1987. The initiation of natural drainage networks. Progress in Physical
Geography 11, 205–245.
Jones, K.H., 1998. A comparison of algorithms used to compute hill slope as a
property of the DEM. Computers and Geosciences 24, 315–323.
Katzenbeisser, R., 2003. On the calibration of LiDAR sensors. In: Maas, H-G.,
Vosselman, G., Streilein, A. (Eds.), 3-D Reconstruction from Airborne
Laserscanner and InSAR Data. Institute of Photogrammetry and Remote Sensing.
University of Twente, Enschede, The Netherlands, pp. 59–64.
Kheir, R.B., Wilson, J.P., Deng, Y.X., 2007. Use of terrain variables for mapping
gully erosion susceptibility in Lebanon. Earth Surface Processes and Landforms
32, 1770–1782.
Kienzle, S., 2004. The effect of DEM raster resolution on first order, second order,
and compound terrain derivatives. Transactions in GIS 8, 83–111.
Klir, G.J., Yuan, B., 1995. Fuzzy Sets and Fuzzy Logic: Theory and Applications.
Prentice-Hall, Upper Saddle Creek, NJ.
Koons, P., Upton, P., Barker, A.D., 2012. The influence of mechanical properties
on the link between tectonic and topographic evolution. Geomorphology 137,
168–180.
Lagacherie, P., Andrieux, P., Bouzigues, R., 1996. Fuzziness and uncertainty of soil
boundaries: From reality to coding in GIS. In: Burrough, P.A., Frank A.U. (Eds.),
Geographic Objects with Indeterminate Boundaries. Taylor and Francis, London,
pp. 275–286.
LaLonde, T., Shortridge, A., Messina, J., 2010. The influence of land cover on
Shuttle Radar Topography Mission (SRTM) elevations in low-relief areas.
Transactions in GIS 14, 461–479.
Lea, N.L., 1992. An aspect driven kinematic routing algorithm. In: Parsons, A.J.,
Abrahams, A.D. (Eds.), Overland Flow: Hydraulics and Erosion Mechanics.
Chapman and Hall, New York, pp. 147–175.
Li, Z., Zhu, Q., Gold, C., 2005. Digital Terrain Modeling: Principles and
Methodology. CRC Press, Boca Raton, FL.
Lindsay, J.B., 2006. Sensitivity of channel mapping techniques to uncertainty in
digital elevation data. International Journal of Geographical Information Science
20, 669–692.
Lindsay, J.B., Creed, I.F., 2005a. Removal of artifact depressions from digital
elevation models: towards a minimum impact approach. Hydrological Processes
19, 3113–3126.
Lindsay, J.B., Creed, I.F., 2005b. Sensitivity of digital landscapes to artifact
depressions in remotely-sensed DEMs. Photogrammetric Engineering and
Remote Sensing 71, 1029–1036.
Lindsay, J.B., Creed, I.F., 2006. Distinguishing between artifact and real depressions
in digital elevation data. Computers and Geosciences 32, 1194–1204.
Lindsay, J.B., Evans, M.G., 2006. The influence of elevation error on the
morphometrics of channel networks extracted from DEMs and the implications
for hydrological modeling. Hydrological Processes 22, 1588–1603.
Lindsay, J.B., Rothwell, J.J., Davies, H., 2008. Mapping outlet points used for
watershed delineation onto DEM-derived stream networks. Water Resources
Research 44, W08442.
184 Geomorphometry
Mackay, D.S., Samanta, S., Ahl, D.E., Ewers, B.E., Gower, S.T., Burrows, S.N., 2003.
Automated parameterization of land surface process models using fuzzy logic.
Transactions in GIS 7, 139–153.
MacMillan, R.A., Shary, P.A., 2009. Landforms and landform elements in
geomorphometry. In: Hengl, T., Reuter, H.I. (Eds.), Geomorphometry: Concepts,
Software, Applications. Elsevier, Amsterdam, pp. 227–254.
MacMillan, R.A., Pettapiece, W.W., Nolan, S.C., Goddard, T.W., 2000. A generic
procedure for automatically segmenting landforms into landform elements using
DEMs, heuristic rules and fuzzy logic. Fuzzy Sets and Systems 113, 81–109.
Martz, L.W., Garbrecht, J., 1999. An outlet breaching algorithm for the treatment of
closed depressions in a raster DEM. Computers and Geosciences 25, 835–844.
Martz, L.W., de Jong, E., 1988. CATCH: a Fortran program for measuring catchment
area from digital elevation models. Computers and Geosciences 14, 627–640.
McBratney, J.A.B., Odeh, I.O.A., 1997. Application of fuzzy sets in soil science:
fuzzy logic, fuzzy measurements, and fuzzy decisions. Geoderma 77, 85–113.
Mészároš, I., 1998. Modelovanie prı́konu slnečnej energie na horské povodie. Acta
Hydrologica Slovaca 1, 68–75.
Miklánek, P., 1993. The Estimation of Energy Income in Grid Points Over the Basin
Using a Simple Digital Elevation Model. Springer, Berlin, European Geophysical
Society, Annales Geophysicae 11, Suppl. II, Berlin.
Miller, A.M., White, R.A., 1998. A conterminous United States multilayer soil
characteristics database for regional climate and hydrology modeling. Earth
Interactions 2, 1–26.
Mitášová, H., Mitas, L., Brown, W.M., Gerdes, D.P., Kosinovsky, I., Baker, T., 1995.
Modeling spatially and temporally distributed phenomena: new methods and
tools for GRASS GIS. International Journal of Geographical Information Systems
9, 433–446.
Molnar, P., England, P., 1990. Late Cenozoic uplift of mountain ranges and global
climate change: Chicken or egg? Nature 346, 29–34.
Montgomery, D.R., Dietrich, W.E., 1989. Source areas, drainage density, and
channel initiation. Water Resources Research 25, 1907–1918.
Montgomery, D.R., Dietrich, W.E., 1992. Channel initiation and the problem of
landscape scale. Science 255, 826–830.
Montgomery, D.R., Dietrich, W.E., 1994. A physically-based model for the
topographic control on shallow landsliding. Water Resources Research 30,
1153–1171.
Moore, I.D., Burch, G.J., 1986. Modeling erosion and deposition: topographic
effects. Transactions of the American Society of Agricultural Engineers 29,
1624–1630, 1640.
Moore, I.D., Burch, G.J., MacKenzie, D.H., 1988. Topographic effects on the
distribution of surface soil water and the location of ephemeral gullies.
Transactions of the American Society of Agricultural Engineers 31, 1098–1117.
Moore, I.D., Grayson, R.B., Ladson, A.R., 1991. Digital terrain modeling: a review of
hydrological, geomorphological, and biological applications. Hydrological
Processes 5, 3–30.
Moore, I.D., Lewis, A., Gallant, J.C., 1993a. Terrain attributes: estimation
methods and scale effects. In: Jakeman, A.J., Beck, M.B., McAleer, M.J. (Eds.),
Modeling Change in Environmental Systems. John Wiley and Sons, New York,
pp. 189–214.
Moore, I.D., Norton, T.W., Williams, J.E., 1993b. Modeling environmental
heterogeneity in forested landscapes. Journal of Hydrology 150, 717–747.
Moore, I.D., Wilson, J.P., 1992. Length-slope factors for the Revised Universal Soil
Loss Equation: simplified method of estimation. Journal of Soil and Water
Conservation 47, 423–428.
Moore, I.D., Wilson, J.P., 1994. Reply to ‘Comment on length-slope factors for the
Revised Universal Soil Loss Equation: Simplified method of estimation’ by G.R.
Foster. Journal of Soil and Water Conservation 49, 174–180.
Morris, D., Heerdegen, R., 1988. Automatically derived catchment boundaries and
channel networks and their hydrological applications. Geomorphology 1,
131–141.
Nelson, A., Reuter, H.I., Gessler, P., 2009. DEM production methods and sources.
In: Hengl, T., Reuter, H.I. (Eds.), Geomorphometry: Concepts, Software, and
Applications. Elsevier, Amsterdam, pp. 65–85.
Nguyen, T.M., 2011. Uncertainty in geomorphometry. University of Southern
California, Unpublished PhD dissertation.
Nguyen, T.M., Wilson, J.P., 2010. Sensitivity of the quasi-dynamic topographic
wetness index to choice of DEM resolution, flow routing algorithm and soil
variability. Proceedings of the Ninth International Symposium on Spatial
Accuracy Assessment in Natural Resources and the Environmental Sciences
(Accuracy 2010), Leicester, UK.
O’Callaghan, J.F., Mark, D.M., 1984. The extraction of drainage networks from
digital elevation data. Computer Vision, Graphics and Image Processing 28,
323–344.
O’Loughlin, E.M., 1986. Prediction of surface saturation zones in natural catchments
by topographic analysis. Water Resources Research 22, 794–804.
Olaya, V., 2009. Basic land-surface parameters. In: Hengl, T., Reuter, H.I. (Eds.),
Geomorphometry: Concepts, Software, Applications. Elsevier, Amsterdam, pp.
141–169.
Peckham, S.D., 1998. Efficient extraction of river networks and hydrologic
measurements from digital elevation data. In: Barndorff-Nielsen, O.E., Gupta,
V.K., Pérez-Abreu, V., Waymire, E. (Eds.), Stochastic Methods in Hydrology:
Rain, Landforms, and Floods. World Scientific, Singapore, pp. 173–203.
Phillips, J.D., 1990. A saturation-based model of relative wetness for wetland
identification. Water Resources Bulletin 26, 333–342.
Pike, R.J., 1995. Geomorphometry: progress, practise and prospect. Zeitschrift für
Geomorphologie 101, 221–238.
Pike, R.J., 2000. Geomorphometry: diversity in quantitative surface analysis.
Progress in Physical Geography 24, 1–20.
Pike, R.J., Evans, I.S., Hengl, T., 2009. Geomorphometry: a brief guide. In: Hengl,
T., Reuter, H.I. (Eds.), Geomorphometry: Concepts, Software, Applications.
Elsevier, Amsterdam, pp. 3–30.
Planchon, O., Darboux, F., 2001. A fast, simple and versatile algorithm to fill the
depressions of digital elevation models. Catena 46, 159–176.
Proy, C., Tanre, D., Deschamps, P.Y., 1989. Evaluation of topographic effects in
remotely sensed data. Remote Sensing of the Environment 30, 21–32.
Qi, F., 2004. Knowledge discovery from area-class resource maps: data
preprocessing for noise reduction. Transactions in GIS 8, 297–308.
Qi, F., Zhu, A.-X., 2003. Knowledge discovery from soil maps using inductive
learning. International Journal of Geographical Information Science 17,
771–795.
Qi, F., Zhu, A.-X., Harrower, M., Burt, J.E., 2006. Fuzzy soil mapping based on
prototype category theory. Geoderma 136, 774–787.
Qin, C., Zhu, A-X., Pei, T., Li, B., Zhou, C., Yang, L., 2007. An adaptive approach
to selecting the flow partition exponent for multiple flow direction algorithms.
International Journal of Geographical Information Science 21, 443–458.
Quinn, P.F., Beven, K.J., Chevallier, P., Planchon, O., 1991. The prediction of
hillslope paths for distributed hydrological modeling using digital terrain
models. Hydrological Processes 5, 59–79.
Quinn, P.F., Beven, K.J., Lamb, R., 1995. The ln(a/tan b) index: how to calculate it
and how to use it within the TOPMODEL framework. Hydrological Processes 9,
161–182.
Quinn, T., Zhu, A-X., Burt, J.E., 2005. Effects of detailed soil spatial information on
watershed modeling across different model scales. International Journal of
Applied Earth Observation and Geoinformation 7, 324–338.
Raaflaub, L.D., Collins, M.J., 2006. The effect of error in gridded digital elevation
models on the estimation of topographic parameters. Environmental Modeling
and Software 21, 710–732.
Raper, J., Livingstone, D., 1995. Development of a geomorphological spatial model
using object-oriented design. International Journal of Geographical Information
Systems 9, 359–383.
Rawls, W.J., 1983. Estimating soil bulk density from particle size analysis and
organic matter. Soil Science 135, 123–125.
Reuter, H.I., Hengl, T., Gessler, P., Soille, P., 2009. Preparation of DEMs for
geomorphometric analysis. In: Hengl, T., Reuter, H.I. (Eds.), Geomorphometry:
Concepts, Software, and Applications. Elsevier, Amsterdam, pp. 87–120.
Reuter, H.I., Kersebaum, K.C., Wendroth, O., 2005. Modeling of solar radiation
influenced by topographic shading: evaluation and application for precision
farming. Physics and Chemistry of the Earth 30, 139–149.
Reiger, W., 1992. Automated river line and catchment area extraction for DEM data.
Proceedings of the Seventeenth International Society for Photogrammetry and
Remote Sensing Congress, Washington, DC, pp. 642–649.
Robinson, V.B., 2003. A perspective on the fundamentals of fuzzy sets and their use
in Geographical Information Systems. Transactions in GIS 7, 3–30.
Rodriguez, E., Morris, C., Belz, J.E., 2006. A global assessment of the SRTM
performance. Photogrammetric Engineering and Remote Sensing 72, 249–260.
Rodriguez, E., Morris, C.S., Belz, J.E., Chapin, E.C., Martin, J.M., Daffer, W.,
Hensley, S., 2005. An Assessment of the SRTM Topographic Products. Jet
Propulsion Laboratory, Pasadena, California.
Rossi, R.E., Dungan, J.L., Beck, L.R., 1994. Kriging in the shadows: Geostatistical
interpolation for remote sensing. Remote Sensing of the Environment 49,
32–40.
Ruhe, R.V., 1960. Elements of the soil landscape. Proceedings of the Ninth
Congress of the International Society of Soil Science, Madison, WI, pp.
165–170.
Ruiz, M., 1997. A causal analysis of error in viewsheds from USGS digital elevation
models. Transactions in GIS 2, 85–94.

Saunders, W.K., Maidment, D.R., 1996. A GIS Assessment of Nonpoint Source
Pollution in the San Antonio-Nueces Coastal Basin. University of Texas Center
for Research in Water Resources Report No. 96-1, Austin, TX.
Saxton, K.E., Rawls, W.J., 2006. Soil water characteristic estimates by texture and
organic matter for hydrologic solutions. Soil Science Society of America
Agronomy Journal 70, 1569–1578.
Saxton, K.E., Rawls, W.J., Romberger, J.S., Papendick, R.I., 1986. Estimating
generalized soil-water characteristics from texture. Transactions of the American
Society of Agricultural Engineers 50, 1031–1035.
Schmidt, J., Dikau, R., 1999. Extracting geomorphometric attributes and objects
from digital elevation models: semantics, methods, and future needs. In: Dikau,
R., Saurer, H. (Eds.), GIS for Earth Science Systems: Analysis and Modeling of
the Natural Environment. Schweizbart’sche Verlagsbuchhandlung, Stuttgart,
Germany, pp. 153–173.
Schmidt, J., Hennrich, K., Dikau, R., 1998. Scales and similarities in runoff
processes with respect to geomorphometry. Proceedings of the Third
International Conference on GeoComputation, Bristol, UK.
Schmidt, J., Hewitt, A., 2004. Fuzzy land element classification from DTMs based
on geometry and terrain position. Geoderma 121, 243–256.
Schoorl, J.M., Sonneveld, M.P.W., Veldkamp, A., 2000. Three-dimensional
landscape process modeling: the effect of DEM resolution. Earth Surface
Processes and Landforms 25, 1025–1034.
Shary, P.A., 1995. Land surface in gravity points classification by a complete
system of curvatures. Mathematical Geology 27, 373–390.
Shary, P.A., Sharaya, L.S., Mitusov, A.V., 2002. Fundamental quantitative methods of
land surface analysis. Geoderma 107, 1–32.
Shary, P.A., Sharaya, L.S., Mitusov, A.V., 2005. The problem of scale-specific and
scale-free approaches to geomorphometry. Geografia Fisica e Dinamica
Quaternaria 28, 81–101.
Shary, P.A., Stepanov, I.N., 1991. On the second derivative method in geology.
Doklady AN SSSR 319, 456–460, (in Russian).
Shi, X., Zhu, A-X., Wang, R-X., 2005. Deriving fuzzy representations of some
special terrain features based on their typical locations. In: Petry, F.E., Robinson,
V.B., Cobb, M.A. (Eds.), Fuzzy Modeling with Spatial Information for Geographic
Problems. Springer-Verlag, New York, pp. 233–252.
Shortridge, A., 2006. Shuttle Radar Topography Mission elevation data error and its
relationship to land cover. Cartography and Geographic Information Science 33,
65–75.
Sivapalan, M., Beven, K.J., Wood, E.F., 1987. On hydrologic similarity: 2, a
scaled model of storm runoff production. Water Resources Research 23,
2266–2278.
Skidmore, A.K., 1989. A comparison of techniques for calculating gradient and
aspect from a gridded digital elevation model. International Journal of
Geographical Information Systems 3, 323–334.
Skidmore, A.K., Ryan, P.J., Short, D., Dawes, W., 1991. Forest soil type mapping
using an expert system with Landsat Thematic Mapper data and a digital
elevation model. International Journal of Geographical Information Systems 5,
431–445.
Slater, J.A., Heady, B., Kroenung, G., et al., 2009. Evaluation of the New ASTER
Global Digital Elevation Model. National Geospatial-Intelligence Agency,
Washington, DC.
Smith, M.P., Zhu, A.X., Burt, J.E., Stiles, C., 2006. The effects of DEM resolution
and neighborhood size on digital soil survey. Geoderma 137, 58–68.
Soille, P., 2004. Optimal removal of spurious pits in grid digital elevation models.
Water Resources Research 40, W12509.
Soille, P., Gratin, C., 1994. An efficient algorithm for drainage networks extraction
on DEMs. Journal of Visual Communication and Image Representation 5,
181–189.
Soille, P., Vogt, J., Colombo, R., 2003. Carving and adaptive drainage enforcement
of grid digital elevation models. Water Resources Research 39, 1366–1378.
Speight, J.G., 1968. Parametric description of land form. In: Stewart, G.A. (Ed.),
Land Evaluation. Papers of a CSIRO Symposium, MacMillan, Melbourne,
pp. 239–250.
Speight, J.G., 1990. Landform. In: MacDonald, R.C., Isbell, R.F., Speight, J.G.,
Walker, J., Hop, M.S. (Eds.), Australian Soil and Land Survey Field Handbook.
Inkata Press, Melbourne, pp. 9–57.
Sulebak, J.R., Hjelle, Ǿ., 2003. Multi-resolution spline models and their
applications in geomorphology. In: Evans, I.S., Dikau, R., Tokunaga, R., Ohmori,
H., Hirano, M. (Eds.), Concepts and Modeling in Geomorphology: International
Perspectives. Terra Publications, Tokyo, pp. 221–237.
Tang, G., Shi, W., Zhao, M., 2002. Evaluation of the accuracy of hydrologic data
derived from DEMs of different spatial resolution. In: Hunter, G.J., Lowell, K.
(Eds.), Proceedings of the Fifth International Symposium on Spatial Accuracy
Geomorphometry 185
Assessment in Natural Resources and Environmental Sciences (Accuracy 2000).
RMIT University, Melbourne, pp. 204–213.
Tarboton, D.G., 1997. A new method for the determination of flow directions and
upslope areas in grid digital elevation models. Water Resources Research 33,
309–319.
Temme, A.J.A.M., Heuvelink, G.B.M., Schoorl, J.M., Claessens, L., 2009.
Geostatistical simulation and error propagation in geomorphometry. In: Hengl,
T., Reuter, H.I. (Eds.), Geomorphometry: Concepts, Software, and Applications.
Elsevier, Amsterdam, pp. 121–140.
Temme, A.J.A.M., Schoorl, J.M., Veldkamp, A., 2006. An algorithm for dealing with
depressions in dynamic landscape evolution models. Computers and
Geosciences 32, 452–461.
Thompson, J.A., Bell, J.C., Butler, C.A., 2001. Digital elevation model resolution:
effects on terrain attribute calculation and quantitative soil-landscape modeling.
Geoderma 100, 67–89.
Usery, E.L., 1996. A conceptual framework and fuzzy set implementation for
geographic features. In: Burrough, P.A., Frank, A.U. (Eds.)., Geographic Objects
with Indeterminate Boundaries. Taylor and Francis, London, pp. 71–85.
Van Niel, K., Laffan, S.W., Lees, B.G., 2004. Effect of error in the DEM on
environmental variables for predictive vegetation modeling. Journal of Vegetation
Science 15, 747–756.
Waltman, W.J., Waltman, S.W., Palecki, M.A., 2003. A GIS-based spatial pattern
analysis model for eco region mapping and characterization. International
Journal of Geographical Information Science 17, 445–462.
Wang, L., Liu, H., 2006. An efficient method for identifying and filling surface
depressions in digital elevation models for hydrologic analysis and modeling.
International Journal of Geographical Information Science 20, 193–213.
Warren, S.D., Hohmann, M.G., Auerswald, K., Mitášová, H., 2004. An evaluation of
methods to determine slope using digital elevation data. Catena 58, 215–233.
Webster, T.L., Dias, G., 2006. An automated GIS procedure for comparing GPS and
proximal LIDAR elevations. Computers and Geosciences 32, 713–726.
Wilson, J.P., 2012. Digital terrain modeling. Geomorphometry 137, 107–121.
Wilson, J.P., Aggett, G.R., Deng, Y.X., Lam, C.S., 2008. Water in the landscape: a
review of contemporary flow routing algorithms. In: Zhou, Q., Lees, B.G., Tang,
G-A. (Eds.), Advances in Digital Terrain Analysis. Springer Lecture Notes in
Geoinformation and Cartography, Berlin, pp. 213–236.
Wilson, J.P., Burrough, P.A., 1999. Dynamic modeling, geostatistics, and fuzzy
classification: new sneakers for a new geography? Annals of the Association of
American Geographers 89, 736–746.
Wilson, J.P., Gallant, J.C., 2000a. Digital terrain analysis. In: Wilson, J.P., Gallant,
J.C. (Eds.), Terrain Analysis: Principles and Applications. John Wiley and Sons,
New York, pp. 1–27.
Wilson, J.P., Gallant, J.C., 2000b. Secondary topographic attributes. In: Wilson, J.P.,
Gallant, J.C. (Eds.), Terrain Analysis: Principles and Applications. John Wiley
and Sons, New York, pp. 51–85.
Wilson, J.P., Inskeep, W.P., Wraith, J.M., Snyder, R.D., 1996. GIS-based solute
transport modeling applications: scale effects of soil and climate input data.
Journal of Environmental Quality 25, 445–453.
Wilson, J.P., Lam, C.S., Deng, Y.X., 2007. Comparison of performance of flow
routing algorithms used in Geographic Information Systems. Hydrological
Processes 21, 1026–1044.
Wilson, J.P., Repetto, P.L., Snyder, R.D., 2000. Effect of data source, grid resolution,
and flow-routing method on computed topographic attributes. In: Wilson, J.P.,
Gallant, J.C. (Eds.), Terrain Analysis: Principles and Applications. John Wiley
and Sons, New York, pp. 133–161.
Wise, S.M., 1998. The effect of GIS interpolation errors on the use of digital
elevation models in geomorphology. In: Lane, S.N., Richards, K.S., Chandler,
J.H. (Eds.), Landform Monitoring, Modeling, and Analysis. John Wiley and
Sons, Chichester, pp. 139–164.
Wise, S.M., 2000. Assessing the quality of hydrological applications of digital
elevation models derived from contours. Hydrological Processes 14,
1909–1929.
Wolock, D.M., McCabe, G.J., 1995. Comparison of single and multiple flow
direction algorithms for computing topographic parameters in TOPMODEL.
Water Resources Research 31, 1315–1324.
Wood, R., Sivapalan, M., Robinson, J., 1997. Modeling the spatial variability of
surface runoff using a topographic index. Water Resources Research 33,
1061–1073.
Yokoyama, R., Shirasawa, M., Pike, R.J., 2002. Visualizing topography by
openness: a new application of image processing to digital elevation models.
Photogrammetric Engineering and Remote Sensing 68, 257–265.
Zandbergen, P., 2008. Applications of Shuttle radar topography mission elevation
data. Geography Compass 2, 1404–1431.
186 Geomorphometry

Zhang, W.H., Montgomery, D.R., 1994. Digital elevation model grid size, landscape
representation, and hydrological simulations. Water Resources Research 30,
1019–1028.
Zhou, Q., Lees, B.G., Tang, G-A. (Eds.), 2008. Advances in Digital Terrain Analysis.
Springer Lecture Notes in Geoinformation and Cartography, Berlin.
Zhou, Q., Liu, X., 2002. Error assessment of grid-based flow routing algorithms
used in hydrological models. International Journal of Geographical Information
Science 16, 819–842.
Zhou, Q., Liu, X., 2004. Analysis of errors of derived slope and aspect related to
DEM data properties. Computers and Geosciences 30, 369–378.
Zhu, A-X., 1997a. A similarity model for representing soil spatial information.
Geoderma 77, 217–242.
Zhu, A-X., 1997b. Measuring uncertainty in class assignment for natural resource
maps under fuzzy logic. Photogrammetric Engineering and Remote Sensing 63,
1195–1202.
Zhu, A-X., 1999. A personal construct-based knowledge acquisition for natural resource
mapping. International Journal of Geographical Information Science 13, 119–141.
Zhu, A-X., Band, L.E., Vertessy, R., Dutton, B., 1997. Derivation of soil properties
using a soil land inference model (SoLIM). Soil Science Society of America
Journal 61, 523–533.
Zhu, A-X., Mackay, D.S., 2001. Effects of spatial detail of soil information on
watershed modeling. Journal of Hydrology 248, 54–77.

Biographical Sketch

Dr. John P Wilson is a professor of spatial sciences at the University of Southern California (USC), where he heads
the Spatial Sciences Institute as well as the Geographic Information Science & Technology Graduate Programs and
GIS Research Laboratory, and also holds appointments as professor in the Dornsife College’s Department of
Sociology, the School of Architecture and in the Viterbi School of Engineering’s Departments of Computer Science
and Civil & Environmental Engineering. His current research is focused on the use of geospatial data and tools for
characterizing hydrologic systems and the relationships linking environmental exposures, societal forces, and
human health outcomes.

Dr. Michael P Bishop is a professor and Haynes Chair of Geosciences in the Department of Geography at Texas
A&M University. He received his PhD at Indiana State University (1987) in physical geography with a focus on
remote sensing and geographic information science. His areas of expertise are in remote sensing, geographic
information systems (GIS), geomorphometry, numerical modeling, and mountain geomorphology. He has
published over 30 articles in scientific journals, three books, and numerous book chapters on topics including
radiation transfer, image and terrain analysis, surface processes and landforms, climate and glacier change, and
landscape evolution modeling. Furthermore, he has presented over 200 national and international professional
papers on various remote sensing, GIS, and mountain geomorphology-related topics. Financial support for his
research has been obtained through the National Geographic Society, National Science Foundation, NASA, DOE,
USGS, and numerous foundations. His current research is focused on the use of satellite imagery and terrain
analysis for characterizing surface processes and landforms in complex mountain environments.
 3.8 Remote Sensing and GIScience in Geomorphological Mapping
J Napieralski, University of Michigan-Dearborn, Dearborn, MI, USA
I Barr, Queen Mary University of London, London, UK
U Kamp, The University of Montana, Missoula, MT, USA
M Kervyn, Vrije Universiteit Brussel, Brussels, Belgium
r 2013 Elsevier Inc. All rights reserved.

3.8.1 Introduction 188

3.8.2 Background 190
3.8.2.1 Remotely Sensed Data 190
3.8.2.1.1 Aerial photographs 190
3.8.2.1.2 Satellite images 190
3.8.2.1.3 Digital elevation models 191
3.8.2.1.4 Supplementary data 192
3.8.2.2 GIScience 192
3.8.2.2.1 Mapping approaches 194
3.8.3 Glacial Landscapes and Landforms 194
3.8.3.1 Glacial Landforms 195
3.8.3.2 Data Processing and Visualization 195
3.8.3.2.1 Aerial photographs 195
3.8.3.2.2 Satellite images 196
3.8.3.2.3 Digital elevation models 197
3.8.3.3 GIS-based Mapping 198
3.8.4 Volcanic Terrain and Landforms 199
3.8.4.1 Satellite Images 201
3.8.4.2 Digital Elevation Models 203
3.8.4.3 GIS-Based Mapping 206
3.8.5 Landslide Mapping 207
3.8.5.1 Terrestrial and Airborne Photography 207
3.8.5.2 Satellite Imagery 208
3.8.5.3 Digital Elevation Models 209
3.8.5.3.1 LiDAR 209
3.8.5.3.2 Radar 209
3.8.5.3.3 ASTER 211
3.8.5.3.4 Geomorphometric analysis 211
3.8.5.4 GIS Modeling 212
3.8.5.5 GIS-based Landslide Inventories 212
3.8.5.6 GIS-based Landslide Assessment Maps 212
3.8.6 Fluvial Landscapes and Landforms 213
3.8.6.1 Aerial Photographs 213
3.8.6.2 Airborne And Satellite Sensors 214
3.8.6.3 Active Sensors 215
3.8.6.4 Geographic Information Systems 215
3.8.7 Conclusion 218
References 219

Glossary most commonly acquired using visible light from either

Aerial photography The act of photographing Earth’s vertical (e.g., useful for photogrammetry) or oblique
surface from an elevated position. Aerial photographs are perspectives.

Napieralski, J., Barr, I., Kamp, U., Kervyn, M., 2013. Remote sensing and
GIScience in geomorphological mapping. In: Shroder, J. (Editor in Chief),
Bishop, M.P. (Ed.), Treatise on Geomorphology. Academic Press, San Diego,
CA, vol. 3, Remote Sensing and GIScience in Geomorphology, pp. 187–227.

Treatise on Geomorphology, Volume 3 http://dx.doi.org/10.1016/B978-0-12-374739-6.00050-6 187

188 Remote Sensing and GIScience in Geomorphological Mapping

Digital elevation model A digital elevation model (DEM)
is generally a land-surface model that attempts to accurately
portray the altitude of the topography. In geomorphology,
it commonly takes the form of a raster data layer
representing a field of square tessellations. The resolution of
the grid cells is generally determined based upon the source
data utilized and the desired scale for representing the
topography.
Fuzzy logic Fuzzy theory deals with approximations
rather than exactness. Fuzzy logic is used in
geomorphological mapping because of the difficulty
associated with mapping exact boundaries of landforms.
Geographic information science The discipline of
GIScience is concerned with the theory behind the
development, implementation, and application of GIS.
Geographic information systems A geographic
information system (GIS) is used to store, manipulate,
analyze, and display geographically referenced data.
Geomorphology An Earth-science discipline that focuses
on studying and understanding surface processes and
the morphological characteristics of the topography, via
the classification, measurement, and description of
landforms.
Geomorphometry The discipline that is concerned
with the science of quantitative land-surface
characterization.
Photogrammetry The science and technology of
obtaining quantitative measurements and other
geometrically reliable information from photographs.
An advanced version of photogrammetry is
stereophotogrammetry, which allows for calculations of
relief (a valuable landform mapping parameter).
Raster data A data structure that is used to represent a
field data model to characterize surface attributes. It
represents a field of tessellations organized into rows and
columns where each cell/pixel contains an attribute value
(e.g., elevation, reflectance).
Remote sensing The science and technology of obtaining
information about objects and planetary surfaces using
sensors that record variations in surface matter/energy
interactions. It also involves the symbolic transformation of
data into useable information that can be used to study the
Earth and solve problems.
Scale The concept of scale can be defined in many ways.
It is most commonly used to characterize the geographic
coverage and level of detail associated with maps,
photographs, and imagery. A large scale indicates a
relatively large representative fraction (e.g., relatively
more detail and less spatial coverage compared to small
scale).
Semantics The relationship between computer
representations and the corresponding real-world feature
within an application domain. Semantic models can be
used for landform mapping, and they formalize the
landform attributes and characteristics that permit GIS-
based landform mapping.
Vector data Spatial cartographic primitives are used to
characterize discrete features (i.e., points, lines, and
polygons). Vector data also include attribute information
that is referenced to the spatial information.

Abstract

The rapid proliferation of remote sensing and geographic information systems (GIS) into geomorphologic mapping
has increased the objectivity and efficiency of landform segmentation, measurement, and classification. The near
ubiquitous presence of Earth-observing satellites provides an array of perspectives to visualize the biophysical characteristics
of landscapes, access inhospitable terrain on a predictable schedule, and study landscape processes when conditions
are hazardous. GIS technology has altered the analysis, visualization, and dissemination of landform data due to the
shared theoretical concepts that are fundamental to geomorphology and GIScience. The authors review geospatial tech-
nology applications in landform mapping (including emerging issues) within glacial, volcanic, landslide, and fluvial
research.

3.8.1 Introduction Earth’s surface. Therefore, landform mapping can be used to

Geomorphological studies serve to improve our understand-
ing and problem-solving capabilities involving surface pro-
cesses, landscape evolution, landform mapping, hazard
prediction, inventory, planning/management, and ecological
analyses (Drăgut- and Blaschke, 2006). Surveying and map-
ping technologies are critical to geomorphological and map-
ping studies, as they provide a primary source of data to
represent bare-Earth terrain, and permit information ex-
traction for product generation. In particular, geomorph-
ologists frequently view landforms as geometric signatures, or
evidence that specific processes modified the shape of the
study dynamic processes based upon landform distributions,
terrain morphology, and terrain composition. The techniques
and approaches by which landforms are delineated, measured,
and classified have evolved rapidly, from point observations,
surveys, and qualitative interpretations, to a dependence on
digital data from airborne- and space-based sensors, advanced
geostatistical analysis and spatial modeling, and scientific
visualization techniques. Although field-based surveying and
qualitative methods continue to support geomorphic studies,
contemporary geomorphologists rely on an ever-expanding
range and profusion of digital data because of the relative ease
by which inaccessible terrain can be studied, the availability
 Remote Sensing and GIScience in Geomorphological Mapping 189

and low-cost of topographic and spectral data, and the

access to automation and efficient quantitative assessment
(Millington and Townshend, 1987; Walsh et al., 1998; Smith
and Pain, 2009). Geographic information technologies (GIT)
that include digital image processing, geographic information
systems (GIS), and global positioning systems (GPS), have
equipped geomorphologists with the capability to solve more
sophisticated and interdisciplinary problems. This includes
revisiting more traditional concepts and problems, addressing
new concepts and issues (Hynek and Phillips, 2003; Mullen
and Kellet, 2007), developing algorithms and analysis pro-
cedures (Chandler et al., 2005), and evaluating new algo-
rithms and emerging technologies in landform studies (Lejot
et al., 2007; Wilford, 2009).
With the rapid proliferation of GIT, geomorphologists are
routinely utilizing GIS to analyze the topography and map
and study landforms. The use of GIT, however, will not
completely replace traditional field methods of mapping, as
first-hand knowledge of a study area can yield information
that is not readily accessible solely from spatial data and GIS-
based analysis. Rather, the complexity of landscape mapping
and geomorphology requires a comprehensive approach to
fully understand landform morphology and landform-process
linkages. Figure 1 illustrates the value of combining multiple
datasets, along with field knowledge, to identify landforms
and understand geomorphic processes. Aerial photos provide
an easy-to-interpret perspective of the Western Michigan
landscape (Figure 1(a)). Topographic variations are difficult
to detect in the photograph, so a contoured elevation model
(Figure 1(b)), derived from the U.S. National Elevation
Dataset (NED), illustrates the ‘lay of the land,’ which includes
topographic highs (e.g., south and north of the Herring Lakes,
and Lake Michigan shoreline) and topographic lows (Herring
Lakes). Segmenting this landscape into individual units
or landforms is significantly enhanced when using multi-
spectral imagery (i.e., regions of the spectrum beyond visible
wavelengths). In particular, landforms such as sand dunes,
water bodies, and sediment-filled depressions are enhanced
using a Landsat Thematic Mapper (TM) image of the area
(Figure 1(c)). It is only when information derived from these
datasets are combined, along with a contextualization of
geologic history based upon fieldwork and local knowledge,
however, that one can relate landform characteristics to
geomorphic processes. The ridges (topographic highs in
Figure 1(b)) are actually glacial moraines, formed during the
retreat of the Last Glacial Maximum (LGM). Due to sediment
transportation and deposition from the longshore current, the
bay mouth has shifted from the filled depression adjacent to
Upper Herring Lake (dark red in Figure 1(c)), to Upper
Herring and then to Lower Herring Lake, where geologic, cli-
matic, and biophysical processes continue to modify the
western coast of Michigan. This simple, qualitative assessment Figure 1 Western Michigan viewed from different data sources. (a)
using spatial data typifies many of the spatial problems (and Aerial photography from the US National Agriculture Imagery Program
solutions) in geomorphology, as GIT offers countless per- (NAIP). (b) Contoured digital elevation model (DEM) from the U.S.
spectives and methods to map and study landforms, although NED (30 m). (c) Landsat ETM þ false-color composite (FCC; 4, 3, 2).
limitations can exist regarding data, methodological design,
and user experience. structures), this chapter summarizes the use of GIT to map
Although a vast array of landform mapping applications and assess glacial, volcanic, fluvial, and mass movement
rely on remote sensing, GIS, and GPS (e.g., hazards assess- landforms. The integration of GIT in glacial geomorphology
ment, glacier mapping, impact craters, aeolian dunes, coastal has enhanced the ability to map the surface topography
190 Remote Sensing and GIScience in Geomorphological Mapping

and the ice-velocity fields of glaciers and ice sheets (Smith

et al., 2001; Greenwood and Clark, 2008; Barr and Clark,
2009; Hughes et al., 2010). Spatial information can be further
combined with numerical models to better understand land-
scape evolution in glaciated terrain (Napieralski et al., 2007a).
Remote sensing and GIS have also improved efforts to map
and distinguish between volcanic lava flows (Kahle et al.,
1988; Wadge et al., 1995), characterize the slope and dynamic
growth of volcanic edifices (Kervyn et al., 2008; Favalli et al.,
2009; Grosse et al., 2009), discriminate between lava flows
and different lithologies, predict rates of erosion, and generate
hazard maps (Marti and Felpeto, 2010). Geospatial technolo-
gies are also required to continuously monitor active geo-
logical areas characterized by slope stability, as inventory of
landform structures are needed (Owen et al., 2008) to predict
mass movements (Xie et al., 2004; Refice and Capolongo,
2002), risk assessment is required (Davis et al., 2000; Dahal
et al., 2008; Hasegawa et al., 2009), and rapid dissemination
of data and information to scientists and the general public is
critical (Kim et al., 2004; Chang and Park, 2004). Further-
more, dynamic river environments exhibit a wide range of
biophysical characteristics that make them a challenge to map,
but the integration of geospatial technologies in fluvial geo-
morphology has improved our understanding of channel
migration, sinuosity, and migration (e.g., Winterbottom and
Gilvear, 2000; O’Connor et al., 2003; Jordan and Fonstad,
2005), floodplain extent (Marston et al., 1995; Thoma et al.,
2005; Hamilton et al., 2006), stream-bank stability and ero-
sion (Micheli and Kirchner, 2002), sediment entrainment, and
facilitates paleochannel reconstruction (Gurnell et al., 1994;
Gurnell, 1997).
3.8.2 Background
3.8.2.1 Remotely Sensed Data
Remote sensing involves the acquisition and analysis of in-
formation about the landscape through the use of sensor
systems that record the magnitude of matter–energy inter-
actions. Biophysical and thematic information can be pro-
duced that greatly facilitates geomorphological assessment
and landform mapping. Key landscape information includes
spectral data in the form of aerial photographs, satellite im-
ages, ground photography, and videography, as well as key
biophysical information of the near-surface environment, in-
cluding topographic, surface temperature, albedo, and sub-
surface information. These sources can be used alone or, more
commonly, in conjunction with one another or with field-
based data, to enable real-time or comprehensive geo-
morphological mapping (e.g., Heyman et al., 2008; Hughes
et al., 2010; Karatson et al., 2010b). In addition, GPS receivers
can be used to survey ‘on-the-fly’ in the field or to remotely
monitor rugged landscapes affected by tectonic forces (e.g.,
Higgitt and Warburton, 1999; Burbank and Anderson, 2001;
Baldi et al., 2008).
3.8.2.1.1 Aerial photographs
The earliest application of remote-sensing methods for land-
form mapping involved the qualitative (visual) analysis of
aerial photographs (e.g., Prest et al., 1968). Such photographs
Figure 2 Oblique aerial photo (15 October 2010) of lava streams
flowing from the Piton de la Fournaise volcano on Réunion Island.
Aerial photography offers a relative safe vantage point during
volcanic eruptions, floods, and landslides. Note helicopter and
people along right of photograph for scale. Photograph from
Richard Bouhet/AFP/Getty Images.
remain a valuable source to study local-scale landforms,
and are commonly used to augment field mapping, or used
in conjunction with other data (Hall et al., 2009). Aerial
photographs are a reliable data source for detailed mapping
and are frequently used to produce baseline information
for conducting change-detection studies. Visual analyses of
aerial photographs also support assessments of flooding
events, lava flows (Figure 2), and landslide hazards. Large-
scale information (i.e., high spatial resolution), however, is
counterbalanced by the limited spatial coverage of individual
photographs, limiting their utility for local to regional
mapping. The analysis of large areas (e.g., mountains, glaci-
ated, extensive stream and watershed systems) generally
requires numerous photographs, and is therefore both
time-consuming and expensive (Smith et al., 2006). Neverthe-
less, digital cameras and video systems are rapidly replac-
ing ground and aerial photography to provide additional
benefits including spectral analysis and quantitative analysis/
modeling.
3.8.2.1.2 Satellite images
Satellite images are available from a multitude of sensors
and can be used for geomorphological mapping, although
their formal utility for specific types of landform mapping
has not yet been formally defined in digital geomorphological
mapping (Bishop et al., 2012). In comparison to aerial
photographs, they have the advantage of covering large areas
(typically 100 100 km), allowing rapid, systematic, and low-
cost mapping (Millington and Townshend, 1987; Walsh et al.,
1998; Smith and Pain, 2009). They are ideal for mapping large
areas at a wide-range of scales, which can elucidate landforms
(commonly large-scale features) that may be unidentifiable
using traditional field methods or aerial photographs
(Figure 3). Some satellite data are now easily obtained from
on-line sources (e.g., http://www.landcover.org).
The potential of using satellite images for geomorpho-
logical mapping was first realized through the initiation of the
Landsat program in the 1970s, as Landsat images remain one
 Remote Sensing and GIScience in Geomorphological Mapping
of the most commonly used sources for mapping geo-
morphological features (Figure 4). Various platforms have
been launched since the inaugural Landsat mission and sen-
sors vary with respect to their spatial, spectral, and radiometric
resolutions. Such parameters determine the utility of imagery
for mapping landforms (Table 1). Satellite images typically
have a spatial resolution from 5–120 m, with some imagery
comparable to aerial photographs (e.g., IKONOS, Quickbird,
WorldView, and GeoEye), though coverage and cost often
limit their use for detailed landform mapping. Image charac-
teristics have different merits and limitations, so to ensure
comprehensive mapping, more than one type of imagery is
commonly used within a single study (i.e., multisensor
imagery).
Figure 3 Advanced Spaceborne Thermal Emission and Reflection
Radiometer (ASTER) False-color composite image showing the linear
dunes in Rub’ al Khali, or the Empty Quarter, of Saudi Arabi. The
dunes appear yellow due to the presence of iron oxide minerals,
whereas the inter-dune areas are comprised of clays and silts, which
appear bright due to a high reflectance in ASTER’s band 1. Image
covers an area of 37 km by 28 km and is from NASA/GSFC/METI/
ERSDAC/JAROS, and U.S./Japan ASTER Science Team (http://
asterweb.jpl.nasa.gov).
Figure 4 Mount McKinley (Alaska) viewed using a Landsat ETM þ false-color composites (FCC) using bands 5, 4, 3 (a) and 4, 3, 2 (b). Ice
surface area, crevasses, lateral moraines, and valley characteristics are easily detected from FCC images of glaciated terrain.
191
3.8.2.1.3 Digital elevation models
The Earth’s topography is optimally represented using digital
elevation models (DEMs), that are derived from a variety of
data sources, including terrestrial surveying, laser scanning,
topographic maps, interpretations from stereographic imagery,
and radar measurements (Raseman et al., 2004). Morpho-
logical information (e.g., slope, slope azimuth, curvature) is
fundamental for identifying and mapping landforms, and
many new terrain analysis approaches permit unique charac-
terization of the topography. Consequently, geomorphometry
is considered by some to be superior to satellite image inter-
pretation and analysis (e.g., Clark, 1997), although surface
compositional information is best obtained from spectral
analysis. Such assessments are the result of the need for in-
ferential interpretation strategies due to the disconnect be-
tween land-cover patterns and landforms, and our inability to
effectively utilize biophysical information to describe land-
form compositional variations in some environments (Bishop
et al., 2012). Their use has flourished partly because of the
development of new methods of analyzing the elevation of the
Earth’s surface; in fact, the increased availability and accessi-
bility of DEMs has perhaps been the greatest advance in data
collection within the field of mountain geomorphology over
the past decade. With world-wide coverage and rapidly im-
proving measurement scales, the production, analysis, and
visualization of DEMs have become a standard practice in
landform mapping.
In general, geomorphologists use DEMs to analyze continu-
ous land-surface morphometric conditions (i.e., land-surface
parameters) or partition landscapes into discrete surface features
(Pike et al., 2009; i.e., land-surface object). This quantitative
characterization of the topography is referred to as geomor-
phometry, a multidisciplinary field of study that has produced
an immense amount of data and information regarding
processes, landform morphometry, and landscape evolution
(Figure 5). The very rapid development of sensors, GIS, and new
advanced algorithms to characterize land-surface parameters and
objects complicates landform research, but also offers endless
opportunities to merge knowledge from different domains
(e.g., software, mathematics, engineering, geomorphology) to
study landscapes in a more scientific manner.
192 Remote Sensing and GIScience in Geomorphological Mapping

Table 1 Summary of available satellite sensors used in landform mapping

Satellite Sensor Bands Spatial resolution (m) Spectral range (mm)

Landsat 1–3 MSS 4 80 VNIR

Landsat 4, 5 TM 6 30 VNIR, SWIR
1 120 TIR
Landsat 7 ETM 1 15 VNIR
6 30 VNIR, SWIR
1 60 TIR
EO-1 Hyperion 220 30 Hyperspectral
Ali 1 10 Panchromatic
9 30 VNIR
SPOT 1–3 HRV 1 10 VNIR
3 20 VNIR
SPOT 4 HRVIR 1 10 VIS
4 20 VNIR, SWIR
SPOT 5 HRG 1 5, 2.5 VIS, VNIR
3 10 VNIR
IKONOS Panchromatic 1 1 VNIR
MSS 4 4 VNIR
Quickbird Panchromatic 1 0.61 VNIR
MSS 4 2.44 VNIR
Terra ASTER 3 15 VNIR
6 30 SWIR
5 90 TIR
Geoeye HRG 1 0.4 Panchromatic
4 1.65 VNIR
Envisat ASAR 2 30 Microwave
ERSI/2 1 30 Microwave
TerraSAR-X 1 1,3,18 Microwave

geophysical data into this category, as gravity, aeromagnetic,

electrical, and seismic information can provide valuable in-
formation about near-surface composition and structure.
These datasets may be compared to mapping results using
spatial overlay techniques to ensure that each feature has been
interpreted correctly with respect to identification and
delineation.

Figure 5 Negative-openness image generated from Shuttle Radar
Topography Mission (SRTM) elevation data for the Karakoram
Himalaya that highlights the hydrological network. Terrain openness
is an angular measure of the relation between surface relief and
horizontal distance (in negative openness, bright areas correspond to
valley floors).
3.8.2.1.4 Supplementary data
When generating landform maps from aerial photographs,
satellite images, and DEMs, it is generally necessary to con-
sider additional data sources to provide some means of
evaluating mapping accuracy (e.g., Greenwood and Clark,
2008; Hughes et al., 2010). Such supplementary data might
include topographic maps and Quaternary geology maps that
distinguish sediment types from exposed bedrock, solid, and
structural geological datasets, permafrost maps, and land-
cover and land-use data. It is also important to include
3.8.2.2 GIScience
Geographic information science refers to the conceptual and
technological issues associated with the representation, an-
alysis, modeling, and visualization of space-time data and
information (Goodchild, 1992a, b; Bishop and Shroder, 2004;
Bishop et al., 2012). In this context, GIScience represents
much more than just a cartographic tool to map landforms
(Wright et al., 1997). Rather, it represents an interdisciplinary
field of study focused on the development of the ‘philo-
sophical, cognitive, and scientific treatments of spatial theory
and concepts that can be used to address information science-
and Earth- science-related problems’ (Bishop and Shroder,
2004). These objectives align with many key concepts and
issues in geomorphology and landform mapping, including
space-time and multidimensional representation, preservation
of spatial properties and relationships (i.e., topology), and the
development of empirical and predictive models (Napieralski
et al., 2007a, b). It is not unreasonable to assume that the
theoretical developments occurring in the rapidly growing
 Remote Sensing and GIScience in Geomorphological Mapping 193

field of GIScience will contribute to a more efficient and in-
tegrated treatment of spatial theory, advance data assimilation,
and analytical capabilities, and facilitate new approaches to
visualization that will permit more scientific treatments of
geomorphological mapping, as described in Bishop et al.
(2012).
Conducting terrain analyses and delineating landforms
into spatial entities is not without challenges, especially when
relying on GIT. The demarcation of a landform boundary
should theoretically rely on the principle of homogeneity of
spatial entities, which states: ‘borders are expressions of
maximizing internal homogeneity while maximizing external
heterogeneity’ (Eisank et al., 2010). These circumstances,
however, are not always realistic, as there are a number of
factors that influence the capability to identify landform
boundaries, including topographic relief (i.e., landforms tend
to be easier to extract with higher topographic relief), land-
form properties (e.g., size, shape, relative relief), scale, differ-
ences between landform orientation and illumination
orientation associated with remote-sensing-derived data (i.e.,
azimuth bias), and the application of knowledge-based clas-
sification rules (Smith et al., 2001; Eisank et al., 2010). In fact,
landform boundaries are not characteristically discrete; rather
they are transitional. So overcoming these limitations require
an established, universally accepted set of classification rules,
in addition to an interdisciplinary integration of spatial in-
formation designed to approach landform mapping from a
rigorous, but also realistic, perspective. As an example, fuzzy
logic can be used to represent landform boundaries as spatial
fuzzy uncertainty approximations, rather than discrete, exact
boundaries. Many fuzzy techniques have recently contributed
to landform segmentation and classification (Burrough et al.,
2000; MacMillian et al., 2000; Fisher et al., 2004; Mark and
Smith, 2004; Moody and Katz, 2004). They are considered
effective for mapping landforms that have transitional
boundaries. Conversely, object-orientated analysis involves
characterizing each landform as a discrete component in a
complex system, and this is ideal for investigations related to
process mechanics and space-time representation (Drăgut- and
Blaschke, 2006; Drăgut- et al., 2009; Anders et al., 2009;
Ghosh et al., 2010; Eisank et al., 2010), although fuzzy
boundaries could be represented as a discrete object. The
framework for characterizing and measuring landforms con-
tinues to evolve, but it progresses faster when geomorph-
ologists integrate principles and concepts that are central to
the underpinnings of GIScience theory.
The ability to resolve landform entities and understand
landscape evolution is largely controlled by operational scales
(i.e., spatio-temporal scales at which surface processes oper-
ate). Spatial scales range from picorelief forms (e.g., glacial
striations) that measure millimeters in length and width to
megarelief (e.g., continental) features (Table 2; Dikau, 1989).
The spatial scale-dependency of landform entities within a
landscape signifies the presence of ‘nested-form assemblages
that are organized in a spatial hierarchy’ (Raseman et al.,
2004). Therefore, scale-dependent analysis is vital in deter-
mining landform boundaries (based on the principle of
homogeneity). Software tools, such as the Estimation of Scale
Parameters (ESP), are designed specifically for finding the
appropriate segmentation scales (Drăgut- et al., 2009). Further
complicating landform identification and extraction is the fact
that landforms evolve over varying lengths of time, which
frequently leaves behind a palimpset (i.e., smaller landforms
superimposed on larger features, younger modifying older)
landscape (Chorley et al., 1984; Brunsden, 1993; Raseman
et al., 2004). Therefore, the spatio-temporal characterization
of landforms require objective analysis with repeatable
results (Table 2). This can potentially be accomplished using
advances in GIScience and well-defined taxonomic schemes
developed by geomorphologists to reduce subjectivity and
facilitate landform classification.
The importance of understanding landform patterns is re-
flected in the wide interest in segmenting landscapes into dis-
tinct units. Landform patterns influence the flow of sediment,
nutrients, and water (e.g., Dalrymple et al., 1968), create cor-
ridors that impact biodiversity and environmental quality
(Hunsaker et al., 1992; Gordon et al., 1994; López-Blanco and
Villers-Ruiz, 1995; O’Neill et al., 1997; Butler, 2001), and limit
the spatial distribution of geologic resources. Of interest to
geomorphologists is the linkage between landform patterns and

Table 2 A classification of geomorphological features by spatio-temporal characteristics

Order Classification Area (km2) Time (yr) Landforms

8 9
0 Megarelief 5.098  10 5  10 Lithosphere, cryosphere, hydrosphere, atmosphere, ecosphere
1 107–108 109–108 Continents, ocean basins, tectonic plates
2 Macrorelief 107–105 108 Physiographic provinces, shields, large volcanoes, mountain ranges
3 105–103 108–107 Medium-scale tectonic units, mountain massifs, fault blocks, grabens
4 Mesorelief 103–102 107 Small-scale tectonic units, fault blocks, sackung
5 Microrelief 102–101 107–106 Large-scale erosional/depositional landforms, major valleys, piedmonts, deltas, landslides
6 101–101 106–105 Medium-scale erosional/depositional landforms, cirques, moraines, floodplains, alluvial fans
7 Nanorelief 101–103 105–104 Small-scale erosion/depositional landforms, ridges, terraces, dunes, slump blocks, talus
8 Picorelief 10–3–105 104–103 Larger geomorphic process units, hillslopes, stream channel reaches, talus, small debris flows
9 10–5–107 103–102 Medium geomorphic process units, riffles/pools, river bars, slope facets, solution pits, gullies
10 10–7–109 o102 Smaller geomorphic process units, ripple marks, glacial striae, rills, raindrop impact pits
11 10–9–1012 Clast grain morphologies, clay mineral structure

Source: Modified from Shroder, Jr. J.F., Bishop, M.P., 2004. Mountain geomorphic systems. In: Bishop, M.P., Shroder, J.F. (Eds.), Geographic Information Science and Mountain
Geomorphology. Springer-Praxis, Berlin Heidelberg, pp. 33–66, and Raseman, S., Schimdt, J., Schrott, L., Dikau, R., 2004. Geomorphometry in mountain terrain. In: Bishop, M.P.,
Shroder, J.F. (Eds.), Geographic Information Science and Mountain Geomorphology. Springer-Praxis, Berlin, pp. 101–137.
194 Remote Sensing and GIScience in Geomorphological Mapping
process mechanisms or landscape evolution. Pattern recog-
nition techniques range from more tedious manual and semi-
automated approaches (digitizing boundaries of images before
comparing properties to neighboring features), to more ad-
vanced methods that are fundamental in computer science. In
particular, artificial intelligence (AI) has improved the decision-
making process for detecting patterns and spatial correlations
between landforms. The process by which landforms are
identified or patterns recognized (i.e., decision rules) in a vir-
tual system can be designed to simulate the process used by the
human brain (i.e., artificial neural networks). Input data may
include landform geometry, size, texture, curvature, or even
spatial topological characteristics. Thus the semantic modeling
of landforms (i.e., the relationship between computer repre-
sentations of landforms and real-world features) requires
comprehensive formalization of landform properties to enable
landform classification (Bishr, 1998; Eisank et al., 2010). For
example, Eisank et al. (2010) provide five steps used for se-
mantics-based landform classification: (1) identification of
landforms observable in DEM; (2) conceptualization of land-
forms based on geomorphological knowledge; (3) scale de-
tection to define the threshold of the maximum increase in
heterogeneity (recall the principle of homogeneity of spatial
entities); (4) semantic modeling, which is how the geomorphic
concepts are represented within the computer; and (5) software
implementation, such as the use of object-based image classi-
fication (OBIA).
Once landform morphology and pattern are described and
understood, these results are commonly linked to surface pro-
cesses and feedback mechanisms to gain knowledge of land-
scape evolution. The physical characteristics of landforms can be
classified according to the dominant surface processes (Volker
et al., 2007; Perron et al., 2008). For example, Volker et al.
(2007) classified alluvial landforms as either dry debris flows, or
fluvial (wet) sediment transport based on local-relief, proving
that higher local-relief most likely resulted from a dry debris
flow. These results can be used to link topographic features
(i.e., fingerprints) to dominant surface process. Finally, geo-
morphologists have evaluated the factors that potentially con-
trol landscape evolution (Bishop et al., 2003; Walcott and
Summerfield, 2008). As expected, landscapes (e.g., mountain,
hillslopes) are formed by scale-dependent interactions between
surface processes, climate, and tectonics. Eliciting these inter-
actions is key to understanding some of the most classic prob-
lems pursued by geomorphologists, including quantifying the
driving and resisting forces behind topographic production and
modification over varying temporal and spatial scales.
Finally, Walsh et al. (1998), Bishop and Shroder (2004),
and Bishop et al. (2012) summarized several GIScience issues
and geospatial concepts relevant to landform and geomorphic
studies that include:
• Viewing landscapes through an array of spatial, spectral, and
temporal perspectives.
• Issues of scale related to surface processes, feedback mech-
anisms, system dynamics, and landform features.
• Exploration of data through a variety of attribute and geo-
graphic operations.
• Conversion of data to information, including thematic
layers and composites.
• Linkage to causal and multivariate models and pattern
indices.
• Efficient display and dissemination of data and maps (e.g.,
cartographic modeling).
3.8.2.2.1 Mapping approaches
Traditionally, landform mapping required intensive fieldwork,
qualitative interpretations of aerial photography and topo-
graphic maps, or quantitative assessments using photogram-
metric techniques. For example, some early landform
mapping techniques relied on extracting and measuring
landforms using contour lines off topographic maps, fre-
quently to indicate the boundary of a landform (e.g., lowest
enclosed contour indicated boundary of glacial landform).
However, this was time-intensive, limited to map scale and the
horizontal and vertical error, and was subjective and relatively
inconsistent. The rapid proliferation of GIT has created an
opportunity to integrate data from various sources, to conduct
objective, reproducible studies, and to increase the efficiency
of landform identification, extraction, and measurement.
Clark and Wilson (1994) noted that automated techniques
associated with computer programming and GIS could dras-
tically increase the efficiency of landform mapping relative to
traditional methods (e.g., they could measure 4800 features
using automated methods, and acknowledged that in the
same time, only 50–1600 might be measured using traditional
techniques).
Landforms can be mapped using either field or entity data
models. Raster-based DEMs can be used to generate land-
surface parameters, and topographic information can be
overlayed with other raster layers to assign raster cells to a
particular landform class. In contrast, landforms can be
mapped using points (indicating landform location), lines
(the location and orientation of linear or nonlinear land-
forms), or polygons (demarking the landform boundaries that
permit efficient calculations of area or perimeter). The type of
data structure used to represent a landscape or individual
landform units is influenced by a desire to have boundary-
based representation (i.e., vector) or accept topographic
gradation (Plewe, 1997). This ontological predicament is
critical in geomorphology. As an example, Mark and Smith
(2004) noted that mountain peaks are typically marked at the
highest point, but rarely are they illustrated with boundaries
due to the challenge of identifying where mountains and
peaks begin and end. It is clear that spatial scale and
perception (e.g., structural geologist vs. geomorphologist) in-
fluence how and where landforms are delimited (Deng, 2007),
and one way around this problem is to develop multiscale,
quantitative rules for landform identification. As an example,
Lucieer and Stein (2005) defined mountain peak member-
ships by characterizing multiscale topographic variance, lead-
ing to the definition of peak prototypes and multiscale peak
memberships (Deng, 2007).
3.8.3 Glacial Landscapes and Landforms
Investigations of glacier and ice-sheet extent, volume, and
behavior frequently require spatial data covering large areas,
 Remote Sensing and GIScience in Geomorphological Mapping
high latitudes, or high altitudes. Additionally, the surface
topography of glaciers adjusts relative to the glacial mass
balance, which can occur rapidly in some situations, thus re-
quiring the acquisition of spatial data on a relatively short
time interval. As a result, the acquisition and dissemination of
spatio-temporal glacial data is central to the productivity and
effectiveness of local and global glacial surface and landform
studies. As an example, the Global Land Ice Measurements
from Space (GLIMS) project (Bishop and Shroder, 2004;
Kargel et al., 2005) involves assessing and mapping the Earth’s
glaciers and provides access to spatio-temporal data through
either a web-based GIS server or a traditional text-based
interface that allows for more advanced searches. Organiza-
tions such as GLIMS increase interdisciplinary collaborations
to monitor and map glacier surface characteristics, but they
also reduce discrepancies in data format, minimize geometric
distortion between datasets, and augment the development of
universally accepted taxonomic schemes.
Mapping glaciers and glacial landforms is one of the key
approaches to reconstructing the extent and dynamics of for-
mer ice masses (Andrews, 1982), and has been the focus of
research for well over a century (Close, 1867; Charlesworth,
1928). Studies have traditionally involved field-based analyses
of landforms such as glaciers, drumlins, moraines, and eskers,
but are now commonly based on remote-sensing studies
(Boulton and Clark, 1990; Bishop and Shroder, 2004; Smith
et al., 2006). In fact, remote sensing investigations have been
so successful that some consider them to have largely super-
seded the ‘traditional’ (field-based) approach (e.g., Smith
et al., 2006). Remote sensing has been particularly instru-
mental in acquiring large inventories of glacial landforms to
reconstruct ice-sheet extent and regional flow characteristics
in the British Isles (Clark and Meehan, 2001; Jansson and
Glasser, 2005; Bradwell et al., 2008), North America (Boulton
and Clark, 1990; Clark et al., 2000; Jansson et al., 2002;
Kleman et al., 2002; Stokes and Clark, 2003), and Scandinavia
(Punkari, 1993; Kleman and Borgström, 1996; Kleman et al.,
1997; Hättestrand and Stroeven, 2002).
Some of the limitations inherent to the field-based ap-
proaches are that it is often expensive, time-consuming, and
the data produced generally relate to small areas – leading to
difficulties with assimilating information to derive a country-
or region-wide assessment or overview (see Clark et al., 2004;
Evans et al., 2005). The remote-sensing approach, by contrast,
provides a time- and cost-effective way for a single user or user
group to systematically map large areas, some of which are
remote and inhospitable, at a range of scales, while main-
taining consistency in mapping style (Punkari, 1982; Clark,
1997). It is then possible to collate these data in a GIS together
with other spatial data (e.g., soils, geological maps). The
synoptic coverage of remote-sensing data also reveals some
landforms (e.g., megascale glacial lineations (MSGLs)) and
patterns that are not easily identifiable by traditional means
(Clark, 1993, 1997).
The above advantages clearly make remote sensing critical
for mapping glacial geomorphology, and one that has de-
veloped rapidly over recent years. The intention here is to focus
upon the data sources and data processing and visualization
methods most widely used when employing a remote sensing
and GIS approach to the mapping of glacial landforms.
195
3.8.3.1 Glacial Landforms
Mapping landforms with the intention of understanding
glacierization and glaciation relies upon the identification and
interpretation of field evidence indicating ice geometry
and thickness. Ice-marginal landforms such as terminal, end,
and lateral moraines are commonly used to infer the hori-
zontal dimensions of former ice masses, though the extent of
till may also be used for this task. The vertical dimensions
of former ice masses are generally more difficult to constrain
(see Ng et al., 2010), though trimlines, where identifiable
(and this is characteristically difficult), are used in some cases,
but not unambiguously (e.g., Ballantyne, 1997). Bedforms
(i.e., features formed beneath an ice mass) such as drumlins,
crag-and-tails, flutes, and ribbed moraines are used as indi-
cators of former ice-flow direction and velocity, and also
provide information on former ice mass extent and evolution.
Robust paleo-ice sheet reconstructions therefore require a GIS
to integrate and map cross-cutting sets of bedforms of varying
scales and from a wide range of sources before any interpret-
ations can be considered.
3.8.3.2 Data Processing and Visualization
Techniques for processing and visualizing remotely sensed
datasets in order to maximize their effectiveness for mapping
glacial landforms have been reviewed by Clark (1997) and
Smith and Clark (2005). These relatively simple techniques
are generally implemented in a GIS and, when judiciously
applied, can increase the detectability of glacial landforms
significantly (Smith et al., 2000; Jansson and Glasser, 2005).
3.8.3.2.1 Aerial photographs
Methods of mapping glacial landforms from aerial photo-
graphs have varied very little since the first use of these
data (Woodruff and Evenden, 1962; Aniya and Welch, 1981;
Lindner et al., 1985), and a qualitative approach, whereby
features are manually mapped either in a GIS or directly onto
aerial photographs, remains widely used. Relative to satellite
imagery, the revisit period, swath width, spatial resolution,
and variable flying altitude associated with aerial photography
is more flexible and allows for a customized perspective of
small, localized glaciers, and glacial landscapes (Gao and Liu,
2001). Despite the advantages of using aerial photography,
there are still few established rules for interpreting glacial
landforms and features, especially since some landforms may
be extracted from physical copies of the aerial photo (or in
combination with a topographic map), whereas others may
use on-screen digitizing with a GIS (Hubbard and Glasser,
2005). Additionally, interpretations derived from analog
photographs (e.g., scanning, then digitizing) can be laborious
and the issue of distortion in high relief areas is far more
prevalent with aerial photography than satellite imagery
(Sidjak and Wheate, 1999).
Identifying glacial landforms in such photographs depends
upon obtaining largely cloud-free, high-quality images, ac-
quired during periods of low solar elevation, as many glacial
features (e.g., drumlins) are identified by a break-of-slope,
which is easier to identify with relatively high solar zenith
angles (Clark, 1997). Glacial landforms mapped and extracted
196 Remote Sensing and GIScience in Geomorphological Mapping
from mosaics of aerial photos lend support to paleoglaciolo-
gical reconstructions of basal conditions, flow direction, and
extent, and include erosion (e.g., polished bedrock, melt-water
channels, cirques) and depositional (e.g., moraines, glacio-
fluvial deposits) features (Sahlin and Glasser, 2008). A com-
mon approach to managing such large assemblages of data is
to segregate the landform data into ‘flow sets’ that act as
building blocks in the reconstruction of previous ice sheets
and glaciers.
Although some of the advantages of using aerial photog-
raphy in glacial studies discussed earlier have diminished due
to improved satellite sensor capabilities, aerial photos are
valuable when monitoring and mapping glacier extent and
glacial landform distribution. Fluctuations in glacier extent
(e.g., Field, 1966; Meier, 1966; Petrie and Price, 1966; Espizúa,
1986), surface motion (Evans, 2000), and crevasse migration
(Haefeli, 1966; Harper et al., 1998; Aniya et al., 2002) have
relied on repeat aerial photography due to relatively high
spatial and temporal resolutions. Aniya and Welch (1981)
utilized photogrammetric techniques to analyze glacial cir-
ques, such that the crest line and lowermost limit, gradient,
aspect, and shape of cirques were compiled and compared
between glaciated regions to link the topographic, climatic,
and tectonic conditions that influence the evolution of glacial
cirques (Federici and Spagnolo, 2004). Mosaics of aerial
photos also have offered confirmation or rejection of inter-
pretations derived from satellite data. Clark (1993) compiled
glacial lineation data from over 75 000 aerial photographs and
Landsat MSS images to analyze the spatial pattern of glacial
landforms (i.e., flutes, drumlins, megaflutes) to assess possi-
bilities for subglacial landform evolution.
3.8.3.2.2 Satellite images
As with aerial photographs, satellite images should be cloud-
free, and acquired with comparatively low solar elevations,
ideally below 201 (Smith et al., 2006). In some regions, this
necessitates obtaining images with a winter acquisition date,
because sensors such as Landsat always acquire data for a
given region at a particular time of day. Obtaining images with
low solar elevations is easier in the high-Arctic, as any images
with a summer acquisition date are likely to meet this cri-
terion, but at lower latitudes obtaining cloud-free data cap-
tured during periods of low solar elevation is difficult (partly
because satellite images cover large areas), and concessions
must commonly be made (e.g., using inferior data). Another
approach employed to enhance glacial landform identification
from satellite images is to view data acquired during a period
of uniform snow cover (e.g., Eyton, 1989; Skoye and Eyton,
1992). As emphasized by Clark (1997), however, this is an
approach rarely and opportunistically applied, as cloud-free
images, with uniform snow-cover, but which also have suf-
ficient illumination levels, are scarce.
Once suitable satellite images have been acquired, pro-
cessing prior to visualization is a comparatively straight-
forward task. The approach advocated by Clark (1997) is to
select a subset of an image, covering a region where the dis-
tribution of glacial landforms is well known, then experiment
with image-enhancement techniques, such as contrast
stretching and histogram equalization, until the most useful
enhancements for landform detectability are found. These
techniques can then be applied to the remainder of the sat-
ellite image if the global probability distribution does not
adversely affect the results. Some have also applied high-pass
filtering to further enhance landform boundary clarity, but
with varying degrees of success (Jansson and Glasser, 2005),
as this can also produce image artifacts.
The most widely used satellite images for glacial landform
mapping, Landsat 7 ETM þ , come in a variety of spectral
bands, ranging in wavelength from 0.45 mm (visible blue) to
12.50 mm (thermal infrared). These images are typically
viewed as false-color composites (FCCs) using a variety of
band combinations (Jansson and Glasser, 2005), or as hybrid
(pansharpened) images with information in a color composite
transferred onto the higher resolution panchromatic image
(e.g., Greenwood and Clark, 2008; Heyman et al., 2008).
The optimum combination of bands within a FCC varies
with season and location, but as vegetation (type and quan-
tity) is generally an important factor in distinguishing glacial
ice and landforms from the surrounding landscape, ETM
band 4 is generally used, as this is the near-infrared waveband
which allows differences in vegetation to be best detected, and
has limited atmospheric effects (Figure 6; Clark, 1997).
Using Landsat 7 ETM þ images, Jansson and Glasser
(2005) compared the panchromatic band 8, different FCC
combinations (RGB bands 4-3-2, 7-4-2, and 4-5-6), and the
Ordnance Survey Panoramas digital terrain model (DTM,
50 m spatial resolution) to assess which is best suited to the
compilation of regional-scale maps of glacial lineaments in
Wales and similar environments. They displayed all FCC im-
ages, and establish band combinations on initial experiments
to determine the suitability of different combinations. They
found that only 9% of lineaments within their study area were
identified when the panchromatic band alone was considered;
25% were identified when the FCC combination 4-3-2 alone
was considered; 22% with the FCC combination 7-4-2 alone;
36% with the FCC combination 4-5-6 alone; and 29% when
the DEM alone was considered. The number of features
identified increased significantly when FCC images were
overlaid on the DEM, with the FCC combination 4-5-6
proving the most effective. On this basis, Jansson and Glasser
(2005) recommended that to obtain the best possible results
when mapping glacial lineaments in this type of environment,
a FCC Landsat 7 ETM þ combination 4-3-2 should be draped
over a DEM. In addition, they suggested viewing a non DEM-
draped FCC combination 4-5-6. This work demonstrates the
value of image enhancement, data fusion, and visualization
for mapping, as it has been found to be effective elsewhere
(Greenwood and Clark, 2008).
Satellite imagery permits the calculation of surface area
(Aniya et al., 1996; Sidjak and Wheate, 1999), measurements
of surface motion (Bindschadler and Scambos, 1991; Scambos
et al., 1992; Whillans and Tsend, 1995), and ice-snow surface
classifications (Bronge and Bronge, 1999). Image processing
techniques such as principal component analysis (PCA) and
the normalized difference snow index, reduce data redun-
dancy, and increase capabilities to distinguish changes in ice
surface characteristics, respectively. Sidjak and Wheate (1999)
analyzed Landsat TM scenes of the Illecillewaet icefield
(British Columbia) using PCA, image ratioing, and image
differencing to produce a land-cover classification map
 Remote Sensing and GIScience in Geomorphological Mapping
177°0′0″E 178°0′0″E
66°0′0″N
(a)
65°0′0″N
177°0′0″E 178°0′0″E
Figure 6 Moraines in the Anadyr Lowland of Eastern Siberia, depicted (a) in a Landsat ETM þ false-color composite image (FCC; 5, 4, and 2),
and (b) using ASTER GDEM data. This is a region outside that covered by SRTM data.
separating the icefield into subsets of glaciated and non-
glaciated terrain. It was also ascertained that the lower order
components of the analysis likely indicated ice-surface topo-
graphy. The assessment of sequential satellite images can also
reveal the ice-flow velocity fields, especially with surging ice or
active alpine glaciers (e.g., Scambos and Bindschadler, 1993).
3.8.3.2.3 Digital elevation models
When mapping using topographic information, the user
generally has little choice regarding acquisition date or con-
ditions. This is less important than with aerial photographs or
satellite images, however, as GIS-based simulations of solar
geometry can be used to produce shaded-relief maps that
highlight the topographic structure. For example, it has been
found that such illumination variations improve feature
detectability (see Clark and Meehan, 2001; Smith and Clark,
2005). More specifically, varying the solar azimuth can high-
light features that might otherwise remain undetected (e.g.,
Clark and Meehan, 2001; Greenwood and Clark, 2008), with
low (o201) solar elevations typically maximizing detectability
(Smith et al., 2001; Jansson and Glasser, 2005). Thus, DEMs
have been successfully used to visualize glacial landforms
based upon relief shading, with solar illumination geometry
defined by the user (e.g., Pike, 1992). Other visualization
methods include local contrast stretching and enhancing the
vertical exaggeration of altitudes, as well as examining slope
gradient and various curvature metric variations (Clark and
Meehan, 2001; Smith and Clark, 2005; Smith et al., 2006).
Smith and Clark (2005) analyzed 12 separate DEM visual-
ization techniques and found no single method to allow
complete and unbiased mapping of glacial lineaments. They
recommended that initial mapping should be conducted by
197
177°0′0″E 178°0′0″E
66°0′0″N
(b)
65°0′0″N
177°0′0″E 178°0′0″E
viewing relief-shaded visualizations illuminated from two
different azimuths (preferably orthogonal to one another). In
fact, in order to minimize illumination bias some have used
images illuminated from three different locations, commonly
including one with the sun directly overhead (e.g., Hughes
et al., 2010).
There are numerous DEM analyses that would be of
interest to glacial geomorphologists, including some that
have been developed for other purposes, but modified spe-
cifically for use in glacial landform mapping. For example,
Duncan et al. (1998) modified drainage extraction techni-
ques to estimate modern and historic glacier extents in
central Nepal. They computed hillslope angles to generate
links to define network channels and determine morpholo-
gic parameters of valley form (e.g., distinguishing between
U- and V-shaped valleys using flatness of valley floor or
curvedness of valley walls). This was estimated in part using
the flow accumulation scheme frequently used to delineate
watersheds and stream networks and to simulate runoff. In
other circumstances, DEMs contribute to landform recog-
nition and identification, even if it is a reevaluation of
previously identified features. For example, Moraines have
been historically mapped with some subjectivity and com-
bining high-resolution DEMs and supplemental data can
reduce the amount of bias in detecting moraine crests versus
moraine outlines (e.g., Clark et al., 2004). Morphological
features can also be highlighted using terrain exaggeration
of shaded relief derived from LiDAR, which filters out
vegetation to produce a high-resolution DEM (Salcher
et al., 2010). Mapping minor topographic variations in al-
pine environments using LiDAR produces a more complete
picture of landform morphology, which can potentially be
used to generate information regarding landform genesis
198 Remote Sensing and GIScience in Geomorphological Mapping
(e.g., sequence of proglacial outwash plains subsequently
covered by ice due to glacial advances).
Mapping the topographic characteristics of glacier sur-
faces also necessitates DEMs derived from satellites with
worldwide coverage, or those designed specifically for ice
surface characterization. The Ice, Cloud, and Land Elevation
Satellite (ICESat) (National Aeronautics and Space Adminis-
tration – NASA) and CryoSat (European Space Agency – ESA)
both provide radar altimetry data of Earth’s polar and glaci-
ated regions that can be used to map ice sheet topography
(Zwally et al., 2002; Drinkwater et al., 2003). Changes in land-
surface elevation reflect changes in glacial mass balance con-
ditions, and multitemporal DEM analysis facilitates estimates
of regional mass balance. Data from ICESat, CryoSat, ASTER,
and the Shuttle Radar Topography Mission (SRTM) can be
used. For example, Sauber et al. (2005) correlated changes in
surface ice elevation with rates of ice thinning for the
Malaspina Glacier, Alaska using ICESat and SRTM data. Fur-
thermore, multitemporal synthetic-aperture radar (SAR)
interferometry techniques are also used to measure surface ice-
flow velocities (Goldstein et al., 1993; Joughin et al.,
1995,1996; Bamber et al., 2000; Li et al., 2008a, b; Koehl
et al., 2009), and even estimate three-dimensional (3D) vel-
ocity fields (Joughin et al., 1998; Mohr et al., 1998). In both
approaches, data representing surface elevation and ice
thickness are vital to the production of accurate surface ice-
flow velocity maps (Høgda et al., 2010). Even airborne laser
scanning can produce surface elevation approximations and
velocity fields, detect crevasses, distinguish between glacier
surface types, and map glacier surface roughness to use as
input for surface energy-balance modeling (e.g., Geist and
Støtter, 2010).
The influence of DEM grid-cell size on glacial landform
mapping accuracy has recently been tested. Using statistical
process control charts (SPCC), Napieralski and Nalepa (2010)
found an optimum grid cell size to delineate and measure
drumlin length, width, and orientation (and thus area,
elongation, Rose curve). Although the drumlin sample size
was relatively small, the results indicated that the use of a
30 m resolution DEMs caused a significant loss in drumlin
count, size, and shape (Figure 7). The authors warned, how-
ever, that higher resolution does not necessarily equate to
increased accuracy, as there was no statistically significant
difference between morphological characteristics calculated
from a 10 m versus a 1 m DEM. Therefore, the combination of
DEM resolution and landform size and shape represents a
mapping issue.
3D perspectives represent another way of visualizing image
and topographic information. Visualizations based on ele-
vation allow the topographic-context of landforms to be better
accounted for, although direct digitizing from oblique views is
difficult, and the process does not lend itself to rapid mapping
(Smith and Clark, 2005). Morén et al. (2011) used Google
TM
Earth to visualize 3D glacial landforms (i.e., valleys, mar-
ginal moraines, lineations, and hummocky terrain) generated
from a SRTM DEM and supplemental data (Figure 8). Their
ability to distinguish between glaciation and nonglaciated
terrain around the periphery of the Tibetan Plateau contrib-
utes to the belief that glaciers have had limited extent, and
thus influence, on landscape evolution there.
3.8.3.3 GIS-based Mapping
The mapping of glacial landforms from remotely sensed
sources typically involves on-screen digitization of features
using a digital-image processing system or a GIS. Features are
usually mapped as points, lines or polygons depending
upon the feature type and the scale of the resulting map
relative to the size of the mapped features (Smith et al., 2001;
Greenwood and Clark, 2008). Emphasis is commonly placed
on mapping the crest-lines of individual landforms or out-
lining the break-of-slope around their margins (e.g., Hughes
et al., 2010). Mapping is conducted at a range of scales, again
depending upon the data- and landform-type in question.
A repeat-pass approach is commonly adopted, whereby each
region is viewed on numerous occasions at a range of scales,
preferably using various data sources (Greenwood and Clark,
2008; Hughes et al., 2010). Systematic and comprehensive
mapping may be performed in a comparatively rapid and
cost-effective manner. This approach, however, depends to a
large extent on the observer’s ability and experience (Clark,
1997; Smith et al. 2001), representing a subjective and time-
consuming approach.
GIS-based research focuses on the development and
evaluation of automated, or semiautomated techniques.
Automated techniques, including pattern-recognition ap-
proaches, have been used to extract geomorphological infor-
mation directly from remotely sensed data and DEMs, but
despite improvements in the sophistication of pattern-recog-
nition algorithms and performance, these methods do not
formalize landform classification and are not universally ap-
plicable, as results typically depended upon data selection and
the generation of an appropriate feature space, given the
landforms of interest. Consequently, reliance on the human-
visualization system and digitization continue to be favored.
Landform patterns are generally distinctive in previously
glaciation terrain, as analogous landforms tend to cluster as a
result of topography, subglacial hydrology and thermal con-
ditions, and sedimentological characteristics. Similar land-
forms are commonly grouped according to morphological
criteria (e.g., size, orientation, shape indices) and these land-
form sets typically reflect subglacial conditions during for-
mation, acting as building blocks for inverse reconstructions
of previous glaciations (Figure 9). Many of these landform
sets contain digitized landforms that display distinct patterns,
such that their relative alignment within the set (i.e., parallel
conformity), the number of landforms per unit area (land-
form density), and surface area of landforms per unit area (i.e.,
packing) can be calculated and established as criteria for
identifying similar landforms elsewhere (Stokes and Clark,
2003). Dunlop and Clark (2006) digitized over 33 000 land-
forms with the objective of developing a clear definition of a
ribbed moraine using morphological characteristics (i.e., ridge
width, length, height, wavelength, shape), moraine accord-
ance (e.g., ridge summit heights within field, jigsaw match-
ing), and position of ridges relative to ice-sheet margin or
divide and topography. In particular, they repositioned a series
of digitized ridges (polygons) to test the theory that the
landforms were once connected, and subsequently split due to
overlying ice motion (Figure 10). The capability to move and
rotate polygons is efficiently accomplished using GIS and
 Remote Sensing and GIScience in Geomorphological Mapping 199

80 m 50 m 30 m 20 m 10 m 5m 1m

500

Meters

Figure 7 Drumlin shape and size are influenced by the resolution of a digital elevation model. Note only subtle changes occur between 1 and
20 m, but larger grid cells then influence landform characteristics at 30 m (e.g., Landsat TM) and coarser. Reproduced from Napieralski, J.A.,
Nalepa, N., 2010. The role of resolution and contour interval in drumlin delineation and measurement. Computers & Geosciences 36, 222–230.

results from morphological studies of glacial landforms, such
as this, will contribute to our understanding of landform-
process dynamics.
The dimensions and temporal nature of paleoice sheets is
determined through subglacial landform characterization.
Napieralski et al. (2007a) developed spatial analytical GIS
tools to assist in the systematic improvement and refinement
of numerical ice-sheet models through a comparison between
landform sets (i.e., lineations, end moraines) and model
output for the LGM in Scandinavia. The Automated Proximity
and Conformity Analysis (APCA) (Li et al., 2008a, b) was
designed to measure agreement between lineations and model
output indicating ice extent (commonly mapped and analyzed
in vector format), whereas the Automated Flow Direction
Analysis (AFDA) (Li et al., 2007) compared the orientation of
streamlined landforms (i.e., surrogate for ice-flow direction)
with simulated ice-flow direction (landform data converted to
raster format). In this example, the landform maps were de-
veloped in GIS and then compared against model data to elicit
the model parameters that generated ice-sheet configurations
that best matched field evidence. In addition to merging field
and model data to reconstruct ice-sheet extent and behavior,
geostatistical tools within GIS have also been utilized to re-
construct Quaternary landscapes using strandline (i.e., raised
shorelines) evidence (Leverington et al., 2002). Strandline
data can be interpolated from isobase data and then sub-
tracted from modern topographic and bathymetric data
to generate a rasterized approximation of paleotopography.
Leverington et al. (2002) noted the efficiency at which the
datasets could be generated using relatively basic GIS func-
tionalities, as well as the ability to integrate and visualize
extensive data for a large area of the Canadian Arctic.
3.8.4 Volcanic Terrain and Landforms
Volcanic landscapes are rapidly evolving geomorphologic
systems. During their growth, volcanoes undergo repeated
phases of eruptive activity associated with the emplacement of
different types of volcanic deposits such lava flows, ash-fall
deposits, and pyroclastic-flow deposits, causing progressive
evolution of edifices with varying shapes and heights. Con-
currently, emplacement of magmatic intrusions at depth
contributes to the deformation and uplift of the surface
topography. Eruptive activity and other volcano-tectonic
processes also lead at times to a rapid destruction of part of a
volcano through caldera formation or flank collapses. Gravi-
tational deformation and erosional processes will act at re-
ducing and dissecting the volcanic relief during and following
the period of active volcano growth. Changes in volcano
200 Remote Sensing and GIScience in Geomorphological Mapping

Figure 8 Landform category examples used to map Tibetan Plateau, including SRTM panels (left), Landsat 7 ETM þ imagery (middle), and
mapped landforms (right). All satellite images are false-color composites of bands 5, 4, and 2, draped with a semitransparent band 8. The DEM
panels are draped by a semitransparent slope image in panel (a) and by semitransparent hill-shade images in panels (b), (c), and (d). North is
toward the top of the maps and each map is 9  9 km. (a) Glacial valley in the upper left. (b) Marginal moraine deposited beyond the reaches of
a glacial valley (lower left corner) and an integral part of the hummocky terrain area. (c) Glacial lineations on the floor of glacial valleys. (d)
Hummocky terrain area. Reproduced from Morén, B., Heyman, J., Stroeven, A.P., 2011. Glacial geomorphology of the central Tibetan Plateau.
Journal of Maps v2011, 115–125. doi: 10.4113/jom.2011.1161, with permission from Taylor and Francis.

geomorphology are associated with, and control the spatial
distribution of volcanic hazards. Remote sensing and GIS offer
a wide range of techniques to document volcano morphology,
its temporal evolution, and the associated hazards.
Individual subaerial volcanoes on Earth range in diameter
from 102 (e.g., spatter or scoria cones) to 105 m (i.e., Hawaiian
shield volcanoes), and from a few meters to B4 km in relief
(e.g., Kilimanjaro). The synoptic view provided by satellite
imagery is used to assess essential geomorphologic questions
such as: (1) how does the size and shape of volcanoes vary
with the type of volcanic activity and the tectonic setting? (2)
what is the spatial distribution and geometric characteristics of
contrasted volcanic deposits within a volcano? (3) what geo-
morphologic, tectonic, and geologic factors contribute to the
 Remote Sensing and GIScience in Geomorphological Mapping 201

(a) (b)

Spacing Length
Spacing Length
(c) (d)

Figure 9 A spatial distribution of glacial bedforms reveals a distinct difference that can be used to discriminate between flow events: (a)
hypothetical lineation pattern; (b) an interpretation that assumes all of the flow evidence is of the same age; (c) an alternative interpretation that
accounts for cross-cutting lineations; and (d) flow event discrimination based on spatial pattern and morphometry. Reproduced from Clark, C.D.,
1997. Reconstructing the evolutionary dynamics of former ice sheets using multitemporal evidence, remote sensing and GIS. Quaternary Science
Reviews 16, 1067–1092.

destabilization of volcanoes? and (4) can the morphology of
eroded volcanoes be used to estimate their age or their
initial shape?
The geomorphology of volcanoes is characteristically dif-
ficult to study in the field due to their large scale, their
inhospitable and remote locations, or due to the hazards as-
sociated with ongoing eruptive activity. Remote-sensing in-
vestigations are therefore essential for assessing and
documenting the rapid evolution of volcanic surface features,
and for characterizing the surface morphology of edifices or
structural features, such as calderas. Great advances in the
study of volcanic processes and the mapping of related haz-
ards were also achieved within the last two decades thanks to
the increased availability of accurate DEMs and the integration
of datasets and models into GIS.
3.8.4.1 Satellite Images
A variety of satellite sensors are readily used to study eruptive
processes, including thermal emissions, identification of vol-
canic clouds, and quantification of ash and gas emitted during
explosive eruptions (Ernst et al., 2008). Newly emplaced vol-
canic material, such as lava or pyroclastic flows, can be easily
mapped based upon thermal emissions/surface temperature.
After cooling, lava flows generally preserve a spectral contrast
relative to older surface features given weathering, making it
possible to discriminate flows using spectral data. Simple
techniques such as spectral-feature extraction, thresholding,
and various other pattern-recognition techniques can be used
to produce thematic maps.
Real-time mapping of newly emplaced material is essential
for the identification of the hazard/risk areas, but also to
rapidly assess the potential of secondary hazards, such as
those caused by pyroclastic flows. Detailed mapping of erup-
tive material is crucial at remote volcanoes where direct ob-
servations and monitoring data are generally lacking, as it
provides key constraints on the eruption style and the volume
of erupted magma. Whereas such mapping is now routine for
lava flows, which cause a drastic and complete change in
surface cover, a future research challenge will be to use satellite
images to map spatial areas affected by ash deposition, and to
derive information on the variation in ash thickness from its
impact on surface spectra. The spatial distribution of ash is an
essential constraint on eruption dynamics, and has impli-
cations related to health hazards (Figure 11).
Optical sensors are of limited utility to assess poster-
uptive geomorphologic changes in tropical countries, due to
the frequent cloud coverage. In these regions, microwave
sensors, with all-weather capabilities, facilitate assessment of
newly emplaced material. For example, the 2002 lava flow of
Nyiragongo caused infrastructure damage and casualties.
Rapid acquisition of a range of satellite data, through the
implementation of the U.N. International Charter for Space
and Major Disasters, highlighted that Synthetic Aperture
Radar (SAR) data were the most useful in characterizing
the extent of the area affected by the lava flow, providing
support to relief operations (www.disasterscharter.org).
Microwave SAR data are generally useful for identifying and
studying volcano-tectonic structures and deposits in tropical
regions like Indonesia (Carn, 1999).
202 Remote Sensing and GIScience in Geomorphological Mapping
Ribbed moraine ridge in
lac naococane, quebec
0 40 km
(a)
(b)
Figure 10 (a) A generalized overview of the Lac Naococane ribbed
moraine ridge in Central Quèbec (Canada) and estimated ice-flow
direction. (b) The connectivity of moraines is determined by sliding
moraines along the axis of ice-flow direction (i.e., solid line) and
rotating the moraines in GIS to achieve a better fit (the shaded
moraines were rotated). Modified from Dunlop, P., Clark, C.D., 2006.
The morphological characteristics of ribbed moraines. Quaternary
Science Reviews 25, 1668–1691.
Detailed mapping of eruptive material using satellite im-
agery is also used to reconstruct the eruptive history of a vol-
cano, and to constrain the spatial variability in eruption
dynamics. Different lava-flow surfaces, associated with con-
trasted flow dynamics and eruption rate, can be discriminated
based upon spectra or surface roughness derived from SAR
backscatter data. Using SPOT and AirSAR data, Rowland
(1996) produced a detailed map of the lava flows of the San
Fernandina volcano in the Galapagos. Radar images proved
especially useful to discriminate between the smooth pahoe-
hoe and the rough as lava surfaces. This map enabled the
identification of two contrasting types of eruption character-
ized by vent type and location, eruption volumes, and lava
morphology. These differences accounted for variations in
slope segments along the volcano flank. Smets et al. (2010)
used a time series of multispectral and radar data to document
the multiple overlap of lava flows at Nyamulagira volcano in
the Democratic Republic of Congo over the last century. The
loss of coherence between successive SAR phase images was
Figure 11 Satellite images captured by the High Resolution
Geometric sensor onboard SPOT 4 satellite over Oldoinyo Lengai
volcano (Tanzania). (a) 10 m resolution false-color composite
acquired on 15 June 2007, highlighting the active crater and the
overflow of natrocarbonatite lava on the North flank, as well as a
fresh 3 km long lava flow on the West flank, emplaced in March
2006; (b) 5 m resolution black and white image acquired on 29
October 2007, after the onset of explosive activity within the active
crater. Recently burned areas are observed on the NW flank and
significant ash deposits affect the entire region west of the summit.
Opposing satellite-viewing angles on the two images account for the
varying topographic distortions.
used to identify newly emplaced flows in zones where spectral
contrast between recent flows was insufficient in the visible
and near-infrared wavelengths.
In a complementary approach, Kahle et al. (1988) ana-
lyzed the evolution of the spectral characteristics of lava flows
in the thermal wavelengths, in relationship to their emplace-
ment age. They showed that the weathering of superficial
lava crust was associated with a detectable variation in their
thermal-infrared emittance spectra. Such an approach opens
interesting perspectives for the systematic mapping and rela-
tive dating of lava flows at volcanoes for which the eruption
history still remains to be documented.
Spectral variations of volcanic materials were also used to
discriminate between lithologies (i.e., dacite lava blocks,
ignimbrite) observed in the deposits of a large debris ava-
lanche at Socompa volcano in Chile (Wadge et al., 1995).
These rock types could be associated with different sectors
of the volcanic edifice prior to collapse, which enabled further
constraints to be put on the sector collapse dynamics.
 Remote Sensing and GIScience in Geomorphological Mapping
Hyperspectral data were also used to document the occurrence
and type of weathered volcanic rocks at active volcanoes like
Vulcano (Mazzarini et al., 2001). At Mt. Rainier in Washington
state, U.S., the identification of weathered rocks provided es-
sential constraints on the potential location of future flank
instabilities (Crowley and Zimbelman, 1997). Destabilization
of these weathered materials have produced repetitive sector
collapses in the past, and the probability of a future event
represents a potential hazard for the densely populated area of
Tacoma, around the volcano.
Multitemporal imagery of volcanic landscapes that docu-
ment eruption events is of great interest to assess erosional
processes rapidly reshaping the newly formed landscape, and
can characterize vegetation colonization, and succession over
new volcanic surfaces. This was documented after the Plinian-
style explosive eruption of Mt. Pinatubo in the Philippines in
1991, which deposited an ash layer, over 200 m thick in pla-
ces, on the volcano flanks. This loose ash was rapidly remo-
bilized during the successive rainy seasons following the
eruption, causing deadly mudflows or lahars in inhabited
plains at the volcano’s base and secondary ash avalanches due
to detachment of poorly indurated ash blocks in the deep
erosion canyons (Torres et al., 2004). Vegetation growth on
this ash layer caused a significant decrease in ash mobilization
and canyon incision within 10 years of the eruption.
Finally, the nature of planetary volcanism (e.g., Mars, Venus,
Io) has been a matter of debate for a long time. One has to rely
almost exclusively on imagery and topographic datasets col-
lected by space missions. Based on these constraints, volcan-
ologists, and planetary scientists are looking for analogous
topographic features on Earth to identify the processes re-
sponsible for the formation of extra-terrestrial morphologies.
This approach applies to a variety of landforms, and was spe-
cifically followed to investigate the volcanogenic nature of
cones on Mars, to identify large-scale gravitational instabilities
on Olympus Mons (e.g., McGovern and Morgan, 2009), or to
suggest the occurrence of magma-water interaction processes,
and therefore the presence of water in the subsurface of Mars
(e.g., Murray et al., 2010).
3.8.4.2 Digital Elevation Models
DEMs are now widely available and used for: (1) overall
visualization of the geomorphology of a volcanic area; (2)
production of background maps to locate field observations
and sampling sites; (3) morphostructural interpretation of
volcano-tectonic interactions; (4) geomorphometric charac-
terization of volcanic features; (5) geometric characteristics of
volcanic deposits or features; and (6) modeling of volcanic
flow processes and assessment of related hazards.
Since geologists recognized the wide range of sizes and
shapes of edifices produced by volcanic activities, classification
schemes are required that account for morphological vari-
ations related to magma composition, eruption dynamics,
or tectonic settings. Systematic quantification of terrestrial
volcano morphologies is also required to compare the
morphologies of potential volcanic structures observed on
other planets (e.g., Pike and Clow, 1981).
Using DEMs, it is now possible to systematically document
the spatial distribution and morphometry of many volcanic
203
structures within a given region or on a worldwide scale.
Kervyn et al. (2008) and Wright et al. (2006) showed that the
grid-cell size of the DEM needs to be adapted to the average
size of the features of interest. A SRTM DEM at 90 m reso-
lution is sufficient to characterize polygenetic stratovolcanoes
(Wright et al., 2006), but a finer resolution is required to ac-
curately characterize the morphometry of pyroclastic cones
which are a few hundred meters wide (Kervyn et al., 2008).
Using GIS-based techniques, a large set of morphometric
data characterizing the size, height, and shape of volcanic
cones can be derived. Such statistical datasets provide key
constraints to interpret the relationship between the relative
crater size and the intensity of the eruptive activity, or the
decrease in the relative cone height caused by erosion. Using
LiDAR data, Favalli et al. (2009) documented all the cones on
the flank of Etna in Italy, and highlighted the great variability
in morphometric ratios relative to previously defined average
values (Figure 12). The spatial distribution of volcanic
vents within a volcanic field can be easily mapped with a
high-resolution DEM, putting constraints on the structures
controlling their spatial distribution and the associated dis-
tribution of hazard. The alignment of these vents and the
morphology of the surrounding volcanic cones can moreover
be used as the main constraint to infer the orientation of
feeding dykes which are clear indicators of the stress fields
within volcanic edifices (e.g., Corazzato and Tibaldi, 2006).
The ASTER GDEM has also provided thematic information
(e.g., caldera floor, incised hydrographic network, various
deposits and flows, flow paths and morphologies such as
ogives, landslide scars) derived from geomorphometric ana-
lyses of cones, thus revealing new volcano-tectonic features
and offering a reconnaissance tool for landform mapping in a
volcanic environment (Figure 13; Zouzia et al., 2011).
Grosse et al. (2009) recently developed the MORVOLC
method, with the objective of characterizing all terrestrial vol-
canoes using a standardized methodology. This method is
based on a set of quantitative parameters describing the size
and shape of an edifice from a planimetric (e.g., ellipticity, ir-
regularity) and cross-sectional perspective (e.g., height/width,
summit/basal width, slope). Based on these parameters, vol-
canoes were classified as cones, subcones, and massifs. It can be
applied to any volcanic structures represented in a DEM, given
an appropriate spatial resolution. Application of this method-
ology to 145 strato-volcanoes from the Andes suggested that
this quantitative method provided strong evidence for different
evolutionary trends of volcano growth, which can be inter-
preted in terms of magma supply, spatial distribution of vents,
edifice strength, gravitational deformation processes, and tec-
tonic controls. A comparable methodology has recently been
applied to the study of a selection of symmetrical strato-
volcanoes, which highlighted two groups with contrasting slope
profiles around the summit, attributed to different material
characteristics and thus different dominant eruptive dynamics
(Karatson et al., 2010a). The key limitation of such systematic
quantification of volcano morphologies is the definition of the
volcano base. So far the delimitation of the volcano base de-
pends upon visual interpretation, introducing subjectivity and
low reproducibility. The definition of common repeated criteria
or the development of automated algorithms to identify the
volcano base is crucial for the production of consistent results.
204 Remote Sensing and GIScience in Geomorphological Mapping

(a) (d)

(b) (e)

(i)

0 200 400 km

(c) (f) N

(g) (h)

Figure 12 Examples of Etnean scoria cones showing typical shape variations. (a) a pre-122 BCE ‘well formed’ cone (minimal modification in
two millennia); (b) a less regular cone (the cone shape is obvious, but the base is irregular); (c) the majority of Etnean cones are even more
poorly formed (cone can be recognized but the base is irregular due to lava flows; (d, e) strongly eroded cones (if there is a well-developed gully
system, these are referred to as ‘heavily dissected cones’); (f) many cones show only little evidence of gully dissection; (g, h) example of how
gully formation is controlled by vegetation, which is related to altitude and exposure; (i) cone shape can change during emplacement due to
eruptive dynamics (e.g., structurally deformed cone). Reproduced from Favalli, M., Karatson, D., Mazzarini, F., Pareschi, M.T., Boschi, E., 2009.
Morphometry of scoria cones located on a volcano flank: a case study from Mt. Etna (Italy), based on high-resolution LiDAR data. Journal of
Volcanology and Geothermal Research 186, 320–330.

Morphological analysis of volcanoes has also proved useful
to reconstruct the past shape of eroded edifices. Combining
terrain analysis with field observations and geochronological
data, Karatson et al. (2010b) proposed a model for the dif-
ferent stages of evolution for Mt. San Francisco in Arizona,
U.S. Rodriguez-Gonzalez et al. (2010) also illustrated how
paleogeomorphologic reconstruction can be conducted
using interpolation to estimate the original volume of tephra
and lava emitted by an eroded volcanic center. Such
reconstruction, when sufficiently well constrained with com-
plementary data sources, provides valuable insights on the
volcanic history of a region.
It is now widely recognized that volcanoes are not simply
growing by accumulation of volcanic material from one or
several vents. They are also affected by internal deformation
due to magmatic, tectonic, or gravitational forces, leading to
 Remote Sensing and GIScience in Geomorphological Mapping 205

780000 783000 786000

0
LLP 4
7
4056000 11
AF Emporios 15
(1) 18
N 22
(2) BR 26
(3) (1) 30
MPE
(5) (4) 33
(4) PV 37
PE
(5) KV 4053000 41
(7) (2) ST 44
(6) TR 48
(3) Nikia 52
(a)
Avlaki 55
59
Degrees

4050000
Nikia lava flows Pyroclastic flows
Avlaki lava flows Lava flow front and limits
Post caldera domes Dome
N-NE-E-SE-S calders wall Number and relative age
Caldera floor
(1) of lava flows and domes
Caldera rim and
Landslide
phreatic crater
Magma feeder systems River
Dome scree deposits Fault
Lava flows Trapdoor fault (PRF)
(b)

Figure 13 Slope image of the Nisyros caldera (Aegean Volcanic Arc). The pixels range between 01 and 58.81 (the darker the pixel, the higher
the slope. Inset (a) displays volcano-morphological features resulting from ash and lava flows, lave flow paths and feeder systems, ogive
structures, younging, and contact relationships, height of lava fronts and width of flow. Inset (b) displays volcano-tectonic and tectonic faults and
resultant features such as radial hydrographic network along radial caldera faults, the Nisyros caldera ring fault, caldera rim, and the Perigussa
trapdoor (PRF). Reproduced from Zouzias, D., Miliaresis, G.C., Seymour, K.S., 2011. Interpretation of Nisyros volcanic terrain using land surface
parameters generated from the ASTER global digial elevation model. Journal of Volcanology and Geothermal Research 200, 159–170. Site names
are provided as acronyms. Consult article for site names.

long-term incremental deformation, or catastrophic and large-
scale ground displacements such as caldera or flank collapses.
All these processes, which are potential sources of hazards, are
currently being studied at many volcanoes, using a variety of
techniques including field-based structural assessment, real-
time ground and space-based monitoring techniques, and
analogue and numerical modeling efforts. In many regions,
application of field monitoring and geological mapping is
restricted by access or cost limitations. DEMs provide essential
data to explore the distribution of volcano-tectonic structures,
generally indicated by km-scale topographic lineaments or
break-in-slopes. Lagmay and Valdivia (2006) illustrated how
the morpho-structural features (e.g., breached crater) identi-
fied on the SRTM DEM for Philipinian strato-volcanoes could
be related to the orientation of the main regional principal
stresses and known strike-slip faults. These observations were
consistent with results from sand-cone analogue models
simulating the effect of strike-slip tectonic faults on the de-
formation of a conical edifice. In the same way, the natural
applicability of analogue models is generally validated pri-
marily based on morpho-structural comparison of the models
with natural volcanoes using terrain analysis. The synoptic
coverage of imagery and topographic information offers the
unique advantage to identify large scale but subdued features,
such as some caldera’s or debris avalanche deposits that can-
not be readily identified by local field observations.
Multitemporal DEMs can be used to characterize the vol-
ume and change in geometry of newly emplaced volcanic
materials. Advances in technologies such as SAR interfer-
ometry, ground, air- or space-borne stereoscopic imaging and
LiDAR, have the potential to rapidly and repetitively produce
DEMs over an evolving eruption area. Lu et al. (2003) used
SAR-based DEMs to estimate the volume of a lava flow at
Okmok Volcano in the Aleutian Islands, U.S., with un-
precedented accuracy (15% error). Bastero and Lagmay (2006)
showed that essential information such as lava’s silicate con-
tent and yield strength can be estimated from high-resolution
DEMs. Such applications are, however, limited by the insuffi-
cient spatial resolution and accuracy of freely available DEMs,
resolving the volume of newly emplaced materials only a few
meters thick. High-resolution LiDAR data proved efficient in
mapping individual lava-flow units, characterizing the chan-
nel geometry at Etna. This opens the potential for more de-
tailed characterization of the geometry and volume of newly
emplaced volcanic materials, as repeated acquisition of LiDAR
data become more readily available and affordable in the
future.
Differential InSAR is another relevant technique to docu-
ment topographic changes at volcanoes, as it is capable of
mapping mm-scale surface displacement related to coseismic
ground displacement or crustal-magma movement. This
technique is based on the calculation of the phase shift be-
tween the radar backscatter registered by a radar antenna in
two successive flights over the same area. In order to isolate the
phase shift attributed to displacement occurring between the
two data acquisitions, an accurate DEM has to be available to
206 Remote Sensing and GIScience in Geomorphological Mapping
retrieve the topographic effect on phase differences. Docu-
mentation of these displacements in volcanic regions provide
essential constraints on the location, volume, and geometries
of intrusive bodies, as well as the location of active faults.
Interpolated over repetitive individual intrusive or seismic
events, the deformation field calculated through differential
InSAR provides an indication on the long-term morphological
evolution that can be attributed to intrusive or tectonic
processes.
Topographic information also contributes to the better
understanding of volcanological processes and volcano
growth, as terrain conditions are required for hazard model-
ing. Some of the greatest hazards posed by volcanic activity are
associated with flow-dominated processes such as pyroclastic
density currents, debris avalanches, lava flows or mud flows
(lahars). Although the physics of each of these processes is
different and often poorly understood, topography always
represents a key constraint on the flow path and the extent of
the affected area. Stevens et al. (2002) illustrated how the
accuracy and resolution of DEMs was crucial in controlling the
results of a lahar-simulation model. Some of the most easily
implemented lava-flow simulation models use topography as
the main control on lava-flow invasion probability. The spatial
variation in the probability of invasion is modeled by inte-
grating the results of multiple iterations of flow-path simu-
lations, that implement a random-height variation of chosen
amplitude into the initial DEM to account for the ability of
lava flows to overflow limited relief (e.g., Felpeto et al., 2001).
The difference between the reconstructed pre- and post-
collapse volcano topography is also a principal constraint of
debris-avalanche models. These examples show that although
it is crucial to constrain the rheology and dynamics of haz-
ardous volcanic flows to accurately model the area they might
affect, much attention should also be paid to the use of ac-
curate and up-to-date topographic datasets with the appro-
priate spatial resolution to obtain reliable hazard maps.
3.8.4.3 GIS-Based Mapping
GIT is used in volcano geomorphology and volcanology to
develop, integrate and manage spatial databases, conduct
spatial analysis and modeling, and to produce a variety of
information products to communicate the results of scientific
research to a nonexpert audience.
Geospatial analytical software tools are used to study the
controlling factors that govern the spatial distribution of
point- or linear-type features in volcanic environments. The
distribution of volcanic cones in flat volcanic provinces can be
analyzed to highlight the orientation of the underlying struc-
tural pattern controlling the propagation of magma through
the crust. For example, Hamilton et al. (2010) analyzed the
nearest-neighboring distance within a field of rootless cones in
Iceland. Through this geospatial analysis, it was shown that
these cones were characterized by a repelled distribution
maximizing the distance to the nearest neighbor, indicating
that the formation of these structures was controlled by a
competitive process for a limited resource, here water.
The study of active volcanic regions for the prediction of
short-term hazards or the understanding of long-term
evolution always requires the integration of a variety of data
and observations including topographic and geological data,
imagery, field observations, geophysical data, and numerical
model results. Prior to an eruption at a large shield volcano,
it is essential to be able to locate the location of earthquake
epicenters, the direction and amplitude of surface defor-
mation recorded by GPS stations, and to examine this infor-
mation contextually in relation to existing fractures, dike
networks, or local topographic lows, to obtain an indication
on the probable location of the eruption outbreak. Analysis
of slope orientation, flow paths, or more advanced flow
simulations cross-correlated with maps of human infra-
structure and settlements, enable elements-at-risk to be rapidly
identified.
Detailed geological maps of active volcanoes can be rapidly
updated once they are integrated in a GIS. During the eruption
of Nyamulagira in the Democratic Republic of Congo in
January 2010, mapping of the advancing lava flow using
thermal imagery was combined with lava-front localization by
helicopter flights in the field to constrain lava flow-path
simulation models. Such near real-time integration of infor-
mation, even in the absence of advanced monitoring techni-
ques, lead to rapid production of updated maps that serve to
communicate to the authorities about an evolving eruptive
crisis, and in refining the risk assessment for the local popu-
lation. At the end of the eruption, the existing GIS-based
geological map directly included the newly emplaced lava
flow, providing key information about its geometry and the
impacted region. Making assumptions on the lava thickness,
its volume can be estimated and the DEM updated accord-
ingly, in order to account for this morphological feature in the
prediction of flow path during future eruptions.
GIS is also essential in the assessment of the spatial dis-
tribution of long-term hazard and risks around a volcano. A
first key step in such assessment, especially for a large com-
posite volcano, is estimating the spatial distribution of the
probability of eruption outbreaks. Marti and Felpeto (2010)
recently proposed a GIS method that integrates several indi-
cators such as the distribution of past eruptive vents, fractures,
and the documented volcano stress field in estimating out-
break probability. Using this probability of occurrence,
plausible eruption scenarios, varying in the eruption dynamics
and intensity can be simulated with appropriate models and
results displayed in a GIS. Separate hazard maps are generally
produced for different eruptive processes such as fallout and
pyroclastic flows, as they respond to contrasting dispersion
dynamics, but hazards are associated with different intensity
of one process (e.g., increasing lava effusion rates) and are
summarized on a single output map. Correlation of these
hazard maps with the vulnerability of the population and
infrastructures in the region leads to the production of syn-
thesized maps localizing the probability of risk from volcanic
eruptions.
A compromise has to be found between the precision of
the data provided by such hazard or risk maps, and its read-
ability for the stakeholders and population that will base their
decisions on it. Using GIS, great efforts can be focused on the
production of the most informative and understandable end-
products: different levels of hazard should be coded in a
few clearly labeled classes and should be displayed on a
 Remote Sensing and GIScience in Geomorphological Mapping
background map or image on which people can easily locate
themselves. The effectiveness of various visual outputs to
convey information about risks was evaluated by Haynes et al.
(2007) showing that 3D perspective views of aerial photos
revealing the areas at risk were judged to be the most effective
by the targeted population. Research on effective dissemin-
ation of scientific spatial information to the general public is
only in its early stage in volcanology, and much more atten-
tion needs to be paid to these issues in the future.
With the rapid development of interactive web-based GIS
platforms, several new developments will become possible.
First, existing process-simulation models will be fully inte-
grated into GIS and made more available. This will make it
possible for the end-user to evaluate the impact of eruption
scenarios and to produce hazard and risk maps most appro-
priate for their needs and area of interest. The integration of
several eruption simulation models into a GIS offers the op-
portunity to simulate the long-term evolution of the morph-
ology of a volcano according to defined scenarios of long-term
eruptive behavior and eruption rate’s evolution; it also allows
volcanologists to test the effectiveness of engineering miti-
gation measures (e.g., Chirico et al., 2009).
3.8.5 Landslide Mapping
Landslides occur frequently and can be very destructive.
Hence, it is imperative to understand what the controlling and
triggering factors are, and be able to conduct hazard and risk
assessments. The use of GIT greatly facilitates the assessment
and mapping of landslides, and in predicting the potential
impacts on the environment and society. Numerous forms of
spatial data are collected and used for local and regional scale
assessments (Alexander, 2008). Detailed maps can be gener-
ated to show past events and predict likely future events.
GIS-based landslide analysis is a relatively new field and the
majority of such GIS-based investigations have occurred since
the 1990s. Carrara and Guzzetti (1995) noted that few bran-
ches of hazard studies have been transformed as completely by
information technology as has landslide research. The study of
geospatial technologies for landslide analysis has rapidly
evolved, and Alexander (2008) acknowledged that, as a result,
landslide occurrence and susceptibility are now generally
known in many areas of the world. Good overviews, reviews,
and edited volumes on GIS and natural hazards and landslides
are those by Varnes (1984), Carrara and Guzzetti (1995),
Coppock (1995), Wohl and Oguchi (2004), Chacon et al.
(2006), Alexander (2008), and Carrara and Pike (2008).
Alexander (2008) revealed that there is a considerable re-
dundancy in landslide research results, and a tendency for
scientists to use GIS in an exploratory, rather than hypothesis-
solving mode. Alexander (2008) argued that the reason for
this misguided trend is due to the use of GIS as mainly a
deductive, rather than an inductive method. Carrara and Pike
(2008) warned that although the increasing user-friendliness
of GIS-software and the widespread use of it reflect a positive
development, it also encourages the perception that GIS-based
landslide assessment is a ‘point-and-click’ undertaking, ac-
complished rapidly with limited resources. Although a large
number of assessment methods are available today, there is a
207
lack of standardization in the use of analysis protocols and
information products (Chacon et al., 2006). Nevertheless, GIS
software tools are routinely used in sustainable land man-
agement and development.
Slope instability in a cultural landscape is a function of
direct and indirect natural and human factors, the so called
controlling or triggering parameters. These generally include
lithology, structure, tectonics, geomorphology, topography,
precipitation, temperature, infiltration, runoff, land cover, and
road construction. Others exist, although they have not been
widely utilized and include topographic stress, rock strength,
and rate of erosion, as system dynamics control slope stability.
Unfortunately, many of these variables effectively represent
associations with the potential for slope instability. Terrain
analysis greatly facilitates an assessment of land-surface mor-
phometry and processes that are related to slope instability,
and provides a better understanding of how and under what
circumstances, events, states, dynamics, or processes trigger
landslides. Chacon et al. (2006) described the types of data
and the methods that are included in these analyses which
include: (1) geotechnical data and slope-hydrological models;
(2) temporal data on failure and mass activity; (3) spatial and
temporal data on stages of development; (4) rainfall and
seismic events; and (6) slope-stability analysis based on
probabilistic-reliability index assessment.
Today, there is a wide range of data that can be used in
landslide research including photography and movies (ter-
restrial, aerial, repeat), surveying data, images from passive
and active sensor systems (LiDAR and Radar), and morpho-
metric parameters and indices based upon terrain analysis of
DEMs. Although terrestrial data sources are mainly employed
to study discrete landslides or smaller landslide complexes,
satellite imagers are used for examining entire landslide re-
gions. Digital imagery from airborne or satellite sensors de-
liver important information about the spatial distribution of
landslides in affected regions (inventory), landslide par-
ameters such as area and run-out distance, and environmental
conditions leading to the failure (event-controlling par-
ameters) such as lithology, vegetation, or roads. Multitemporal
data sets allow for studying and monitoring changes and for
understanding geomorphic processes.
3.8.5.1 Terrestrial and Airborne Photography
The qualitative interpretation and quantitative analysis of
terrestrial photographs and movies from discrete landslides or
landslide complexes are used to reveal causes and processes of
landsliding, and to automatically monitor landslide dis-
placement over longer time periods. Photographs proved to be
useful in documenting landslide movement (e.g., Goguel,
1989) and for rapid assessment of landslide displacement
(e.g., Kalaugher and Grainger, 1997). Iovine and Parise (2002)
developed a graphic method to classify landslide damage in
urban structures. Movies can be taken using stationary 35 mm
cameras for monitoring (e.g., Ballantyne et al., 1987) and
time-lapse studies of landslide movement (e.g., Belknap and
Gilmore, 1987).
Airborne photographic surveys offer a synoptic view of entire
landslide areas, and stereophotogrammetry allows for estimating
208 Remote Sensing and GIScience in Geomorphological Mapping

the 3D coordinates of points on a landslide. Airborne photo-
graphs can be used to study landslide behavior, displacement,
and kinematics (Yoshizawa et al., 1991; Brueckl et al., 2006;
Baldi et al., 2008), and for monitoring landslides (Chandler and
Cooper, 1988; Kajiyama et al., 2005; Walstra et al., 2007). Using
aerial photogrammetry (including aerial photos taken at a var-
iety of flying altitudes), Dvigalo and Melekestsev (2009) mapped
deposit thickness and surface characteristics with the objective of
qualitatively and quantitatively characterizing post-landslide
landforms and forecasting future events in Geyser Valley, Kam-
chatka (Figure 14). Furthermore, repeat terrestrial photographs
can add information about how the landscape changed after an
event (e.g., if the landsliding continued or the affected slope
recovered). Extensive multitemporal landslide inventories can be
used to draw general conclusions for entire landslide-effected
regions (Khattak et al., 2010).
3.8.5.2 Satellite Imagery
The utilization of satellite imagery in analyzing landslides
started after the launch of the first Landsat satellite in the
1970s. Today, satellite imagery is frequently used in landslide
research and landslide-risk assessment that has a spatial
component. Recent technological developments led to the
introduction of very high resolution (VHR) sensors with spa-
tial resolution o0.5 m that allow for the creation of large scale
maps of discrete landslides (Figure 15).
Over the last 40 years, Landsat images have been an im-
portant dataset in landslide research and landslide-risk as-
sessment. For example, Strom and Korup (2006) combined
Landsat ETMþ imagery with SRTM DEMs and declassified
KFA-1000 satellite imagery to study extremely large rockfalls
and rock avalanches. Seismic-triggered landsliding was map-
ped by Evans and Bent (2004), Leroy et al. (2010), and Willige
(2010). Dunning et al. (2006) draped an ETMþ image over a
90 m SRTM DEM to investigate the formation and failure
of a landslide dam. Alcantara-Ayala et al. (2006) related
landsliding to land-cover change by using vegetation frag-
mentation determined from decadal Landsat images (TM
and ETMþ ). MacDonald and Grubbs (1975) predicted land-
slide-prone areas for highway construction, and Aniya et al.
(1985) identified landslides for watershed management.
Other examples include Bencardino (2005) who assessed
landslide hazard and risk in river basins, and Haberler-
Weber et al. (2009) who conceptualized multiscale landslide
monitoring and evaluation. Landslide susceptibility and

Goryashchii Ge
Zhemchuzhnyi ize
Rozovyi konus rn
ay
Velikan aR
.
Neprotayannyi
New Grot
Dammed Fountain Fountain
lake “shchel”
Bolshov Line of detachment of
landslide rocks

Office of Helicopter field

kronotskii Lakes
Thermal patches arising
reserve after the landslide

Preserved geysers in the

Stream beds overlain by landslide
rocks and by the dammed lake
Sestrenka R.
central part of the valley
Lines of equal landslide
deposit thickness (5 m)

Remnants of rocks partly or

completely buried by landslide

Triumfal’nye Thermal anomalies

vorota

Vdp
Sh
um

Pervenets
na
ya
R
.

0 250 500 Meters

Figure 14 Map of landslide thickness (in meters) for the Geyser Valley in Kamchatka, derived from a 2007 aerial photography survey (flying
altitude varied to produce three different scales) and subsequent creation of three-dimensional terrain models. The goal was to measure
thickness, but also to map the transportation path of material. Modified from Dvigalo, V.N., Melekestsev, I.V., 2009. The geological and
geomorphic impact of catastrophic landslides in the geyser valley of Kamchatka: aerial photogrammetry. Journal of Volcanology and Seismology
3, 314–325, with permission from Springer.
 Remote Sensing and GIScience in Geomorphological Mapping
Figure 15 Very high-resolution image of landslides in Nova
Friburgo, Brazil. The image is from the GeoEye-1 satellite (0.5 m
resolution), acquired on 20 January 2011 following heavy rains that
triggered numerous deadly landslides in Nova Friburgo. Note the
intensity of deforestation that is known to increase the probability of
landsliding (image downloaded from www.geoeye.com).
hazard mapping was also conducted by Samarakoon et al.
(1993), Lee and Dan (2005), and Pradhan (2010).
Higher resolution SPOT imagery has also been used in
landslide and hazard studies. For example, Zhou et al. (2001)
and Yang et al. (2004) used imagery to study specific land-
slides. Aste and Girault (1995) and Yamaguchi et al. (1999)
used imagery to detected landslide movement, and Leroi et al.
(1992) and Ostir et al. (2003) monitored landslides. Roessner
et al. (2006) also studied landslides with the potential for dam
formation. Lee and Pradhan (2006), Dymond et al. (2006),
and Schneider (2009) used SPOT data for landslide-risk as-
sessment. The data were utilized to assess the relationship
between tectonics and landslides (e.g., Morelli et al., 2001;
Wang et al., 2003). Vegetation recovery after an earthquake can
also be investigated (e.g., Chou et al., 2009).
The VHR sensors on IKONOS and QuickBird are useful
for large-scale landslide mapping. IKONOS data were used
for slope instability analysis for pyroclastic deposits (Baeza
et al., 2006), extraction of landslide traces (Ochoa-Tejeda and
Parrot, 2007), retrogressive thaw slump activity studies (Lan-
tuit and Pollard, 2008), mapping of earthquake-related land-
sliding (Rathje et al., 2006), investigating urban landslides
(Wang, 2006), and landslide-risk assessment (Nichol et al.,
2006). Using QuickBird data, Delacourt et al. (2004) and
Chadwick et al. (2005) carried out landslide displacement
measurements, and Granica et al. (2007) derived indicators
for landslide detection.
3.8.5.3 Digital Elevation Models
DEMs are now widely used in landslide research (Iwahashi
et al., 2001; Barlow et al., 2003) and landslide-risk assessment
209
(Davis et al., 2000; Dahal et al., 2008; Hasegawa et al., 2009).
They can be generated from a variety of data sources, and
special care must go into the preprocessing that is typically
required before conducting an analysis. We highlight a few of
the data sources and types of analysis commonly used.
3.8.5.3.1 LiDAR
Light detection and ranging (LiDAR) systems have led to sig-
nificant advances in geomorphology, particularly through
their abilities to detect subtle topographic features and pene-
tration through vegetation canopies. The use of LiDAR data in
landslide mapping is a substantial improvement over more
traditional types of data that are used for landslide inventories
(Haugerud et al., 2003; Schulz, 2007; Van den Eeckhaut et al.,
2007). Nevertheless, there is still a degree of subjectivity as-
sociated with landslide mapping based upon this data, as
there is no consensus on which topographic features are part
of, or uniquely characterize landslides (Guzzetti et al., 2000;
Van den Eeckhaut et al., 2005, 2007; Booth et al., 2009).
Terrestrial LiDAR has been used to map morphological
changes on rock falls (Mikos et al., 2005), describe geo-
morphic features of large landslides in mountains (Dunning
et al., 2009), characterize retrogressive landslides in sensitive
clay (Figure 16) and rotational landslides in river banks
(Jaboyedoff et al., 2009), and monitor unstable slopes in
urban environments (Pieraccini et al., 2006). Obviously the
utility of terrestrial LiDAR is limited in spatial extent, as the
primary focus tends to be on discrete slope failures or smaller
failure complexes.
Alternatively, airborne LiDAR offers a high-resolution
synoptic view of entire landslide areas to track landslide
displacements (Dewitte et al., 2008), reveal topographical
changes in earthquake regions (Chen et al., 2006), estimate
mass-wasting processes in active Earth slides (Figure 17;
Corsini et al., 2009), map landslides under forest (Van den
Eeckhaut et al., 2007) and in urban areas (Madin and Burns,
2006), and carry out landslide-susceptibility mapping (Schulz,
2007). McKean and Roering (2004) used DEMs derived from
LiDAR to characterize surface roughness of terrain near
Christchurch, New Zealand. Bedrock and earthflow landslides
tend to be rougher than adjacent unfailed terrain thereby
permitting landslide-boundary mapping. They concluded that
older mass movements are ‘smoothed’ by surface processes
over time, allowing them to differentiate between active zones
of movement and older, inactive failures. Glenn et al. (2006)
also relied on surface roughness, in addition to using slope,
semivariograms, and the fractal dimension to characterize
morphology. The landslide surface characteristics were related
to underlying material type and local motion of the landslide,
although landslide size will influence semivariance values (i.e.,
small scale landslides show less roughness and semivariance
than large slides).
3.8.5.3.2 Radar
Most radar systems are able to detect surface elevation beneath
vegetation cover and be used during any season. Current air-
borne SAR sensors provide ground resolutions to about 10 cm,
and permit the monitoring of subtle landslide changes.
Interferometric SAR (InSAR) uses the scattering properties
210 Remote Sensing and GIScience in Geomorphological Mapping

from two or more SAR images to generate maps of surface

Figure 16 Othophoto obtained from aerial photograph taken on 9
May 2006 and the terrestrial laser scanning (TLS) high-resolution
DEM at ‘Nicolet’ landslide, Quebec, Canada. Contour lines with a
vertical spacing of 1 m (white lines) were derived from the TLS data.
The main and secondary landslide scarps are shown as red and
yellow lines, respectively Reproduced from Jaboyedoff, M., Demers,
D., Locat, J., et al., 2009. Use of terrestrial laser scanning for the
characterization of retrogressive landslides in sensitive clay and
rotational landslides in river banks. Canadian Geotechnical Journal
46, 1379–1390, with permission from Canadian Science Publishing.
deformation and digital elevation of large-scale terrain (Car-
nec et al., 1996; Fruneau et al., 1996; Singhroy et al., 1998;
Rott et al., 1999). For example, Singhroy et al. (1998) inte-
grated SAR and Landsat TM images to characterize and in-
ventory landslides in Canada. They concluded that InSAR
techniques provide better representations of elevation and
slope changes (improving landslide detection capabilities)
than RADARSAT and that a combination of SAR and TM can
characterize retrogressive slope failures and flow features in
regions with relatively low relief. Polarimetric SAR is also
useful in multitemporal landslide analysis.
Numerous radar-based landslide applications exist, in-
cluding earthquake-triggered landsliding (Czuchlewski et al.,
2003), landslide displacement (Tarchi et al., 2003; Noferini
et al., 2006), surface deformation mapping (Finnegan et al.,
2008), monitoring unstable slopes (Colesanti and Wasowski,
2006), and modeling rainfall-induced landslides (Chiang and
Chang, 2009). Bozzano et al. (2008) installed a monitoring
system that continuously measured a landslide interacting
with an under-construction infrastructure, whereas Rott and
Nagler (2006) evaluated the use of radar in landslide-risk
assessment. Tarchi et al. (2003) combined ground-based
InSAR with field instrumentation to derive multitemporal
surface deformation maps of the depletion zone of the Tessina

Elevation difference 2003–1973 0 250 500 750 1000 m

Mass waste in river Accumulation Depletion

Depletion

Depletion

TOE zone TRANSIT zone SOURCE zone

Depletion

–20 –15 –10 –5 –3 –0.5+0.5 +3 +5 +10 +15 +20

(a) Elevation difference (m) 10 m contour lines: 2006 topography

Depletion Accumulation Accumulation

Depletion

Accumulation

TOE zone TRANSIT zone SOURCE zone

Elevation difference 2006–2003 0 250 500 750 1000 m

(b)

Figure 17 Digital elevation model difference images, computed from LiDAR data for the Valoria landslide in the Northern Apennines of Italy:
(a) 2003–1973; (b) 2006–03. Reproduced from Corsini, A., Borgatti, L., Cervi, F., Dahne, A., Ronchetti, F., Sterzai, P., 2009. Estimating mass-
wasting processes in active earth slides: earth flows with time-series of high-resolution DEMs from photogrammetry and airborne LiDAR. Natural
Hazards and Earth System Sciences 9, 433–439.
 Remote Sensing and GIScience in Geomorphological Mapping 211

landslide (Italy) that was used to interpret landslide kin- Table 3 Geomorphometric parameters and objects commonly used
ematics and evolution (short term). The use of InSAR, how- in mountain terrain
ever, has several limitations related to landslide scale and
Parameter/object Application/Mountain landforms
viewing geometry, which influence landslide mapping studies
(Catani et al., 2005; Glenn et al., 2006). ‘Local geomorphic parameters’
Slope angle Insolation/snow cover, permafrost distribution
Curvature Classification of form elements
3.8.5.3.3 ASTER Flow accumulation Delineation of drainage basins
The Advanced Spaceborne Thermal Emission and Reflection
Radiometer (ASTER) sensor can also be used to generate DEMs ‘Linear geomorphic parameters’
(Kääb, 2002). ASTER-derived DEMs offer fine geometric detail, Hillslope profile Hillslopes, toposequences
relatively high spatial and low temporal coverage, and are cost- Valley long profile Valley floor (e.g., steps)
Ridgelines Landform/mountain classification
effective for landslide studies at the global scale (Santini et al.,
2009). These DEMs are commonly used to map landslide ‘Areal geomorphic parameters’
susceptibility and hazard assessment in areas with high relief Form elements Hillslopes, talus
(Kääb, 2002; Liu et al., 2004; Havenith et al., 2006; Fourniadis Basins Drainage system
et al., 2007a, b). Haeberli et al. (2004) studied a massive rock- Landform units Landform/mountain classification
ice slide and subsequent mud-flows on the Kazbek massif ‘Aggregated geomorphic objects’
(Russia) to put the event in a historical context, mapped sur- Landform units Landform/mountain classification
face features indicative of unstable conditions, and inferred Toposequence Hillslopes types
the impact of a valley glacier. Domakinis et al. (2008) mapped
‘Representative geomorphic parameters: dimension/form’
coastal landslides and Giardino et al. (2008) analyzed deep-
Length Cirques
seated slope deformation. Examples of susceptibility and Elongation Cirques
hazard assessments using ASTER DEMs include the work of Circularity Cirques
Ercanoglu (2005), Fourniadis et al. (2007a, b), and Kamp
et al. (2008). ‘Representative geomorphic parameters: internal characteristics’
Relief Cirques
Floor elevation Cirques
3.8.5.3.4 Geomorphometric analysis Steepest gradient Cirques
Although much research has focused on DEM generation Hypsometric integral Valleys
and evaluation in general, the utility of terrain analysis in ‘Representative geomorphic parameters: internal/external topology’
landslide research has tremendous potential. Claessens et al. Surface structure Rock glacier
(2005) showed that DEM resolution effects landslide-hazard Toposequence Hillslopes
modeling. Fabris and Pesci (2005) described precision
Source: Reproduced from Raseman, S., Schimdt, J., Schrott, L., Dikau, R., 2004.
and validation of automated DEM extraction using digital
Geomorphometry in mountain terrain. In: Bishop, M.P., Shroder, J.F. (Eds.),
aerial photogrammetry. Iwahashi et al. (2009) addressed
Geographic Information Science and Mountain Geomorphology. Springer-Praxis,
computational scale issues for calculation of slope using high- Berlin Heidelberg, pp. 101–137.
resolution LiDAR DEMs, and Liu et al. (2007) simulated the
formation of landslides using a DEM. Although new capabil-
ities exist, it is advisable to use caution in landslide mapping,
as researchers do not generally address issues such as digital (Table 3). Hypsometric curves characterize the area/altitude
terrain modeling techniques, DEM error and uncertainty function (e.g., ‘elevation relief ratio’), whereas clinometry de-
issues, as well as the propagation of errors in terrain analysis. scribes the slope angle/altitude function. Slope and curvature
Carrara and Pike (2008) criticized that despite a veritable profiles can include longitudinal, cross-section, and even
‘cottage industry’ of current DEM-based hazard mapping, tangential perspectives on topographic variation that reflect
many of the predictions are either based on unsuitable data or multiple process domains (i.e., erosion, transportation, and
are not properly modeled. deposition), and are thus of interest to landslide scientists
Topographic parameters such as slope angle and azimuth, (Raseman et al., 2004). Cross-sectional assessment of valleys
profile, various curvatures, upslope contributing area, and can distinguish between U-shape and V-shaped valleys, map
catchment area can be event-controlling parameters and as the extent of talus, and indicate valley wall stability.
a result, they are important for mapping landslide features A basic mapping approach that can be used is DEM dif-
and analyzing causes of slope instability. In particular, high- ferencing, where multitemporal DEMS are used to estimate
resolution DEMs offer detailed geomorphometric analyses to the differences in altitude caused by landslide material de-
ascertain scale-dependent processes in landslide susceptible pletion and accumulation. This helps to identify source,
areas (in contrast to slide mechanics or model constraints transit, and toe zones (Corsini et al., 2009). This relatively
developed from InSAR) (Glenn et al., 2006). Thus, landslide simple approach can highlight changes in surface character-
mapping now relies heavily on rigorous geomorphometric istics over time, or following a landslide event (Figure 17).
analyses using high-resolution DEMs to produce new per- Alternatively, the manner in which DEMs are visualized can
spectives on the morphology and triggering mechanisms elucidate older landslides that have been modified by surficial
associated with landslides. Terrain analyses in high-relief processes. Van den Eeckhaut et al. (2005) combined a series of
areas include hypsometric, clinometry, and profile analysis hill-shaded DEMs (either a sun azimuth angle of 451 or 3151)
212 Remote Sensing and GIScience in Geomorphological Mapping
with ‘expert knowledge’ to test the reliability of hill-shading as
an identification tool. Some landslides did not appear on both
hill-shaded DEMs, proving that relying solely on the visual-
ization capabilities of DEMs can be misleading. Developing,
shading, and analyzing the DEMs took less than 1 day, how-
ever, whereas field surveying lasted 100 days.
Most geomorphometric analyses use slope and curvature
parameters to map landslide structures in mountainous regions
(McKean and Roering, 2004; Chadwick et al., 2005; Glenn
et al., 2006; Van den Eeckhaut et al., 2007). Landslide surfaces
tend to exhibit more surface roughness (at local scale of meters)
than unfailed slopes (McKean and Roering, 2004), and the
extent of older, inactive landslides can be mapped based on
surface roughness, as surface processes will reduce overall
topographic variability and the variability of inactive landslide
morphological components (Glenn et al., 2006). Glenn et al.
(2006) generated semivariograms from LiDAR-derived DEMs to
study the morphological and scale-dependent nature of the
topography, and used the fractal dimension to contrast scale-
dependent topographic variability of different landslide com-
ponents. McKean and Roering (2004) also used surface
roughness as a surrogate for detecting landslides by measuring
variability in slope and azimuth in local ‘patches’ of a LiDAR-
derived DEM. They suggested accounting for resolution/feature
scale issues, in order to increase the probability of detection and
high-quality mapping. Several studies have provided insight
into the use of semivariograms and spatial autocorrelation to
analyze topographic variability and surface roughness (Bishop
et al., 1998, 2003; Walsh et al., 2003; Miska and Hjort, 2005).
In addition, signal processing techniques, such as Fourier
transforms and wavelet transforms can quantify landslide
morphologic characteristics and highlight patterns of past slope
instability over large areas (Booth et al., 2009).
Much like landslides, rockfalls have been studied using
GIS and DEMs at varying scales (e.g., Guzzetti et al., 2002;
Baillifard et al., 2003; Günther, 2003; Jaboyedoff et al., 2004;
Derron et al., 2005). Mapping rockfall potential based on
threshold angles at which slopes maintain stability is a rela-
tively simple approach. Angles 4601 (Guzzetti et al., 2003),
4451 (Jaboyedoff and Labiouse, 2003), and 4371 (Frattini
et al., 2008) have been documented, although this depends on
a multitude of variables (e.g., lithology, relief, tectonic setting).
Foliation orientation and major discontinuities can be derived
directly from DEMs, and the spatial arrangement of these
features can be labeled as hazardous structures with a high
probability of falling (Derron et al., 2005). This process has
been enhanced with the development of geomorphometric
software tools that are designed to expedite the analysis of cliff
face and structure (e.g., Günther, 2003). Finally, GIS-based
rockfall prediction is readily being adopted to forecast where a
rockfall might originate. Spatial modeling approaches in-
cluded the use of weighting schemes that make use of topo-
graphic, geomorphological, and geological criteria involving
faults, scree slopes within a short distance, rocky cliffs, steep
slopes, and road cuts (Baillifard et al., 2003).
3.8.5.4 GIS Modeling
Different spatial modeling approaches can be used for
landslide mapping and analysis. Perhaps the most common
type of analysis is landslide susceptibility (or the relative
spatial probability of occurrence) that relies on probability
theory and fuzzy-set theory (Remondo et al., 2003; Lu
and Rosenbaum, 2003; Tangestani, 2004), and bivariate or
multivariate statistical analyses (Santacana et al., 2003;
Fernández et al., 2003; Süzen and Doyuran, 2004). The use
of GIS spatial analysis in landslide studies has shifted from
inventorying and mapping to predicting the time, place, and
extent of a mass movement. This includes the trajectory (i.e.,
runout) of an event (Corominas et al., 2003a, b) or the in-
stability of slopes (e.g., slope-safety factor) (Esaki et al., 2001;
Bhattacharya et al., 2003). For example, the rockfall-activity
indicator metric can be used to estimate rockfall-activity
susceptibility, based on climate variables, morphological
factors, and lithological properties (Marquı́nez et al., 2003).
Artifical neural networks (ANN) have also been used for
mapping (Rumelhardt et al., 1986; Fernandez-Steeger et al.,
2002). Specifically, Ermini et al. (2005) utilized nominally
classified lithology, slope angle, profile curvature, land cover,
and upslope contributing area as input into two different
ANNs (e.g., multilayered perceptron and a probabilistic
neural network). Both ANNs produced satisfactory results,
though validation of the outcome was difficult, as the only
options were a comparison against an inventory map, or to
simply wait and see if the models properly predicted slope
failure. Unfortunately, the complexity of modeling mass-
movement processes means that many GIS-based models still
oversimplify the terrain characteristics and mechanics that
trigger landslides.
3.8.5.5 GIS-based Landslide Inventories
Landslide-inventory maps provide the occurrence and outlines
of landslides and, in the case of larger scale maps, also classify
landslide types (Carrara and Merenda, 1976; Spiker and Gori,
2000; Owen et al., 2008). In particular, geomorphological
inventory maps can be either a landslide single-event inven-
tory, or a historical landslide inventory, which summarizes
a palimpset landscape comprised of many mass movements
(Malamud et al., 2004). Landslide inventories vary in
scale, including small scale (o1:200 000), medium scale
(1:25 000–1:200 000), and large scale (41:25 000; see
Guzzetti et al., 2003). Most inventories produce discrete
boundaries (e.g., crown, depletion zones) of different failures
types (e.g., rockfalls, deep-seated slides, debris flows, etc.) so
that the basic dimensions, such as area or perimeter, can be
automatically calculated. Malamud et al. (2004) warned of the
limitations associated with compiling landslide inventories
using a GIS, including the quality and scale of imagery, mor-
phological and geological complexity, the expertise of the
geomorphologist involved, and the calculations of planar
approximations of 3D features.
3.8.5.6 GIS-based Landslide Assessment Maps
The landslide-hazard map presents the probability (likeli-
hood) of landslide occurrence throughout an area within a
specified period of time (landslide frequency). Such maps
were generated, for example, for rainfall-induced landslides
 Remote Sensing and GIScience in Geomorphological Mapping
(Terlien et al., 1995; Coe et al., 2000; Xie et al., 2004) and
earthquake-triggered landslides (Jibson et al., 1998; Refice
and Capolongo, 2002). In contrast, landslide-risk assessment
maps describe the expected annual cost of landslide damage
throughout an area based on the analysis of all possible
consequences. These maps typically require large-scale datasets
(41:25 000) and are displayed in easy to read formats for
dissemination to the general public. Fell (1994) presented
classifications results where colored zones (from ‘green’ to
‘red’) represent the level of landslide hazard in a specific de-
lineated area (from ‘low’ to ‘high’). It is commonly accepted
that the use of such linguistic expressions creates ambiguity
about the results.
The use of GIT in landslide studies has contributed much
to the knowledge of slope instability, landslide processes, and
landslide-risk assessment. Furthermore, GIT is used to map
morphology, occurrence, and susceptibility in landslide re-
gions in many countries. The rapid developments in geospa-
tial technologies permit new analytical and dissemination
capabilities. For example, field data from landslide sites can be
collected using mobile GIT technology and transferred online
in real time using existing web facilities (Kim et al., 2004;
Chang and Park, 2004), and 3D GIS can help us to better
understand specific landslide types such as rotational slides
(Esaki et al., 2001). Although GIS-based analysis and mapping
can be iterative, the data typically represent cartographic
snapshots in time, and do not characterize dynamic processes
or account for nonlinear variations of change. Furthermore,
existing GIT methods and products can be very subjective
(Hudson, 1992; Budetta et al., 2008; Kamp et al., 2010).
Product quality depends on the skill and experience of the
investigator, the complexity of the study area, and the com-
pleteness and reliability of the available data (Galli et al.,
2008).
3.8.6 Fluvial Landscapes and Landforms
Streams and fluvial environments are complex and dynamic
systems with fluctuating flow properties such as discharge,
competence, and power, and landform assemblages such as
channel form, levees, floodplains, and meanders. Geospatial
technologies can help facilitate river restoration and main-
tenance, inventory aquatic biodiversity and floodplain eco-
systems, and quantify river form, process, and rates of
morphological change. Remote surveying techniques and
digital data are valuable for the contemporary fluvial geo-
morphologist where analyzing and mapping the temporal and
spatial scaling and organization of fluvial processes and
landforms because of the far-reaching capabilities in river
science and rapid development of GIT. As in other fields in
geomorphology, remote sensing investigations are sup-
planting field-based methods (Carbonneau, 2005) to estimate
stream-water depth (Lyon et al., 1992; Winterbottom and
Gilvear, 1997; Legleiter and Roberts, 2005; Carbonneau,
2005), assess stream habitats (Wright et al., 2000; Whited
et al., 2002; Legleiter et al., 2002; Legleiter and Goodchild,
2005), and estimate the amount of suspended sediment
(Mertes, 2002). Although this shift toward digital data is
progressive, it should not be seen as a replacement for field-
213
based data acquisition. It is important to note that field data
are essential to support or augment imagery and topographic
information (Gilvear and Bryant, 2005).
Many remote sensing studies that investigate inchannel
water properties such as water-light interaction originate from
research in lacustrine or near-shore marine environments.
Within the last decade or two, fluvial geomorphologists have
accelerated their development and application of GIT to
study landscape and water properties, inundation patterns,
and channel form and flow characteristics using a variety of
spectral and spatial perspectives (Mertes, 2002). This includes
the early use of analytical photogrammetry for terrain mod-
eling and the extraction of fluvial features (Lane et al., 1994;
Barker et al., 1997; Dixon et al., 1998; Heritage et al., 1998),
to using more advanced techniques of image processing
(Lyon et al., 1992; Gilvear et al., 1995, 1998; Winterbottom
and Gilvear, 1997; Whited et al., 2002; Marcus et al., 2003;
Westaway et al., 2003; Legleiter et al., 2004; Legleiter and
Roberts, 2005) and laser scanning (Hicks et al., 2002; Thoma
et al., 2005).
Numerous challenges exist, however, when using imagery to
extract information that accurately characterizes the river en-
vironment. Within one photograph or image, a river environ-
ment will exhibit a wide range of soil moisture conditions and
image textures, including vegetation and land cover variations,
and topographic conditions, all of which influence surface re-
flectance. Gilvear and Bryant (2005) and Mertes (2002) provide
guidelines on selecting appropriate sensors (Table 1) and dis-
cuss the diverse matter/energy interactions associated with
stream-system surfaces. The challenges associated with remote
sensing water-rich environments are amplified when trying to
penetrate water to study stream bathymetry (e.g., reflectance of
shallow, clear water may be driven by reflectance of bed sedi-
ments and surface roughness) and substream landforms, or
when attempting to link stream dynamics with landforms, such
as distinguishing between pools, glides, and riffles, extracting
data about sediment concentration, amount of organics,
and surface roughness. The objective here is to briefly discuss
data sources, processing, and visualization techniques, along
with recent applications that are most widely used in fluvial
geomorphology.
3.8.6.1 Aerial Photographs
Aerial photograph interpretation in geomorphology requires
an understanding of photographic scale. Small-scale aerial
photographs lack detail but cover large areas, whereas large-
scale photos offer great detail with limited spatial coverage. As
a result, many high-resolution stream studies that depend on
remote sensing only cover limited areas or stream reaches,
which tend to characterize streams as discontinuous systems,
even though it is understood that streams vary continuously at
different scales (Marcus and Fonstad, 2008). Therefore, the
potential of aerial photography is maximized when, among
other things, sensor resolution and spatial scale of landform
and fluvial process are matched. For example, small stream
reaches (stream channel width o20 m) are best mapped and
studied by combining field-based techniques with large-scale
aerial photography (1:2500 or better), or even hand-held
214 Remote Sensing and GIScience in Geomorphological Mapping
cameras. Most analyses of medium sized rivers (channel width
between 20 and 200 m) tend to use airborne remote
sensors and relatively large-scale aerial photography (1:5000
to 1:25 000), whereas the characteristics of large rivers (width
4200 m) should be observed using space-borne sensors
(Gilvear and Bryant, 2005).
Photographic and digital stereoimagery can produce 3D
perspectives on fluvial environments, revealing topographic
variation. Although it is relatively straightforward to interpret
above-water landforms and channel characteristics based on
data derived from aerial photographs, such as studies of
braided, gravel-bed rivers (Westaway et al., 2003), and two-
and 3D channel morphology (Winterbottom, 2000), the de-
velopment of refraction-correction algorithms have opened
new opportunities to study the ongoing evolution of sub-
stream landforms, as it relates to sediment characteristics and
channel shape (Lane et al., 2000; Westaway et al., 2000; Butler
et al., 2002). If stereophotographs are merged with calibrating
camera data and ground-control points, the results are high-
accuracy DEMs, which are most effective in open (i.e., limited
shadow interference), relatively low-relief, shallow and
clear-watered, and coarse-grained braided stream systems
(Westaway et al., 2001, 2003). Clear water allows for greater
penetration of electromagnetic energy, as the percent of or-
ganic material increases energy absorption, and gravel-sized
substrate increases reflectivity. This results in an accurate
rendering of streambed depth below water, although these
characteristics are associated with stream discharge and sedi-
ment concentration (which fluctuate over time and along
stream length). A sequence of these high-resolution DEMs
that are at times subdecimeter could prove useful when
quantifying streambed morphology and channel migration
(Winterbottom and Gilvear, 1997). Again, successful use of
aerial imagery requires a clean line of site (e.g., avoiding
bridges, shadows), thus photogrammetry is somewhat limited
in headwater streams or on streams that have densely vege-
tated riparian zones with obstructed views of the water surface,
or even shadows.
Qualitative interpretations derived from single or mo-
saicked photos are a simple, but effective, method to measure
sinuosity and changes in channel width, so long as the photo
scale is known and consistent. Most applications of aerial
photography include plan-view measurements of bankfull
width, river sinuosity, and lateral channel migration (e.g.,
Brooks and Luft, 1987; Whitesell et al., 1988; Marston et al.,
1995; Leys and Werrity, 1999; Winterbottom and Gilvear,
2000; O’Connor et al., 2003; Mount and Louis, 2005; Hughes
et al., 2006). Temporal sequences of aerial photos, frequently
spanning decades, reveal stream channel change (e.g., me-
ander migration) over time (Brice, 1977; Gurnell et al., 1994;
Gurnell, 1997), potentially providing baseline data for stream
monitoring and management (Micheli and Kirchner, 2002).
For example, digitized stream centerlines, taken from a se-
quence of aerial photos, can be overlayed in GIS to analyze
shifts in channel location, and then relate this to zones of
wetness to predict stream-bank erosion potential. Micheli and
Kirchner (2002) calculated ‘eroded area polygons’ of aerial
photos in Arc/Info to quantify the distance the channel mi-
grated perpendicular to a channel centerline. They concluded
stream-banks with dry vegetation (i.e., vegetation that thrives
in relatively dry conditions) migrate faster than those with wet
vegetation, but also emphasized the value of sequential aerial
photos and geoprocessing tools in GIS in quantifications
of stream-channel migration over time. Photogrammetry of
stream-banks also offers an objective method for calculating
volume of sediment loss due to erosion, as a means of
monitoring stream-bank stability, but also to determine the
relative importance of stream-banks as sediment sources
(Barker et al., 1997).
Due to the desire for fine-resolution, fluvial geomorph-
ologists now consider geospatial error as a necessary aspect of
analysis for the successful application of aerial photography to
study fluvial environments. The positional errors associated
with aerial images and DEMs are just now being recognized as
a critical influence on lateral two-dimensional measurements
from aerial photographs. Error analysis of data is now being
rigorously investigated to provide a conceptual foundation for
future stream studies. This includes testing the number and
type of ground-control points, interpolation issues, the prac-
ticality of using simple metrics like root-mean-square error
(RMSE), and the impact that positional error has on stream
studies (e.g., Mount et al., 2003; Mount and Louis, 2005;
Hughes et al., 2006). Finally, many countries or agencies have
extensive libraries of aerial photographs, and using modern
photogrammetric methods, some of these archived images can
still be used to reevaluate river dynamics and change over the
past 50 years or more (Marcus and Fonstad, 2008; Lane et al.,
2010).
3.8.6.2 Airborne And Satellite Sensors
Early applications of multispectral data on river systems fo-
cused on estimating suspended sediment concentrations in
relatively large rivers using Landsat, Coastal Zone Color
Scanner (CZCS), and AVHRR data (e.g., Amos and Alföldi,
1979; Li, 1993; Mertes et al., 1993). For example, Landsat TM
has a ‘water penetration band’ (0.45–0.54 mm) that can
penetrate 10–20 m of clear water. This allows for bathymetric
mapping of lakes and rivers (the optimum wavelength is
0.48 mm), but the utility of Landsat data is limited by spatial
resolution, which is too large to adequately sample small
and many medium-sized streams. With specific wavelengths,
suspended-sediment texture in rivers can be distinguished
(0.58–0.69 mm), whereas it is possible to use near infrared
wavelengths (0.714–0.880 mm) to estimate the amount of
suspended sediments (Han and Rundquist, 1994). Therefore,
advances in sensor technology (i.e., hyperspectral sensors) and
data availability have exponentially increased the possibilities
to generate deeper insights into the relationships between
stream dynamics and landform evolution. Although digitally
acquired data is used to characterize stream properties and
landforms, remote-sensing applications are still limited to case
studies, rather than operational assessment, or comprehensive
assessment of entire watershed systems.
Hyperspectral data, collected from satellites or aircraft,
cameras suspended over rivers, or hand-held spectrometers, can
be used to elucidate the physical conditions and processes in
stream environments, as biophysical conditions and processes
influence matter/energy interactions. Legleiter et al. (2004)
 Remote Sensing and GIScience in Geomorphological Mapping
combined field-derived spectral data with radiative-transfer
modeling to illustrate the effect stream depth and substrate
characteristics, water-surface roughness, and suspended-sedi-
ment concentration have on spectra. Fonstad and Marcus
(2005) merged airborne hyperspectral images with local
stream-gage data and hydrologic principles such as Manning’s
equation, to develop the hydraulically assisted bathymetry
(HAB) model. The HAB model generates stream-depth maps
based on image brightness, and the authors demonstrated the
capability to integrate a wide range of sensor data, including
airborne multispectral and space-borne imagery such as
IKONOS. They also demonstrate the use of archived imagery
to map 3D changes in river bathymetry. As is the case with
aerial photography, the use of these images is limited to water
turbidity conditions and a clear line-of-sight.
3.8.6.3 Active Sensors
Although the use of hyperspectral data has been explored in
more detail in river science, a number of surface hydrologists
and fluvial geomorphologists are using active sensors to map
stream bottom, stage height, and discharge (Spicer et al., 1997;
Costa et al., 2000; Jones et al., 2007). For example, Jung et al.
(2010) characterized and compared floodplains in the
Amazon and Congo using InSAR from the Japanese Earth
Resources Satellite-1 (JERS-1). They used the spatial and
temporal variations in water-surface levels to describe complex
fluvial processes and describe floodplain-building features
between two different fluvial systems. One reason for this
work was to illustrate the value of, and contribute to the de-
sign of a future radar satellite system that aims to improve
global water monitoring (e.g., Surface Water and Ocean
Topography (SWOT) Mission). Similar research has also relied
on radar altimetry to estimate stream discharge, slope, and
water-surface topography (Koblinsky et al., 1993; Birkett et al.,
2002; Kouraev et al., 2004; Frappart et al., 2008; Calmant and
Seyler, 2006; Leon et al., 2006), even though it is recom-
mended that field-based data still support interpretations from
remote sensing investigations (e.g., cross-sectional profiles).
Laser ranging systems have proven to be a valuable tech-
nology that can be used to monitor subtle landform changes
and measure ongoing fluvial processes such as stream-bank
erosion (Thoma et al., 2005; Notebaert et al., 2009). The de-
tection of depressions and subtle changes in topography, es-
pecially with regards to small channel forms, is facilitated by
the fine resolution typical of LiDAR data (Figure 18). Much
like aerial photography, sequential LiDAR data provides a
temporal perspective on stream processes, including change
migration and stream-bank erosion (Notebaert et al., 2009).
Although LiDAR data are not readily available for many
watersheds, and data-collection missions are expensive, LiDAR
accuracy permits mapping of paleochannels and other low-
relief landforms, and facilitates hydraulic modeling efforts
(e.g., Cobby et al., 2001; French, 2003; Dal Cin et al., 2005)
and studies on morphological change (e.g., Lohani and
Mason, 2001; Challis, 2006; Challis and Howard, 2006;
Nelson et al., 2006). Furthermore, the signal penetrates vege-
tation and LiDAR data can be acquired at all times of the year
in some of the most inaccessible fluvial environments in the
215
world. Despite the high level of topographic detail produced
from LiDAR data, field-based data such as coring and ground-
penetrating radar are still frequently required to properly in-
terpret the importance of mapped features (Notebaert et al.,
2009).
Floodplains, especially those adjacent to large stream
systems (e.g., Amazon, Mississippi), contain an array of
landforms indicative of fluctuations in stream flow and water-
table elevation (i.e., hydroperiod). Yet, despite being dynamic
Earth surface features, the relative relief within a floodplain
can be quite small. To address this limitation, and considering
the dynamic nature of floodplains, geomorphic studies now
integrate data from various sources, at various scales, to sys-
tematically delineate landforms in floodplain environments.
Floodplain features have been distinguished using satellite
imagery (FCC 5, 4, 3 from Landsat ETM þ and the L-band of
JERS-1) and topographic information (Figure 19; Hamilton
et al., 2006). The high backscatter on radar imagery clearly
denoted flooded forests, whereas Landsat imagery character-
ized the meander melts. An object-based analysis then sim-
ultaneously analyzed image elements and topography to
produce homogeneous terrain objects, which were then hier-
archically classified. Unfortunately, most of the imagery did
not capture flooding events, meaning that some parts of the
classification (i.e., flooded terrain) relied more on hypothe-
sized classes.
Mapping by Marston et al. (1995) focused on the inter-
action between the stream channel, floodplain, and bio-
diversity. They produced a landscape change matrix that
captured channel migration and floodplain disturbance along
the Ain River, France between 1945 and 1991 (Figure 20).
Although their goal was more focused on understanding
vegetation development, the spatial distribution of river de-
posits and landforms (i.e., landscape units) reflected the im-
pact humans have on channel migration (e.g., dams, increased
summer discharge to support salmon) and floodplain dis-
turbance. Changes in channel location and behavior then
caused changes in the distribution and diversity of vegetation,
as a decrease in floodplain disturbance increased the growth of
larger shrubs and trees, whereas areas once dominated by
hydrophytic plants were replaced with mixed forests and
hardwood trees.
3.8.6.4 Geographic Information Systems
Fluvial geomorpholgists have integrated topographic, soil,
lithology, and precipitation data into GIS databases to sys-
tematically evaluate the driving forces behind stream pro-
cesses, or to piece together the evolution of fluvial landforms.
Jordan and Fonstad (2005) estimated water-surface slope and
average depth within a GIS so they could use Manning’s
equation to estimate stream velocity, discharge, and power for
the Brazos River in Texas, U.S. Finlayson and Montgomery
(2003) analyzed the influence of DEM resolution and map
projection on predictions of erosion in large, mountainous
stream systems (e.g., Olympic Mountain Range, U.S.). They
concluded that DEM resolution and map projection can
distort area and stream length (typically minimizing), area-
discharge relationships may be relatively weak when involving
216 Remote Sensing and GIScience in Geomorphological Mapping
169 000 169 200
0 0.1 km
> 35 m
162 200
30 m
162 000
(a)
162 200
162 000
(c) 169 000 169 200
Figure 18 Comparison of two LiDAR datasets in Belgium (Dijle and Amblève). (a) Flemish LiDAR at a resolution of 5 m; (b) Walloon LiDAR at a
resolution of 1 m; (c) Hill-shade map of F-LiDAR data (azimuth: 901); (d) Hill-shade map of the W-LiDAR data (azimuth: 901). Modified from
Notebaert, B., Verstraeten, G., Govers, G., Poesen, J., 2009. Qualitative and quantitative applications of LiDAR imagery in fluvial geomorphology.
Earth Surface Processes and Landforms 34, 217–231.
multiple climatic zones, and the stream-power law is sensitive
to DEM resolution (e.g., stream-power per unit area decreased
as DEM resolution decreased). Therefore, numerous meth-
odological issues must be considered before using GIS-based
erosion models to predict denudation for large mountainous
stream systems, which are complicated by topographic vari-
ability, changes in lithology, and abrupt changes in climate.
Nevertheless, such investigations have revealed the value of
integrating data to elicit key factors that influence topographic
evolution in large stream systems, a task that would otherwise
be difficult to complete using more traditional mapping
methods.
Geomorphologists can derive and extract fluvial features
from DEMs, such as stream networks, watershed boundaries,
flow accumulation grids, and nodes or confluences of tribu-
taries. These characteristics are based upon the use of geo-
morphometric parameters that include slope azimuth and
curvature, whereas connectivity, or topology, predict flow
paths and relationships between stream channels. So common
169 000 169 200
162 200
162 000
(b)
162 200
162 000
(d) 169 000 169 200
are these applications that many software programs offer
functions that are specific to stream research (e.g., the Hydro
toolbox in ArcGIS, IDRISI, River Tools).
Advances in remote sensing and GIS provide the ability to
measure river-system behavior at various spatial and temporal
scales. Relative to most field-based techniques, remote sur-
veying is the only practical method to monitor and map flu-
vial environments, especially considering the high-resolution
imagery and DEMs, the temporal frequency of airborne
and spaceborne imagery, and the numerous regions of the
electromagnetic spectrum that can be used to assess the bio-
physical characteristics of inundated or floodplain environ-
ments. Furthermore, topographic information is now being
integrated with spectral data, significantly altering how river
scientists view river-system components and behavior, which
in turn impacts our dependence on GIT for studying fluvial
environments.
The contributions of remote sensing in river science are
substantial. New data and information are being used as input
 Remote Sensing and GIScience in Geomorphological Mapping 217

0 25 km

River channels (open water) Meander belt (early succession herbaceous vegetation) Palm swamp (dry)

Lakes (mostly oxbow) Meander belt (late successional broadleaf canopy) Palm swamp (flooded)

River bars (seasonally exposed gravel, sand or mud) Backswamp (low-lying forest outside meander belts) Clouds & shadows

Figure 19 Image analysis of the Madre de Dios river system in Peru based on a Landsat ETM þ FCC (5, 4, 3) (left), and a floodplain
classification based on SRTM, Landsat ETM þ , and JERS-1 (right). Modified from Hamilton, S.K., Kellndorfer, J., Lehner, B., Tobler, M., 2006.
Remote sensing of flood plain geomorphology as a surrogate for biodiversity in a tropical river system (Madre de Dios, Peru). Geomorphology
89, 23–38.

Unvegetated gravel-cobble deposits

Sand-gravel-cobble deposits (low shrubs)

Sand-gravel-cobble deposits (high shrubs)

Old channels

Dense, diverse, mesophytic shrubs

Mixed forest, dominated by hardwoods

Dry grassland on sandy soils (few shrubs)

Dry grassland on gravel-cobble deposits

Cleared land for mines, campgrounds, agriculture

Water
0 0.5 1.0 km

Figure 20 Landscape unit maps for a portion of the Ain River 100 year floodplain near Mollon, France between 1945 (left) and 1991 (right).
Note reduction in meandering and increase in vegetation that requires drier conditions (i.e., reduced floodplain disturbance). Modified from
Marston, R.A., Girel, J., Pautou, G., Piegay, H., Bravard, J.P., Arneson, C., 1995. Channel metamorphosis, floodplain disturbance, and vegetation
development: Ain River, France. Geomorphology 13, 121–131.
218 Remote Sensing and GIScience in Geomorphological Mapping
for hydrological modeling (spatial hydrology), permit de-
tection and mapping of previously undetected landforms in
densely vegetation floodplains, enable the generation of
quantitative estimates of stream characteristics such as dis-
charge and power, permit the monitoring of stream-channel
migration, and enable the characterization of habitat bio-
diversity based upon stream conditions. Based on these ad-
vances in river science, some have argued for permanent
remote-sensing river observatories (Marcus and Fonstad,
2008), which would provide standardized imagery for a host
of users, while minimizing the impact of ongoing privatiza-
tion of remote-sensing technology and data that is of interest
to geomorphologists.
3.8.7 Conclusion
Rapid advances in remote sensing and GIT have now signifi-
cantly altered the way that geomorphologists analyze data and
study processes and system dynamics. Continued multi-
disciplinary investigations of landform mapping that combine
the theoretical underpinnings that are fundamental to geo-
morphology and GIScience will speed-up and advance our
capabilities for improved mapping and assessment of geo-
morphological systems. The relative ubiquity of spectral data
from unique regions of the electromagnetic spectrum provide
us with a variety of biophysical information that permit in-
novative perspectives to map, extract, and measure landforms.
Advances in representation, geomorphometry, and AI con-
tinue to alter how geomorphic concepts are implemented in
the software domain, and there are mutual challenges com-
mon to geomorphology and GIScience, such as space-time
and multidimensional representation, characterization of
spatial properties and topological relationships, and predictive
modeling and mapping, that ensure that GIT will continue
contributing to the field of geomorphology for years to come.
Although there are other developments in geospatial
technologies related to landform mapping not covered in this
review (e.g., aeolian, coastal, and tectonic geomorphology),
several issues and future directions have emerged that include:
• Scale is critical in landform identification and classification,
including the scaling relationships between land-surface
geometric properties and landform size (Walsh et al., 1998;
Evans, 2000; Bishop et al., 2012). It is now possible to es-
timate the scale at which the spatial organization of the
topography best characterizes landforms (e.g., Eisank et al.,
2010), which can be used to guide computational scale se-
lection and methodological design and interpretation. In
particular, multiscale analysis approaches can be used to
determine the amount and patterns of spatial autocorrela-
tion for various landscape properties. Moreover, it is im-
portant to consider temporal scales as well, given that many
studies are focusing on characterizing the operational scale
of processes and landform evolution.
• Geomorphologists are also beginning to recognize the im-
portance of semantics in landform classification. Eisank et al.
(2010) stressed that semantic modeling offers one possible
solution for relating a particular landform concept to repre-
sentation, analysis, and visualization using GIT. Furthermore,
semantic modeling requires a solid knowledge-based classi-
fication scheme, as noted by Raseman et al. (2004), Dunlop
and Clark (2006), and Grosse et al. (2009). This will require a
rigorous, universal, and quantitative treatment of landform
properties that includes morphology in order to formalize
landform classification relationships and hierarchies (Deng,
2007). Thus, the merging of GIScience into geomorphology is
forcing geomorphologists to revisit landform taxonomic
schemes, as GIScience investigations attempt to standardize
geomorphological mapping approaches based upon a scien-
tific treatment of geomorphological concepts and infor-
mation requirements.
• While it is acknowledged that automating landform identi-
fication, extraction, and classification is an ultimate goal in
geomorphology, this continues to be a challenge for geo-
morphologists. Klingseisen et al. (2008) customized a GIS to
develop LANDFORM, a semiautomated classification of
landform elements (e.g., crest, flats, depressions). Based on
the semiautomated derivation of these elements, a landscape
could potentially be divided into individual morpho-
logical types. Although this particular automation was de-
signed for hydrological structures and drainage patterns,
landform characterization based on basic landform element
properties most likely will contribute to the automation of
landform identification and classification in other fields of
geomorphology.
• Distinguishing spatial and temporal patterns has and con-
tinues to dominate landform-mapping research. This usually
requires the design of spatial metrics or indices that quantify
morphological conditions (Bishop et al., 2012), and com-
position and spatial patterns of landscape units (Walsh et al.,
1998). Many of these can be found in the literature, and some
are built into customized programs or software packages.
Landform patterns, however, are usually a remnant of a par-
ticular process (e.g., subglacial conditions and drumlins),
such that pattern identification and measurement contribute
to our understanding of the dominant processes responsible
for landscape evolution.
• The need to access various forms of data and disseminate
information has grown rapidly within the past decade. Web-
based software tools can now be used for sharing data (e.g.,
TM
Google Earth , NASA World Wind) and permit advanced
cartographic functionalities. The capability to distribute pre-
and postprocessed images and data means that reporting
sources, methods, and positional accuracies should now
become mandatory, so as to reduce error propagation be-
tween studies (Napieralski et al., 2007b) and to develop
universally-accepted methodological protocols for landform
mapping.
Remote sensing and GIT have greatly increased objectivity
and efficiency of geomorphological investigations, but just as
important, it is forcing us to merge different theoretical per-
spectives on space, time, processes, and systems. As illustrated
in Figure 1, the ability to formally segment landscapes into
landforms, and then align the morphological characteristics
with process mechanics, depends on the quality of data, range
of perspectives (e.g., topography, spectral, temporal), and a
much needed geologic contextualization of the area. Land-
form classifications continue to be modified, as the inclusion
 Remote Sensing and GIScience in Geomorphological Mapping
of more modern data and the use of new analytical techniques
(e.g., fuzzy logic, semivariance) reveal previously un-
recognized characteristics. Such progress will continue, al-
though the use of GIScience in geomorphology requires the
user to have experience with an ever-expanding profusion of
digital data, software interfaces, geostatistical tools, spatial
models, visualization approaches, and now space-time theory.
Regardless, it is now almost standard practice to characterize
landforms using geomorphometry and GIS, and this has made
significant contributions to our understanding of the topo-
graphic, climatic, biological, and geologic processes respon-
sible for shaping Earth’s surface.
References
Alcantara-Ayala, I., Esteban-Chavez, O., Parrot, J.F., 2006. Landsliding related to
land-cover change: a diachronic analysis of hillslope instability distribution in
the Sierra Norte, Puebla, Mexico. Catena 65, 152–165.
Alexander, D.E., 2008. A brief survey of GIS in mass-movement studies, with
reflections on theory and methods. Geomorphology 94, 261–267.
Amos, C.L., Alföldi, T.T., 1979. The determination of suspended sediment
concentration in a macrotidal system using Landsat data. Journal of Sedimentary
Petrology 49, 159–174.
Anders, N., Seijmonsbergen, A., Bouten, W., 2009. Multi-scale and object-oriented
image analysis of high-red LiDAR data for geomorphological mapping in Alpine
mountains. In: Purves, R., Gruber, S., Straumann, R., Hengl, T. (Eds.),
Proceedings of Geomorphometry 2009. University of Zurich, Zurich, pp. 61–65.
Andrews, J.T., 1982. On the reconstruction of Pleistocene ice sheets: a review.
Quaternary Science Reviews 1, 1–30.
Aniya, M., Etaya, M., Shimoda, H., 1985. Evaluation of Landsat data for landslide
identification as a means for watershed management. Journal of the Japan
Society of Photogrammetry and Remote Sensing 24, 17–21.
Aniya, M., Naruse, R., Yamaguchi, S., 2002. Utilization of 6  6 cm format vertical
aerial photographs for repetitive mapping of surface morphology and
measurement of flow velocities of a small glacier in a remote area: Glacier Soler,
Hielo Patagónico Norte, Chile. Annals of Glaciology 34, 385–390.
Aniya, M., Sato, H., Naruse, R., Skvarca, P., Casassa, G., 1996. The use of satellite
and airborne imagery to inventory outlet glaciers of the Southern Patagonia
Icefield, South America. Photogrammetric Engineering & Remote Sensing 62,
1361–1369.
Aniya, M., Welch, R., 1981. Morphometric analyses of Antarctic cirques from
photogrammeric measurements. Geografiska Annaler 63A, 41–53.
Aste, J.P., Girault, F., 1995. GIS, SPOT, DEM and morphology of major land
movements. Proceedings of the International Symposium on Landslides 6,
1539–1545.
Baeza, R.C., Esteban-Chavez, O., Marcos-Lopez, J., Pena-Garnica, R.J., Alcantara-
Ayala, I., 2006. Slope instability on pyroclastic deposits: landslide distribution
and risk mapping in Zacapoaxtla, Sierra Norte de Puebla, Mexico. Journal of
Mountain Science 3, 1–19.
Baillifard, F., Jaboyedoff, M., Sartori, M., 2003. Rockfall hazard mapping along a
mountainous road in Switzerland using a GIS-based parameter rating approach.
Natural Hazards in Earth System Science 3, 435–442.
Baldi, P., Cenni, N., Fabris, M., Zanutta, A., 2008. Kinematics of a landslide derived
from archival photogrammetry and GPS data. Geomorphology 102, 435–444.
Ballantyne, C.K., 1997. Periglacial trimlines in the Scottish Highlands. Quaternary
International 38/39, 119–136.
Ballantyne, J.D., Dean, Jr. D.R., Thompson, B.L., 1987. Monitoring landslide
movement with a 35-mm camera. Transportation Research Record 1119, 47–54.
Bamber, J.L., Hardy, R.J., Joughin, I., 2000. An analysis of balance velocities over
the Greenland ice sheet and comparison with synthetic aperture radar
interferometry. Journal of Glaciology 46, 67–74.
Barker, R., Dixon, L., Hooke, J., 1997. Use of terrestrial photogrammetry for
monitoring and measuring bank erosion. Earth Surface Processes and
Landforms 22, 1217–1227.
Barlow, J., Martin, Y., Franklin, S.E., 2003. Detecting translational landslide scars
using segmentation of Landsat ETM þ and DEM data in the northern Cascade
Mountains, British Columbia. Canadian Journal of Remote Sensing 29,
510–517.
219
Barr, I.D., Clark, C.D., 2009. Distribution and pattern of moraines in Far NE Russia
reveal former glacial extent. Journal of Maps 2009, 186–193.
Bastero, C.F., Lagmay, A., 2006. Estimating SiO2 content of lava deposits in the
humid tropics using remotely sensed imagery. Journal of Volcanology and
Geothermal Research 151(4), 357–364.
Belknap, E.P., Gilmore, J.B., 1987. Use of time-lapse movie photography in
landslide monitoring. Transportation Research Record 1119, 39–46.
Bencardino, M., 2005. Landslide hazards and risk assessment at the Basento river
basin using Landsat satellite imagery. Bollettino Della Societa Geografica Italiana
12, 115–130.
Bhattacharya, G., Jana, D., Ojha, S., Chakraborty, S., 2003. Direct search for
minimum reliability index of earth slopes. Computer and Geotechnics 30,
455–462.
Bindschandler, R.A., Scambos, T.A., 1991. Satellite-image-derived velocity field of
an Antarctic ice stream. Science 252, 242–246.
Birkett, C.M., Mertes, A.A.K., Dunne, T., Costa, M.H., Jasinski, M.J., 2002. Surface
water dynamics in the Amazon Basin: application of satellite radar altimetry.
Journal of Geophysical Research 107. http://dx.doi.org/10.1029/2001JD000609.
Bishop, M.P., James, L.A., Shroder, J.F., Walsh, S.J., 2012. Geospatial technologies
and digital geomorphological mapping: concepts, issues and research.
Geomorphology 137, 5–26.
Bishop, M.P., Shroder, Jr. J.F., 2004. GIScience and mountain geomorphology:
overview, feedbacks, and research directions. In: Bishop, M.P., Shroder, J.F.
(Eds.), Geographic Information Science and Mountain Geomorphology. Springer-
Praxis, New York, pp. 1–31.
Bishop, M.P, Shroder, Jr. J.F., Colby, J.D., 2003. Remote sensing and
geomorphometry for studying relief production in high mountains.
Geomorphology 55, 345–361.
Bishop, M.P, Shroder, Jr. J.F., Hickman, B.L., Copland, L., 1998. Scale-dependent
analysis of satellite imagery for characterization of glacier surfaces in the
Karakoram Himalaya. Geomorphology 21, 217–232.
Bishr, Y., 1998. Overcoming the semantic and other barriers to GIS interoperability.
International Journal of Geographical Information Science 12, 299–314.
Booth, A.M., Roering, J.J., Perron, J.T., 2009. Automated landslide mapping using
spectral analysis and high-resolution topographic data: Puget Sound lowlands,
Washington, and Portland Hills, Oregon. Geomorphology 109, 132–147.
Boulton, G.S., Clark, C.D., 1990. A highly mobile Laurentide ice sheet revealed by
satellite images of glacial lineations. Nature 346, 813–817.
Bozzano, F., Mazzanti, P., Prestininzi, A., 2008. A radar platform for continuous
monitoring of a landslide interacting with an under-construction infrastructure.
Italian Journal of Engineering Geology and Environment 8, 71–87. (In Italian).
Bradwell, T., Stoker, M., Krabbendam, M., 2008. Megagrooves and streamlined
bedrock in NW Scotland: the role of ice streams in landscape evolution.
Geomorphology 97, 135–156.
Brice, J., 1977. Lateral Migration of the Middle Sacramento River, California.
U.S. Geological Survey, Water Resources Investigation, Sacramento, CA,
pp. 77–43.
Bronge, L.B., Bronge, C., 1999. Ice and snow-type classification in the Vestfold
Hills, East Antarctica, using Landsat-TM data and ground radiometer
measurements. International Journal of Remote Sensing 20, 225–240.
Brook, G.A., Luft, E.R., 1987. Channel pattern changes along the lower Oconee
River, Georgia, 1805/7 to 1949. Physical Geography 8, 191–209.
Brueckl, E., Brunner, F.K., Kraus, K., 2006. Kinematics of a deep-seated landslide
derived from photogrammetric, GPS and geophysical data; remote sensing and
ground-based geophysical techniques for recognition, characterisation and
monitoring of unstable slopes. Engineering Geology 88, 149–159.
Brunsden, D., 1993. The persistence of landforms. Zeitschrift für Geomorphologie
Supplementband 93, 13–28.
Budetta, P., Santo, A., Vivenzio, F., 2008. Landslide hazard mapping along the
coastline of the Cilento region (Italy) by means of a GIS-based parameter rating
approach. Geomorphology 94, 340–352.
Burbank, D.W., Anderson, R.S., 2001. Tectonic Geomorphology. Wiley-Blackwell,
Hoboken, NJ, pp. 1–274.
Burrough, P.A., van Gaans, P.F.M., MacMillan, R.A., 2000. High-resolution
landform classification using fuzzy k-means. Fuzzy Sets and Systems 113,
37–52.
Butler, D., 2001. Geomorphic process-disturbance corridors: a variation on a
principle of landscape ecology. Progress in Physical Geography 25, 237–248.
Butler, J.B., Lane, S.N., Chandler, J.H., Porfiri, K., 2002. Through-water close range
digital photogrammetry in flume and field environments. Photogrammetric
Record 17, 419–439.
Calmant, S., Seyler, F., 2006. Continental surface waters from satellite altimetry.
Comptes Rendus Geoscience 338, 111–122.
220 Remote Sensing and GIScience in Geomorphological Mapping
Carbonneau, P.E., 2005. The threshold effect of image resolution on image-based
automated grain size mapping in fluvial environments. Earth Surface Processes
and Landforms 30, 1687–1693.
Carn, S.A., 1999. Application of Synthetic Aperture Radar (SAR) imagery to volcano
mapping in the humid tropics: a case study in East Java, Indonesia. Bulletin of
Volcanology 61, 92–105.
Carnec, C., Massonet, D., King, C., 1996. Two examples of the use of SAR
interferometry on displacement fields of small spatial extent. Geophysical
Research Letters 23, 3579–3582.
Carrara, A., Guzzetti, F. (Eds.), 1995. Geographical Information Systems in
Assessing Natural Hazards. Kluwer, Dordrecht, 353 pp.
Carrara, A., Merenda, L., 1976. Landslides inventory in northern Calabria, southern
Italy. Geological Society America Bulletin 87, 1229–1246.
Carrara, A., Pike, R.J., 2008. GIS technology and models for assessing landslide
hazard and risk. Geomorphology 94, 257–260.
Catani, F., Farina, P., Moretti, S., Nico, G., Strozzi, T., 2005. On the application of
SAR interferometry to geomorphological studies: estimatation of landform
attributes and mass movements. Geomorphology 66, 119–131.
Chacon, J., Irigaray, C., Fernandez, T., El Hamdouni, R., 2006. Engineering geology
maps: landslides and geographical information systems. Bulletin Engineering
Geology and Environment 65, 341–411.
Chadwick, J., Dorsch, S., Glenn, N., Thackray, G.D., Shilling, K., 2005. Application
of multitemporal high-resolution imagery and GPS in a study of the motion of a
canyon rim landslide: remote sensing and geospatial information for natural
hazards characterization. Journal of Photogrammetry and Remote Sensing 59,
212–221.
Challis, K., 2006. Airborne laser altimetry in alluviated landscapes. Archaeological
Prospection 13(2), 103–127.
Challis, K., Howard, A.J., 2006. A review of trends within archaeological remote
sensing in alluvial environment. Archaeological Prospection 13(4), 231–240.
Chandler, J., Cooper, M., 1988. Monitoring the development of landslides using
archival photography and analytical photogrammetry. Land and Minerals
Surveying 6, 576–584.
Chandler, J.H., Fryer, J.G., Jack, A., 2005. Metric capabilities of low-cost digital
cameras for close range surface measurement. Photogrammetric Record 20,
12–26.
Chang, Y.S., Park, H.D., 2004. Development of a web-based geographic information
system for the management of borehole and geological data. Computer and
Geosciences 30, 887–897.
Charlesworth, J.K., 1928. The glacial geology of North Mayo and West Sligo.
Proceedings of the Royal Irish Academy 38, 100–115.
Chen, R.-F., Chang, K.-J., Angelier, J., Chan, Y.-C., Deffontaines, B., Lee, C.-T., Lin,
M.-L., 2006. Topographical changes revealed by high-resolution airborne LiDAR
data: the 1999 Tsaoling landslide induced by the Chi-Chi earthquake.
Engineering Geology 88, 160–172.
Chiang, S.-H., Chang, K.-T., 2009. Application of radar data to modeling rainfall-
induced landslides. Geomorphology 103, 299–309.
Chirico, G.D., et al., 2009. Lava flow hazard at Nyiragongo Volcano, DRC. Bulletin
of Volcanology 71(4), 375–387.
Chorley, R.J., Schumm, S.A., Sudgen, D.E., 1984. Geomorphology. Methuen,
London.
Chou, W.-C., Lin, W.-T., Lin, C.-Y., 2009. Vegetation recovery patterns assessment
at landslides caused by catastrophic earthquake: a case study in central Taiwan.
Environmental Monitoring and Assessment 152, 245–257.
Claessens, L., Heuvelink, G.B.M., Schoorl, J.M., Veldkamp, A., 2005. DEM
resolution effects on shallow landslide hazard and soil redistribution modelling.
Earth Surface Processes and Landforms 30, 461–477.
Clark, C.D., 1993. Mega-scale glacial lineations and cross-cutting ice-flow
landforms. Earth Surface Processes and Landforms 18, 1–29.
Clark, C.D., 1997. Reconstructing the evolutionary dynamics of former ice sheets
using multitemporal evidence, remote sensing and GIS. Quaternary Science
Reviews 16, 1067–1092.
Clark, C.D., Evans, D.J., Khatwa, A., et al., 2004. Map and GIS database of
glacial landforms and features related to the last British Ice Sheet. Boreas 33,
359–375.
Clark, C.D., Knight, J.K., Gray, J.T., 2000. Geomorphological reconstruction of the
Labrador sector of the Laurentide Ice Sheet. Quaternary Science Reviews 19,
1343–1366.
Clark, C.D., Meehan, R.T., 2001. Subglacial bedform geomorphology of the Irish Ice
Sheet reveals major configuration changes during growth and decay. Journal of
Quaternary Science 16, 483–496.
Clark, C.D., Wilson, C., 1994. Spatial analysis of lineaments. Computers &
Geosciences 20, 1237–1258.
Close, M.H., 1867. Notes on the general glaciation of Ireland. Journal of the Royal
Geological Society of Ireland 1, 207–242.
Cobby, D.M., Mason, D.C., Davenport, I.J., 2001. Image processing of airborne
scanning laser altimetry data for improved river flood modeling. ISPRS Journal
of Photogrammetry and Remote Sensing 56(2), 121–138.
Coe, J.A., Michael, J.A., Crovelli, R.A., Zavage, W.A., 2000. Preliminary map
showing landslides densities, mean recurrence intervals, and exceedance
probabilities as determined from historic records, Seattle, Washington. USGS
Open-File report 00-303.
Colesanti, C., Wasowski, J., 2006. Investigating landslides with space-borne
synthetic aperture radar (SAR) interferometry: remote sensing and ground-based
geophysical techniques for recognition, characterisation, and monitoring of
unstable slopes. Engineering Geology 88, 173–199.
Coppock, J.T., 1995. GIS and natural hazards: an overview from a GIS perspective.
In: Carrara, A., Guzzetti, F. (Eds.), Geographical Information Systems in
Assessing Natural Hazards. Kluwer, Dordrecht, pp. 21–34.
Corazzato, C., Tibaldi, A., 2006. Fracture control on type, morphology and
distribution of parasitic volcanic cones: an example from Mt. Etna. Italy Journal
of Volcanology and Geothermal Research 158(1–2), 177–194.
Corominas, J., Copons, R., Vilaplana, J.M., Altimir, J., Amigó, J., 2003b. Integrated
landslide susceptibility analysis and hazard assessment in the principality of
Andorra. Natural Hazards 58, 421–435.
Corominas, J., Santacana, N., 2003a. Stability of the Vallcebre translational
slide, Eastern Pyrenees (Spain) by means of a GIS. Natural Hazards 58,
473–485.
Corsini, A., Borgatti, L., Cervi, F., Dahne, A., Ronchetti, F., Sterzai, P., 2009.
Estimating mass-wasting processes in active earth slides: earth flows with time-
series of high-resolution DEMs from photogrammetry and airborne LiDAR.
Natural Hazards and Earth System Sciences 9, 433–439.
Costa, J.E., Spicer, K.R., Cheng, R.T., et al., 2000. Measuring stream discharge by
non-contact methods: a proof of concept experiment. Geophysical Research
Letters 27(4), 553–556.
Crowley, J., Zimbelman, D., 1997. Mapping hydrothermally altered rocks on Mount
Rainier, Washington, with Airborne Visible Infrared Imaging Spectrometer
(AVIRIS) data. Geology 25, 559–569.
Czuchlewski, K.R., Weissel, J.K., Kim, Y., 2003. Polarimetric synthetic aperture radar
study of the Tsaoling landslide generated by the 1999 Chi-Chi earthquake,
Taiwan. Journal of Geophysical Research 108, 6006.
Dahal, R.K., Hasegawa, S., Nonomura, A., Yamanaka, M., Dhakal, S., 2008. DEM-
based deterministic landslide hazard analysis in the lesser Himalaya of Nepal.
Georisk 2, 161–178.
Dal Cin, C., Moens, L., Dierickx, P., Bastin, G., Zech, Y., 2005. An integrated
approach for realtime floodmap forecasting on the Belgian Meuse River. Natural
Hazards 36(1–2), 237–256.
Dalrymple, J.B., Blong, R.J., Conacher, A.J., 1968. A hypothetical nine unit land
surface model. Zeitschrift fur Geomorphologie 12, 60–76.
Davis, A.M., Moore, J.M., Mason, P.J., 2000. A study of landslide hazard in S.E.
Spain using integrated DEM, orthophotography and Landsat ETM imagery.
Proceedings of the Fourteenth International Conference on Applied Geologic
Remote Sensing 14, 271–276.
Delacourt, C., Allemand, P., Casson, B., Vadon, H., 2004. Velocity field of the ‘‘La
Clapiere’’ landslide measured by the correlation of aerial and QuickBird satellite
images. Geophysical Research Letters 31, 5.
Deng, Y., 2007. New trends in digital terrain analysis: landform definition,
representation, and classification. Progress in Physical Geography 31,
405–419.
Derron, M.-H., Jaboyedoff, M., Blikra, L.H., 2005. Preliminary assessment of
rockslide and rockfall using a DEM (Oppstadhornet, Norway). Natural hazards in
Earth System Science 5, 285–292.
Dewitte, O., Jasselette, J.C., Cornet, Y., van den Eeckhaut, M., Collignon, A.,
Poesen, J., Demoulin, A., 2008. Tracking landslide displacements by multi-
temporal DTMs: a combined aerial stereophotogrammetric and LIDAR approach
in western Belgium. Engineering Geology 99, 11–22.
Dikau, R., 1989. The application of a digital relief model to landform analysis in
geomorphology. In: Raper, J. (Ed.), Three Dimensional Applications in
Geographic Information Systems. Taylor & Francis, London, pp. 51–77.
Dixon, L.F.J., Barker, R., Bray, M., et al., 1998. Analytical photogrammetry for
geomorphological research. In: Lane, S., Richards, K., Chandler, J. (Eds.),
Landform Mapping, Modelling, and Analysis. Wiley, Chichester, pp. 49–63.
Domakinis, C., Oikonomidis, D., Astaras, T., 2008. Landslide mapping in the
coastal area between the Strymonic Gulf and Kavala (Macedonia, Greece) with
the aid of remote sensing and geographical information systems. International
Journal of Remote Sensing 29, 6893–6915.
 Remote Sensing and GIScience in Geomorphological Mapping
Drinkwater, M.R., Francis, R., Ratier, G., Wingham, D.J., 2003. The European Space
Agency’s Earth Explorer Mission CryoSat: measuring variability in the
cryosphere. Annals of Glaciology 39, 313–320.
Drăgut- , L., Blaschke, T., 2006. Automated classification of landform elements using
object-based image analysis. Geomorphology 81, 330–344.
Drăgut- , L., Schauppenlehner, T., Muhar, A., Strobl, J., Blaschke, T., 2009.
Optimization of scale and parametrization for terrain segmentation: an
application to soil-landscape modeling. Computers & Geosciences 35,
1875–1883.
Duncan, C.C., Klein, A.J., Masek, J.G., Isacks, B.L., 1998. Comparison of late
Pleistocene and modern glacier extent in central Nepal based on digital
elevation data and satellite imagery. Quaternary Research 49, 241–254.
Dunlop, P., Clark, C.D., 2006. The morphological characteristics of ribbed
moraines. Quaternary Science Reviews 25, 1668–1691.
Dunning, S.A., Massey, C.I., Rosser, N.J., 2009. Structural and geomorphological
features of landslides in the Bhutan Himalaya derived from terrestrial laser
scanning. Geomorphology 103, 17–29.
Dunning, S.A., Rosser, N.J., Petley, D.N., Massey, C.R., 2006. Formation and failure
of the Tsatichhu landslide dam, Bhutan. Landslides 3, 107–113.
Dvigalo, V.N., Melekestsev, I.V., 2009. The geological and geomorphic impact of
catastrophic landslides in the geyser valley of Kamchatka: aerial
photogrammetry. Journal of Volcanology and Seismology 3, 314–325.
Dymond, J.R., Ausseil, A.-G., Shepherd, J.D., Buettner, J., 2006. Validation of a
region-wide model of landslide susceptibility in the Manawatu-Wanganui region
of New Zealand. Geomorphology 74, 70–79.
Eisank, C., Drăgut- , L., Götz, J., Blaschke, T., 2010. Developing a semantic model of
glacial landforms for object-based terrain classification – the example of glacial
cirques. In: Addink, E., Van Coillie, F.M.B. (Eds.), GEOBIA 2010- Geographic
Object-Based Image Analysis. Ghent University, Ghent, Belgium, 29 June to
June 2 July. ISPRS XXXVIII-4/C7, Archives ISSN 1682-1777.
Ercanoglu, M., 2005. Landslide susceptibility assessment of SE Bartin (west Black
Sea region, Turkey) by artificial neural networks. Natural Hazards and Earth
System Sciences 5, 979–992.
Ermini, L., Catani, F., Casagli, N., 2005. Artificial neural networks applied to
landslide susceptibility assessment. Geomorphology 66, 327–343.
Ernst, G.G.J., Kervyn, M., Teeuw, R.M., 2008. Advances in the remote sensing of
volcanic activity and hazards, with special consideration for applications in
developing countries. International Journal of Remote Sensing 29, 6687–6723.
Esaki, T., Xie, M.W., Zhou, G.Y., Mitani, Y., 2001. Monte Carlo method for locating
and evaluating 3D critical slope slip based on GIS Database. In: Sijing, W.,
Bingjunm, F., Zhongkui, L. (Eds.), Frontiers on Rock Mechanics and Sustainable
Development in 21st Century. Balkema, Rotterdam, pp. 17–21.
Espizúa, L.E., 1986. Fluctuations of the Rio del Plomo Glaciers. Geografiska
Annaler 68, 317–327.
Evans, A.N., 2000. Glacier surface motion computation from digital image
sequences. IEEE Transactions on Geoscience and Remote Sensing 38,
1064–1072.
Evans, D.J.A., Clark, C.D., Mitchell, W.A., 2005. The last British Ice Sheet: a review
of the evidence utilised in the compilation of the Glacial Map of Britain. Earth-
Science Reviews 70, 253–312.
Evans, S.G., Bent, A.L., 2004. The Las Colinas landslide, Santa Tecla: a highly
destructive flowslide triggered by the January 13, 2001, El Salvador earthquake.
Geological Society of America Special Paper 375, 25–37.
Eyton, J.R., 1989. Low relief topographic enhancement in a Landsat snow cover
scene. Remote Sensing Environment 27, 105–118.
Fabris, M., Pesci, A., 2005. Automated DEM extraction in digital aerial
photogrammetry: precisions and validation for mass movement monitoring.
Annals of Geophysics 48, 973–988.
Favalli, M., Karatson, D., Mazzarini, F., Pareschi, M.T., Boschi, E., 2009.
Morphometry of scoria cones located on a volcano flank: a case study from Mt.
Etna (Italy), based on high-resolution LiDAR data. Journal of Volcanology and
Geothermal Research 186, 320–330.
Federici, P.R., Spagnolo, M., 2004. Morphometric analysis on the size, shape and
areal distribution of glacial cirques in the maritime Alps (Western French-Italian
Alps). Geografiska Annaler 86, 235–248.
Fell, R., 1994. Landslide risk assessment and acceptable risk. Canadian
Geotechnical Journal 31, 261–272.
Felpeto, A., Arana, V., Ortiz, R., Astiz, M., Garcia, A., 2001. Assessment and
modelling of lava flow hazard on Lanzarote (Canary islands). Natural Hazards
23, 247–257.
Fernández, T., Irigaray, C., El Handouni, R., Chacón, J., 2003. Methodology for
landslide susceptibility mapping by means of a GIS. Application to the
Contraviesa area (Granada, Spain). Natural Hazards 58.2897-308.
221
Fernandez-Steeger, T.M., Rohn, J., Czurda, K., 2002. Identification of landslide areas
with neural nets for hazard analysis. In: Rybar, J., Stemberk, J., Wagner, P.
(Eds.), Proceedings of the 1st European Conference on Landslides, Prague.
Balkema, Rotterdam, pp. 163–168.
Field, W.O., 1966. Mapping glacier termini in southern Alaska, 1931–1964.
Canadian Journal of Earth Sciences 6, 819–825.
Finlayson, D.P., Montgomery, D.R., 2003. Modeling large-scale fluvial erosion in
geographic information systems. Geomorphology 53, 147–164.
Finnegan, N.J., Pritchard, M.E., Lohman, R.B., Lundgren, P.R., 2008. Constraints on
surface deformation in the Seattle, WA, urban corridor from satellite radar
interferometry time-series analysis. Geophysical Journal International 174,
29–41.
Fisher, P., Wood, J., Cheng, T., 2004. Where is Helvellyn? Fuzziness of multi-scale
landscape morphometry. Transactions of the Institute of British Geographers 29,
106–128.
Fonstad, M.A., Marcus, W.A., 2005. Remote sensing of stream depths with
hydraulically assisted bathymetry (HAB) models. Geomorphology 72, 320–339.
Fourniadis, I.G., Liu, J.G., Mason, P.J., 2007a. Regional assessment of landslide
impact in the Three Gorges area, China, using ASTER imagery: Wushan-Zigui.
Landslides 4, 267–278.
Fourniadis, I.G., Liu, J.G., Mason, P.J., 2007b. Landslide hazard assessment in the
Three Gorges area, China, using ASTER imagery: Wushan-Badong.
Geomorphology 84, 126–144.
Frappart, F., Papa, F., Famiglietti, J.S., Prigent, C., Rossow, W.B., Seyler, F., 2008.
Interannual variations of river water storage from a multiple satellite approach: a
case study for the Rio Negro River basin. Journal of Geophysical Research 113.
http://dx.doi.org/10.1029/2007JD009438.
Frattini, P., Crosta, G., Carrar, A., Agliardi, F., 2008. Assessment of rockfall
susceptibility by integrating statistical and physically-based approaches.
Geomorphology 94, 419–437.
French, J.R., 2003. Airborne LiDAR in support of geomorphological and hydraulic
modeling. Earth Surface Processes and Landforms 28(3), 321–335.
Fruneau, B., Achache, J., Delacourt, C., 1996. Observation and modeling of the
Saint-Etienne-de Trinée landslide using SAR interferometry. Tectonophysics 265,
181–190.
Galli, M., Ardizzone, F., Cardinali, M., Guzetti, F., Reichenbach, P., 2008. Comparing
inventory maps. Geomorphology 94, 268–289.
Gao, J., Liu, Y., 2001. Applications of remote sensing, GIS, and GPS in glaciology:
a review. Progress in Physical Geography 25, 520–540.
Geist, T., Støtter, J., 2010. Airborne laser scanning in glacier studies. In: Pellikka,
P., Rees, W.G. (Eds.), Techniques for Topographic, Spatial, and Thematic
Mapping of Glaciers. Taylor & Francis, New York, NY, pp. 179–194.
Ghosh, S., Stepniski, T., Vilalta, R., 2010. Automatic annotation of planetary
surfaces with geomorphic labels. IEEE Transactions on Geoscience and Remote
Sensing 48, 175–185.
Giardino, M., Alberto, W., Perotti, L., 2008. Geomatics and geomorphology
integration for the analysis of morphostructures and deepseated gravitational
slope deformations: application in the Susa Valley (western Alps). Memorie
Descrittive Della Carta Geologica d’Italia 78, 105–112.
Gilvear, D., Bryant, R., 2005. Analysis of aerial photography and other remotely
sensed data. In: Kondolf, G.M., Piégay, H. (Eds.), Tools in Fluvial
Geomorphology. Wiley, Chichester. http://dx.doi.org/10.1002/0470868333.
Gilvear, D.J., Waters, T., Milner, A., 1998. Image analysis of aerial photography to
quantify the effect of gold place mining on channel morphology, Interior Alaska.
In: Lane, S., Richards, K., Chandler, J. (Eds.), Landform Mapping, Modelling,
and Analysis. Wiley, Chichester, pp. 195–216.
Gilvear, D.J., Waters, T.M., Milner, A.M., 1995. Image analysis of aerial photography
to quantify changes in channel morphology and instream habitat following place
mining in interior Alaska. Freshwater Biology 34, 389–398.
Glenn, N.F., Streutker, D.R., Chadwick, D.J., Thackray, G.D., Dorsch, S.J., 2006.
Analysis of LiDAR-derived topographic information for characterizing and
differentiating landslide morphology and activity. Geomorphology 73, 131–148.
Goguel, B., 1989. A mass movement detected by photography: the Friolin landslide,
Savoie. Revue Francaise De Geotechnique 48, 55–64.
Goldstein, R.M., Engelhardt, H., Kamb, B., Frolich, R.M., 1993. Satellite radar
interferometry for monitoring ice sheet motion: application to an Antarctic Ice
Stream. Science 262, 1525–1530.
Goodchild, M.F., 1992a. Geographical data modeling. Computers & Geosciences
18.401-4-8.
Goodchild, M.F., 1992b. Geographical information science. International Journal of
Geographical Information Systems 6(1), 31–45.
Gordon, J.E., Brazier, V., Lees, G., 1994. Geomorphological systems: developing
fundamental principles for sustainable landscape management. In: O’Halloran, D.,
222 Remote Sensing and GIScience in Geomorphological Mapping
Green, J., Harley, M., Stanley, M., Knill, J. (Eds.), Geological landscape
conservation. Geological Society, London, pp. 185–189.
Granica, K., Hirschmugl, M., Proske, H., Schardt, M., Wallner, H., Wurm, M., 2007.
The derivation of indicators for landslide detection using very high resolution
(VHR) satellite imagery. Proceedings of the 9th International Symposium on
High Mountain Remote Sensing Cartography, Grazer Schriften der Geographie
und Raumforschung 43, 59–66.
Greenwood, S.L., Clark, C.D., 2008. Subglacial bedforms of the Irish Ice Sheet.
Journal of Maps 2008, 332–357.
Grosse, P., de Vries, B.V., Petrinovic, I.A., Euillades, P.A., Alvarado, G.E., 2009.
Morphometry and evolution of arc volcanoes. Geology 37, 651–654.
Gurnell, A.M., 1997. Channel change on the River Dee meanders, 1946–1992, from
the analysis of air photographs. Regulated Rivers: Research and Management
13, 13–26.
Gurnell, A.M., Downward, S.R., James, R., 1994. Channel planform change on the
River Dee meanders. Regulated Rivers: Research and Management 9, 187–204.
Guzzetti, F., Cardinali, M., Reichenbach, P., Carrarra, A., 2000. Comparing landslide
maps: a case study in the upper Tiber River basin. Environmental Management
25, 247–263.
Guzzetti, F., Crosta, G., Detti, R., Agliardi, F., 2002. STONE: a computer program for
the three-dimensional simulation of rock-falls. Computers & Geosciences 28,
1079–1093.
Guzzetti, F., Reichenbach, P., Wieczorek, G.F., 2003. Rockfall hazard and risk
assessment in the Yosemite Valley, California, USA. Natural Hazards in Earth
System Science 3, 491–503.
Günther, A., 2003. SLOPEMAP: programs for automated mapping of geometrical
and kinematical properties of hard rock hill slopes. Computers & Geosciences
29, 865–875.
Haberler-Weber, M., Eichhorn, A., Kahmen, H., Theilen-Willige, B., 2009. Concept
of a multi-scale monitoring and evaluation system for landslide disaster
prediction. Proceedings of the 2007 IAG General Assembly, International
Association of Geodesy Symposia 133, 669–677.
Haeberli, W., Huggel, C., Kaeaeb, A., et al., 2004. The Kolka-Karmadon rock/ice
slide of 20 September 2002: an extraordinary event of historical dimensions in
north Ossetia, Russian Caucasus. Journal of Glaciology 50, 533–546.
Haefeli, R., 1966. Some notes on glacier mapping and ice movement. Canadian
Journal of Earth Sciences 3, 863–876.
Hall, S.R., Farber, D.L., Ramage, J.M., et al., 2009. Geochronology of Quaternary
glaciations from the tropical Cordillera Huayhuash, Peru. Quaternary Science
Reviews 28(25–26), 2991–3009.
Hamilton, C.W., Fagents, S.A., Thordarson, T., 2010. Explosive lava-water
interactions II: self-organization processes among volcanic rootless eruption
sites in the 1783–1784 Laki lava flow, Iceland. Bulletin of Volcanology 72(4),
469–485.
Hamilton, S.K., Kellndorfer, J., Lehner, B., Tobler, M., 2006. Remote sensing of
floodplain geomorphology as a surrogate for biodiversity in a tropical river
system (Madre de Dios, Peru). Geomorphology 89, 23–38.
Han, L., Rundquist, D.C., 1994. The response of both surface reflectance and the
underwater light field to various levels of suspended sediments: preliminary
results. Photogrammetric Engineering & Remote Sensing 60, 1463–1471.
Harper, J.T., Humphrey, N.F., Pfeffer, W.T., 1998. Crevasse patterns and the strain-
rate tensor: a high resolution comparison. Journal of Glaciology 44, 68–76.
Hasegawa, S., Dahal, R.K., Nishimura, T., Nonomura, A., Yamanaka, M., 2009.
DEM-based analysis of earthquake-induced shallow landslide susceptibility.
Geotechnical and Geological Engineering 27, 419–430.
Hättestrand, C., Stroeven, A.P., 2002. A relict landscape in the centre of
Fennoscandian glaciation: geomorphological evidence of minimal Quaternary
glacial erosion. Geomorphology 44, 127–143.
Haugerud, R.A., Harding, D.J., Johnson, S.Y., Harless, J.L., Weaver, C.S., Sherrod,
B.L., 2003. High resolution Lidar topography of the Puget Lowland, Washington
– a bonanza for earth science. GSA Today 13, 9.
Havenith, H.B., Strom, A., Caceres, F., Pirard, E., 2006. Analysis of landslide
susceptibility in the Suusamyr region. Tien Shan: Statistical and Geotechnical
Approach. Landslides 3, 39–50.
Haynes, K., Barclay, J., Pidgeon, N., 2007. Volcanic hazard communication
using maps: an evaluation of their effectiveness. Bulletin of Volcanology 70,
123–138.
Heritage, G.L., Fuller, I.C., Charlton, M.E., Brewer, P.A., Passmore, D.P., 1998. CDW
photogrammetry of low relief fluvial features: accuracy and implications for
reach-scale sediment budgeting. Earth Surface Processes and Landforms 23,
1219–1233.
Heyman, J., Hättestrand, C., Stroeven, A.P., 2008. Glacial geomorphology of the
Bayan Har sector of the NE Tibetan Plateau. Journal of Maps 2008, 42–62.
Hicks, D.M., Duncan, M.J., Walsh, J.M., Westaway, R.M., Lane, S.N., 2002. New
views of the morphodynamics of large braided rivers from high-resolution
topographic surveys and time-lapse video. In: The Structure, Function and
Management Implications of Fluvial Sedimentary Systems, International
Association of Hydrological Sciences (IAHS) Publication 276, pp. 373–380.
Higgitt, D.L., Warburton, J., 1999. Applications of differential GPS in upland fluvial
geomorphology. Geomorphology 29, 121–134.
Hubbard, B., Glasser, N., 2005. Field Techniques in Glaciology and Glacial
Geomorphology. Wiley, Chichester, 400 pp.
Hudson, J.A., 1992. Rock Engineering Systems: Theory and Practice. High Plains
Press (JAH), Chichester, 185 pp.
Hughes, L.C., Clark, C.D., Jordan, C.J., 2010. Subglacial bedforms of the last
British Ice Sheet. Journal of Maps 2010, 543–563.
Hughes, M.L., McDowell, P.F., Marcus, W.A., 2006. Accuracy assessment of
georectified aerial photographs: implications for measuring lateral channel
movement in GIS. Geomorphology 74, 1–16.
Hunsaker, C.T., Levine, D.A., Timmins, S.P., Jackson, B.L., O’Neill, R.V., 1992.
Landscape characterization for assessing regional water quality. In: McKenzie,
D.H., Hyatt, D.E., McDonald, V.J. (Eds.), Ecological Indicators. Elsevier, New
York, pp. 997–1006.
Hynek, B.M., Phillips, R.J., 2003. New data reveal mature, integrated drainage
systems on Mars indicative of past precipitation. Geology 31, 757–760.
Høgda, K.A., Storvold, R., Lauknes, T.R., 2010. SAR imaging of glaciers.
In: Pellikka, P., Rees, W.G. (Eds.), Techniques for Topographic, Spatial, and
Thematic Mapping of Glaciers. Taylor & Francis, New York, NY, pp. 153–178.
Iovine, G., Parise, M., 2002. Graphic method to classify and survey landslide
damage in buildings of urban areas. Memorie Della Societa Geologica Italiana
57, 595–603. (In Italian).
Iwahashi, J., Kamiya, I., Yamagishi, H., 2009. Estimation of the most
suitable window size of the slope gradient and convexo-concave index for the
assessment of shallow landslides using high resolution Lidar DEM. Transactions
Japanese Geomorphological Union 30, 15–27.
Iwahashi, J., Watanabe, S., Furuya, T., 2001. Landform analysis of slope
movements using DEM in Higashikubiki area, Japan. Computers and
Geosciences 27, 851–865.
Jaboyedoff, M., Baillifard, F., Couture, R., Locat, J., Locat, P., 2004. New insight of
geomorphology and landslide prone area detection using DEM. In: Lacerda, W.,
Ehrlich, M., Fontoura, A., Sayaõ, A. (Eds.), Landslides: Evaluation and
Stabilization. Taylor & Francis, London, pp. 191–198.
Jaboyedoff, M., Demers, D., Locat, J., et al., 2009. Use of terrestrial laser scanning
for the characterization of retrogressive landslides in sensitive clay and rotational
landslides in river banks. Canadian Geotechnical Journal 46, 1379–1390.
Jaboyedoff, M., Labiouse, V., 2003. Preliminary assessment of rockfall hazards
based on GIS data. 10th International Congress on Rock Mechnics ISRM 2003-
Technology roadmap for rock mechanics, South African Institute of Mining and
Metallurgy, Johannesburgh, South Africa, pp. 575–578.
Jansson, K.N., Glasser, N.F., 2005. Using Landsat 7 ETM þ imagery and digital
terrain models for mapping glacial lineaments on former ice sheet beds.
International Journal of Remote Sensing 26, 3931–3941.
Jansson, K.N., Kleman, J., Marchant, D., 2002. The succession of ice-flow patterns
in north-central Québec-Labrador, Canada. Quaternary Science Reviews 21,
503–523.
Jibson, R.W., Harp, E.L., Michael, J.A., 1998. A method for producing digital
probabilistic seismic landslide hazard maps. USGS Open-file report 98–113,
17 pp.
Jones, A.F., Brewer, P.A., Johnstone, E., Macklin, M.G., 2007. High-resolution
interpretative geomorphological mapping of river valley environments using
airborne LiDAR data. Earth Surface Processes and Landforms 32(10),
1574–1592.
Jordan, D.C., Fonstad, M.A., 2005. Two dimensional mapping of river bathymetry
and power using aerial photography and GIS on the Brazos River, Texas.
Geocarto International 20(3), 1–8.
Joughin, I., Kwok, R., Fahnestock, M., 1996. Estimation of ice-sheet motion using
satellite radar interferometry: method and error analysis with application to
Humboldt Glacier, Greenland. Journal of Glaciology 42, 564–575.
Joughin, I.R., Kwok, R., Fahnestock, M.A., 1998. Interferometric estimation of three-
dimensional ice-slow using ascending and descending passes. IEEE
Transactions on Geoscience and Remote Sensing GE 36, 25–37.
Joughin, I.R., Winebrenner, D.P., Fahnestock, M.A., 1995. Observations of ice-sheet
motion in Ggreenland using satellite radar interferometry. Geophysical Research
Letters 22, 571–574.
Jung, H.C., Hamski, J., Durand, M., et al., 2010. Characterization of complex fluvial
systems using remote sensing of spatial and temporal water level variations in
 Remote Sensing and GIScience in Geomorphological Mapping
the Amazon, Congo, and Brahmaputra Rivers. Earth Surface Processes and
Landforms 35, 294–304.
Kääb, A., 2002. Monitoring high-mountain terrain deformation from repeated
air – and spaceborne optical data: examples using digital aerial imagery and
ASTER data. ISPRS Journal of Photogrammetry and Remote Sensing 57,
39–52.
Kahle, A.B., Gillespie, A.R., Abbott, E.A., Abrams, M.J., Walker, R.E., Hoover, G.,
Lockwood, J.P., 1988. Relative dating of Hawaiian lava flows using multispectral
thermal infrared images : a new tool for geologic mapping of young volcanic
terranes. Journal of Geophysical Research 93(B12), 15239–15251.
Kajiyama, A., Mitamura, M., Luan, T.X., Okudaira, T., Ha, N.V., 2005. Landslide
monitoring by photogrammetry in Mongset area, northwestern Vietnam.
International Journal of Geoinformatics 1, 125–132.
Kalaugher, P.G., Grainger, P., 1997. A photographic method for rapid assessment of
landslide displacement. The Quarterly Journal of Engineering Geology 30,
193–196.
Kamp, U., Growley, B.J., Khattak, G.A., Owen, L.A., 2008. GIS-based landslide
susceptibility mapping for the 2005 Kashmir earthquake region. Geomorphology
101, 631–642.
Kamp, U., Owen, L.A., Growley, B.J., Khattak, G.A., 2010. Back analysis of landslide
susceptibility zonation mapping for the 2005 Kashmir earthquake: an assessment
of the reliability of susceptibility zoning maps. Natural Hazards 54, 1–25.
Karatson, D., Favalli, M., Tarquini, S., Fornaciai, A., Worner, G., 2010a. The regular
shape of stratovolcanoes: a DEM-based morphometrical approach. Journal of
Volcanology and Geothermal Research 193, 171–181.
Karatson, D., Telbisz, T., Singer, B., 2010b. Late-stage volcano geomorphic
evolution of the Pleistocene San Francisco Mountain, Arizona (USA), based on
high-resolution DEM analysis and Ar-40/Ar-39 chronology. Bulletin of
Volcanology 72, 833–846.
Kargel, J.S., Abrams, M.J., Bishop, M.P., et al., 2005. Multispectral imaging
contributions to global land ice measurements from space. Remote Sensing of
Environment 99(1–2), 187–219.
Kervyn, M., Ernst, G.G.J., Goossens, R., Jacobs, P., 2008. Mapping volcano
topography with remote sensing: ASTER vs SRTM. International Journal of
Remote Sensing 29, 6515–6538.
Khattak, G.A., Owen, L.A., Kamp, U., Harp, E.L., 2010. Evolution of earthquake-
triggered landslides in the Kashmir Himalaya, northern Pakistan. Geomorphology
115, 102–108.
Kim, J.W., Park, S.S., Kim, C.S., Lee, Y., 2004. The efficient web-based mobile GIS
service system through reduction of digital map. In: Laganà, A., Gavrilova, M.L.,
Kumar, V., Mun, Y., Tan, C.J.K., Gervasi, O. (Eds.), ICCSA 2004, LNCS 3043.
Springer, Berlin, pp. 410–417.
Kleman, J., Borgström, I., 1996. Reconstruction of palaeo-ice sheets: the use of
geomorphological data. Earth Surface Processes and Landforms 21, 893–909.
Kleman, J., Fastook, J., Stroeven, A.P., 2002. Geologically and geomorphologically
constrained numerical model of Laurentide Ice Sheet inception and build-up.
Quaternary International 95–96, 87–98.
Kleman, J., Hättestran, C., Borgström, I., Stroeven, A., 1997. Fennoscandian
palaeoglaciology reconstructed using a glacial geological inversion model.
Journal of Glaciology 43, 283–299.
Klingseisen, B., Metternicht, G., Paulus, G., 2008. Geomorphometric landscape
analysis using a smie-automated GIS-approach. Environmental Modelling &
Software 23, 109–121.
Koblinsky, C.J., Clarke, R.T., Brenner, A.C., Frey, H., 1993. Measurement of river
level variations with satellite altimetry. Water Resources Research 29(6),
1839–1848.
Koehl, M., Mombrun, L., Gay, M., et al., 2009. SAR interferometry and GPS
measurements on Argentiere’s glacier: comparisons and validation. Traitement
Du Signal 26, 109–126.
Kouraev, A.V., Zakharova, E.A., Samain, O., Mognard, N.M., Cazenave, C., 2004. Ob
river discharge from TOPEX/Poseidon satellite altimetry (1992-2002). Remote
Sensing of the Environment 93, 238–245.
Lagmay, A.M.F., Valdivia, W., 2006. Regional stress influence on the opening
direction of crater amphitheaters in Southeast Asian volcanoes. Journal of
Volcanology and Geothermal Research 158(1-2), 139–150.
Lane, S.N., 2000. The measurement of river channel morphology using digital
photogrammetry. Photogrammetric Record 16(96), 937–957.
Lane, S.N., Chandler, J.H., Richards, K.S., 1994. Developments in monitoring and
modeling small-scale river bed topography. Earth Surface Processes and
Landforms 19, 349–368.
Lane, S.N., Widdison, P.E., Thomas, R.E., et al., 2010. Quantification of braided
river channel change using archival digital image analysis. Earth Surface
Processes and Landforms 35, 971–985.
223
Lantuit, H., Pollard, W.H., 2008. Fifty years of coastal erosion and retrogressive
thaw slump activity on Herschel Island, southern Beaufort Sea, Yukon Territory,
Canada. Geomorphology 95, 84–102.
Lee, S., Dan, N.T., 2005. Probabilistic landslide susceptibility mapping in the Lai
Chau province of Vietnam: focus on the relationship between tectonic fractures
and landslides. Environmental Geology 48, 778–787.
Lee, S., Pradhan, B., 2006. Probabilistic landslide hazards and risk mapping of
Penang Island, Malaysia. Journal of Earth System Science 115, 661–672.
Legleiter, C.J., Goodchild, M.F., 2005. Alternative representations of in-stream
habitat: classification using remote sensing, hydraulic modeling, and fuzzy logic.
International Journal of Geographic Information Science 19(1), 29–50.
Legleiter, C.J., Marcus, W.A., Lawrence, R., 2002. Effects of sensor resolution on
mapping in-stream habitats. Photogrammetric Engineering and Remote Sensing
68(8), 801–807.
Legleiter, C.J., Roberts, D.A., 2005. Effects of channel morphology and sensor
spatial representation on image-derived depth estimates. Remote Sensing of
Environment 95(2), 231–247.
Legleiter, C.J., Roberts, D.A., Marcus, W.A., Fonstad, M.A., 2004. Passive optical
remote sensing of river channel morphology and in-stream habitat: Physical
basis and feasibility. Remote Sensing of Environment 93, 493–510.
Lejot, J., Delacourt, C., Piegay, H., Fournier, T., Termelo, M.-L., Allemand, P., 2007.
Very high spatial resolution imagery for channel bathymetry and topography
from an unmanned mapping controlled platform. Earth Surface Processes and
Landforms 33, 4–24.
Leon, J.G., Calmant, S., Seyler, F., et al., 2006. Rating curves and estimation of
average water depth at the upper Negro River based on satellite altimeter data
and modeled discharges. Journal of Hydrology 328, 481–496.
Leroi, E., Rouzeau, O., Scanvic, J.Y., Weber, C.C., Vargas Cuervo, G., 1992. Remote
sensing and GIS technology in landslide hazard mapping in the Colombian
Andes. Episodes 15, 32–35.
Leroy, S., Markus, A.G., Schwab, J., Costa, P.J.M., 2010. Seismic influence on the
last 1500-year infill history of Lake Sapanca (north Anatolian Fault, NW Turkey).
Tectonophysics 486, 15–27.
Leverington, D.W., Teller, J.T., Mann, J.D., 2002. A GIS method for reconstruction
of late Quaternary landscapes from isobase data and modern topography.
Computers & Geosciences 28, 631–639.
Leys, K.F., Werrity, A., 1999. River channel planform change: software for historical
analysis. Geomorphology 29, 107–120.
Li, S.S., Benson, C., Gens, R., Lingle, C., 2008a. Motion patterns of Nabesna
Glacier (Alaska) revealed by interferometric SAR techniques. Remote Sensing of
the Environment 112, 3628–3638.
Li, X., 1993. A united model for quantitative remote sensing of suspended sediment
concentration. International Journal of Remote sensing 14(14), 2665–2676.
Li, Y., Napieralski, J., Harbor, J., 2008b. A revised automated proximity and
conformity analysis method to compare predicted and observed spatial
boundaries of geologic phenomena. Computers & Geosciences 34, 1806–1814.
Li, Y., Napieralski, J., Harbor, J., Hubbard, A., 2007. Identifying patterns of
correspondence between modeled flow directions and field evidence: an
automated flow direction analysis. Computers & Geosciences 33, 141–150.
Lindner, L., Marks, L., Ostaficzuk, S., Pekala, K., Szczesny, R., 1985. Application of
photogeological mapping to studies of glacial history of south Spitsbergen.
Earth Surface Process and Landforms 10, 387–399.
Liu, G.-R., Shi, Y.-C., Feng, W.-K., 2007. Simulation of the formation of Jinjiang
landslide by DEM. Hydrogeology and Engineering Geology 34, 74–7680.
Lohani, B., Mason, D.C., 2001. Application of airborne scanning laser altimetry to
the study of tidal channel geomorphology. ISPRS Journal of Photogrammetry
and Remote Sensing 56(2), 100–120.
López-Blanco, J., Villers-Ruiz, L., 1995. Delineating boundaries of environmental
units for land management using a geomorphological approach and GIS – a
study in Baja California, Mexico. Remote Sensing of the Environment 53,
109–117.
Lu, P., Rosenbaum, M.S., 2003. Artificial neural networks and grey systems for the
prediction of slope stability. Natural Hazards 30, 383–398.
Lu, Z., Fielding, E., Patrick, M., Trautwein, C., 2003. Estimating lava volume by
precision combination of multiple baseline spaceborne and airborne
interferometric synthetic aperture radar: the 1997 eruption of Okmok volcano.
IEEE Transactions on Geoscience & Remote Sensing 41, 1428–1436.
Lucieer, A., Stein, A., 2005. Texture-based landform segmentation of LiDAR imagery.
International Journal of Applied Earth Observation and Geoinformation 6,
261–270.
Lui, J.G., Mason, P.J., Clerici, N., et al., 2004. Landslide hazard assessment in the
Three Gorges area of the Yangtze river using ASTER imagery. Geomorphology
61, 171–187.
224 Remote Sensing and GIScience in Geomorphological Mapping
Lyon, J.G., Lunetta, R.S., Williams, D.C., 1992. Airborne multispectral scanner data
for evaluating bottom sediment types and water depths of the St. Marys River,
Michigan. Photogrammetric Engineering and Remote Sensing 61(2), 71–82.
MacDonald, H.C., Grubbs, R.S., 1975. Landsat imagery analysis; an aid for
predicting landslide-prone areas for highway construction. NASA Technical
Memorandum, X-58168 Vol.I-B, 769–778.
MacMillan, R.A., Pettapiece, W.W., Nolan, S.C., Goddard, T.W., 2000. A generic
procedure for automatically segmenting landforms in elements using DEMs,
heuristic rules and fuzzy logic. Fuzzy Sets and Systems 113, 81–109.
Madin, I.P., Burns, W.J., 2006. Map of the Landslide Geomorphology of Oregon
City, Oregon and Vicinity Interpreted from Lidar Imagery and Aerial
Photographs. Oregon Department of Geology and Mineral Industries, Portland,
USA.
Malamud, B.D., Turcotte, D.L., Guzzetti, F., Reichenbach, P., 2004. Landslide
inventories and their statistical properties. Earth Surface Processes and
Landforms 29, 687–711.
Marcus, W.A., Fonstad, M.A., 2008. Optical remote mapping of rivers at sub-meter
resolutions and watershed extents. Earth Surface Processes and Landforms 33,
3–24.
Marcus, W.A., Legleiter, C.J., Aspinall, R.J., Boardman, J.W., Crabtree, R.L., 2003.
High spatial resolution hyperspectral mapping of in-stream habitats, depths, and
woody debris in mountain streams. Geomorphology 55, 363–380.
Mark, D., Smith, B., 2004. A science of topography: From qualitative ontology to
digital representations. In: Bishop, M.P., Shroder, J.F. (Eds.), Geographic
Information Science and Mountain Geomorphology. Springer, Berlin Heidelberg,
pp. 75–100.
Marquı́nez, J., Duarte, R.M., Farias, P., Sánchez, M.J., 2003. Predictive GIS-based
model of rockfall activity in mountain cliffs. Natural Hazards 58, 341–360.
Marston, R.A., Girel, J., Pautou, G., Piegay, H., Bravard, J.P., Arneson, C., 1995.
Channel metamorphosis, floodplain disturbance, and vegetation development:
Ain River, France. Geomorphology 13, 121–131.
Marti, J., Felpeto, A., 2010. Methodology for the computation of volcanic
susceptibilityan example for mafic and felsic eruptions on Tenerife (Canary
Islands). Journal of Volcanology and Geothermal Research 195(1), 69–77.
Mazzarini, F., Pareschi, M.T., Sbrana, A., Favalli, M., Fulignati, P., 2001. Surface
hydrothermal alteration mapping at Vulcano island using MIVIS data.
International Journal of Remote Sensing 22(11), 2045–2070.
McGovern, P.J., Morgan, J.K., 2009. Volcanic spreading and lateral variations in the
structure of Olympus Mons, Mars. Geology 37(2), 139–142.
McKean, J., Roering, J., 2004. Objective landslide detection and surface
morphology mapping using high-resolution airborne laser altimetry.
Geomorphology 57, 331–351.
Meier, M., 1966. Some glaciological interpretations of remapping programs of
South Cascade, Nisqually, and Klawatti Glaciers, Washington. Canadian Journal
of Earth Sciences 6, 811–818.
Mertes, L.A.K., 2002. Remote sensing of riverine landscapes. Freshwater Biology
47, 799–816.
Mertes, L.A.K., Smith, M.O., Adams, J.B., 1993. Estimating suspsended sediment
concentrations in surface waters of the Amazon River wetlands from Landsat
images. Remote Sensing of the Environment 43, 281–301.
Micheli, E.R., Kirchner, J.W., 2002. Effects of wet meadow riparian vegetation on
streambank erosion. 1. Remote Sensing Measurements of Streambank Migration
and Erodibility. Earth Surface Processes and Landforms 27, 627–639.
Mikos, M., Vidmar, A., Brilly, M., 2005. Using a laser measurement system for
monitoring morphological changes on the Strug rock fall, Slovenia. Natural
Hazards and Earth System Sciences 5, 143–153.
Millington, A.C., Townshend, J.R.G., 1987. The potential of satellite remote sensing
for geomorphological investigations: an overview. In: Gardiner, V. (Ed.),
International Geomorphology. Wiley, Chichester, pp. 331–342.
Miska, L., Hjort, J., 2005. Evaluation of current statistical approaches for predictive
geomorphological mapping. Geomorphology 67, 299–315.
Mohr, J.J., Reeh, N., Madsen, S.N., 1998. Three-dimensional glacial flow and
surface elevation measured with radar interferometry. Nature 391, 273–276.
Moody, A., Katz, D.B., 2004. Artificial intelligence in the study of mountain
landscapes. In: Bishop, M.P., Shroder, J.F. (Eds.), Geographic Information Science
and Mountain Geomorphology. Springer, Berlin Heidelberg, pp. 219–247.
Morelli, M., Piana, F., Polino, R., 2001. Comparison between remotely sensed
lineament intersection density and landslide distribution in the Langhe area
(southern Piemonte). Geoingegneria Ambientale e Mineraria 38, 251–256. (In
Italian).
Morén, B., Heyman, J., Stroeven, A.P., 2011. Glacial geomorphology of the central
Tibetan Plateau. Journal of Maps v2011, 115–125. http://dx.doi.org/10.4113/
jom.2011.1161.
Mount, N.J., Louis, J., 2005. Estimation and propogation of error in measurements
of river channel movement from aerial imagery. Earth Surface Processes and
Landforms 30, 635–643.
Mount, N.J., Louis, J., Teeuw, R.M., Zukiski, P.M., Stott, T., 2003. Estimation of
error in bankfull width comparisons from temporally sequences raw and
corrected aerial photographs. Geomorphology 56, 65–77.
Mullen, I., Kellet, J., 2007. Groundwater salinity mapping using airborne
electromagnetics and borehole data within the lower Balonne catchment,
Queensland, Australia. International Journal of Applied Earth Observation and
Geoinformation 9, 116–123.
Murray, J.B., van Wyk de Vries, B., Marquez, A., Williams, D.A., Byrne, P., Muller, J.-P.,
Kim, J.-R., 2010. Late-stage water eruptions from Ascraeus Mons volcano, Mars:
implications for its structure and history. Earth and Planetary Science Letters
294(3–4), 479–491.
Napieralski, J., Harbor, J., Li, Y., 2007b. Glacial geomorphology and geographic
information systems. Earth Science Reviews 85, 1–22.
Napieralski, J., Hubbard, A., Li, Y., et al., 2007a. Toward a GIS assessment of
numerical ice sheet model performance using geomorphological data. Journal of
Glaciology 53, 71–83.
Napieralski, J.A., Nalepa, N., 2010. The role of resolution and contour interval in
drumlin delineation and measurement. Computers & Geosciences 36, 222–230.
Nelson, P.A., Smith, J.A., Miller, A.J., 2006. Evolution of channel morphology and
hydrologic response in an urbanizing drainage basin. Earth Surface Processes
and Landforms 31(9), 1063–1079.
Ng, F., Barr, I.D., Clark, C.D., 2010. Using the surface profiles of modern ice
masses to inform palaeo-glacier reconstructions. Quaternary Science Reviews
29(23–24), 3240–3255.
Nichol, J.E., Shaker, A., Wong, M.-S., 2006. Application of high-resolution stereo
satellite images to detailed landslide hazard assessment. Geomorphology 76,
68–75.
Noferini, L., Pieraccini, M., Mecatti, D., Macaluso, G., Luzi, G., Atzeni, C., 2006.
Long term landslide monitoring by ground-based synthetic aperture radar
interferometer. International Journal of Remote Sensing 27, 1893–1905.
Notebaert, B., Verstraeten, G., Govers, G., Poesen, J., 2009. Qualitative and
quantitative applications of LiDAR imagery in fluvial geomorphology. Earth
Surface Processes and Landforms 34, 217–231.
Ochoa-Tejeda, V., Parrot, J.-F., 2007. Automatic extraction of landslide traces
utilizing digital terrain models and high-resolution IKONOS satellite imagery:
example of the Sierra Norte de Puebla, Mexico. Revista Mexicana De Ciencias
Geologicas 24, 354–367.
Ostir, K., Veljanovski, T., Podobnikar, T., Stancic, Z., 2003. Application of remote
sensing and GIS in Mt. Mangart landslide observation (Slovenia). Geologija 46,
385–390.
Owen, L.A., Kamp, U., Khattak, G., Harp, E., Keefer, D.K., Bauer, M., 2008.
Landslides triggered by the October 8, 2005, Kashmir earthquake.
Geomorphology 94, 1–9.
O’Connor, J.E., Jones, M.A., Haluska, T.L., 2003. Flood plain and channel dynamics
of the Quinault and Queets Rivers, Washington, USA. Geomorphology 51,
31–59.
O’Neill, R.V., Hunsaker, C., Jones, K.B., et al., 1997. Monitoring environmental
quality at the landscape scale. Bioscience 47, 513–519.
Perron, J.T., Kirchner, J.W., Dietrich, W.E., 2008. Spectral signatures of
characteristic spatial scales and nonfractal structure in landscapes. Journal of
Geophysical Research 113, F04003. http://dx.doi.org/10.1029/2007JF000866.
Petrie, G., Price, R.J., 1966. Photogrammetric measurements of the ice wastage and
morphological changes near the Casement Glacier, Alaska. Canadian Journal of
Earth Sciences 6, 827–840.
Pieraccini, M., Noferini, L., Mecatti, D., Atzeni, C., Teza, G., Galgaro, A., Zaltron, N.,
2006. Integration of radar interferometry and laser scanning for remote
monitoring of an urban site built on a sliding slope. IEEE Transactions on
Geoscience & Remote Sensing 44, 2335–2342.
Pike, R., 1992. Machine visualization of synoptic topography by digital image
processing. In: Wiltshire, D.A. (Ed.), Selected Papers in the Applied Computer
Sciences. US Geological Survey Bulletin, vol. 2016, pp. 1–12.
Pike, R.J., Clow, G.D., 1981. Revised classification of terrestrial volcanoes and a
catalog of topographic dimensions with new results on edifice volume. U.S.
Geological Survey Open-File Report, OF 81-1038.
Pike, R.J., Evans, I.S., Hengl, T., 2009. Geomorphometry: a brief guide. In: Hengl,
T., Reuter, H.I. (Eds.), Geomorphometry–Concepts, Software, Applications.
Developments in Soil Science. Elsevier, Amsterdam, vol. 33, pp. 3–30.
Plewe, B., 1997. A representation-oreinted taxonomy of gradation. In: Hirtle, S.C.,
Frank, A.U. (Eds.), Spatial information theory: a theoretical basis for GIS,
Lecture Notes in Computer Sciences 1329. Spring-Verlag, Berlin, pp. 121–136.
 Remote Sensing and GIScience in Geomorphological Mapping
Pradhan, B., 2010. Remote sensing and GIS-based landslide hazard analysis and
cross-validation using multivariate logistic regression model on three test areas
in Malaysia. Advances in Space Research 45, 1244–1256.
Prest, V.K., Grant, D.R., Rampton, V.N., 1968. The Glacial Map of Canada.
Geological Survey of Canada, Ottawa.
Punkari, M., 1982. Glacial geomorphology and dynamics in the eastern parts of the
Baltic Shield interpreted using landsat imagery. Photogrammetric Journal
Finland 9, 77–93.
Punkari, M., 1993. Modelling of the dynamics of the Scandinavian ice sheet using
remote sensing and GIS methods. In: Aber, J. (Ed.), Glaciotectonics and
Mapping Glacial Deposits. Proceedings of the INQUA commission of formation
and properties of glacial deposits. Canadian Plains Research Center, University
of Regina, Canada, pp. 232–250.
Raseman, S., Schimdt, J., Schrott, L., Dikau, R., 2004. Geomorphometry in
mountain terrain. In: Bishop, M.P., Shroder, J.F. (Eds.), Geographic
Information Science and Mountain Geomorphology. Springer-Praxis, Berlin,
pp. 101–137.
Rathje, E.M., Kayen, R., Woo, K.-S., Towhata, I., 2006. Remote sensing
observations of landslides and ground deformation from the 2004 Niigata Ken
Chuetsu earthquake. Soils and Foundations 46, 831–842.
Refice, A., Capolongo, D., 2002. Probabilistic modelling of uncertainties in
earthquake-induced landslide hazard assessment. Computers and Geosciences
28, 735–749.
Remondo, J., González-Dı́ez, Dı́az de Terán, J.R., Cendrero, A., 2003. Landslide
susceptibility models utilizing spatial data analysis techniques. A case study
from the Lower Deba Valley, Guipuzcoa (Spain). Natural Hazards 58, 267–279.
Rodriguez-Gonzalez, A., Fernandez-Turiel, J.L., Perez-Torrado, F.J., Gimeno, D.,
Aulinas, M., 2010. Geomorphological reconstruction and morphometric
modelling applied to past volcanism. International Journal of Earth Sciences 99,
645–660.
Roessner, S., Wetzel, H.U., Kaufmann, H., Sarnagoev, A., 2006. Satellite remote
sensing and GIS for analysis of mass movements with potential for dam
formation. Italian Journal of Engineering Geology and Environment, Special
Issue 2006, 103–114.
Rott, H., Nagler, T., 2006. The contribution of radar interferometry to the assessment
of landslide hazards. Advances in Space Research 37, 710–719.
Rott, H., Scheuchl, B., Siegel, A., Grasemann, B., 1999. Monitoring very slow slope
movements by means of SAR interferometry: a case study from a mass waste
above a reservoir in the Ötztal Alps, Austria. Geophysical Research Letters
26(11), 1629–1632.
Rowland, S.K., 1996. Slope, lava flow volumes, and vent distribution on Volcano
Fernandina, Galapagos Islands. Journal of Geophysical Research 101,
27657–27672.
Rumelhart, H., Hinton, G.E., Williams, R.J., 1986. Learning internal representation
by error propagation. In: Rumelhart, H., McClelland, J.L. (Eds.), Parallel
Distributed Processing. MIT Press, Cambridge, pp. 318–362.
Sahlin, E.A.U., Glasser, N.F., 2008. A geomorphological map of Cadair Idris, Wales.
Journal of Maps v2008, 299–314. http://dx.doi.org/10.4113/jom.2008.1011.
Salcher, B.C., Hinsch, R., Wagreich, M., 2010. High-resolution mapping of glacial
landforms in the North Alpine Foreland, Austraia. Geomorphology 122,
283–293.
Samarakoon, L., Ogawa, S., Ebisu, N., Lapitan, R., Kohki, Z., 1993. Inference of
landslide susceptible areas by landsat thematic mapper data. IAHS-AISH
Publication 217, 83–90.
Santacana, N., Baeza, B., Corominas, J., de Paz, A., Marturiá, J., 2003. A GIS-
based multivariate statistical analysis for shallow landslide susceptibility
mapping in La Pobla de Lillet area (Eastern Pyrenees), Spain. Natural Hazards
58, 281–295.
Santini, M., Grimaldi, S., Nardi, F., Petroselli, A., Rulli, M.C., 2009. Pre-processing
algorithms and landslide modeling on remotely sensed DEMs. Geomorphology
113, 110–125.
Sauber, J., Molnia, B., Carabajal, C., Luthcke, S., Muskett, R., 2005. Ice elevations
and surface change on the Malaspina Glacier, Alaska. Geohysical Research
Letters 32, L23S01. http://dx.doi.org/10.1029/2005GL023943.
Scambos, T.A., Bindschadler, R.A., 1993. Complex ice stream flow revealed by
sequential satellite imagery. Annals of Glaciology 17, 177–182.
Scambos, T.A., Dutkiewicz, M.J., Wilson, J.C., Bindschadler, R.A., 1992. Application
of image cross-correlation to the measurement of glacier velocity using satellite
image data. Remote Sensing of the Environment 42, 177–186.
Schneider, J.F., 2009. Seismically reactivated Hattian slide in Kashmir, Northern
Pakistan. Journal of Seismology 13, 387–398.
Schulz, W.H., 2007. Landslide susceptibility revealed by LIDAR imagery and
historical records, Seattle, Washington. Engineering Geology 89, 67–87.
225
Shroder, Jr. J.F., Bishop, M.P., 2004. Mountain geomorphic systems. In: Bishop,
M.P., Shroder, J.F. (Eds.), Geographic Information Science and Mountain
Geomorphology. Springer-Praxis, Berlin Heidelberg, pp. 33–66.
Sidjak, R.W., Wheate, R.D., 1999. Glacier mapping of the Illecillewaet icefield,
British Columbia, Canada, using Landsat TM and digital elevation data.
International Journal of Remote Sensing 20, 273–284.
Singhroy, V., Mattar, K.E., Gray, A.L., 1998. Landslide characterization in Canada
using interferometric SAR and combined SAR and TM images. Advances in
Space Research 21, 465–476.
Skoye, K.R., Eyton, J.R., 1992. Digital analysis of low relief topography in a
Landsat snow-cover scene in south central Alberta. Canadian Journal of Remote
Sensing 18, 143–150.
Smets, B., Wauthier, C., d’Oreye, N., 2010. A new map of the lava flow field of
Nyamulagira (D.R. Congo) from satellite imagery. Journal of African Earth
Sciences 58, 778–786.
Smith, G.R., Woodward, J.C., Heywood, D.I., Gibbard, P.L., 2000. Interpreting
Pleistocene glacial features from SPOT HRV data using fuzzy techniques.
Computers and Geosciences 26, 479–490.
Smith, M.J., Clark, C.D., 2005. Methods for the visualisation of digital elevation
models for landform mapping. Earth Surface Processes and Landforms 30,
885–900.
Smith, M.J., Clark, C.D., Wise, S.M., 2001. Mapping glacial lineaments from
satellite imagery: an assessment of the problems and development of best
procedure. Slovak Geological Magazine 7, 263–274.
Smith, M.J., Pain, C.F., 2009. Applications of remote sensing in geomorphology.
Progress in Physical Geography 33(4), 568–582.
Smith, M.J., Rose, J., Booth, S., 2006. Geomorphological mapping of glacial
landforms from remotely sensed data: an evaluation of the principal
data sources and an assessment of their quality. Geomorphology 76,
148–165.
Spicer, K.R., Costa, J.E., Placzek, G., 1997. Mesuring flood discharge in unstable
stream channels using ground penetrating radar. Geology 25(5), 423–426.
Spiker, E.C., Gori, P.L., 2000. National landslide hazards mitigation strategy: a
framework for loss reduction. Department of Interior, USGS, Open-file Report
00-450, 49 pp.
Stevens, N.F., Manville, V., Heron, D.W., 2002. The sensitivity of a volcanic flow
model to digital elevation model accuracy: experiments with digitised map
contours and interferometric SAR at Ruapehu and Taranaki volcanoes, New
Zealand. Journal of Volcanology and Geothermal Research 119, 89–105.
Stokes, C.R., Clark, C.D., 2003. Giant glacial grooves detected on Landsat ETM þ
satellite imagery. International Journal of Remote Sensing 24, 905–910.
Strom, A.L., Korup, O., 2006. Extremely large rockslides and rock avalanches in the
Tien Shan mountains, Kyrgyzstan. Landslides 3, 125–136.
Süzen, M.L., Doyuran, V., 2004. Data driven bivariate landslide susceptibility
assessment using geographical information system: a method and application to
Asarsuyu catchment, Turkey. Engineering Geology 71, 303–321.
Tangestani, M.H., 2004. Landslide susceptibility mapping using the fuzzy gamma
approach in a GIS, Kahan catchment area, southwest Iran. Australian Journal of
Earth Sciences 51, 439–450.
Tarchi, D., Casagli, N., Fanti, R., et al., 2003. Landslide monitoring by using
ground-based SAR interferometry: an example of application to the Tessina
landslide in Italy. Engineering Geology 68, 15–30.
Terlien, M.T.J., Van Westen, C.J., Van Asch, T.W.J., 1995. Deterministic modelling
in GIS-based landslide hazards assessment. In: Carrara, A., Guzzetti, F. (Eds.),
Geographical Information Systems in Assessing Natural Hazards. Kluwer,
Dordrecht, pp. 57–77.
Thoma, D.P., Satish, T., Gupta, C., Bauer, M.E., Kirchoff, C.E., 2005. Airborne laser
scanning for riverbank erosion assessment. Remote Sensing of Environment 95,
493–501.
Torres, R., et al., 2004. Monitoring the evolution of the Pasig-Potrero alluvial fan,
Pinatubo Volcano, using a decade of remote sensing data. Journal of
Volcanology and Geothermal Research 138, 371–392.
Van den Eeckhaut, M., Poesen, J., Verstraeten, G., Vanacker, V., Moeyersons, J.,
Nyssen, J., van Beek, L.P.H., 2005. The effectiveness of hillshade maps and
expert knowledge in mapping old deep-seated landslides. Geomorphology 67,
351–363.
Van den Eeckhaut, M., Poesen, J., Verstraeten, G., et al., 2007. Use of LIDAR-
derived images for mapping old landslides under forest. Earth Surface Processes
and Landforms 32, 754–769.
Varnes, D.J. 1984. Landslide Hazard Zonation: a Review of Principles and Practice.
UNESCO Natural Hazards Series 3, Paris, 63 pp.
Volker, H.X., Wasklewicz, T.A., Ellis, M.A., 2007. A topographic fingerprint to
distinguish alluvial fan formative processes. Geomorphology 88, 34–45.
226 Remote Sensing and GIScience in Geomorphological Mapping

Wadge, G., Francis, P.W., Ramirez, C.F., 1995. The Socompa collapse and
avalanche event. Journal of Volcanology and Geothermal Research 66, 309–336.
Walcott, R.C., Summerfield, M.A., 2008. Scale dependence of hypsometric integrals:
an analysis of southeast African basins. Geomorphology 96, 174–186.
Walsh, S.J., Bian, L., McKnight, S., Brown, D.G., Hammer, E.S., 2003. Solifluction
Stepps and risers, Lee Ridge, Glacier National Park, Montana, USA: a scale and
pattern analysis. Geomorphology 55, 381–398.
Walsh, S.J., Butler, D.R., Malanson, G.P., 1998. An overview of scale, pattern,
process relationships in geomorphology: a remote sensing and GIS persepctive.
Geomorphology 21, 183–205.
Walstra, J., Chandler, J.H., Dixon, N., Dijkstra, T.A., 2007. Aerial Photography and
Digital Photogrammetry for Landslide Monitoring; Mapping Hazardous Terrain
Using Remote Sensing. Geological Society, London, Special Publication 283,
53-63.
Wang, W., Wu, H., Nakamura, H., Wu, S., Ouyang, S., Yu, M., 2003. Mass
movements caused by recent tectonic activity: the 1999 Chi-Chi earthquake in
central Taiwan. Island Arc 12, 325–334.
Wang, Z.-H., 2006. Digital landslide technology and its application in investigation
of Tiantai town landslide. Chinese Journal of Geotechnical Engineering 28,
516–520.
Westaway, R.M., Lane, S.N., Hicks, D.M., 2000. Development of an automated
correction procedure for digital photogrammetry for the study of wide, shallow
gravel-bed rivers. Earth Surface Processes and Landforms 25(2), 200–226.
Westaway, R.M., Lane, S.N., Hicks, D.M., 2001. Airborne remote sensing of clear
water, shallow, gravel-bed rivers using digital photogrammetry and image
analysis. Photogrammetric Engineering and Remote Sensing 67(11), 1271–1281.
Westaway, R.M., Lane, S.N., Hicks, D.M., 2003. Remote survey of large-scale
braided, gravel-bed rivers using digital photogrammetry and image analysis.
International Journal of Remote Sensing 24(4), 795–815.
Whillans, I.M., Tsend, Y.-H., 1995. Automatic tracking of crevasses on satellite
images. Cold Regions Science Technology 23, 201–421.
Whited, D., Stanford, J.A., Kinball, J.S., 2002. Application of airborne multispectral
digital imagery to quantify riverine habitats at different base flows. River
Research and Applications 18(6), 583–594.
Whitesell, B.L., Vitek, J.D., Butler, D.R., 1988. Changes in the planform of the Red River,
McCurtain County, Oklahoma, 1938-84. Oklahoma Geology Notes 48, 194–211.
Wilford, J., 2009. Using airborne geophysics to define the 3D distribution and
landscape evolution of Quaternary valley-fill deposits around the Jamestown
area, South Australia. Australian Journal of Earth Sciences 56, 567–588.
Willige, B., 2010. Detection of local site conditions influencing earthquake shaking
and secondary effects in southwest-Haiti using remote sensing and GIS-
methods. Natural Hazards and Earth System Sciences 10, 1183–1196.
Winterbottom, S.J., 2000. Medium and short-term channel planform changes on the
Rivers Tay and Tummel, Scotland. Geomorphology 34, 195–208.
Winterbottom, S.J., Gilvear, D.J., 1997. Quantification of channel-bed morphology
in gravel-bed rivers using airborne multispectral imagery and aerial
photography. Regulated Rivers: Research and Management 13(6), 489–499.
Winterbottom, S.J., Gilvear, D.J., 2000. A GIS-based approach to mapping
probabilities of river bank erosion: regulated River Tummel, Scotland. Regulated
Rivers: Research and Management 16, 127–140.
Wohl, E., Oguchi, T., 2004. Geographic information systems and mountain hazards.
In: Bishop, M.P., Shroder, J.F. (Eds.), Geographic Information Science and
Mountain Geomorphology. Springer-Praxis, Chichester, pp. 309–342.
Woodruff, J.F., Evenden, L.J., 1962. Geomorphic measurement from aerial photos.
Professional Geographer 14, 23–26.
Wright, A., Marcus, W.A., Aspinall, R.J., 2000. Applications and limiations of using
multispectral digital imagery to map geomorphic stream units in a lower order
stream. Geomorphology 33(1/2), 107–120.
Wright, D.J., Goodchild, M.F., Proctor, J.D., 1997. Demystifying the persistent
ambiguity of GIS as ‘‘tool’’ versus ‘‘science’’. Annals of the Association of
American Geographers 87(2), 346–362.
Wright, R., Garbeil, H., Baloga, S.M., Mouginis-Mark, P.J., 2006. An assessment of
shuttle radar topography mission digital elevation data for studies of volcano
morphology. Remote Sensing of Environment 105(1), 41–53.
Xie, M., Esaki, T., Cai, M., 2004. A time-space based approach for mapping
rainfall-induced shallow landslide hazard. Environmental Geology 46,
840–850.
Yamaguchi, Y., Tanaka, S., Odajima, T., Tsuchida, S., Kamai, T., ERIM International,
1999. Detection of landslide movement using SPOT HRV and ADEOS AVNIR
data in Azumayama area, Japan. Proceedings of the 13th International
Conference on Applied Geologic Remote Sensing 13, II.97-II.103.
Yang, M.D., Yang, Y.F., Hsu, S.C., 2004. Application of remotely sensed data to the
assessment of terrain factors affecting the Tsao-Ling landslide. Canadian Journal
of Remote Sensing 30, 593–603.
Yoshizawa, N., Nishizawa, S., Misawa, T., Negishi, R., 1991. Application of
photogrammetry to analysis of landslide behavior. Journal of the Japan Society
of Photogrammetry and Remote Sensing 30, 8–20.
Zhou, C.H., Yue, Z.Q., Lee, C.F., Zhu, B.Q., Wang, Z.H., 2001. Satellite image
analysis of a huge landslide at Yi Gong, Tibet, China. Quarterly Journal of
Engineering Geology and Hydrogeology 34, 325–332.
Zouzias, D., Miliaresis, G.C., Seymour, K.S., 2011. Interpretation of Nisyros volcanic
terrain using land surface parameters generated from the ASTER global digial
elevation model. Journal of Volcanology and Geothermal Research 200,
159–170.
Zwally, H.J., Schutz, B., Abdalati, W., et al., 2002. ICESat’s laser measurements of
polar ice, atmosphere, ocean, and land. Journal of Geodynamics 34, 405–445.

Biographical Sketch

Associate Professor Napieralski completed his PhD in Earth and Atmospheric Sciences at Purdue University, West
Lafayette, IN, USA in 2005 with a dissertation focused on the development of geospatial tools to link numerical
ice-sheet model output with landform assemblages. After completing his PhD, he joined the Department of
Natural Sciences at the University of Michigan–Dearborn, MI, USA as an Assistant Professor of Geology, where he
continues to integrate spatial theory and applications into geomorphology both in research and in the classroom.
He was recently a US Fulbright Scholar to Cyprus to study soil and water conservation and was awarded the
Distinguished Teaching Award at the University of Michigan–Dearborn. His research includes alpine geography,
fluvial and glacial geomorphology, environmental science, GIScience, and field-based science education.
 Remote Sensing and GIScience in Geomorphological Mapping 227

Dr. Barr is currently a lecturer in Physical Geography at Queen Mary University of London, UK, having obtained
his PhD in Palaeoglaciology from the University of Sheffield, UK, in 2009. Dr. Barr’s research interests are in the
use of remote sensing and geographic information systems methods to better understand the distribution, di-
mensions, and dynamics of both modern and paleoglaciers. To date, his research has largely focused on the glacial
history of Eastern Siberia and Kamchatka.

Associate Professor Kamp received his PhD in Geography with an emphasis on Geomorphology from the
Technische Universität Berlin, Germany, in 1999. In the following years, he had positions at Freie Universität
Berlin, University of Nebraska–Omaha, and DePaul University in Chicago before joining The University of
Montana in 2005. Dr. Kamp’s research is on alpine geomorphology and glaciology, in particular, Quaternary and
recent glacial change. He carried out fieldwork in the mountains of South and Central Asia, and in the Rocky
Mountains and the Andes.

Dr. Kervyn holds a Masters in Geography from the University Catholique de Louvain, Belgium, and a PhD in
Geology from Ghent University, Belgium. He is currently working as a lecturer in physical geography, geo-
morphology, and natural hazards at the Vrije Universiteit Brussel. His research focuses on the geomorphology and
spatial distribution of volcanic edifices, on the effect of volcano-tectonic, gravitational, and erosion processes on
volcano morphology, and on monitoring and assessing hazards in active volcano-tectonic regions of Africa. His
research combines observations from digital elevation model and satellite images with analog experiments and
field observations.
 3.9 GIS-Based Soil Erosion Modeling
H Mitasova, North Carolina State University, Raleigh, NC, USA
M Barton and I Ullah, Arizona State University, Tempe, AZ, USA
J Hofierka, Pavol Jozef Safarik University, Kosice, Slovakia
RS Harmon, North Carolina State University, Raleigh, NC, USA
r 2013 Elsevier Inc. All rights reserved.

3.9.1 Introduction 229

3.9.2 Background 230
3.9.2.1 Erosion Processes 230
3.9.2.2 Spatial Variability 230
3.9.2.3 Temporal Variability 230
3.9.2.4 GIS-Based Erosion Modeling 232
3.9.3 Foundations in Erosion Modeling 234
3.9.3.1 Sediment Transport and Net Erosion/Deposition Equations 234
3.9.3.2 Detachment and Sediment Transport Capacities 235
3.9.4 Simplified Models of Erosion Processes 237
3.9.4.1 Detachment Capacity Limited Case 237
3.9.4.2 Transport Capacity Limited Case 238
3.9.4.3 Process-Form Relationship 239
3.9.4.4 Path-Sampling Transport Modeling 240
3.9.4.5 Gully Erosion 241
3.9.4.6 Statistical Modeling 242
3.9.4.7 Landscape Evolution Modeling 242
3.9.5 GIS Implementation 243
3.9.5.1 Coupling GIS and Models 243
3.9.5.2 Derived Model Parameters 244
3.9.5.3 Analysis and Visualization 244
3.9.6 Case Studies 244
3.9.6.1 North Carolina Piedmont 245
3.9.6.2 Mediterranean Landscape Evolution 248
3.9.7 Conclusion and Future Directions 253
Acknowledgments 254
References 254

Symbols r (m1) first-order reaction term

(x, y) (m) georeferenced coordinates dependent on land cover
t (s) time properties
qs(x, y, t) (kg m3) sediment mass density Kt(s) effective sediment transport
h(x, y, t) (m) water flow depth capacity coefficient
qs(x, y, t) (kg (ms)1) direction and rate of sediment Kd (s m1) effective erodibility (detachment
flow per unit width capacity coefficient)
qs ¼ 9qs9 magnitude of sediment flow rate s(Pa¼ kg m2) shear stress
per unit width s0 (Pa) critical shear stress
= divergence of a flow vector field a, b, m, n empirical exponents
ds(x, y, t) (kg m2s1) net erosion or deposition rate b (deg) surface slope angle
q(x, y, t) (m2 s1) direction and rate of water flow S ¼ tanb surface slope (rise over run)
per unit width gw ¼ qwg hydrostatic pressure of water with
Tc (kg (ms)1) sediment transport capacity the unit height
Dc (kg m2 s1) detachment capacity g ¼ 9.81 (m s2) gravitational acceleration

Mitasova, H., Barton, M., Ullah, I., Hofierka, J., Harmon, R.S., 2013. GIS-
based soil erosion modeling. In: Shroder, J. (Editor in Chief), Bishop, M.P.
(Ed.), Treatise on Geomorphology. Academic Press, San Diego, CA, vol. 3,
Remote Sensing and GIScience in Geomorphology, pp. 228–258

228 Treatise on Geomorphology, Volume 3 http://dx.doi.org/10.1016/B978-0-12-374739-6.00052-X

 GIS-Based Soil Erosion Modeling 229

qw ¼ 103 (kg m3) mass density of water a (deg) aspect angle equivalent to the
v (m s1) stream velocity direction of flow
n Manning’s coefficient jp (m1) profile curvature (normal
x (J (ms)1) the unit stream power curvature in gradient direction,
D (kg m2 yr1) average annual soil loss rate of change in slope)
R (M J mm (ha hr yr)1) rainfall factor jt (m1) tangential curvature (normal
K (ton ha hr soil erodibility factor curvature in the direction
(ha M J mm)1) perpendicular to the gradient, rate
LS dimensionless topographic of change in aspect)
(length-slope) factor jh (m1) plan curvature (tangential
L (m) hillslope length curvature projected to
L0 ¼22.23 m length of the standard USLE horizontal plane)
experimental plot Pe channel erosion index
C dimensionless land-cover factor w topographic wetness index
P dimensionless prevention z (m) elevation
measures factor c (m2 s1) sediment transport diffusion
U (m2 m1) upslope contributing area per constant
unit width Kg (m(1000 yr)1) long term diffusion coefficient
A (m2) upslope contributing area r (m) rainfall excess
s0 ¼ (cosa, sina) unit vector in the steepest slope
direction

Glossary Rill erosion Removal of soil particles by water flowing in

Detachment capacity Maximum potential soil
detachment by overland flow.
Erosion by overland flow Detachment of soil particles by
raindrop impact and their removal downslope by water
flowing overland as a sheet or in small concentrated flow
channels called rills.
Gully erosion Detachment of soil by surface water flow
concentrated in rapidly developing channel via headwall or
knickpoint migration.
Landscape evolution The change in the altitude and
morphology of the topography over time given variations in
erosion and deposition caused by numerous surface
processes.
Net erosion and deposition Soil mass that is lost or
gained at a unit area for a unit time due to removal,
transport, or deposition by water flow.
small concentrated flow channels called rills.
Sediment transport capacity Maximum potential
sediment transport by overland flow.
Shear stress Force of moving water against the bed of the
channel.
Stream power Rate of energy dissipation against the bed
and banks of a stream per unit downstream length.
Topographic erosion factor Measure of terrain impact on
erosion rates, computed as a function of water flow proxy
(hillslope length or contributing area) and slope angle.
Universal Soil Loss Equation Simple empirical equation
for estimation of annual soil loss rate for hillslopes with
simple geometry.
Water Erosion Prediction Project Process-based
continuous time modeling system for prediction of
sediment yield and erosion/deposition in small watersheds.

Abstract

This chapter explains the theory and methods for GIS-based modeling of soil erosion, sediment transport, and deposition
by surface water flow. The mathematical foundations of erosion models are introduced and simplified equations,
suitable for GIS implementation, are derived. The presented methods cover modeling of hillslope erosion and deposition,
gully formation, and landscape evolution processes. Coupling of erosion models with GIS is discussed, followed by
examples of GIS implementation of simple and advanced models. The concepts and methods are illustrated using two case
studies, that focus on feedbacks between the human activity and landscape processes.

3.9.1 Introduction spatial and temporal scales. Erosion processes that form the
land surface are extremely complex, poorly understood, and
The Earth’s surface, exposed to gravitational forces, wind, hard to predict quantitatively over large landscapes (Finlayson
water, and ice action, continuously evolves over wide range of and Montgomery, 2003). Remote sensing provides important
230 GIS-Based Soil Erosion Modeling
data that allow us to gain insights into interactions between
physical processes and environmental conditions that control
erosion and landform evolution. Recent advances in mapping
technologies, such as Light Detection and Ranging (LiDAR),
hyperspectral imaging, and ground penetrating radar have
dramatically increased the spatial and temporal resolution of
Earth surface and shallow subsurface monitoring. The new,
more detailed data indicate that fundamental changes in the
underlaying theory of erosion processes may be needed to
align it with the new observations. Geospatial information
science (GISc)-based analysis and modeling plays an import-
ant role in integrating observations and models, and improves
understanding and prediction capabilities aimed at min-
imizing negative impacts of erosion and sedimentation.
3.9.2 Background
3.9.2.1 Erosion Processes
Erosion encompasses a broad range of processes that involve
soil detachment and transport due to forces that act upon
Earth surface. Remote sensing and field-based geospatial
technology provides effective tools for detecting and mapping
specific landforms that are created by different driving forces,
including (Figure 1):
• soil erosion by overland flow (sheet, rill, gully),
• channel erosion by fluvial processes,
• gravitational erosion, landslides, debris flow,
• wind erosion,
• coastal erosion by surge and waves,
• glacial erosion.
Fundamentals of erosion processes and quantitative mod-
eling are covered in the relevant volumes of this series. Al-
though all types of erosion have been, at least to some extent,
analyzed or modeled using remote sensing data and geographic
information system (GIS) tools, most focus has been on soil
erosion by overland flow, due to its impact on agricultural
productivity, water quality, and sustainable land management.
Therefore, in this chapter, the emphasis is on erosion, sediment
transport, and deposition driven by surface water flow.
3.9.2.2 Spatial Variability
GIS-based modeling of soil erosion investigates spatial pat-
terns of soil detachment, transport and deposition, and their
impact on landscape evolution. Several approaches are used to
develop models of erosion processes: ‘Empirical or statistical
models’ derive the governing equations from monitoring data
or field experiments using statistical methods such as re-
gression. ‘Rule-based models’ are constructed from general
observations that relate a combination of inputs to observed
result, implemented as logical operations such as overlays.
‘Physics-based models’ use equations derived from natural
laws, such as continuity of mass and energy conservation.
Specific model implementations commonly combine ap-
proaches, for example, physics-based, sediment transport
routing with empirically derived parameters (Renschler, 2003;
Mitasova et al., 2005b). Spatial variability in landscape-scale,
soil-erosion processes requires discretization of landscape
representation in GIS-based models.
Spatially averaged hydrologic units are commonly em-
ployed in rule-based models (indexes and conditional over-
lays) or spatially averaged, process-based models. For example,
when modeling erosion by surface water flow, units with
simple geometry (tilted planes or polylines) represent hill-
slope segments, watershed hierarchies, channels, and stream
networks (Figure 2). The processes are then described by unit-
to-unit transport rules or by ordinary differential equations for
quasiunivariate transport. This approach is very effective for
systems that include anthropogenic features (ditches, sedi-
ment control structures), but selection of suitable spatial units,
their network topology and hierarchies require substantial
expertise and can significantly influence the results (Arabi
et al., 2006). This is true especially for complex, natural
environments that cannot be easily described by simple geo-
metrical features and where spatially averaged models have
limited capabilities to identify precise location of sediment
sources and sinks, and the pattern of their propagation
through landscapes. By reducing the size of the discrete units,
especially in locations with complex topography, spatial pat-
tern of erosion can be captured at the level of detail that ap-
proaches the fully distributed models. For example, the
combined Water Erosion Prediction Project (WEPP) and
GeoWEPP model supports modeling at three different levels of
spatial detail – watershed, user-defined hillslope segments,
and flow-paths (Renschler, 2003).
Spatially distributed models represent the input variables and
modeled values as continuous fields, usually discretized as
regular grids or irregular meshes. As opposed to the spatially
averaged models that predict erosion rates for an area of a dis-
crete spatial unit, the distributed models predict the modeled
values for any point in space and provide detailed spatial patterns
of sediment sources and sinks (Figure 2). On the other hand,
these models are complex in terms of data inputs and calibration,
and the quantitative summary predictions (e.g., total sediment
loads at a watershed outlet) are not necessarily more accurate
than the predictions by simpler, spatially averaged models.
Erosion and sediment transport is a multiscale process and
different processes require modeling at different spatial reso-
lutions. At hillslope scale, sheet and rill erosion dominate and
their accurate spatial modeling requires submeter resolution.
Watershed scale uses averaged sheet and rill erosion repre-
sentation and generally captures large gullies and channels.
Depending on the size of the modeled watershed and its
channels, 1–10 m resolution is required. Regional scale mod-
eling of large watersheds (thousands of square km), averages
sheet, rill and gully erosion, and uses simplified channel
representations and spatially averaged hydrologic units. The
resolutions of input data and results range from 30 m to
hundreds of meters. Irregular meshes and hydrologic units
provide means to adjust the level of detail to the complexity of
landscape features and support spatially variable scale.
3.9.2.3 Temporal Variability
Simple empirical erosion models assume relatively steady
erosion over time and estimate long-term, average annual

(a)
(b)
(c)
Figure 1 Erosion caused by surface water flow: (a) rills, (b) gullies, (c) stream bank erosion.
erosion rates based on empirical factors derived from long-
term observations. In reality, erosion rates change rapidly due
to the highly dynamic nature of the driving forces, such as
rainfall and water flow, over multiple temporal scales. The full
dynamics of erosion processes are difficult to capture due to
complex, multiscale interactions and limited temporal reso-
lution of the input data. Therefore, modeling is commonly
performed for the steady state case associated with peak flows
during individual erosion events. Sophisticated modeling
systems then perform continuous time simulations of several
interacting processes that incorporate steady-state impacts of
rainfall events over several years. For example, the WEPP
GIS-Based Soil Erosion Modeling 231
model simulates daily changes in soil and vegetation (crops)
and when rainfall occurs, the plant and soil characteristics are
used to determine whether surface runoff will occur or not. If
runoff is predicted, sheet, rill, and channel sediment detach-
ment and deposition are computed. The model thus simulates
impacts of climate (rainfall, temperature, solar radiation),
hydrology (infiltration, depressional storage and runoff),
water balance (evapotranspiration, percolation, and drain-
age), vegetation growth (cropland, rangeland, and forest),
erosion (interrill, rill, channel), and deposition (in rills,
channels, and impoundments). In addition to continuous
time simulations, WEPP can be used to assess impact of a
232 GIS-Based Soil Erosion Modeling

3.2 5.8

2.5 1.4

(a) (b)
500 m

Deposition
20
Erosion
10
0
(c) (d)

Figure 2 Representation of spatial variability: (a) spatially averaged representation: study area is partitioned into hydrologic units; (b) complex
topography within hydrologic units is replaced by simple geometry of a tilted plane with uniform slope and soil erosion is estimated as uniform
value within each unit; (c) spatially distributed representation estimates topographic erosion index for each grid cell using raster-based flow
routing; (d) spatially distributed erosion and deposition rates estimated by GeoWEPP, red area shows high erosion rates due to vegetation
removal (Moore et al., 2007).

single storm as well as an average long-term impact over
several years.
3.9.2.4 GIS-Based Erosion Modeling
GIS implementation of erosion models supports efficient
management of georeferenced data, computation of input
parameters for different scenarios, spatial analysis of the
modeling results, and effective visualization. GIS also in-
corporates tools for statistical analysis and modeling of ero-
sion processes captured by remotely sensed data.
In early 1990s, Geographic Resources Analysis and Support
System (GRASS) (Neteler and Mitasova, 2008) provided an
environment for pioneering work in integrating GIS and
hydrologic as well as erosion modeling (Rewerts and Engel,
1991; Mitchell et al., 1993). Most of the geospatial erosion
modeling development and applications focused on agriculture,
soil conservation, sediment pollution control, sustainable
military land management (Harmon and Doe, 2001; Gaffer
et al., 2008), and forestry, especially post-wildfire impacts.
Among the first implementations of hillslope erosion
modeling within GIS was the computation of the Universal
Soil Loss Equation (USLE) (Wischmeier and Smith, 1978),
and derivation of its topographic parameters from digital
elevation models (DEMs). Moore and Burch (1986) and later
Moore and Wilson (1992) paved the way for the USLE ap-
plications for landscapes with complex topography by de-
riving the relationship between the unit stream power theory
and USLE. Although the work generated heated discussions, it
stimulated development in the direction that resulted in nu-
merous implementations of USLE with GIS support for com-
plex topographic conditions (Mitasova et al., 1996; Desmet
and Govers, 1996). More recent GIS applications of USLE
cover wide range of scales including large watersheds with
land cover mapped from remote sensing imagery (Suri et al.,
2002; Cebecauer and Hofierka, 2008; Pandey et al., 2009a, b;
Jain and Das, 2010).
USLE and its updated, improved versions (Revised Universal
Soil Loss Equation RUSLE and RUSLE2) have been used to es-
timate soil detachment within watershed-scale models that focus
on nonpoint source pollution in agricultural, environmental,
and engineering applications, including well known models
such as ANSWERS, AGNPS, and SWAT (Rewerts and Engel, 1991;
Mitchell et al., 1993; Lim et al., 2005; Kim et al., 2009). These
models have been coupled with proprietary and open source
GIS to process and manage modeling inputs and analyze the
outputs (Table 1). Several watershed modeling systems have
been recently implemented as on-line tools (Park et al., 2009).
Although USLE was developed as a simple tool for farmers
to encourage soil conservation and thus its representation of
erosion processes has been greatly simplified, the needs for
research level modeling stimulated development of more
complex models and their coupling with GIS. A process-based,
continuous time model, based on the work by Foster (1982),
was developed as WEPP (Flanagan and Nearing, 1995;
Flanagan et al., 2007). Adaptation of WEPP to GIS, which led
to the development of Geo WEPP (Renschler, 2003), traces the
flow of sediment along flow paths on hillslopes generated
from a DEM. Integration of water and wind erosion models
 GIS-Based Soil Erosion Modeling 233

Table 1 Selected soil erosion models and their coupling with GIS; s, r – sheet and rill erosion; g – gullies; ch – channel; fl – fluvial; dz –
elevation change; see also US Department of Agriculture, Agricultural Research Service (USDA ARS) at http://www.ars.usda.gov/Research/
docs.htmdocid=5971, SWAT at http://swatmodel.tamu.edu/, EUROSEM at http://www.eurosem-soil-erosion.org/, LISEM at http://www.itc.nl/lisem/

Models Processes Spatial scale and representation Temporal scale GIS implementation Reference

USLE s, r hillslope – profile annual map algebra USDA ARS

RUSLE s, r hillslope – profile annual – event map algebra USDA ARS USDA
GeoWEPP s, r, ch small watershed – raster continuous time – event extension, Web ARS, NSERL USDA
AnnAGNPS s, r, ch watershed – custom grid continuous time – event custom GIS ARS
ArcSWAT s, r, ch watershed – hydrologic unit continuous time – event extension USDA ARS, TAMU
MapWinSWAT s, r, ch watershed – hydrologic unit continuous time – event custom Web GIS USDA ARS, TAMU
openLisem s, r, ch watershed – raster event PCRaster De Roo et al. (1996)
SIBERIA s, r, ch, dz landscape – raster continuous time data input CSMDS (2011)
CHILD s, r, ch, dz landscape – mesh continuous time data input CSMDS (2011)
USPED s, r small watershed – raster annual – event map algebra Neteler and Mitasova (2008)
SIMWE s, r small watershed – raster event GIS module Neteler and Mitasova (2008)
r.landscape.evol s, r, ch, dz landscape – raster continuous time GIS module Barton et al. (2010a)

within the WEPP modeling framework is under development.
It represents an important step toward understanding of
combined impacts of water and wind erosion on soil con-
servation and agricultural sustainability.
Several erosion modeling systems with strong geospatial
components have been developed in Europe. European Soil
Erosion Model (EUROSEM) (Morgan et al., 1998) introduced
a dynamic approach for predicting sediment transport from
small watersheds. The LImburg Soil Erosion Model (LISEM)
(De Roo et al., 1996; Sheikh et al., 2010) is a physically based
runoff and erosion model for research, planning, and con-
servation purposes. It simulates the spatial effects of rainfall
events on small watersheds and it uses the free GIS PCRaster
Environmental Software (Karssenberg et al., 2001). As one of
the most comprehensive geospatial erosion modeling tools,
LISEM incorporates rainfall, interception, surface storage in
microdepressions, infiltration, vertical movement of water in
the soil, overland flow, channel flow in man-made ditches,
detachment by rainfall and throughfall, transport capacity,
and detachment by overland flow. Influence of compaction on
the hydrological and soil erosion processes and gully incision
are also included.
The SIMulation of Water Erosion (SIMWE) model (Mitas
and Mitasova, 1998; Mitasova et al., 2005b) was developed as
a bivariate generalization of hillslope erosion model used by
WEPP to capture the spatial pattern of erosion, sediment
transport, and deposition under spatially diverse conditions.
Its robust path sampling algorithm and integration within GIS
supports simulations of water and sediment flow patterns at
high spatial resolutions, including impacts of small terrain
variations such as depressions, ditches, or, checkdams.
Although several GISc-based implementations of erosion
models predict locations of high erosion due to concentrated
water flow, specific models that focus on ephemeral gully ero-
sion were also developed (DeRose et al., 1998; Woodward,
1999). These models rely on field observations and modified
channel sediment transport equations to improve the quantita-
tive estimates of sediment eroded by ephemeral gully formation.
Time series of aerial photography combined with multitemporal
DEMs were also used to map the location, evolution and erosion
rates of gullies (Martı́nez-Casasnovas et al., 2004).
In addition to the models focused on short term erosion
(single event to multiple years), landscape evolution models
have been developed to capture impact of erosion on land-
forms over hundreds or thousands of years. These models
commonly use GIS data as inputs, but they have been de-
veloped and used mostly outside GIS and are covered in re-
spective volumes of this series. Well known models include
SIBERIA (Willgoose et al., 1991; Hancock et al., 2002;
Willgoose, 2004), CHILD (Tucker et al., 2001a, b), or
rule-based cellular automata models for simulation of land-
forms (Luo, 2001), braided rivers evolution (Murray and
Paola, 1994) and sand dunes (Pelletier et al., 2009). The
WILSIM landform evolution model (Luo et al., 2004) was
implemented as a web-based simulation tool. Recently, the
development of landscape evolution models has been co-
ordinated by the Community Surface Dynamics Modeling
System (CSMDS, 2011) program.
Statistical approaches have been commonly combined with
remote sensing techniques to improve the results of erosion
mapping. For example, a logistic regression-based erosion
index was developed to map probability of erosion resulting
from concentrated water flow from high resolution DEMs
(Pike et al., 2009). Tree-based regression models were used to
identify the topographic parameters that explain the variability
in field gully measurements (Kheir et al., 2007). The most re-
cent GIS-based models study pattern of erosion and deposition
rates at very high resolutions using repeated scans of landscape
by terrestrial LiDAR and by differencing the resulting DEMs.
Several recent papers indicate that ‘‘there are major weak-
nesses in the current understanding and data underpinning
existing models’’ (Govers et al., 2007; Wainwright et al., 2008;
Polyakov et al., 2004; Finlayson and Montgomery, 2003; Jetten
et al., 2003). For example, Van Oost et al. (2005) concluded
that performance of process-based erosion and hydrological
models is extremely sensitive to parameter estimations and that
predictions are generally poor. The difficulties associated with
accuracy of spatially distributed soil erosion models are com-
monly due to the spatial and temporal variability of erosion
processes and uncertainty associated with the model par-
ameters. Jetten et al. (2003) suggest that model performance
may be improved by using more complete spatial information
234 GIS-Based Soil Erosion Modeling
for model calibration and validation instead of the data re-
stricted to the measurements at watersheds outlets. Spatially
and temporally distributed observations may provide the data
to elucidate the poorly understood interactions between sedi-
ment load and detachment rates, and provide insights into
scaling of erosional processes and mechanisms. Therefore, ad-
justments to the theoretical foundations of erosion modeling
presented in the next section can be expected in future as new
monitoring and sensing technologies provide more complete
experimental data and field observations.
3.9.3 Foundations in Erosion Modeling
Spatial and temporal patterns of erosion and deposition
are results of complex interactions between the Earth
system processes such as rainfall, surface, subsurface and
ground water flow, vegetation growth, soil detachment,
transport, and deposition. The focus of this section is on soil
and sediment erosion and transport while treating rainfall,
runoff, vegetation cover, and soil properties as inputs without
going into detail about the estimation of these parameters’
values. Here, general mathematical representation of erosion
and sediment transport processes is described and relation
between models with different levels of complexity and
backgrounds is derived, highlighting the common principles.
3.9.3.1 Sediment Transport and Net Erosion/Deposition
Equations
Soil, detached by raindrop impact and shearing force of overland
flow is transported by flowing water until its transport capacity
decreases, leading to sediment deposition (Haan et al., 1994).
The fundamental relationship that governs sediment transport
by overland flow is continuity of mass. It can be described by the
sediment continuity equation, which relates the change in
sediment storage over time, and the change in sediment flow rate
along 3D hillslope to effective sources and sinks (Haan et al.,
1994; Govindaraju and Kavvas, 1991; Foster and Meyer, 1972).
To capture the spatial pattern of sediment transport over com-
plex topography the bivariate form of the continuity of sediment
mass equation (Hong and Mostaghimi, 1997) is used
q ½rs ðx,y,tÞhðx,y,tÞ
þ r  qs ðx,y,tÞ ¼ sources  sinks
qt
¼ ds ðx,y,tÞ,
where (x, y) are georeferenced coordinates, t (s) is time,
rs(x, y, t) (kg m3) is sediment mass density, h(x, y, t) (m) is
water flow depth, qs(x, y, t) (kg (ms)1) is the vector that
represents the direction and rate of sediment flow per unit
width (unit sediment load), r denotes divergence of the
sediment flow vector field, and ds(x, y, t) (kg m2 s1) is the
net erosion or deposition rate. The sediment flow rate is a
function of water flow and sediment concentration:
qs ðx,y,tÞ ¼ rs ðx,y,tÞqðx,y,tÞ,
where q(x, y, t) (m2 s1) represents the direction and rate of
water flow per unit width (unit flow discharge vector) that can
be expressed as a function of water depth h(x, y, t), where the
form of this function depends on the channel cross-section
and flow conditions (Haan et al., 1994).
A simplified, steady state form of the continuity equation
can be derived for the conditions when, at the given location,
the change in water flow and sediment concentrations over
time is close to zero. The net erosion or deposition rate ds(x, y)
is then computed as a divergence of steady state sediment flow
rate per unit width qs(x, y):
q ½rs ðx,y,tÞhðx,y,tÞ
¼ 0 - r  qs ðx,yÞ ¼ ds ðx,yÞ: ½3
qt
Assuming steady rainfall excess rates, eqn [3] applies to
the peak flow and peak concentration conditions when
the water depth and sediment concentrations remain
constant over time. To simplify equations, the (x, y) notation
indicating spatially distributed variables represented by
bivariate continuous functions is omitted in the rest of this
chapter.
The sources and sinks term in eqn [1] is derived from the
assumption that the detachment and deposition rates are pro-
portional to the difference between the sediment transport cap-
acity and the actual sediment flow rate (Foster and Meyer, 1972):
ds ¼ s½Tc  qs , ½4
where Tc (kg(ms)1) is the sediment transport capacity,
qs ¼ 9qs9 is the magnitude of sediment flow rate per unit width,
and s (m1) is the first-order reaction term dependent on soil
and cover properties. The expression for s can be obtained
from the following relationship (Foster and Meyer, 1972):
ds qs
þ ¼1 ½5
Dc Tc
which states that the ratio of the erosion rate to the deta-
chment capacity Dc(kg m2 s1) plus the ratio of the sediment
flow rate to the sediment transport capacity is a conserved
quantity (unity). Equation [5] is based on the observed rela-
tionship between soil detachment and transport when the finite
amount of energy available in the overland flow is applied
proportionally to detaching and to transporting the sediment so
that the total relative available energy is unity (Haan et al.,
1994). The s coefficient thus controls how close is the sediment
transport to the detachment limited (erosion only) or transport
½1 limited (maximum extent of deposition) regime (Figure 3). The
detachment capacity Dc is then proportional to the transport
capacity Tc:
Dc ¼ sTc : ½6
This leads to the expression of net erosion and deposition as:
 
qs
ds ¼ Dc 1  : ½7
Tc
½2 This concept is used in several erosion models including
WEPP (Haan et al., 1994; Flanagan and Nearing, 1995) and
SIMWE (Mitas and Mitasova, 1998). To estimate deposition
of particles with different settling velocities s may be
 GIS-Based Soil Erosion Modeling 235

 = 0.001
0 1.0 10.0 100.0

Sediment flow rate Erosion Deposition

(a)

 = 100.0
300 m
(b)

Figure 3 Impact of s on spatial distribution of sediment flow rate (visualized as a surface) and erosion and deposition (draped as a color map
over DEM): (a) low value of s ¼0.001, with Dc{Tc leads to prevailing erosion, close to the detachment limited regime, (b) high value of s ¼100,
DccTc leads to large extent of deposition, close to the transport capacity limited regime.

approximated by (Foster, 1982): maximum potential detachment rate, respectively. Numerous

Vs
s¼ , ½8
2q
where Vs (m s1) is settling velocity for the given particle size
and q (m2 s1) is flow discharge per unit width. It is possible
to use other frameworks for estimation of s. For example,
Govers et al. (2007) proposed to express the detachment DL as
function of unit length sheer force G:
 
qs
DL ¼ aG 1  , ½9
Tc
where a is an empirical coefficient.
It is important to note, however, that the relationship be-
tween the sediment detachment and transport capacities and
the actual detachment and sediment loads is not fully
understood (Govers et al., 2007; Nearing et al., 1997). Further
experimental research is needed to elucidate this relationship
and develop the underlying theory and equations.
3.9.3.2 Detachment and Sediment Transport Capacities
To solve the equation for sediment transport rate, it is neces-
sary to estimate the transport and detachment capacities. The
sediment transport capacity Tc and detachment capacity Dc
represent the maximum potential sediment flow rate and the
simplified empirical equations represent these rates under
different conditions (Julien and Simons, 1985). They are often
expressed as functions of shear stress (Foster and Meyer,
1972):
Tc ¼ Kt ta , ½10
Dc ¼ Kd ðt  t0 Þb , ½11
where Kt(s) is the effective sediment transport capacity
coefficient, Kd(s m1) is the effective erodibility (detachment
capacity coefficient), t(Pa¼ kg m2) is the shear stress,
t0(Pa) is the critical shear stress, and a and b are empirical
exponents. The shear stress t is function of water depth h and
surface slope angle b(deg) such that
t ¼ gw RS; ½12
where gw ¼ rwg is the hydrostatic pressure of water with the
unit height, g ¼ 9.81 (m s2) is the gravitational acceleration,
rw ¼ 103 (kg m3) is the mass density of water, S¼ tanb is
surface slope and b (deg) is the steepest slope angle (see Haan
et al. (1994), Moore and Burch (1986) for discussion on re-
lationship between water depth and wetted perimeter used in
the shear stress and stream power equations).
Transport capacity, especially in channels, streams and
rivers, can be expressed as a function of unit stream power
236 GIS-Based Soil Erosion Modeling
o (J (ms)1) (Moore and Burch, 1986) that represents the rate
of energy dissipation against the bed and banks of a channel.
If the stream velocity v (ms1) is estimated using Manning’s
equation (Haan et al., 1994; Dingman, 2002), then
v ¼ n1 h0:6 S0:5 ;
where n is Manning’s coefficient, then the unit stream power
can be expressed as:
o ¼ tv ¼ gw n1 h1:6 S1:5 ;
Sediment transport capacity is then computed as
Tc ¼ Ks o ¼ Ks n1 gw hm Sn ,
where Ks is transport capacity coefficient. Several studies in-
dicate that stream power is a better hydraulic predictor vari-
able for detachment and sediment yield than is shear stress
(Nearing et al., 1997, 1999). The equation for sediment
transport can significantly influence the magnitude and spatial
pattern of sediment flow rates as well as the predicted erosion
and deposition (Figure 4). Field observations are generally
needed for selection of the sediment transport equation and
its parameters that adequately represents the specific modeled
landscape.
Julien and Simons (1985) analyzed numerous sediment
transport equations and derived the following general equa-
tion
qs ¼ jid qm Sn ð1  t=t0 Þe ,
where q is unit water discharge, i is rainfall intensity, and j, d,
e, m, n are experimental or physically based coefficients that
depend on type of flow.
When the critical shear stress t0 is negligible and for
channels where d ¼ 0 and qs is not dependent on rainfall in-
tensity the equation simplifies to
qs ¼ jqm Sn :
15
10
5
0
N
(a) (b)
Figure 4 Impact of the sediment transport equation on spatial distribution of sediment flow rate: (a) shear stress eqn [10] with a ¼0.6 predicts
sediment transport increasing slowly with concentrated flow; (b) with a ¼1.5 sediment transport increases rapidly with concentrated water flow;
(c) stream power eqn [15] further increases the difference between sediment transport by sheet and concentrated flow.
As presented above, the stream power, shear stress, and
consequently transport and detachment capacities are gener-
ally expressed as power functions of water flow properties and
slope. For a given rainfall excess rate and surface roughness,
these two variables can be derived from DEM using the GIS
½13 functions for topographic analysis to compute the slope, and
the flow routing tools to compute the upslope contributing
area as an input for estimating unit water flow and
water depth.
The parameters Kt, Kd, t0 used for the estimation of Dc and
Tc are functions of soil and land-cover properties, and are
½14 much more difficult to estimate accurately than the topo-
graphic parameters. They have been derived using empirical
equations or directly from experiments for a wide range of
soils, cover, agricultural, and erosion prevention practices
½15 within the WEPP model (Flanagan and Nearing, 1995), but
the values are valid only for the conditions similar to those in
the experiments. Moreover, for landscape scale modeling, the
soil data are rarely available at the level of detail comparable
with DEMs, and temporal changes in soil properties are gen-
erally neglected. Land-cover data can be obtained at very high
resolutions from airborne or even satellite imagery, however,
the relationships between the type and density of vegetation
cover and the soil detachment and transport parameters are
not very well understood.
In addition to the difficulties of accurate estimation of
detachment and transport capacity parameters, the values of
exponents a, b in the eqns [10] and [11] or m, n in eqn [16]
depend on the type of flow (Julien and Simons, 1985) and
substantially influence model behavior. For example, the
WEPP model uses a¼ 1 and b ¼ 1.5 which means that with
½16 increasing water flow, transport capacity increases faster than
detachment, which is not always the case. Uniform values of
these exponents are commonly not general enough for ero-
sion modeling at landscape scale with different types of flow
due to variability in land-cover properties and spatially vari-
able exponents may be needed (Mitas and Mitasova, 1998).
The solution of eqn [1] can be obtained by several partial
differential equation solvers, including finite element methods
(Hong and Mostaghimi, 1997), finite difference methods
(Desmet and Govers, 1995), or path sampling (Mitas and
½17 Mitasova, 1998; Mitasova et al., 2005b). Detailed description
(c) 500 m

of these methods is beyond the scope of this chapter, however,
in the following sections, various simplifications of the full
solution useful for GIS-based soil erosion modeling are
described.
3.9.4 Simplified Models of Erosion Processes
To satisfy the need for models which are easy to compute and
for which data are readily available, equations for simplified
models of erosion processes can be derived. Depending on
interactions between rainfall, runoff, and local soil and cover
conditions, erosion and sediment transport varies between
two limiting cases (Foster and Meyer, 1972; Tucker et al.,
2001a):
• detachment capacity limited,
• sediment transport capacity limited.
Deriving equations for these limiting cases by simplifying
the general sediment transport equation, leads to well-known
soil erosion models that are useful for estimates of average
annual soil erosion rates or total sediment loads at watershed
outlets. In addition to the limiting cases of sheet and rill
erosion, gully formation and a landscape evolution model
that integrates several erosion processes are also discussed.
3.9.4.1 Detachment Capacity Limited Case
When transport capacity of overland flow exceeds its detach-
ment capacity over the entire studied landscape, erosion, and
sediment transport is detachment capacity limited (i.e., no
deposition occurs). For example, this case is typical for sedi-
ment transported by large amount of water over compacted
soil. The detachment capacity limited case is represented by
TccDc leading to s - 0. Assuming that the critical shear stress
is negligible t0 ¼ 0 the net erosion will be equal to the de-
tachment capacity:
ds EDc ¼ Kd tb ¼ Kd ðgw hSÞb : ½18
This equation can then be used to compute the detachment
limited, steady state erosion with water depth approximated,
for example, from upslope contributing area. If more accurate
estimate is needed, water depth distribution can be computed
by a hydrologic model.
If the estimate of water depth is further simplified by as-
suming a planar hillslope with no water flow divergence or
convergence, upslope contributing area per unit width can be
replaced by hillslope length, leading to expression that is the
basis for the most common erosion model USLE and its
revised version RUSLE
D ¼ RKLSCP, ½19
where D (kg m2 yr1) is average annual soil loss, R (M J mm
(ha hr yr)1) is rainfall factor, K (ton ha hr (ha M J mm)1) is
soil erodibility factor, LS is a dimensionless topographic
(length-slope) factor, C is a dimensionless land-cover factor,
and P is a dimensionless prevention measures factor. The
length-slope factor has been derived from experiments and in
GIS-Based Soil Erosion Modeling 237
the original USLE has the form (Wischmeier and Smith,
1978):
LS ¼ ðL=L0 Þm 64:5 sin2 b þ 4:56 sin b þ 0:0654, ½20
where L (m) is the hillslope length, L0 ¼22.1 m is the length of
the standard USLE experimental plot, m is exponent with
values ranging between 0.2 for slopes less than 1% and 0.5 for
slopes steeper than 5%. RUSLE expands the applicability of
this equation by introducing additional empirical equations
for computation of LS factor for different slope steepness
intervals and provides updated values and equations for
computation of all factors, including the R-factor for indi-
vidual storms (Renard et al., 1994).
The USLE/RUSLE has been developed for estimation of
average sheet and rill erosion on uniform fields with simple,
relatively planar geometry. Numerous efforts and implemen-
tations have extended its applications to large watersheds and
fields with complex topography. For example, USLE is used in
several nonpoint source pollution models for estimation of
averaged soil detachment in hydrologic units, with the de-
tached soil then routed through the watershed.
An important modification of USLE/RUSLE has been de-
rived by Moore and Burch (1986) and applied within GRASS
GIS (Mitasova et al., 1996). This improvement replaced the
hillslope length by upslope contributing area (Figure 5),
leading to the following equation:
 m  
U sinb n
D ¼ RKCPðm þ 1Þ , ½21
L0 S0
where D(kg m2 yr1) is the average annual soil detachment
(soil loss) rate, U(m2 m1) is the upslope contributing area
per unit width (a proxy for discharge), and S0 ¼ 0.09 is the
slope of the standard USLE plot. Single storm and monthly R
are also available, making eqn [21] suitable for estimation of
soil loss for single storms and for modeling of monthly soil
loss distribution over a year (Haan et al., 1994).
Exponents m, n depend on the prevailing type of erosion
(sheet, rill) and the typical values are m ¼ 0.4–0.6 and
n ¼ 1.0–1.3. Replacement of slope length, used in the original
formulation of USLE/RUSLE, by the upslope area predicts
increased erosion due to the concentrated flow without the
need to a priori define these locations as inputs for the model.
It is important to note that the USLE/RUSLE model is
commonly applied within GIS for conditions for which it has
not been originally designed and at scales at which its validity
is uncertain. For example, the original values of the USLE/
RUSLE factors have been derived from experiments using plots
only 22.13 m long, however, many published GIS applications
of USLE used the equations at resolutions with much larger
grid cell size and at locations with much steeper topography
than the 9% slope of the USLE plots (Pandey et al., 2009a).
The authors generally justify such applications by the fact that
comparable experiments were not performed for the studied
conditions and USLE was the best option available to them.
Moreover, in complex topography, detachment limited ero-
sion is rather rare because concave areas at toes of hillslopes
reduce transport capacity, leading to sediment deposition,
further limiting the applicability of the original USLE/RUSLE
238 GIS-Based Soil Erosion Modeling

t ha–1 yr–1
0
6 Modeled soil erosion
12
18

(a) Observed sediment deposition

0 300 m
(b)

Figure 5 Spatial pattern of topographic erosion factor based on: (a) flowpath length (eqn [19]), which does not capture increased erosion due
to convergent water flow, (b) flow accumulation (eqn [21]), which incorporates impact of water flow convergence. The topographic factor maps
are draped over elevation surface. The cross-section shows observed deposited material, indicating locations where these topographic erosion
factors are not applicable for modeling long term net erosion/deposition pattern. Images based on data provided by Prof. K. Auerswald,
Technische University Muenchen.

(Figure 5). In spite of its limitations, the USLE/RUSLE models,
as well as their modifications and applications in GIS have
played an important role in soil conservation efforts and
sustainable land management (Renard et al., 1994; Haan
et al., 1994; Hammad et al., 2004; Cebecauer and Hofierka,
2008).
The transport capacity limited case with spatially variable
soil and cover properties can be modeled using the idea ori-
ginally proposed by Moore and Burch (1986). It combines the
USLE/RUSLE parameters and upslope contributing area per
unit width to estimate the sediment flow at sediment transport
capacity:

qs ETc ERKCPU m ðsinbÞn : ½23

3.9.4.2 Transport Capacity Limited Case
If the soil detachment significantly exceeds the sediment
transport capacity of overland flow, the sediment flow rate will
be at the sediment transport capacity. This means that the
magnitude of sediment flow can be approximated by Tc and
net erosion/deposition rate can be computed as a change in
the sediment transport capacity derived from eqn [3]:
qðTc cosaÞ qðTc sinaÞ
ds ¼ r  qs Er  ðTc s0 Þ ¼ þ ,
qx qy
where s0 ¼ (cos a, sin a) is the unit vector in the steepest slope
direction given by a(deg), the aspect angle of the terrain sur-
face equivalent to the direction of flow. As opposed to the
detachment capacity limited case, the transport capacity lim-
ited formulation can predict spatial pattern of both erosion
and deposition (Figure 6).
The net erosion/deposition ds(kg m2 s1) is then com-
puted as a divergence of sediment flow vector field given by
eqn [22].
The exponents m, n control the relative influence of water
and slope terms and reflect the impact of different types of
flow. The observed extent of colluvial deposits indicated that
an exponent m ¼ 1 reflects the pattern of compounded, long
term impact of both rill and sheet erosion (Mitas and Mita-
½22
sova, 1998). Averaging over a long-term sequence of large and
small events was important for capturing the observed spatial
extent of deposition as well as channel incision due to con-
centrated flow (Warren et al., 2005). Spatially variable m(x, y),
n(x, y), approximated by a continuous function, can be used
to account for the differences in surface flow (Figure 7).
Models representing limiting cases of erosion are simple
to compute in a GIS using the standard flow routing,
 GIS-Based Soil Erosion Modeling 239

t ha−1 yr−1
6
Deposition
0
Erosion
–6

0 300 m

Observed sediment deposition

Figure 6 Topographic potential for net erosion and deposition, estimated as a change in sediment transport capacity.

Erosion
Deposition

(a) (b) (c) 500 m

Figure 7 Impact of exponents m, n in eqn [23] on resulting erosion and deposition rates pattern: (a) m¼ 1.0, n¼1.0 predicts erosion in
87% and deposition in 13% area, including deposition in concave forms of headwaters; (b) m¼ 1.6, n¼1.0 predicts erosion in 92% area
whereas limiting deposition to 8% area mostly in low, broader concave forms; (c) spatially variable m¼ 1 at lower and m ¼1.6 at higher
elevations.

topographic analysis modules combined with map algebra.
They can be used to estimate spatial distribution of soil
detachment and net erosion/deposition rates for a single
storm, as well as monthly and annual averages. Caution
should be used when interpreting the results from modifi-
cations of USLE/RUSLE that incorporate water flow con-
vergence/divergence and erosion/deposition, because the
soil and cover parameters were developed for simple planar
fields and detachment limited erosion. To accurately predict
erosion and deposition rates for complex terrain and spatially
variable land-cover conditions these models need to be
calibrated for a specific geographic area.
deposition as a difference between sediment inflow and out-
flow along the segments of this path (Moore and Wilson,
1992; Desmet and Govers, 1995; Mitasova et al., 1996). As-
suming uniform rainfall, soil and cover conditions, and a
transport capacity limiting case with qsETc, the net erosion/
deposition rate along a flow path (univariate model) can be
derived as directional derivative of sediment transport capacity
in the direction of steepest slope (aspect):
dTc
ds ¼ ¼ rTc  s0 ½24
ds
ds ¼ Ke ðrh  s0 sinb  hkp Þ, ½25

where kp(m1) is profile curvature that measures the rate

3.9.4.3 Process-Form Relationship
As shown in the section 3.9.4.2 topography controls the spa-
tial pattern of sediment transport and erosion/deposition
through upslope contributing area and slope angle. In this
section, the role of terrain curvature in distribution of erosion
and deposition is demonstrated.
Most sediment transport models route sediment in single
direction along flow paths and compute the net erosion and
of change in slope in the gradient direction, Ke ¼ Kt rw g, and
ds40 represents the net erosion rate and dso0 represents
the net deposition rate. The univariate formulation includes
the impact of water flow convergence/divergence through the
water depth term and flow acceleration/deceleration through
the profile curvature.
Within the bivariate formulation, given by eqn [22], the
net erosion and deposition rate is estimated as a divergence of
240 GIS-Based Soil Erosion Modeling

the sediment flow vector field qs (Mitas and Mitasova, 1998):
ds ¼ r  qs ¼ r  ðTc s0 Þ
ds ¼ Ke ½rh  s0 sinb  hðkp þ kt Þ,
where kt (m1) is the tangential curvature that measures the
change in aspect (curvature in the direction perpendicular to
the gradient, i.e., the direction tangential to a contour line
projected to the normal plane).
When eqns [25] and [27] are compared, the fundamental
difference is in incorporation of tangential curvature. Ac-
cording to the bivariate eqn [27], the spatial distribution of
erosion/deposition is controlled by the change in the overland
flow depth rh and by the local geometry of terrain including
both profile and tangential curvatures. Equation [27] thus
demonstrates that the local acceleration of flow in both the
gradient and tangential directions play equally important roles
in spatial distribution of erosion/deposition.
The impact of the tangential curvature is therefore twofold
(Figure 8). First, kt influences the water depth through its control
of water flow convergence and divergence, with tangential con-
cavity leading to rapid increase in water depth and increase in
erosion rates. Second, kt causes a local change in sediment flow
velocity with an opposite effect, with tangential concavity re-
ducing the sediment transport, creating conditions for de-
position. Therefore, it is the interplay between the magnitude of
water flow change and both terrain curvatures in eqn [27] which
determines whether erosion or deposition will occur.
When the results of the univariate and the bivariate models
were compared with the observed pattern of deposits (Mitas
and Mitasova, 1998), the model that computes the sediment
load change along a flowline (eqn [25]) failed to predict
deposition observed in areas where kpE0 but kto0.
½26 The prediction by eqn [27] was in better agreement with
the observed pattern of deposition in these areas (Figure 8). It
½27 is important to note, however, that the total sediment load at
the outlet is the same for both formulations and it is only the
spatial pattern of sediment redistribution within the water-
shed that is influenced by the univariate versus bivariate
solution.
3.9.4.4 Path-Sampling Transport Modeling
Evolution of sediment transport over complex landscapes can
be simulated using a path-sampling approach (Mitas and
Mitasova, 1998; Mitasova et al., 2005b) that is based on the
concept of duality between discrete particles and continuous
fields. The approach solves the steady state sediment flow
eqn [3] that is rewritten to include a small diffusion term
pr2r:
g
 r2 r þ r  ðrvÞ þ rs9v9 ¼ sTc , ½28
2
where r¼ rsh(kg m2) is the mass of sediment carried by
water per unit cross-section area, g(m2 s1) is the diffusion
constant, and v(m s1) is the water flow velocity estimated
by Manning or Chezy equation (Dingman, 2002). On the left
hand side of eqn [28] the first term describes local diffusion,
the second term is a drift driven by the water flow whereas the
third term represents a velocity dependent ‘potential’ acting
on r. The size of the diffusion constant is about one order of
magnitude smaller than the reciprocal Manning’s constant so
that the impact of the diffusion term is relatively small.

+ +
(a) (b) (c)
80.00
Erosion Deposition 50.00
25.00
0.00

300 m
(d) (e) (f)

Figure 8 Computing net erosion and deposition using univariate and bivariate formulation (eqns [25] and [27]): (a) term representing change in
water-flow depth (rh  s0 sinb); (b) term that combines water-flow depth with profile curvature (hkp); (c) term that combines water-flow depth with
tangential curvature (hkt); (d) net erosion/deposition computed as a change in sediment flow along flow-path is a sum of (a) and (b); (e) net erosion/
deposition as sediment flow divergence is a sum of (a), (b) and (c); (f) observed deposited sediment with highlighted locations where divergence
improves the modeling result. Figure based on Mitas and Mitasova (1998) using data provided by Prof. Auserwald, Technical University Muenchen.

0.2
0.1
0.02
0.001
(a)
kg m–1s–1
10
1
Erosion
0.1
0.01
(c)
Figure 9 GIS-based erosion modeling using path sampling method: (a) C-factor at the study site with disturbed land shown in red; (b) particle
representation of soil detachment (only 1% of particles is displayed), particle density is higher in locations with disturbed land; (c) raster
representation of sediment flow rate computed as a function of particle density per grid cell; (d) net erosion and deposition computed as a
divergence of sediment flow.
It represents local dispersion processes caused by micro-
topography which is not captured by the DEM.
The sediment flow described by eqn [28] can be solved by
the path-sampling stochastic method (Mitasova et al., 2005b)
by propagation of particles according to the continuity equa-
tion. Sediment concentrations are then computed based on
particle density (Figure 9). This approach has been used to
develop the SIMWE model which was implemented in GRASS
GIS as a module ‘r.sim.sediment’ (Neteler and Mitasova, 2008).
3.9.4.5 Gully Erosion
Traditionally, modeling of gully and channel erosion has been
performed outside of a GIS because most of the required data,
especially the DEMs, did not have sufficient resolution to
capture the gully or channel geometry. With high resolution
DEMs (submeter grid cell size) now available, gully and
channel erosion models can be implemented and applied
using GIS-based approaches.
Gullies develop under specific conditions when the tractive
force of the water flow exceeds a certain threshold value and
GIS-Based Soil Erosion Modeling 241
(b)
1 km
t ha–1 yr–1
Depo- >30
sition
0
<–30
(d)
shallow flow starts to concentrate in the channel. This transi-
tion occurs abruptly at a knickpoint or headwall. In fact,
several such transitions may occur, eventually merging the
knickpoints and forming one incised channel (Haan et al.,
1994). The gully then develops via headwall migration and
channel widening. The process of initiation requires a rela-
tively large rainfall event so that the rainfall excess generates
water flow exceeding the threshold value needed to detach and
transport large amounts of sediment. The immediate mor-
phological effect of the gullying process is a substantial change
in topography with a sharp change in slope in a headwall and
channel banks. This contributes to a further gully develop-
ment until a new equilibrium is found.
The difference between modeling the rill and gully erosion
lies in the treatment of elevation represented by a DEM. In the
case of landscape-scale rill erosion modeling, the elevation
surface is considered as fixed throughout the erosion event.
Actual net erosion/deposition rates change the elevation sur-
face only marginally and the morphological effect can be
observed only after many similar events. Gully erosion, how-
ever, has an immediate effect on topography and the elevation
change must be taken into account during the simulation.
242 GIS-Based Soil Erosion Modeling
(a)
120 0 120 300 m
(b)
Figure 10 Temporal evolution of a gully, modeled in GRASS GIS: (a) elevation surface with evolving gully, color indicates spatial pattern of
erosion/deposition rates in (kg m  2 s  1), (b) evolving elevation surface: creation of multiple knickpoints, incision of the channel, upslope
headwall migration and channel widening, color represents the elevation values in (m).
This inherently requires a dynamical erosion model reflecting
the changes in topography. To simulate the gully formation,
it is assumed that erosion starts before water flow reaches
steady state, the sediment transport is close to detachment
limited conditions, and all eroded material is transported
outside the gully. Using eqn [1] to estimate erosion rates at
time t(s) driven by the unit water flow qt (at nonsteady state
for the time steps t0, t1,y, ts, where ts is time of concentration)
the change in elevation Dz(x, y, t) (m) due to net erosion can
be estimated as
Dzðx,y,tÞ ¼ Dt  ds ðx,y,tÞ=rs ;
where rs(kg m3) is soil density and Dt ¼ ti  ti1 is the time
interval. Using the map algebra in a GIS, the initial DEM0 is
modified using the changes in elevations Dz(x, y, t). A new set
of topographic parameters is then derived from the updated
DEMi to reflect new flow gradients used by the water flow and
soil erosion model in the next iteration. Figure 10 shows a
time-series of a developing gully modeled by the SIMWE
model (Mitasova et al., 2005b) in a GRASS GIS environment
(Koco, 2009). It illustrates creation of multiple knickpoints,
incision of the channel and then an upslope headwall
migration and channel widening.
3.9.4.6 Statistical Modeling
As previously shown, even the physics-based erosion models
require a set of empirical parameters that are derived from
field or laboratory experiments using statistical methods.
Statistical modeling can also be used to derive empirical
models or indices that relate landscape surface attributes to
erosion rates. The resulting models require field or remotely
–0.4
–0.2
0
0.2
0.4
350
320
190 m
sensed data and are applicable only to locations with con-
ditions similar to those for which they have been developed.
For example, Pike et al. (2009) used logistic regression and
neural networks to derive a probabilistic model that predicts
the occurrence of channel erosion using the index Pe:
Pe ¼ ð1 þ exp½3:63þ1:11LSþ0:217w12:1kh  Þ1 , ½30
where LS is the estimated length-slope factor, w is the topo-
graphic wetness index, and kh is the plan curvature (tangential
curvature projected to horizontal plane). Because this is an
empirical model, the authors make it clear that the model
½29
should only be applied if the same methods were used to
create the DEMs and calculate terrain attributes as in their
application.
An alternative statistical approach explored the capabilities
of three tree-based regression models to explain gully erosion
field measurements using topographic parameters. The in-
vestigated parameters included elevation, upslope contrib-
uting area, aspect, slope, plan, profile, and tangential
curvature; flow direction, flow width, flow path length, rate of
change of specific catchment area along the direction of flow,
steady-state, quasidynamic topographic wetness; and sediment
transport capacity (Kheir et al., 2007). The best regression tree
model combined the steady-state topographic wetness and
sediment transport capacity indices, which explained 80% of
the variability in field gully measurements.
3.9.4.7 Landscape Evolution Modeling
Most GIS-based erosion models focus on mapping the spatial
pattern of erosion and deposition without taking into account
changes in elevation surface. Landscape evolution models
simulate erosion processes including their impact on elevation

surface and their feedback on erosion and deposition patterns.
Landscape evolution process can be described by partial dif-
ferential equations (Willgoose, 2004) and most models are
implemented outside a GIS using customized code or numer-
ical modeling environment such as MATLAB. An exception is a
landscape evolution model that simulates impact of land use
and climate change on landscape evolution r.landscape.evol that
has been fully integrated with GRASS GIS (Barton et al., 2010a).
Simulation of landscape evolution over long periods of
time requires integration of several erosion processes. These
include change in elevation surface due to gravitational hill-
slope erosion modeled using diffusion equation, processes
induced by surface water flow, including hillslope erosion and
deposition, gully erosion, channel incision, and meandering
in regions with low topography.
An approach inverse to the forward landscape evolution
modeling was used to reconstruct ancient topography (Peeters
et al., 2006) by computing soil erosion and deposition based
on change in sediment flow rates estimated from slope and
unit contributing area. Then, the topography was iteratively
updated back in time by adding the elevation associated with
erosion rates and subtracting elevation in areas with de-
position. Comparison with field measurements of historical
soil erosion and sediment deposition volumes shows that it is
possible to simulate realistic soil redistribution patterns. Fur-
ther research is necessary, however, to address simulation ar-
tifacts produced by the routing and elevation change
computation used in this study (Peeters et al., 2006).
Several papers suggest the importance of large events in
landscape evolution (Peeters et al., 2006; Mitasova et al.,
1999) when long periods of relatively small events, dominated
by transport capacity limited erosion and deposition, are
disrupted by rare large events that cause flushing of deposits
and transport of sediment over large distances. These large
events are generally associated with gully and channel incision
leading to significant changes in elevation surface.
3.9.5 GIS Implementation
Implementation of the simple erosion models in GIS is rela-
tively straightforward and involves creating a workflow that
includes input data processing, model computation, and an-
alysis of results. The workflows can be developed using model
builders available in GIS and saved in the form of scripts that
support fully automated simulations. Complex physics-based
models usually require partial differential equation solvers
and can be fully integrated with GIS using the system’s lib-
raries for managing the geospatial data. More often, such
models are only loosely linked to GIS.
3.9.5.1 Coupling GIS and Models
Erosion modeling can be coupled with GIS through data ex-
change, graphical user interface (GUI), or it can be fully inte-
grated as set of modules or scripts (Mitasova and Mitas, 2002).
‘Loose coupling’ links a model with GIS through import
and export of data. The model is developed independently
from a GIS and different GIS software packages can be used for
GIS-Based Soil Erosion Modeling 243
data preprocessing, analysis, and visualization of modeling
results. Standardization in georeferenced data formats and
improvements in software interoperability have made loose
coupling a routine procedure. Several landscape evolution and
erosion models are available with this type of GIS coupling
(CSMDS, 2011).
‘Tight coupling’ integrates a model and a GIS using shared
GUI which guides the user through input data processing,
modeling, and analysis. The interface also allows the user to
visualize the results using both the GIS display tools and
specialized graphical and numerical outputs. This type of in-
tegration has proven to be effective for hydrologic and non-
point source pollution modeling systems (Rewerts and Engel,
1991; Renschler, 2003; Di Luzio et al., 2002) and several
models have been coupled with more than one GIS. Increas-
ingly, the interface is provided through on-line tools via the
Web Processing Service (WPS).
‘Full integration’ involves spatial models which are useful
for a wide range of GIS applications. These models are de-
veloped and implemented within a GIS, using its program-
ming tools such as Application Programming Interface (API),
scripting language, or map-algebra operations. The spatial
model is then run as a GIS function or command, with the
inputs and outputs stored in a GIS database (no data transfer
is needed). Portability of the model is restricted, and the en-
hancements as well as the maintenance of the model is
dependent on the GIS. This type of model development is
further supported by customization and application of de-
velopment tools, extensions to map algebra (Wesseling et al.,
1996) and visual modeling tools.
Full GIS integration of complex models involving solutions
of coupled partial differential equations has been limited, in
spite of several successful implementations (Mitas and Mita-
sova, 1998). Even with all the necessary capabilities available
in the GIS, the biggest disadvantage of full integration of
complex models is that the models become too dependent on
the development and fate of a particular GIS. Changes in the
GIS data structures, functionality, interface, libraries or pro-
gramming tools, may require time consuming changes in the
models or the models become incompatible with the latest
version of the GIS software. Also, the fully integrated model is
less portable and users have to install entire GIS even if they
need the model only for a one-time application. Some of these
issues have been addressed by open source software and
community systems development.
Large, professional modeling systems, most commonly
aimed at engineering applications, use both loose coupling
with a GIS and their own, specialized GIS capabilities. An
external GIS is generally used for storing, managing, and
processing of basic topographic data and for generating the
cartographic output. The modeling system itself includes
support for GIS functions where tight coupling with the model
is necessary, such as the design of a conceptual model for the
given site, adjustment of finite-element grids and meshes, as
well as modifications of the model parameters (conditions of
simulations) based on the simulation results.
With the explosive growth of the Internet, erosion models
useful for a wider range of users, such as farmers, land owners,
city planners, or public land managers are being implemented
as web-based applications. The successful applications include
244 GIS-Based Soil Erosion Modeling

not only the modeling tools, but also the databases with input model calibration and validation. Erosion and sediment
data and model parameters so that the user does not have to transport modeling involves substantially more uncertainty
deal with the time consuming tasks of finding, processing, and than modeling in disciplines where most experiments
submitting the input data for the model runs. Generally, only are done in laboratory under controlled conditions. Model
selection of the location and land-use management scenario is calibration and validation are, therefore, an important com-
needed from the given set of options. Spatial data are stored in a ponent of the modeling effort and are generally handled by
GIS on the server and the digital maps or animations, repre- external statistical tools loosely coupled with models and GIS
senting the inputs and model results, are served using the on- (Poeter et al., 2005). Calibration of erosion models and val-
line map serving technology (Park et al., 2009; Luo et al., 2004). idation of the modeling results has been commonly limited by
difficulties of data collection. Recent advances in temporal
acquisition of high-resolution airborne or terrestrial LIDAR
3.9.5.2 Derived Model Parameters data provide new opportunities to test the accuracy of pre-
dicted patterns of erosion and deposition based on DEM
Finding, importing, integrating, and processing the data ne-
differencing. This, in turn, can be used to evaluate the sensi-
cessary for erosion modeling can be a time-consuming task.
tivity of model parameters for obtaining accurate erosion
Therefore, a well maintained, consistent GIS database makes
magnitudes and distributions.
data preparation for modeling and generating alternative
In addition to summary statistics, maps representing spa-
scenarios highly efficient. The following input data are gen-
tial distribution of soil detachment, net erosion and de-
erally required for GIS-based erosion modeling.
position, and sediment transport rate are among the most
Elevation data are widely available as raster DEMs. Accurate
important outputs of erosion modeling that communicate the
flow routing and slope estimation requires resolutions with
spatial pattern of sediment sources, highlight high erosion risk
grid size of 10 m and smaller and vertical precision at least in
areas and locations with damaging rates of deposition. Per-
centimeters. If a raster DEM is not available, contours, or point
spective views of erosion modeling results draped over DEMs
data, such as LiDAR point clouds, can be interpolated to create
are especially effective in highlighting the relationship be-
the DEM (see Chapter 3.6). Elevation data are used to compute
tween landforms and the spatial pattern of erosion and de-
slope, direction of flow, and upslope contributing area for
position (Figures 3 and 8). Although the perspective 3D views
physics-based models as well as numerous other parameters
of topography do not replace an accurate 2D map, they are
such as curvatures or wetness indices for statistical models (see
useful in interactive mode for visual analysis of complex
Chapter 3.7).
spatial relations. As static images, they can be used to improve
Land cover data are generally derived from aerial or satel-
the perception of 3D features and their impact on erosion that
lite imagery as a raster map layer (see Chapter 3.4). If raster
may be difficult to capture by 2D maps. Animations are
data are not available, polygon areas can be transformed to
commonly used to represent the process dynamics and this
raster at desired resolution. Land cover data are used to esti-
can include animated 2D images as well as dynamic surfaces
mate the land-cover C-factor for USLE and for surface rough-
in perspective 3D views (Mitas et al., 1997).
ness, detachment capacity and sediment transport capacity
Recent advances in human–computer interaction technol-
coefficients in process-based models. These parameters have
ogy provide environments for enhanced, collaborative ex-
empirically derived values and are commonly available in
ploration of landscape models. In addition to multitouch
literature and reference tables (Haan et al., 1994).
screens and immersive 3D environments, interaction with 3D
Soil data are generally available as polygons and have to be
objects is emerging as a powerful tool for investigation of
transformed to raster or hydrologic units using standard GIS
landscape properties and processes. The Tangible Geospatial
data model transformations. If soil samples are available,
Modeling System (TanGeoMS) couples a 3D laboratory laser
spatial interpolation using geostatistics methods is applied to
scanner and a physical landscape model with a video projector
compute the raster representation of soil properties that are
and GIS to support exploration of terrain change impacts on
the basis for computation of erosion related factors. Soil data
topographic parameters and land surface processes (Mitasova
are used to derive the soil erodibility K-factor for USLE and
et al., 2006; Tateosian et al., 2010). The flexible clay land-
detachment capacity and sediment transport capacity co-
scape model can be modified by multiple users. When a
efficients in process-based models.
modified landscape is scanned, the selected terrain parameter
Rainfall data are available in the form of isoline maps for
or process simulation is recomputed based on the scanned
R-factor or from databases used for RUSLE. For field and small
data and the result is projected over the modified surface
watershed scales, a single value is sufficient; for regional
(Figure 11). The projected image or animation provides the
modeling the rainfall factor should be given as raster map and
users with feedback on impact of their terrain modifications
can be obtained using Tropical Rainfall Mapping Mission data
and guides the future exploration.
(see Chapter 3.1). Rainfall factors can also be derived from
measured rainfall data (Haan et al., 1994).

3.9.6 Case Studies

3.9.5.3 Analysis and Visualization
A GIS provides powerful tools for analysis and communi- Two case studies were selected to illustrate the GIS-based
cation of modeling results, and it is commonly combined with erosion modeling and highlight its focus on spatial patterns of
advanced statistical packages and specialized tools to perform erosion and deposition.
 GIS-Based Soil Erosion Modeling 245

90
S P
20

0°

(a) (b) (c)

Figure 11 Tangible Geospatial Modeling System: (a) flexible landscape model M is scanned by an overhead 3D laser scanner S while GIS data
(e.g., orthphoto) and simulation results are projected over the model using a projector P; (b) several users can modify the landscape model by
adding buildings or creating road tracks; (c) perspective views of initial and modified landscape models; slope maps draped over the DEMs
provide feedback on the modification impact on slope angle values.

Grass
Dirt path
North Carolina
Wetland
Crop field
USA Vineyard
Location
of study site Dirt road
Atla

Paved road
ntic

N Bare soil
Oce

Buildings
an

Forest
Pond
100 m Watershed boundary
(a) (b) (c)

Figure 12 Case study site: (a) location in the central North Carolina, (b) aerial photo draped over DEM with watershed boundaries and two
monitoring sites, and (c) simplified land cover used in simulations.

3.9.6.1 North Carolina Piedmont
A small watershed at the North Carolina State University
Sediment and Erosion Control Research and Education Facility
(SECREF) (McLaughlin et al., 2001) has been used for erosion
model testing and sediment transport monitoring (Figures 12
and 13). This site is used to demonstrate advantages and
limitations of various approaches to GIS-based erosion mod-
eling. The 9.4 ha watershed has a mean slope of 3.41 and a
maximum flow-path length of 530 m. The soil type is mostly
Cecil, and land use includes agricultural fields, grass, small
forested area, vineyard, buildings, and experimental ponds.
Disturbances are limited to tilling at a 6 ha agricultural field.
The area was mapped by airborne LiDAR in the year 2001
with point density of 1 point for each 2 m resolution grid cell.
Bare-ground points were used to compute a 1 m resolution
DEM, a slope, and a direction of flow maps using regularized
smoothing spline with tension (RST) method (Mitasova et al.,
2005a). Several approaches were tested for mapping of flow
accumulation and estimation of unit water flow, including
single direction flow routing with 8 directions (SDR-D8 im-
plemented in the GRASS module r.watershed), single direction
flow routing with infinite number of directions (SDR-Dinf,
module r.flow), and multiple direction flow routing with 8
directions (MDR-D8 module r.watershed and r.terraflow).
Additional parameters for the erosion models were derived
from land-cover, soil and rainfall data stored in a GIS and in
the related WEPP databases.
Baseline erosion modeling was performed for simplified
land cover (Figure 12(c)) to obtain estimates of soil erosion
and deposition using the following models:
• on-line version of GeoWEPP to estimate runoff and sedi-
ment yield;
• GIS based version of detachment limited erosion model to
estimate soil detachment;
• GIS based erosion/deposition model to estimate net erosion
and deposition rates;
• overland water flow and sediment transport model SIMWE
to estimate runoff and sediment flow rates;
246 GIS-Based Soil Erosion Modeling

>0.1 m

0.01

0.001

1.0

0.5

N
0.01 kg m–1s–1 N 300 m
(a) (b)

Figure 13 Study site after high intensity storms: (a) runoff with high concentration of sediment, (b) flooding in depressions and over a service
road. GIS-based simulation of sediment flow rate (a) and overland flow depth (b) using a LiDAR-based DEM and the SIMWE model implemented
in GRASS GIS reflects the observed water and sediment flow patterns. Photo courtesy R. McLaughlin, North Carolina State University.

• landscape evolution model r.landevol to simulate change in
elevation.
The GeoWEPP on-line tool was the easiest to use because
the GIS data were provided along with the model interface.
The results were limited by resolution of the available data
(in this case 30 m) and only annual average soil loss and
sediment yield were computed (Figure 14). Using the default
parameters for central North Carolina, total annual runoff
from the watershed was estimated as 2372 m3 from hillslopes,
and sediment yield as 4.9 ton yr1 with erosion rate of 0.5 t
(ha yr)1. Deposition was predicted in less than 10% of the
area, most likely due to the low resolution of the DEM that did
not adequately represent shallow concave landforms.
Annual soil loss and net erosion/deposition was also esti-
mated using the detachment limited and transport capacity
limited models combined with the USLE rainfall, soil, and
land-cover parameters (R ¼ 220 and soil erodibility factors
ranging between K¼ 0.15–0.28 were used). Both models were
run in GRASS GIS using map algebra applied to 1 m reso-
lution raster data. The results depicted the spatial pattern of
soil detachment (Figure 15(a)) and net erosion and de-
position (Figure 15(b)). The results show significant potential
for large rills to develop in the agricultural fields and de-
position to occur in the lower section of the valley. Subtle
terraces with alternating erosion and deposition in the west-
ern, agricultural areas required closer inspection to ensure that
they are not artifacts of point cloud measurements or
interpolation. The fact that the patterns do not follow the
data sampling distribution indicates that the terraces are not
data-processing artifacts. The results from terrestrial LiDAR
surveys confirm that the subtle terracing is due to the con-
figuration of agricultural fields and direction of tillage.
The detachment capacity limited model (eqn [21]) pre-
dicted average annual soil detachment rate of 3.95 t (ha yr)1
for exponent values m¼ 1.1 and n ¼ 0.6. These estimates are in
line with the USLE-based Natural Resources Conservation
Service (NRCS) erosion rates published for North Carolina,
but they are much higher than the values predicted by the
GeoWEPP model and those actually measured at the observed
values at the watershed outlet. This is due to the fact that the
detachment capacity limited models represent soil loss with-
out considering sediment deposition. The estimates from the
transport capacity limited model indicate that most of the
eroded soil can be deposited within the watershed (i.e., 6.3 ha
has a potential for net erosion and deposition will occur over
3.1 ha). The predicted mean deposition rates are higher than
mean erosion rates, compensating for the smaller area with
potential for deposition. The mean net erosion rate was esti-
mated as 1.4 t (ha yr)1 and mean net deposition rate was
3.9 t (ha yr)1 when m ¼ 1 and n ¼ 1 were used. The net soil
loss (total erosion minus total deposition) was estimated at
0.48 t (yr1) assuming vegetation cover, indicating that rela-
tively small amount of sediment will leave the watershed
under transport capacity limited conditions.
The SIMWE model was used for single-storm simulations
to estimate runoff and sediment transport. The model cor-
rectly predicted ponding in the depression created by the road
(Figure 13). It also provides more realistic pattern of erosion/
deposition along the boundaries between the tilled field and
 GIS-Based Soil Erosion Modeling 247

Figure 14 Simulation of hillslope erosion and deposition at 30 m resolution using the Web interface of the GeoWEPP model.

t ha–1 yr–1 t ha–1 yr–1

>30 Depo- >30
sition
15 0
Erosion
<–30
0

(a)
100 m (b)

Figure 15 Spatial distribution of (a) soil detachment rate estimated from eqn [21] that includes prediction of high erosion rates due to
concentrated water flow; (b) net erosion and deposition rates indicating formation of gullies and subtle terraces predicted from eqn [23]. Higher
rates are predicted in areas with crops, dirt roads, and bare soil (see Figure 12).

vegetated strip, by allowing gradual deposition over longer input, additional channel cross-section geometries, conversion
flow path, rather than instant deposition, as is the case with of outputs into commonly used units and measures, and
the steady-state models. However, the model requires more validation with spatially and temporally distributed field data
development and research to support rainfall time-series that are only now becoming available from terrestrial laser
248 GIS-Based Soil Erosion Modeling
scanning technology. It is also computationally much more
demanding than the steady-state models.
Although only small changes in elevation are observed
during the years with average precipitation, severe storms have
occasionally caused major erosion and sediment transport in
this area, altering the topography. The landscape evolution
model was used to simulate the change in elevation over time
due to erosion and deposition by overland flow and diffusive
processes between the rainfall events. It was applied to vari-
able land-cover conditions (Figure 16) with 40% area in
agricultural fields. The results show potential for developing
large rills in the fields with subsequent deposition at the edge
of the fields and in concave terrain. Under bare ground con-
ditions, the formation of a large gully is predicted along the
valley due to concentrated water flow. The simulated erosion
and deposition pattern is in general agreement with obser-
vations. As in the case of the SIMWE model, repeat terrestrial-
laser surveys will provide opportunity to calibrate the model
and evaluate its capability to predict the magnitudes and
patterns of erosion and deposition.
TanGeoMS was used to explore the impact of various
land-design alternatives on surface runoff and soil erosion
(Figure 17). GIS layers, such as orthophotography, footprints
of structures, and flow-accumulation maps were projected
over the 1:1200 scale flexible clay terrain model to guide the
model modifications. Then various sediment control alter-
natives were created by modifying the plasticine surface by
hand. The modified model was scanned, runoff and erosion
130 m
Deposition
120
Erosion
110
(a) 100 m
Deposition
Erosion
(b)
Figure 16 Landscape evolution under different conditions: (a) during small rainfall events erosion rates are low and the valley is filling with
deposits, (b) during high intensity rainfall gullies form in locations with concentrated flow. Perspective views of resulting DEMs with elevation
color maps are on the left, erosion/deposition pattern is on the right.
were computed in GRASS GIS and resulting animations of
water and sediment flow were projected back onto the model
to provide rapid feedback on the effects of the proposed
change and to guide the next modification (Figure 17).
3.9.6.2 Mediterranean Landscape Evolution
The goal of Mediterranean Landscape Dynamics (MedLanD)
project was to build a sophisticated modeling laboratory to
carry out virtual experiments on the long-term, recursive
interactions between society, land-use, and environmental
change. Because it aims to simulate high-resolution, real-
world landscape dynamics and land-use practices, the
MedLanD Modeling Laboratory (MML) is a hybrid modeling
environment that tightly couples different modeling ap-
proaches (Mayer and Sarjoughian, 2008; Mayer, 2009; Mayer
et al., 2006). These include: (1) a GIS-based landscape evo-
lution model that simulates hydrology, erosion/deposition,
and vegetation succession; (2) stochastic GIS-based models
and dynamic agent-based models (ABM) of farming house-
holds and their land-use practices; and (3) regression-based
paleoclimate and paleovegetation models (Figure 18). Details
of the MML are published elsewhere (Barton et al., 2010a, b;
Ullah, 2011; Ullah and Bergin, 2011) and only summarized
here with emphasis on the landscape evolution modeling
component.
In the MML, agropastoral land-use can be modeled sto-
chastically in a GIS or as rule-based actions of independent
0.3 m
0
−0.3
1m
0
−1
 GIS-Based Soil Erosion Modeling 249

130 m >0.1 m 30
120 0.01 15
110 0.001 0

(a)

0 300 m
N

(b)

Figure 17 Exploring terrain modification impact on water flow and sediment transport: (a) DEM, water flow depth and topographic erosion
potential for the initial model, (b) adding buildings and bio-swales to the model, resulting water flow depth, and topographic erosion potential for
the modified model. Model design by B. Harmon (Tateosian et al. 2010).

Vegetation
model
(regression)

Communities Agropastoral
& land use socioecology
model (ABM) model

Landscape
Climate model evolution
(regression) model
(GIS/CA)

Figure 18 Schematic of the component structure and model couplings for the MedLand Modeling Laboratory.

computer agents in an ABM platform. When modeled sto-
chastically, farming and grazing patches are randomly dis-
tributed within catchments, calculated using GIS routines to
account for terrain and suitability for farming or herding,
around communities (Barton et al., 2010a; Ullah, 2011). Al-
ternatively, individual house-holds can be simulated as virtual
agents, organized into villages. Agents choose land to farm
or graze on the basis of their need for farming returns (calories
that affect birth and death rate), the suitability of land
for particular agricultural activities, and costs to use the
land, including access on foot and clearance of vegetation
(Barton et al., 2010a; Mayer, 2009; Mayer et al., 2006; Ullah and
Bergin, 2011). Agents can also collect fuel wood, the amount of
which varies according to household size and activities.
Whether modeled stochastically or as agent behavior, house-
hold land-use can alter the vegetation cover and soil charac-
teristics. These anthropogenic changes to landscapes in turn
impact the results of other surface processes on landscapes
location and intensity of erosion and deposition. These
are simulated in the MML with a landscape evolution model
implemented in GRASS GIS as a Python-based script
r.landscape.evol.py developed by the MedLanD project team.
250 GIS-Based Soil Erosion Modeling
The model iteratively calculates the amount of erosion
and deposition that occurs across the landscape over time.
Because each of the process equations described perform
better under different topographic conditions, the script uses
different process equations for different landforms, and im-
plements them in a manner that optimizes the ratio of
model run-time to accuracy of erosion/deposition calcula-
tions. The module implements a diffusion equation for areas
near drainage divides, a three-dimensional transport capacity
limited method for hillslopes and gully heads (eqn [22]), and
an equation based on the reach-average shear stress for
channels (eqn [10]). It is important to choose the optimal
locations on the terrain for the transition between surface
process models to ensure that the most appropriate process
equation is used for each cell of the DEM. Although these
transition points can vary with overall watershed geometry,
area, and topographic relief, and also can change during a
hydrologic event (e.g., during a storm) as a function of
rainfall intensity and duration, they can be estimated in a GIS
on the basis of upslope contributing area and topographic
profile curvature.
The model employs the advanced flow accumulation and
basin analysis module r.watershed to calculate upslope con-
tributing area using a multiple flow direction (MFD) algo-
rithm that produces much smoother and more realistic
patterns of flow convergence and divergence than does a
simpler single flow direction (SFD) D-8 algorithm. Upslope
contributing area is then plotted against the profile curvature
for each cell (Figure 19). Positive values of profile curvature
indicate a marked increase in slope (a convex profile), negative
values indicate a marked decrease in slope (a concave profile),
whereas values close to zero represent cells where there is little
change in slope. Drainage divides have little accumulation and
100 000
10 000
Upslope accumulated area (A)
1000
100
10
1
–0.004 –0.002 0 0.002
(a) Profile curvature (pc)
Figure 19 Plot of upslope contributing area vs. profile curvature (pc). Transition point 1 indicates the value of contributing area where the
transition from diffusion to sheetwash occurs, and transition point 2 indicates the value of contributing area where transition from sheet wash to
rilling/gullying occurs; (b) a plot of contributing area versus tangential curvature (tc). Transition point 3 indicates the value of contributing area
where the transition from rilling/gullying to channel flow occurs.
little change of slope, plotting near 0 on the x and y axes;
hillslopes also have little change in slope, but have higher
accumulation, plotting near 0 on the x-axis but higher on the
y-axis. The transition between the drainage divide and hill-
slopes has the maximum convex profile curvature and rela-
tively low values of upslope contributing area, whereas the
transition between hillslopes and gully heads at the base of
slopes has concave profile curvature and higher contributing
area values. Hence, for a given landscape and hydrologic re-
gime, the authors use the value of contributing area for the
maximum value of profile curvature for the transition from
the diffusion equation to transport capacity limited erosion by
overland flow with exponents m, n for sheet wash processes.
The value of contributing area the minimum value of profile
curvature is used for the transition from sheet wash to trans-
port capacity limited erosion with higher exponents m, n for
rill/gully flow (Figure 19(a)).
In a similar way, contributing area can be plotted against
the tangential curvature of each cell to identify the accumu-
lation value for the transition from the transport capacity
limited erosion by overland flow to a shear-stress equation for
channelized flow. The beginning of channelized flow can be
identified as the location where very low negative values of
tangential curvature (concave) are associated with high values
of contributing area. Negative values of tangential curvature
that are associated with lower values of contributing area
represent the larger gullies and gully-heads (i.e., that occur
higher in the drainage network than the real stream channels),
and even higher values of contributing area where tangential
curvature has decreased indicate a widening channel carrying
more water (Figure 19(b)). Figure 20 shows the locations of
the transition points identified in Figure 19 on the DEM for
which they were derived.
Transition point 3
Transition point 2
Transition point 1
0.004 –0.004 –0.002 0 0.002 0.004
(b) Tangential curvature (tc)

Diffusion
Sheetwash
Rill/Gully
Channel
Figure 20 Map showing the location of the process transition points (1, 2, and 3) derived from Figure 19 draped over a DEM. Actual gullies
and channels will be located within the outlined respective areas.
The diffusion equation used by the MML is well-known
and simulates soil creep, the movement of soil downslope due
to the effect of gravity and particle movement from rainsplash,
bioturbation, and other local factors, on portions of the
landscape where there is not enough accumulated runoff for
overland flow (Tucker and Bradley, 2010):
qz
¼ Kg r2 z;
qt
where the rate of elevation change is controlled by the dif-
fusion coefficient Kg(m(1000 yr)1), for a soil density of 1,
which is an empirically derived constant for different climate
and vegetation regimes.
The transport capacity limited erosion eqns [22] and [23]
used in the MML hillslope model were derived from concepts
described by Kirkby (1971), adapted for complex topography
by Moore and Burch (1986), and operationalized in GRASS
GIS (Mitasova et al., 1996; Mitas and Mitasova, 1998). The
approach combines the USLE/RUSLE parameters, slope and
upslope contributing area per unit contour width to estimate
sediment flow at sediment transport capacity, and calculate net
erosion and deposition at the center of each grid cell. Imple-
mentation in a GRASS script combines the results of GIS
modules for calculating slope, aspect, and flow accumulation
using map algebra. Input data for the script include a raster
DEM of initial surface topography, soil erodibility (K-factor as
a constant for uniform soil or a raster map for variable soil),
vegetation cover (C-factor as a constant or raster map), and
rainfall intensity (R-factor as a constant only). An underlying
bedrock topography DEM is also created to provide a limit
the total depth of unconsolidated sediment that can be ero-
ded; when bedrock is reached, soil erodibility is set to near
zero. K-factor, C-factor, and R-factor values for RUSLE have
been calculated empirically for a variety of settings in the
Mediterranean (Boellstorff and Benito, 2005; Essa, 2004;
Hammad et al., 2004; Martinez-Casasnovas and Sanchez-
Bosch, 2000; Renard et al., 1997; Renard and Freimund,
1994).
For flow in channels, the MML employs a variation of
eqn [10], where the shear stress is approximated using eqn
[12], but unit stream power eqns [14] and [15] are also
GIS-Based Soil Erosion Modeling 251
considered. In these equations, transport capacity in channels
is largely a function of the local slope and the depth of flow.
However, depth of flow is difficult to estimate accurately be-
cause it is dependent on the local channel geometry and
changes over the course of a rainfall event. Therefore, MML
estimates flow depth in each cell from an idealized unit
hydrograph for that cell using the following equation:
½31
h ¼ 2rA=1:25t, ½32
where r (m) is the excess rainfall (precipitation minus infil-
tration) during a hydrologic event, and A(m2) is the upslope
contributing area. Then rA(m3) is the volume of water that
passed over the cell during the simulation interval and t(s) is
the number of hydrologic instants in the simulation interval.
The hydrologic instant is the time it takes water to cross one
cell of a raster DEM, which can be determined by multiplying
the average velocity of flowing water in the watershed (e.g., as
derived with Manning’s eqn [13]) by the cell resolution.
The MML then computes net erosion and deposition rates
as divergence in sediment transport capacity (eqn [22]). To
approximate the transition between the zones dominated by
different processes, sediment flow divergence is computed
independently for the zones in which each process is mod-
eled, and the results are then combined into integrated spatial
representation of erosion and deposition rates.
For the overland flow driven processes, the net erosion and
deposition rates are calculated as units of mass per area per
year. To model terrain evolution over time, these rates must be
converted to the depth of lost or gained sediment per cell
Dz(x, y, t) (m) for a given time period Dt based on soil density
(eqn [29]). Soil density is approximated using the method
outlined by Rawls (1983) combining the percentages of sand,
silt, clay, and organic matter, and estimated for Terra Rossa
soils (i.e., for Mediterranean landscapes) on the basis of em-
pirical studies by Onori et al. (2006). Similarly as in the gully
modeling, the estimated change in elevation is then added to
(for deposition) or subtracted from (for erosion) the initial
DEM0, to create a new DEM1 after a cycle of land-use and
terrain change. This process can be iterated repeatedly to
simulate decades to millennia of landscape evolution. In
addition to the terrain evolution component, other aspects of
252 GIS-Based Soil Erosion Modeling
the MML simulate vegetation regrowth and recovery of soil
fertility for fallowed patches of the landscape. Together, these
provide feedbacks to the land-use ABM, affecting subsequent
decisions by households about which land to farm and graze.
Initial results using the MML to study the dynamics of
ancient land-use and terrain evolution have been promising.
In one set of modeling experiments, the MML was para-
meterized with values derived from ancient farming com-
munities of northern Jordan (Figure 21) and used for
experiments on the consequences of varying land-use (inten-
sive vs. shifting cereal cultivation, and ovicaprine grazing vs.
no grazing) and community size on soil loss and vegetation
over the course of two and ten generations, 40 and 200 years
respectively (Barton et al., 2010a, b). Some of the modeling
results confirm widely held expectations about such land-use:
shifting cultivation can cause greater soil loss than intensive,
repeated cultivation of a few plots (e.g., with manuring); ex-
tensive ovicaprine grazing will cause more erosion than
France
Spain
1000 km
N
Figure 21 Location of Mediterranean landscapes used in this case study.
Figure 22 Forty years of cumulative hillslope erosion/deposition around modeled prehistoric farming village (location marked by blue star) in
the Wadi Ziqlab drainage of northern Jordan. Dark line marks maximum extent of ovicaprine grazing catchment; grey line marks maximum extent
of zone of shifting cultivation around village.
farming without associated grazing; larger communities with
more people farming and grazing more land will have a
greater impact on the landscape than smaller communities
(Figure 22). Other results were less intuitively obvious be-
cause of the complex interactions between land-use and
landscape dynamics. Notably, when community size is below
a threshold (whose value is determined by local environ-
mental conditions) the amount of soil loss can be sub-
stantially offset by soil accumulation (i.e., by the redeposition
of sediments eroded from other parts of a catchment), so that
the economic impacts of mixed agropastoral land-use can be
negligible or even beneficial. However, if communities pass
this threshold size, the consequences change qualitatively such
that soil loss greatly exceeds soil accumulation within a land-
use catchment. This imbalance continues over the long-term,
with the potential for leaving a catchment unsuitable for
farming. One mitigating strategy is, not surprisingly, to reduce
community size through emigration or fissioning. Another
Turkey
Jordan
Egypt
>3 m
0m
<–3 m
5 km

4
3 1
0 10 km
Figure 23 Map of the Rio Penaguila and Rio Serpis valleys showing the different locations and topographic settings of village for each set of
experiments: (1) in an alluvial plain (for easy access to land for farming and grazing), (2) in a canyon bottom (for seclusion), (3) at the base of a
cliff (for defensibility), and (4) on a topographic prominence (for maximum visibility).
less obvious solution discovered in these experiments is to
increase the area devoted to grazing relative to cultivation,
moving zones of soil loss into uncultivated uplands and
providing more sediment for redeposition in the areas around
farmed fields. Conservation measures, like terracing, could
also be instituted but would require some degree of social
reorganization to ensure the availability of sufficient labor for
terrace construction and long-term maintenance. This kind of
investment in landesque capital and intensification of land
use has often been accompanied by the growth of inequalities
in social power and prestige. Comparison with the known
archaeological record of this region indicates good agreement
between modeling results and both the long-term impacts of
Neolithic farming and the social responses to these impacts
(Kuijt and Goring-Morris, 2002; Legge and Harris, 1996;
Martin, 1999; Quintero et al., 2004; Rollefson and Kohler-
Rollefson, 1992; Rosen et al., 2008; Simmons, 2007; Twiss,
2007).
A second set of experiments involved studying the results
of situating a small farming village in different topographic
contexts in the Rio Penaguila and upper Rio Serpis Valleys of
eastern Spain (Figure 21), the location of one of the earliest
known farming communities in the Iberian Peninsula
(Bernabeu et al., 2003; Bernabeu and Kohler, 2005). In
four different experimental runs (Figure 23): a simulated vil-
lage, populated by household agents, was set alternatively in
an alluvial plain (for easy access to land for farming and
grazing); in a canyon bottom (for seclusion); at the base of a
cliff (for defensibility); and on a topographic prominence (for
maximum visibility).
In each locale, all other initializing parameters besides
geographic setting were kept the same for the village. The
agents farmed and grazed the land around each site for
100 years and resulting data were collected on population
size, economy, vegetation cover, and erosion/deposition
(Figure 24). When the village was located in the alluvial plain
GIS-Based Soil Erosion Modeling 253
2
1400
1200
1000
800
N 600
400 m
it was more successful initially than when placed in the other
settings, as measured in terms of population growth and
agricultural returns. However, this success also led to a recur-
sive, self-amplifying growth cycle of increasing population,
expanding cultivation and grazing, soil degradation and loss,
and even more expansion of cultivation and grazing. When
situated in the other locales, the same village grew more
slowly and experienced more variable economic returns.
However, the smaller and more stable population also had
much less detrimental impact on the surrounding landscape
(Ullah and Bergin, 2011).
3.9.7 Conclusion and Future Directions
This chapter has presented the underlying theory and methods
for GIS-based modeling of soil erosion, sediment transport,
and deposition by surface water flow. Although the dynamic
models of sediment transport are complex, the topographic
controls for most approaches can be approximated using a
power function of upslope contributing area and slope. These
topographic parameters can be readily extracted from DEMs
using GIS tools and they constitute the foundation of GIS-
based erosion modeling.
Multitemporal elevation data are becoming available at
ever increasing accuracies and at levels of detail that can cap-
ture submeter features over large areas, thus providing sig-
nificant opportunities for improvements in modeling and
mapping of soil erosion and deposition. In spite of advances
in sensor technologies, spatial representations of soil prop-
erties, land cover, and rainfall excess remain weak components
of erosion modeling. Uncertainty in these critical input par-
ameters impacts the accuracy of process-based erosion models
and the practical solutions require combination of simplified
empirical and physics-based models. These models can be
calibrated to provide acceptable quantitative summary
254 GIS-Based Soil Erosion Modeling

1 2

3 4

N 10 km

Bare Shrubs Open woodland Woodland

Figure 24 Landcover maps for each modeling experiment with village locations indicated in Figure 23.

estimates, such as total volumes or spatially averaged erosion
rates, as well as depict the spatial patterns of erosion and de-
position. Nevertheless, the accuracy at any point in space and
time is often limited. The GIS implementation extends the
value and usefulness of these approaches by simulating spatial
patterns of erosion and deposition under different existing
and future conditions. Such information is critical for erosion
and sediment control and invaluable for development of
sustainable land-use management practices.
The case studies presented here highlight the use of
GIS-based erosion modeling to study the interactions and
feedbacks between human activities and erosion processes
at different spatial and temporal scales. The first study illus-
trated the differences in outputs from models at annual
and event-based temporal scales. It also introduced an inter-
active environment - TanGeoMS - for exploration of terrain
change impacts on landscape processes with applications for
runoff and sediment control design. The second study repre-
sents a unique application of erosion and landscape evolution
modeling to investigation of relationship between geo-
morphology and evolution of prehistoric agricultural com-
munities. The simulations highlighted the complexity of
feedback mechanisms between the location and size of com-
munities, landforms and their evolution. These examples
Arabi, M., Govindaraju, R., Hantush, M., Engel, B., 2006. Role of watershed
demonstrate the role of GIS-based erosion modeling in prac-
tical problem-solving and in research involving anthropogenic
forcing.
Further improvements in the quantitative prediction of
sediment transport in space and time may require probabil-
istic rather than deterministic modeling, as well as a wider use
of data assimilation techniques and model ensembles. Spatial
analysis and modeling, coupled with advances in laser scan-
ning and hyperspectral technologies provide unique oppor-
tunities to address the gaps in our understanding of erosion
processes and fundamentally improve both the theory, mod-
els, and their GIS implementation.
Acknowledgments
Research results presented in this paper were supported by the
US Army Research Office, grant W911NF-04-D-0003-0011,
National Science Foundation, grant BCS-410269 North Car-
olina Water Resources Research Institute. Assistance with data
and modeling at the SECREF location by Robert Austin and
Richard McLaughlin is greatly appreciated.
References
subdivision on modeling the effectiveness of best management practices with
SWAT1. JAWRA Journal of the American Water Resources Association 42(2),
513–528.

Barton, C., Ullah, I., Mitasova, H., 2010a. Computational modeling and neolithic
socioecological dynamics: a case study from southwest Asia. American Antiquity
75(2), 364–386.
Barton, C.M., Ullah, I.I., Bergin, S., 2010b. Land use, water and mediterranean
landscapes: modeling long-term dynamics of complex socio-ecological systems.
Philosophical Transactions of the Royal Society A: Mathematical, Physical and
Engineering Sciences 368(1931), 5275–5297.
Bernabeu, J., Kohler, T., 2005. Mas d’Is (Penaguila, alicante): un recinto
monumental del VI milenio cal AC. In: Arias, P., Ontanon, R., Garcia- Monco, C.
(Eds.), III Congreso sobre el Neolitico en la Peninsula Iberica (Santander, 2003).
Instituto Internacional de Investigaciones Prehistoricas de Cantabria, Santander,
pp. 485–495.
Bernabeu, J., Kohler, T., Diez, A., Gomez, M., 2003. Mas d’Is (Penaguila, alicante):
aldeas y recintos monumentales del neoltico antiguo en el valle del serpis.
Trabajos de Prehistoria 60(2), 39–59.
Boellstorff, D., Benito, G., 2005. Impacts of set-aside policy on the risk of soil
erosion in central spain. Agriculture, Ecosystems & Environment 107(2–3),
231–243.
Cebecauer, T., Hofierka, J., 2008. The consequences of land-cover changes on soil
erosion distribution in Slovakia. Geomorphology 98(3–4), 187–198.
CSMDS, 2011. Community surface dynamics modeling system.
/http://csdms.colorado.edu/S.
De Roo, A., Jetten, V., Wesseling, C., Ritsema, C., 1996. LISEM: a physically based
hydrological and soil erosion model incorporated in a GIS. IAHS Publication
235, 395–403.
DeRose, R., Gomez, B., Marden, M., Trustrum, N., 1998. Gully erosion in Mangatu
Forest, New Zealand, estimated from digital elevation models. Earth Surface
Processes and Landforms 23(11), 1045–1053.
Desmet, P., Govers, G., 1995. GIS-based simulation of erosion and deposition
patterns in an agricultural landscape: a comparison of model results with soil
map information. Catena 25(1–4), 389–401.
Desmet, P., Govers, G., 1996. A GIS procedure for automatically calculating the
USLE LS factor on topographically complex landscape units. Journal of Soil and
Water Conservation 51(5), 427–433.
Di Luzio, M., Srinivasan, R., Arnold, J., 2002. Integration of watershed tools and
SWAT model into BASINS. Journal of the American Water Resources
Association 38(4), 1127–1142.
Dingman, S., 2002. Physical Hydrology. Prentice Hall, Upper Saddle River, NJ. vol.
646.
Essa, S., 2004. GIS modeling of land degradation in Northern-Jordan using landsat
imagery. In: Proceedings of XXth ISPRS Congress. Istanbul.
Finlayson, D., Montgomery, D., 2003. Modeling large-scale fluvial erosion in
geographic information systems. Geomorphology 53(1–2), 147–164.
Flanagan, D., Gilley, J., Franti, T., 2007. Water erosion prediction project(WEPP):
development history, model capabilities, and future enhancements. Transactions
of the ASAE 50(5), 1603–1612.
Flanagan, D., Nearing, M., 1995. USDA-water erosion prediction project: technical
documentation. NSERL Rep 10.
Foster, G., Meyer, L., 1972. A closed form soil erosion equation for upland areas.
In: Shen (Ed.), Sedimentation: Symposium to Honor Prof. H.A. Einstein.
pp. 12.1–12.19.
Foster, G.R., 1982. Modeling the erosion processes. In: Haan, C.T., Johnson, H.D.,
Brakensiek, D.L. (Eds.), Hydrologic modeling of small watersheds. ASAE
Monograph No. 5, ASAE, St. Joseph, MI, pp. 197–380.
Gaffer, R., Flanagan, D., Denight, M., Engel, B., 2008. Geographical information
system erosion assessment at a military training site. Journal of Soil and Water
Conservation 63(1), 1.
Govers, G., Giménez, R., Van Oost, K., 2007. Rill erosion: exploring the relationship
between experiments, modeling and field observations. Earth- Science Reviews
84(3-4), 87–102.
Govindaraju, R., Kavvas, M., 1991. Modeling the erosion process over steep
slopes: approximate analytical solutions. Journal of Hydrology 127(1–4),
279–305.
Haan, C., Barfield, B., Hayes, J., 1994. Design Hydrology and Sedimentology for
Small Catchments. Academic Press, New York.
Hammad, A.A., Lundekvam, H., Barresen, T., 2004. Adaptation of RUSLE in the
eastern part of the mediterranean region. Environmental Management 34(6),
829–841. http://dx.doi.org/10.1007/s00267-003-0296-7.
Hancock, G., Willgoose, G., Evans, K., 2002. Testing of the SIBERIA land-scape
evolution model using the Tin Camp Creek, Northern Territory, Australia, field
catchment. Earth Surface Processes and Landforms 27(2), 125–143.
Harmon, R., Doe, W., 2001. Landscape Erosion and Evolution Modeling. Kluwer/
Plenum Publishers, New York.
GIS-Based Soil Erosion Modeling 255
Hong, S., Mostaghimi, S., 1997. Comparison of 1-D and 2-D modeling of overland
runoff and sediment transport. JAWRA Journal of the American Water Resources
Association 33(5), 1103–1116.
Jain, M., Das, D., 2010. Estimation of sediment yield and areas of soil erosion and
deposition for watershed prioritization using GIS and remote sensing. Water
Resources Management 24(10), 2091–2112.
Jetten, V., Govers, G., Hessel, R., 2003. Erosion models: quality of spatial
predictions. Hydrological Processes 17(5), 887–900.
Julien, P., Simons, D., 1985. Sediment transport capacity of overland flow.
Transactions of the ASAE-American Society of Agricultural Engineers (USA)
28(3), 755–762.
Karssenberg, D., Burrough, P., Sluiter, R., De Jong, K., 2001. The PCRaster
software and course materials for teaching numerical modeling in the
environmental sciences. Transactions in GIS 5(2), 99–110.
Kheir, R., Wilson, J., Deng, Y., 2007. Use of terrain variables for mapping gully
erosion susceptibility in Lebanon. Earth Surface Processes and Land-Forms
32(12), 1770–1782.
Kim, J., Park, Y., Yoo, D., et al., 2009. Development of a SWAT patch for better
estimation of sediment yield in steep sloping watersheds1. JAWRA Journal of
the American Water Resources Association 45(4), 963–972.
Kirkby, M.J., 1971. Hillslope process-response models based on the continuity
equation. Institute of British Geographers Special Pub 3, 15–30.
Koco, Š. 2009. The simulation of gully erosion by geographic information system.
Ph.D. thesis, University of Presov, Department of Geography and Regional
Development.
Kuijt, I., Goring-Morris, N., 2002. Foraging, farming, and social complexity in the
Pre-Pottery neolithic of the southern levant: a review and synthesis. Journal of
World Prehistory 16(4), 361–440.
Legge, T., Harris, D., 1996. The beginnings of caprine domestication in south-west
asia. The Origins and Spread of Agriculture and Pastoralism in Eurasia. UCL
Press, London, pp. 238–262.
Lim, K., Sagong, M., Engel, B., Tang, Z., Choi, J., Kim, K., 2005. GIS-based
sediment assessment tool. Catena 64(1), 61–80.
Luo, W., 2001. LANDSAP: a coupled surface and subsurface cellular automata
model for landform simulation. Computers and Geosciences 27, 363–367.
Luo, W., Duffin, K., Peronja, E., Stravers, J., Henryb, G., 2004. A web-based
interactive landform simulation. Computers & Geosciences 30, 215–220.
Martin, L., 1999. Mammal remains from the eastern jordanian neolithic, and the
nature of caprine herding in the steppe. Palorient 25(2), 87–104.
Martı́nez-Casasnovas, J., Ramos, M., Poesen, J., 2004. Assessment of sidewall
erosion in large gullies using multi-temporal DEMs and logistic regression
analysis. Geomorphology 58(1–4), 305–321.
Martinez-Casasnovas, J.A., Sanchez-Bosch, I., 2000. Impact assessment of changes
in land use/conservation practices on soil erosion in the Peneds-Anoia vineyard
region (NE spain). Soil and Tillage Research 57(1–2), 101–106.
Mayer, G.R., Jun. 2009. Composing hybrid discrete event system and cellular
automata models. Ph.D. dissertation, Arizona State University.
Mayer, G.R., Sarjoughian, H.S., 2008. A composable discrete-time cellular
automaton formalism. First International Workshop on Social Computing,
Behavioral Modeling, and Prediction. Springer, Phoenix, AZ, pp. 187–196.
Mayer, G.R., Sarjoughian, H.S., Allen, E.K., Falconer, S.E., Barton, C.M.,
2006. Simulation modeling for human community and agricultural landuse.
In: Agent-Directed Simulation, Proceedings of the Agent-Directed Simulation
Multi-Conference, Huntsville, Alabama. Huntsville, AL, pp. 65–72, April 2–4,
2006.
McLaughlin, R.A., Hunt, W.F., Rajbhandari, N., Ferrante, D.S., Sheffield, R.E., 2001.
The sediment and erosion control research and education facility at NC State
University. In: Ascough, II J.C., Flanagan, D.C. (Eds.), Soil Erosion Research for
the 21st Century. Am. Soc. Ag. Eng., St. Josepth, MI, 2001.
Mitas, L., Brown, W., Mitasova, H., 1997. Role of dynamic cartography in
simulations of landscape processes based on multi-variate fields. Computers
and Geoscience 23, 437–446.
Mitas, L., Mitasova, H., 1998. Distributed soil erosion simulation for effective
erosion prevention. Water Resources Research 34(3), 505–516.
Mitasova, H., Hofierka, J., Zlocha, M., Iverson, L., 1996. Modeling topographic
potential for erosion and deposition using GIS. International Journal of
Geographical Information Science 10(5), 629–641.
Mitasova, H., Mitas, L., 2002. Modeling physical systems. In: Clarke, K.C., Parks,
B.O., Crane, M.P. (Eds.), Geographic Information Systems and Environmental
Modeling. Prentice Hall, pp. 189–210.
Mitasova, H., Mitas, L., Brown, W., Johnston, D., 1999. Terrain modeling and soil
erosion simulations for Fort Hood and Fort Polk test areas. Research report,
University of Illinois, Urbana-Champaign, IL.
256 GIS-Based Soil Erosion Modeling
Mitasova, H., Mitas, L., Harmon, R., 2005a. Simultaneous spline interpolation and
topographic analysis for LiDAR elevation data: methods for open source GIS.
IEEE GRSL 2(4), 375–379.
Mitasova, H., Mitas, L., Ratti, C., Ishii, H., Alonso, J., Harmon, R., 2006. Real-time
landscape model interaction using a tangible geospatial modeling environment.
IEEE Computer Graphics and Applications 26(4), 55–63.
Mitasova, H., Thaxton, C., Hofierka, J., McLaughlin, R., Moore, A., Mitas, L.,
2005b. Path sampling method for modeling overland water flow, sediment
transport, and short term terrain evolution in Open Source GIS. In: Miller, C.,
Farthing, M., Gray, V., Pinder, G. (Eds.), Proceedings, XVth International
Conference on Computational Methods in Water Resources (CMWR XV).
Elsevier, pp. 1479–1490.
Mitchell, J., Engel, B., Srinivasan, R., Wang, S., 1993. Validation of AGNPS for
small watersheds using an integrated AGNPS/GIS system. JAWRA Journal of the
American Water Resources Association 29(5), 833–842.
Moore, A., McLaughlin, R., Mitasova, H., Line, D., 2007. Calibrating wepp model
parameters for erosion prediction on construction sites. Transactions of the
ASABE 50(2), 507–516.
Moore, I., Burch, G., 1986. Modeling erosion and deposition: topographic effects.
Transactions ASAE 29, 1624–1640.
Moore, I., Wilson, J., 1992. Length-slope factors for the revised universal soil loss
equation: simplified method of estimation. Journal of Soil and Water
Conservation 47, 423–428.
Morgan, R., Quinton, J., Smith, R., et al., 1998. The European Soil Erosion Model
(EUROSEM): a dynamic approach for predicting sediment transport from fields
and small catchments. Earth Surface Processes and Landforms 23(6), 527–544.
Murray, A., Paola, C., 1994. A cellular model of braided rivers. Nature 371, 54–57.
Nearing, M., Norton, L., Bulgakov, D., Larionov, G., West, L., Dontsova, K., 1997.
Hydraulics and erosion in eroding rills. Water Resources Research 33(4),
865–876.
Nearing, M., Simanton, J., Norton, L., Bulygin, S., Stone, J., 1999. Soil erosion by
surface water flow on a stony, semiarid hillslope. Earth Surface Processes and
Landforms 24(8), 677–686.
Neteler, M., Mitasova, H., 2008. Open Source GIS: A GRASS GIS Approach.
Springer Verlag, New York.
Onori, F., De Bonis, P., Grauso, S., 2006. Soil erosion prediction at the basin scale
using the revised universal soil loss equation (RUSLE) in a catchment of Sicily
(southern Italy). Environmental Geology 50(8), 1129–1140.
Pandey, A., Chowdary, V., Mal, B., 2009a. Sediment yield modeling of an
agricultural watershed using MUSLE, remote sensing and GIS. Paddy and Water
Environment 7(2), 105–113.
Pandey, A., Mathur, A., Mishra, S., Mal, B., 2009b. Soil erosion modeling of a
Himalayan watershed using RS and GIS. Environmental Earth Sciences 59(2),
399–410.
Park, Y., Kim, J., Engel, B., et al., 2009. Development of web-based SWAT System.
Proceedings of ASABE, Reno, Nevada, June 2009.
Peeters, I., Rommens, T., Verstraeten, G., Govers, G., Van Rompaey, A., Poesen, J.,
Van Oost, K., 2006. Reconstructing ancient topography through erosion
modeling. Geomorphology 78(3–4), 250–264.
Pelletier, J., Mitasova, H., Harmon, R., Overton, M., 2009. The effects of interdune
vegetation changes on eolian dune field evolution: a numerical-modeling case
study at Jockey’s Ridge, North Carolina, USA. Earth Surface Processes and
Landforms 34(9), 1245–1254.
Pike, A., Mueller, T., Schorgendorfer, A., Shearer, S., Karathanasis, A., 2009.
Erosion index derived from terrain attributes using logistic regression and neural
networks. Agronomy Journal 101(5), 1068–1079.
Poeter, E., Hill, M., Banta, E., Mehl, S., Christensen, S., 2005. UCODE 2005 and
six other computer codes for Universal Sensitivity Analysis, calibration, and
uncertainty evaluation, technical report, USGS.
Polyakov, V., Nearing, M., Shipitalo, M., 2004. Tracking sediment redistribution in a
small watershed: implications for agro-landscape evolution. Earth Surface
Processes and Landforms 29(10), 1275–1291.
Quintero, L.A., Wilke, P.J., Rollefson, G.O., Bienert, H., Gebel, H.G., Neef, R., 2004.
Highland towns and desert settlements: origins of nomadic pastoralism in the
jordanian neolithic. Central Settlements in Neolithic Jordan. ex oriente, Berlin,
pp. 201–213.
Rawls, W.J., 1983. Estimating soil bulk denisty from particle size analysis and
organic matter content. Soil Science 135(2), 123.
Renard, K., Foster, G., Yoder, D., McCool, D., 1994. RUSLE revisited: status,
questions, answers, and the future. Journal of Soil and Water Conservation
49(3), 213.
Renard, K.G., Foster, G.R., Weesies, G.A., McCool, D.K., Yoder, D.C., 1997.
Predicting soil erosion by water: a guide to conservation planning with the
revised universal soil loss equation (RUSLE). Agriculture Handbook. US
Department of Agriculture, Washington, DC, vol. 703, p. 1251.
Renard, K.G., Freimund, J.R., 1994. Using monthly precipitation data to estimate
the r-factor in the revised USLE. Journal of Hydrology 157(1–4), 287–306.
Renschler, C., 2003. Designing geo-spatial interfaces to scale process models: the
GeoWEPP approach. Hydrological Processes 17(5), 1005–1017.
Rewerts, C., Engel, B., 1991. ANSWERS on GRASS: integrating a watershed
simulation with a GIS. ASAE paper, pp. 91–2621.
Rollefson, G.O., Kohler-Rollefson, I., 1992. Early neolithic exploitation patterns in
the levant: cultural impact on the environment. Population and Environment 13,
243–254, NSF Biocomplexity 2002.
Rosen, S.A., Barnard, H., Wendrich, W., 2008. Desert pastoral nomadism in the
longue dure: a case study from the negev and the southern levantine deserts.
The Archaeology of Mobility: Old World and New World Nomadism. Cotsen
Institute of Archaeology. University Of California, Los Angeles, pp. 115–140.
Sheikh, V., van Loon, E., Hessel, R., Jetten, V., 2010. Sensitivity of LISEM predicted
catchment discharge to initial soil moisture content of soil profile. Journal of
Hydrology 393(3–4), 174–185.
Simmons, A.H., 2007. The Neolithic Revolution in the Near East: Transforming the
Human Landscape. University of Arizona Press, Tucson.
Šúri, M., Cebecauer, T., Hofierka, J., Fulajtár, E, 2002. Erosion assessment of
Slovakia at a regional scale using GIS. Ecology (Bratislava) 21(4), 404–422.
Tateosian, L., Mitasova, H., Harmon, B., Fogleman, B., Weaver, K., Harmon, R.,
2010. TanGeoMS: tangible geospatial modeling system. Visualization and
Computer Graphics, IEEE Transactions on 16(6), 1605–1612.
Tucker, G., Bradley, D., 2010. Trouble with diffusion: reassessing hillslope erosion
laws with a particle-based model. Journal of Geophysical Research 115,
F00A10.
Tucker, G., Lancaster, S., Gasparini, N., Bras, R., 2001a. The channel-hillslope
integrated landscape development model (CHILD). In: Harmon, R.S., Doe, W.W.
(Eds.), Landscape Erosion and Evolution Modeling. Kluwer/Plenum Publishers,
New York, pp. 349–388.
Tucker, G., Lancaster, S., Gasparini, N., Bras, R., Rybarczyk, S., 2001b. An object-
oriented framework for distributed hydrologic and geomorphic modeling using
triangulated irregular networks. Computers & Geosciences 27(8), 959–973.
Twiss, K.C., 2007. The neolithic of the southern levant. Evolutionary Anthropology
1, 24–35.
Ullah, I., 2011. A GIS method for assessing the zone of human-environmental
impact around archaeological sites: a test case from the late neolithic of wadi
ziqab, jordan. Journal of Archaeological Science 38(3), 623–632.
Ullah, I.I., Bergin, S., 2011. Modeling the consequences of village site location:
least cost path modeling in a coupled GIS and Agent-Based model of village
agropastoralism in eastern spain. In: White, D.A., Surface-Evans, S.L. (Eds.),
Least Cost Analysis of Social Landscapes: Archaeological Case Studies for
Beginners and Experts Alike. University of Utah Press, Salt Lake City.
Van Oost, K., Govers, G., Cerdan, O., et al., 2005. Spatially distributed data for
erosion model calibration and validation: The Ganspoel and Kinderveld datasets.
Catena 61(2–3), 105–121.
Wainwright, J., Parsons, A., Muller, E., Brazier, R., Powell, D., Fenti, B., 2008. A
transport-distance approach to scaling erosion rates: 3. Evaluating scaling
characteristics of Mahleran. Earth Surface Processes and Land-forms 33(7),
1113–1128.
Warren, S., Mitasova, H., Hohmann, M., Landsberger, S., Iskander, F., Ruzy-cki, T.,
Senseman, G., 2005. Validation of a 3-D enhancement of the Universal Soil
Loss Equation for prediction of soil erosion and sediment deposition. Catena
64(2–3), 281–296.
Wesseling, C.G., Karssenberg, D., Van Deursen, W.P.A., Burrough, P.A., 1996.
Integrating dynamic environmental models in GIS: the development of a dynamic
modelling language. Transactions in GIS 1, 40–48.
Willgoose, G., 2004. Mathematical modeling of whole landscape evolution. Annual
Review of Earth and Planetary Sciences 33(1), 443.
Willgoose, G., Bras, R., Rodriguez-Iturbe, I., 1991. A coupled channel network
growth and hillslope evolution model: 2. Nondimensionalization and
applications. Water Resources Research 27(7), 1685–1696.
Wischmeier, W., Smith, D., 1978. Predicting Rainfall Erosion Losses: A Guide to
Conservation Planning United States. Dept. of Agriculture. Agriculture Handbook,
Washington, DC.
Woodward, D., 1999. Method to predict cropland ephemeral gully erosion. Catena
37(3–4), 393–399.
 GIS-Based Soil Erosion Modeling 257

Biographical Sketch

Helena Mitasova is an Associate Professor at the Department of Marine, Earth, and Atmospheric Sciences, North
Carolina State University in Raleigh, NC, USA. She co-authored the first book on open source GRASS GIS and
published more than 50 papers on methods and applications of topographic̀ analysis, erosion modeling, coastal
dynamics, and visualization. Her PhD is from the Slovak Technical University, Bratislava, Slovakia.

C. Michael Barton is a Director of the Center for Social Dynamics and Complexity and a Professor of Anthro-
pology in the School of Human Evolution and Social Change at Arizona State University. Barton’s research
interests center around long-term human ecology and the dynamics of Quaternary landscapes. He heads an
international, multidisciplinary team studying the long-term socioecology of the beginnings of agriculture. He
also is active in the development and application of spatial technologies in archaeology and serving on the open
source GRASS GIS development team. He holds a PhD from the University of Arizona in anthropology and
geosciences.

Isaac Ullah is a PhD candidate in the School of Human Evolution and Social Change at Arizona State University,
Tempe, USA. He is a member of the Mediterranean Land-scape Dynamics project, and specializes in geological
applications in archaeology using GIS and modeling techniques. He studies the long term effects of human land
use decisions and how potential anthropogenic environmental degradation may have affected settlement patterns.
He holds an MA in Anthropology from the University of Toronto.

Jaroslav Hofierka is an Associate Professor at the Institute of Geography, Faculty of Sciences, Pavol Jozef Safarik
University, Kosice, Slovakia. His research activities include digital terrain modeling, geomorphometry, erosion risk
analysis as well as solar radiation and 3D city modeling. He has been active in open source GIS development and
has contributed several core modules to GRASS GIS. He received a PhD degree in Cartography and GIScience from
the Comenius University, Bratislava, Slovakia.
258 GIS-Based Soil Erosion Modeling

Russell S. Harmon is an Adjunct Associate Professor at North Carolina State University and he has been a Program
Manager at the Army Research Office, where he is responsible for the ARO extramural basic research program in
terrestrial sciences. He is a Fellow of the Army Research Laboratory, Geological Society of America, and National
Speleological Society and is current Past-President of the International Association of GeoChemistry. Dr. Harmon
is actively engaged in research in landscape dynamics and geospatial analysis, riverine geo-chemistry, and laser-
induced breakdown spectroscopy. He holds a MS from the Pennsylvania State University, and PhD from the Mc-
Master University.
 3.10 Remote Sensing and GIS for Natural Hazards Assessment and Disaster
Risk Management
CJ Van Westen, Faculty ITC, University of Twente, The Netherlands
r 2013 Elsevier Inc. All rights reserved.

3.10.1 Introduction 264

3.10.2 Background 266
3.10.2.1 Trends in Disaster Statistics 266
3.10.2.2 Disaster-Risk Management Framework 268
3.10.2.3 Risk-Analysis Framework 270
3.10.3 Hazard Assessment 271
3.10.3.1 Scale and Hazard Assessment 272
3.10.3.2 Spatial Data for Hazard Assessment 273
3.10.3.2.1 Hazard inventories and triggering events 273
3.10.3.2.2 Environmental factors 276
3.10.3.3 Hazard Assessment Examples 278
3.10.3.3.1 Global hazard assessment 278
3.10.3.3.2 (Inter)National hazard assessment 279
3.10.3.3.3 Provincial and municipal level 279
3.10.3.3.4 Community level 280
3.10.4 Elements-At-Risk and Vulnerability 281
3.10.4.1 Elements-At-Risk Information 281
3.10.4.1.1 Collaborative mapping and Mobile-GIS 281
3.10.4.1.2 Population data 281
3.10.4.1.3 Building data 283
3.10.4.2 Vulnerability 283
3.10.5 Multi-Hazard Risk Assessment 286
3.10.5.1 Qualitative Approaches 287
3.10.5.2 Quantitative Approaches 288
3.10.5.3 Spatial Risk Visualization 289
3.10.6 Conclusions 290
Acknowledgements 293
References 293

Glossary member countries, to build safe communities, and to create

AAL Average annual loss, or expected annual loss
(EAL), is the long-term loss rate that can be obtained
by summing the product of each discrete loss state with
its annual frequency of occurrence over all loss
states.
ADPC The Asian Disaster Preparedness Center is a
non-profit organization from Bangkok, supporting the
advancement of safer communities and sustainable
development through implementing programs and projects
that reduce the impact of disasters on countries and
communities in Asia and the Pacific.
ADRC The Asian Disaster Reduction Center in Kobe,
Japan has a mission to enhance disaster resilience of the
a society where sustainable development is possible.
AEP Annual exceedance probability is the estimated
probability that an event of specified magnitude will be
exceeded in any year.
ALOS Advanced Land Observing Satellite is a land
observing satellite launched in January 2006 by Japan
Aerospace Exploration Agency (JAXA).
APRSAF Asia-Pacific Regional Space Agency Forum was
established in 1993 to enhance space activities in the
Asia-Pacific region.
ArcGIS GIS software by ESRI.
ArcPad GIS software developed by ESRI for use on
Mobile-GIS devices.

Van Westen, C.J., 2013. Remote sensing and GIS for natural hazards
assessment and disaster risk management. In: Shroder, J. (Editor in Chief),
Bishop, M.P. (Ed.), Treatise on Geomorphology. Academic Press, San Diego,
CA, vol. 3, Remote Sensing and GIScience in Geomorphology, pp. 259–298.

Treatise on Geomorphology, Volume 3 http://dx.doi.org/10.1016/B978-0-12-374739-6.00051-8 259

260 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management
ASTER Advanced Spaceborne Thermal Emission and
Reflection Radiometer is a Japanese sensor that is one of
five remote-sensing devices on board of the Terra sate-
llite launched by NASA in 1999. ASTER provides
high-resolution images of the Earth in 15 different bands of
the electromagnetic spectrum, ranging from visible to
thermal infrared light.
AVHRR Advanced Very High Resolution Radiometer is a
space-borne sensor on the National Oceanic and
Atmospheric Administration (NOAA) family of polar
orbiting platforms.
Capacity The positive managerial capabilities of
individuals, households, and communities to confront the
threat of disasters (e.g., through awareness raising, early
warning, and preparedness planning).
CAPRA Central American Probabilistic Risk Assessment,
initiative developed by the Worldbank.
Cartosat A stereoscopic Earth observation satellite in a
sun-synchronous orbit. The satellite was built, launched,
and maintained by the Indian Space Research Organisation
(ISRO).
CLC CORINE Land Cover. See CORINE
Consequence The expected losses in a given area as a
result of a given hazard scenario.
CORINE The Corine Land Cover project is a joint
initiative of the EU Commission and EU Environment
Agency, which developed an available database now
including ortho-rectified Landsat 7 ETM satellite images of
the European territory, Land Cover, and Change
definition maps.
CRED The Centre for Research on the Epidemiology of
Disasters, of the Université catholique de Louvain, Belgium,
specializes in the fields of international disaster and conflict
health studies, with research and training activities linking
relief, rehabilitation, and development, and maintains the
EM-DAT database.
CVA The Capacity and Vulnerability Assessment is used as
a diagnostic tool to understand problems and their
underlying causes, related to the vulnerability and capacity
of local communities to natural hazards.
DEM A digital elevation model is a digital model or 3-D
representation of a terrain’s surface created from terrain
elevation data.
DGPS A differential global positioning system is an
enhancement to the global positioning system that
uses a network of fixed, ground-based reference stations
to broadcast the difference between the positions
indicated by the satellite systems and the known fixed
positions.
Disaster A serious disruption of the functioning of a
community or a society causing widespread human,
material, economic, or environmental losses that exceed the
ability of the affected community or society to cope using its
own resources.
DNCA Damage, needs, and capacity assessment involves
a participatory analysis of the disaster event focussing on
the damages caused, the immediate needs and priorities of
the affected community, and of the remaining capacities
people use to cope with the adverse effects.
DRI Disaster Risk Index produced a model of factors
influencing levels of human losses from natural hazards at
the global scale, by the United Nations Development
Programme.
DRM Disaster-risk management is the systematic process
of using administrative directives, organizations, and
operational skills and capacities to implement strategies,
policies, and improved coping capacities in order to lessen
the adverse impacts of hazards and the possibility of
disaster.
DRR Disaster-risk reduction is the concept and practice of
reducing disaster risks through systematic efforts to analyse
and manage the causal factors of disasters, including
through reduced exposure to hazards, lessened vulnerability
of people and property, wise management of land and the
environment, and improved preparedness for adverse
events.
DSM A digital surface model is a digital model or 3-D
representation of the Earth’s surface and includes all
objects on it.
DTM Digital terrain model is a digital model or 3-D
representation of the bare ground surface without any
objects like plants and buildings.
EDM Electronic distance measurement is an electronic
theodolite integrated with an electronic distance meter to
read slope distances from the instrument to a
particular point.
EFD European Flood Directive requires EU Member States
to engage their government departments, agencies, and
other bodies to draw up a Preliminary Flood Risk
Assessment.
EFFIS The European Forest Fire Information System
supports the services in charge of the protection of forests
against fires in the EU countries and provides the European
Commission services and the European Parliament with
updated and reliable information on wildland fires in
Europe.
EIA An environmental impact assessment is an assessment
of the possible positive or negative impacts that a proposed
project may have on the environment, together consisting
of the natural, social, and economic aspects.
Elements-at-risk Population, properties, economic
activities, including public services, or any other defined
values exposed to hazards in a given area. Also referred to as
‘‘assets’’. The amount of elements-at-risk can be quantified
either in numbers (of buildings, people, etc.), in monetary
value (replacement costs, market costs, etc.), area, or
perception (importance of elements-at-risk).
EM-DAT Emergency Events Database, the international
database on natural hazards, maintained by CRED.
ENVISAT Polar orbiting spacecraft operated by the
European Space Agency possessing several instruments
used to monitor the Earth’s environment. Parameters
measured include ozone concentration, aerosols,
surface stress for earthquake potential, sea-level heights,
and fires.
EO Earth observation is the gathering of information
about planet Earth’s physical, chemical, and biological
systems, using remote sensing. It is used to monitor and
 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management
assess the status of, and changes in, the natural
environment and the built environment.
ERS European radar satellite, developed and maintained
by the European Space Agency (ESA).
ESRI The world’s largest GIS company that develops
geographic information systems (GIS) solutions that
function as an integral component in nearly every type of
organization.
ETOPO Topographic dataset from the NOAA NGDC,
of topography and bathymetry for the entire Earth’s
surface.
EWS Early Warning System is the set of capacities needed
to generate and disseminate timely and meaningful warning
information to enable individuals, communities, and
organizations threatened by a hazard to prepare and to act
appropriately and in sufficient time to reduce the possibility
of harm or loss.
Exposure Exposure indicates the degree to which
the elements-at-risk are exposed to a particular hazard.
The spatial interaction between the elements-at-risk and the
hazard footprints are depicted in a GIS by simple map
overlaying of the hazard map with the elements-
at-risk map.
FEMA Federal Emergency Management Agency, the
national organization for disaster-risk management in
the USA.
Formosat Earth observation satellite operated by the
National Space Organization (NSPO) of the Republic of
Taiwan.
Frequency A measure of likelihood expressed as the
number of occurrences of an event in a given time.
GDACS Global Disaster Alert and Coordination System,
by Joint Research Center (JRC) of the European
Commission.
GeoEye A commercial satellite imagery company based in
Dulles, Virginia, USA.
GEM The Global Earthquake Model is a public-private
partnership initiated in 2006 by the Global Science Forum
of the OECD to develop global, open-source risk
assessment software and tools.
Geological hazard A geological process or phenomenon
that may cause loss of life, injury or other health impacts,
property damage, loss of livelihoods and services, social and
economic disruption, or environmental damage.
GIS A geographic information system, or geospatial
information system is a system designed to capture, store,
manipulate, analyze, manage, and present all types of
geographically referenced data.
GMES Global Monitoring for Environment and Security
initiative of the European Commission and the European
Space Agency (ESA).
G-MOSAIC GMES services for Management of
Operations, Situation Awareness, and Intelligence for
regional Crises.
GMOSS Global Monitoring for Security and Stability is a
project in the aeronautics and space priority of the EU 6th to
integrate Europe’s civil security research and to develop and
maintain an effective capacity for global monitoring using
satellite earth observation.
261
GPS Global positioning systems is a space-based global
navigation satellite system (GNSS) that provides location
and time information in all weather anywhere on or near
the Earth, where there is an unobstructed line of sight to
four or more GPS satellites.
GRUMP Global Rural-Urban Mapping Project is a project
for generation of a gridded model with a grid cell resolution
of 30 arc-seconds for global population distribution
developed by the Socioeconomic Data and Applications
Center (SEDAC).
GSHAP The Global Seismic Hazard Assessment Project
was launched in 1992 by the International Lithosphere
Program (ILP) with the support of the International
Council of Scientific Unions (ICSU), and produced small-
scale standardized seismic hazard maps for all seismic
regions of the world.
Hazard A dangerous phenomenon, substance, human
activity, or condition that may cause loss of life, injury or
other health impacts, property damage, loss of livelihoods
and services, social and economic disruption, or
environmental damage.
Hazard zoning The subdivision of the terrain in zones
that are characterized by the temporal probability of
occurrence of hazardous events of a particular size and
volume within a given period of time.
HAZUS Hazards US is the loss estimation software
developed by the Federal Emergency Management Agency
(FEMA) together with the National Institute of Building
Sciences (NIBS).
HVCA Hazards, vulnerability, and capacity assessment is a
method that involves a participatory analysis of historical
hazard trends and present threats undertaken at the level of
the community (hazard assessment). It is combined with an
understanding of the underlying reasons why hazards
become disasters (vulnerability assessment) and of the
available resources an affected community uses to cope
(capacity assessment).
Hydrometeorological hazard Process or phenomenon of
atmospheric, hydrological, or oceanographic nature that
may cause loss of life, injury or other health impacts,
property damage, loss of livelihoods and services, social and
economic disruption, or environmental damage.
ICG International Centre for Geohazards, Norway.
IFRC International Federation of Red Cross and Red
Crescent Societies.
IKONOS A commercial Earth-observation satellite, and
was the first to collect publicly available high-resolution
imagery at 1- and 4-meter resolution. It offers multispectral
(MS) and panchromatic (PAN) imagery.
ILWIS The Integrated Land and Water Information
System is an open-source combined GIS and image
processing software developed by ITC, and maintained by
53North.
InSAR Interferometric SAR is a radar technique used in
geodesy and remote sensing. This geodetic method uses two
or more synthetic aperture radar (SAR) images to generate
maps of surface deformation or digital elevation, using
differences in the phase of the waves returning to the
satellite.
262 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management
IRS Indian Remote Sensing satellites are a series of Earth
Observation satellites, built, launched, and maintained by
Indian Space Research Organisation. The IRS series provides
many remote sensing services to India.
ITC The Faculty of Geo-Information Science and
Observation of the University of Twente is an organization
that provides international postgraduate education,
research, and project services in the field of geo-information
science and earth observation using remote sensing and
GIS, located in Enschede, the Netherlands.
LANDSAT The longest running enterprise for acquisition
of imagery of Earth from space. The first Landsat satellite
was launched in 1972; the most recent, Landsat 7, was
launched on April 15, 1999. Landsat 7 data has eight
spectral bands with spatial resolutions ranging from 15 to
60 m; the temporal resolution is 16 day.
LiDAR Light Detection And Ranging is an optical remote
sensing technology that can measure the distance to, or
other properties of, a target by illuminating the target with
light, often using pulses from a laser.
LIMES Land and Sea Integrated Monitoring for
Environment and Security is an Integrated Project co-
funded by the European Commission within the 6th
Framework Programme – Aeronautics&Space/GMES
Security.
MARSOP MARS Crop Yield Forecasting System (MCYFS)
carried out by AGRI4CAST and FOODSEC actions within
the Institute for the Protection and the Security of the
Citizen (EU JRC).
MERIS Medium Resolution Imaging Spectrometer is one
of the main instruments on board of the European Space
Agency (ESA)’s Envisat platform, to observe the color of the
ocean, both in the open ocean and in coastal zones.
Mobile-GIS A GIS that is running on a mobile, hand held
device, which is linked to a GPS for collecting spatial and
attribute data in digital format directly in the field.
MODIS Moderate Resolution Imaging Spectro-radiometer
is a sensor launched into Earth’s orbit by NASA in 1999 on
board of the Terra (EOS AM) Satellite, and in 2002 on
board of the Aqua (EOS PM) satellite, designed to provide
measurements in large-scale global dynamics including
changes in Earth’s cloud cover, radiation budget, and
processes occurring in the oceans, on land, and in the lower
atmosphere.
Natural hazard A potentially damaging physical event,
phenomenon, or human activity that may cause loss of life
or injury, property damage, social and economic disruption,
or environmental degradation. This event has a probability
of occurrence within a specified period of time and within a
given area, and has a given intensity.
NDVI Normalized Differential Vegetation Index is a
simple numerical indicator that can be used to analyze
remote sensing measurements, typically but not necessarily
from a space platform, and assess whether the target being
observed contains live green vegetation or not.
NOAA National Oceanic and Atmospheric
Administration, USA.
OneGeology It is an international initiative of the
geological surveys of the world, launched in 2007 as a
contribution to the International Year of Planet Earth, with
the aim to create a web-based geological map of the world.
ORCHESTRA One of the European Union’s major
research and innovation projects for risk management, to
help national and local governments predict and react to
natural disasters by joining up national and local
information systems and applications.
OSM Open Street Map is a collaborative project to create a
free editable map of the world. The maps are created using
data from portable GPS devices, aerial photography, other
free sources, or simply from local knowledge.
PAGER Prompt Assessment of Global Earthquakes for
Response, a tool for the rapid assessment of peoples
exposed after an earthquake, developed by the USGS.
PALSAR Phased Array type L-band Synthetic Aperture
Radar sensor on board of the Advanced Land Observing
Satellite (ALOS) developed by the National Space
Development Agency of Japan (NASDA).
PGA Peak ground acceleration is a measure of earthquake
acceleration on the ground and an important input
parameter for earthquake engineering.
PGIS Participatory GIS is encompassing participatory
approaches to planning of spatial information and
communication management using mobile GIS.
PML Probable maximum loss is the anticipated value of
the largest loss that could result from the destruction and
the loss of use of property.
Preparedness The knowledge and capacities developed by
governments, professional response and recovery
organizations, communities, and individuals to effectively
anticipate, respond to, and recover from the impacts of
likely, imminent, or current hazard events or conditions.
Prevention The avoidance of adverse impacts of hazards
and related disasters.
PREVIEW Prevention, Information, and Early Warning
pre-operational services to support the management
of risks.
PRISM Panchromatic Remote-sensing Instrument for
Stereo Mapping, on board of the Advanced Land Observing
Satellite (ALOS).
Probability A measure of the degree of certainty. This
measure has a value between zero (impossibility) and 1.0
(certainty). It is an estimate of the likelihood of the
magnitude of the uncertain quantity, or the likelihood of
the occurrence of the uncertain future event.
PSInSAR Permanent Scatterers Radar Interferometry is a
relatively recent development from conventional InSAR,
and relies on studying pixels that remain coherent over a
sequence of interferograms.
Qualitative risk analysis An analysis that uses word form
and descriptive or numeric rating scales to describe the
magnitude of potential consequences and the likelihood
that those consequences will occur.
Quantitative risk analysis An analysis based on
numerical values of the probability, vulnerability, and
consequences, and resulting in a numerical value of the risk.
QuickBird A high-resolution commercial Earth-
observation satellite, owned by DigitalGlobe and launched
in 2001.
 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management
RADARSAT The RADARSAT constellation is a pair of
Canadian remote-sensing satellites. The constellation
consists of RADARSAT-1 launched in 1995 and RADARSAT-
2 launched in 2007.
RADIUS Risk assessment tools for diagnosis of urban
areas against seismic disasters.
Remote sensing The acquisition of information about an
object or phenomenon without making physical contact
with the object. In modern usage, the term generally refers
to the use of aerial sensor technologies to detect and classify
objects on Earth (both on the surface and in the
atmosphere and oceans) by means of propagated signals
(e.g., electromagnetic radiation emitted from aircraft or
satellites).
Resilience The ability of a system, community, or society
exposed to hazards to resist, absorb, accommodate to, and
recover from the effects of a hazard in a timely and efficient
manner, including through the preservation and restoration
of its essential basic structures and functions.
Resourcesat The Indian Remote Sensing (IRS) P6 satellite
operated by ISRO, and includes the two remote sensing
instruments known as LISS-III and AWiFS.
Response The provision of emergency services and public
assistance during, or immediately after, a disaster in order to
save lives, reduce health impacts, ensure public safety, and
meet the basic subsistence needs of the people affected.
Risk The probability of harmful consequences or expected
losses (deaths, injuries, property, livelihoods, economic
activity disrupted, or environment damaged), resulting
from interactions between various types of hazards and
vulnerable conditions in a given area and over a specific
time period.
Risk analysis The use of available information to estimate
the risk to individuals or populations, property, or the
environment from hazards. Risk analysis generally contains
the following steps: hazard identification, hazard
assessment, elements-at-risk/exposure analysis,
vulnerability assessment, and risk estimation.
Risk assessment The process of risk analysis and risk
evaluation.
Risk control or risk treatment The process of decision
making for managing risks, and the implementation, or
enforcement of risk-mitigation measures and the re-
evaluation of its effectiveness from time to time, using the
results of risk assessment as one input.
Risk evaluation The stage at which values and judgements
enter the decision process, explicitly or implicitly, by
including consideration of the importance of the estimated
risks and the associated social, environmental, and
economic consequences, in order to identify a range of
alternatives for managing the risks.
Risk management The complete process of risk
assessment and risk control (or risk treatment).
Risk perception The way people/communities/authorities
judge the severity of the risk based on their personal
situation, social, political, cultural, and religious
background, economic level, their level of awareness, the
information they have received regarding the risk, and the
way they rate the risk in relation with other problems.
263
RiskScape New Zealand software and methodology for
multi-hazard risk assessment.
Risk transfer The process of formally or informally
shifting the financial consequences of particular risks from
one party to another whereby a household, community,
enterprise, or state authority will obtain resources from the
other party after a disaster occurs, in exchange for on-going
or compensatory social or financial benefits provided to
that other party.
SAFER Services and applications for emergency
Response, an EU GMES Emergency Response Service, with
the aim to provide space-based products in support of
European decision-makers facing natural and technological
disasters.
SAR Synthetic aperture radar is a form of radar whose
defining characteristic is its use of relative motion between
an antenna and its target region to provide distinctive long-
term coherent-signal variations that are exploited to obtain
finer spatial resolution than is possible with conventional
beam-scanning means. It originated as an advanced form of
side-looking airborne radar (SLAR).
SDI Spatial data infrastructure is a framework of spatial
data, metadata, users, and tools that are interactively
connected in order to use spatial data in an efficient and
flexible way.
SELENA SEimic Loss EstimatioN using a logic tree
Approach, an open risk assessment package consisting of
the two separate software tools SELENA (Seismic Loss
Estimation using a Logic Tree Approach) and RISe (Risk
Illustrator for SELENA), developed by the University of
Alicante, NOSAR, and ICG.
ShakeMaps A GIS-based tool for earthquake hazard
assessment, developed by the USGS in cooperation with
regional seismic-network operators. ShakeMaps provides
near-real-time maps of ground motion and shaking
intensity after important earthquakes.
SMCE Spatial Multi-Criteria Evaluation, a tool for
decision-aid and a mathematical tool allowing the
comparison of different alternatives or scenarios according
to many criteria, often conflicting, in order to guide the
decision maker toward a judicious choice.
Societal risk The risk of multiple fatalities or injuries
in society as a whole: one where society would have to
carry the burden of a landslide causing a number of
deaths, injuries, financial, environmental, and other
losses.
SPOT Satellite Pour l’Observation de la Terre, a high-
resolution, optical imaging Earth observation satellite
system from France.
SRTM Shuttle Radar Topography Mission (SRTM), an
international research effort that obtained digital elevation
models on a near-global scale from 561S to 601N to
generate a high-resolution digital topographic database of
the Earth.
Technological hazard A hazard originating from
technological or industrial conditions, including accidents,
dangerous procedures, infrastructure failures, or specific
human activities that may cause loss of life, injury, illness or
other health impacts, property damage, loss of livelihoods
264 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management

and services, social and economic disruption, or
environmental damage.
TMPA TRMM Multi-satellite Precipitation Analysis.
TRMM Tropical Rainfall Measuring Mission, a joint NASA
and JAXA mission to monitor and study tropical rainfall
from satellite imagery.
UN-ISDR United Nations International Strategy for
Disaster Risk Reduction.
UN-SPIDER United Nations Platform for Space-based
Information for Disaster Management and Emergency
Response.
USGS United States Geological Survey.
Ushahidi Free open-source software for co-ordinating
citizen reports on a national crisis.
VDV Virtual Disaster Viewer, a crowdsourcing tool for
disaster-damage assessment developed by ImageCat.
Vulnerability The conditions determined by physical,
social, economic, and environmental factors or processes,
which increase the susceptibility of a community to the
impact of hazards. Can be subdivided in physical, social,
economical, and environmental vulnerability.
WASP Weighted Anomaly of Standardized Precipitation
gives an estimate of the relative deficit or surplus of
precipitation for different time intervals ranging from 1- to
6-months.
Web-GIS It is the process of designing, implementing,
generating, and delivering maps on the World Wide Web
with an emphasis on analysis and processing of project
specific geodata and exploratory aspects.
WorldView Commercial high-resolution panchromatic
satellite owned by DigitalGlobe. WorldView-2 provides
commercially available panchromatic imagery of 0.5 m
resolution, and eight-band multispectral imagery with
1.8 m resolution.

Abstract

The world has experienced an increasing impact of disasters in the past decades. Many regions are exposed to natural
hazards, each with unique characteristics. The main causes for this increase can be attributed to a higher frequency of
extreme hydro-meteorological events, most probably related to climate change and an increase in a vulnerable population.
To reduce disaster losses, more efforts should be applied toward disaster-risk management, with a focus on hazard as-
sessment, elements-at-risk mapping, and vulnerability and risk assessment, all of which have an important spatial com-
ponent. Multi-hazard assessment involves the assessment of relationships between different hazards, especially
concatenated or cascading hazards. The use of Earth observation (EO) products and geographic information systems (GIS)
has become an integrated approach in disaster-risk management. Hazard and risk assessments are carried out at multiple
scales, ranging from global to community levels. These levels have their own objectives and spatial data requirements for
hazard inventories, environmental data, triggering or causal factors, and elements-at-risk. This chapter provides an overview
of various forms of spatial data, and examines the approaches used for hazard and risk assessment. Specifically, hazard
examples include earthquakes, windstorms, drought, floods, volcanic eruptions, landslides, and forest fires. Several ap-
proaches are also treated that have been developed to generate elements-at-risk databases with emphasis on population and
building information, as these are the most used categories for loss estimation. Furthermore, vulnerability approaches are
discussed, with emphasis on the methods used to define physical vulnerability of buildings and population, and indicator-
based approaches used for a holistic approach, also incorporating social, economic, and environmental vulnerability, and
capacity. Finally, multi-hazard risk approaches and spatial-risk visualization are addressed. Multi-hazard risk assessment is a
complicated procedure, which requires spatial data on many different aspects and a multi-disciplinary approach. Within
this procedure, geographers and in particular geomorphologists can play a key role, as they are able to integrate spatial
information from various disciplines. The research challenge for geomorphologists is to focus on the integrated modeling
of multi-hazards that share the same triggering event or occur as cascading hazards.

3.10.1 Introduction
Disasters are headline news almost every day. Most happen in
far-away places and are rapidly forgotten. Others keep the
attention of the world media for a longer period of time. The
events that receive maximum media attention are those that
hit instantaneously and cause widespread losses and human
suffering, such as earthquakes, tsunamis, hurricanes, and
floods. Recent examples are the Indian Ocean tsunami (2004),
the earthquakes in Pakistan (2005), Indonesia (2006), China
(2008), Haiti (2010), and Japan (2011), and the hurricanes in
the Caribbean and the USA (2005, 2008). However, there
are many serious geomorphologic hazards that have a slow
onset, such as the recent drought in the Horn of Africa (2011),
soil erosion, land degradation, desertification, glacial retreat,
sea-level rise, loss of biodiversity, etc. These processes and
related events may cause local, regional, and global impacts in
the long run, but generally receive less attention.
Disasters are defined by the United Nations International
Strategy for Disaster Risk Reduction (UN-ISDR, 2004) as ‘‘a
serious disruption of the functioning of a community or a
society causing widespread human, material, economic or
environmental losses which exceed the ability of the affected
community or society to cope using its own resources’’. Al-
though the term ‘‘natural disasters’’ in its strict sense is not
correct, as disasters are a consequence of the interaction
between hazards and vulnerable societies, the term is used
extensively in the literature and also in daily use.
A hazard is defined as ‘‘a potentially damaging physical
event, phenomenon or human activity that may cause the loss
 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management
of life or injury, property damage, social and economic dis-
ruption or environmental degradation’’. This event has a
probability of occurrence within a specified period of time and
within a given area, and has a given intensity (UN-ISDR,
2004). Hazards can be single, sequential, or combined in their
origin and effects. Each hazard is characterized by its location,
area affected (size or magnitude), intensity, speed of onset,
duration, and frequency. Hazards can be classified in several
ways. A possible subdivision is between natural, human-
induced, and human-made hazards. Natural hazards are
natural processes or phenomena in the Earth’s system (litho-
sphere, hydrosphere, biosphere, or atmosphere) that may
constitute a damaging event (e.g., earthquakes, volcanic
eruptions, and hurricanes). A subdivision of natural hazards
relates to the main controlling factors of the hazards leading
to a disaster. They may be hydro-meteorological (including
floods and wave surges, storms, droughts, and related disasters
such as extreme temperatures and forest/scrub fires, landslides,
and snow avalanches), geophysical hazards (resulting from
anomalies in the Earth’s surface or subsurface, such as
earthquakes, tsunamis, and volcanic eruptions), or biological
hazards (related to epidemics and insect infestations).
Human-induced hazards are those resulting from modifi-
cations of natural processes in the Earth’s system caused by
human activities that accelerate/aggravate the damage poten-
tial (e.g., land degradation, landslides, and forest fires).
Human-made hazards originate from technological or indus-
trial accidents, dangerous procedures, infrastructure failures,
or certain human activities, which may cause the loss of life
or injury, property damage, social and economic disruption,
or environmental degradation (e.g., industrial pollution,
nuclear activities and radioactivity, toxic wastes, dam failures,
and transport, industrial, or technological accidents such as
explosions, fires, and oil spills).
The objective of a hazard assessment is to identify the
various types of hazards that may threaten a territory, and to
partition the landscape in spatial zones that are characterized
by different expected intensities and frequencies of hazardous
processes. In a hazard assessment, several aspects should be
evaluated: 1) the triggering event; 2) the areas where hazards
are likely to initiate; 3) the areas where the hazards are likely
to spread; and 4) the expected intensity of the hazard and its
associated frequency or probability of occurrence. Hazards
may impact vulnerable societies, and may result in physical
and other types of damage to so called ‘‘elements-at-risk’’. With
certain exceptions (e.g., ash clouds affecting air-traffic or oil-
spills affecting marine flora and fauna), these impacted
elements-at-risk are mostly located on the Earth’s surface.
Therefore, the relationship between the hazard events and
surface processes, landforms, and materials is a key com-
ponent to study. Therefore, geomorphology plays a key role in
hazard and risk analysis. Geomorphology is the study of
landforms and surface materials and of the processes that have
formed or reshaped them. Geomorphologists investigate the
evolution of landscapes, and study the history and the dy-
namics of landforms and the processes responsible for cre-
ating or modifying them, through a combination of field
observations, physical experiments, and numerical modelling.
Numerous processes that have shaped the Earth’s surface can
be potentially dangerous if they exceed a certain threshold
265
(e.g., they may result in instability and erosion on slopes and
catastrophic flooding in river or coastal areas).
The study of the endogenic (volcanic eruptions or earth-
quakes) and exogenic (extreme meteorological) triggering
events causing hazardous processes may sometimes be outside
of the scope of geomorphology, as this is the domain of
seismologists, volcanologists, meteorologists, and climat-
ologists, although geomorphology is becoming more multi-
disciplinary. Nevertheless, geomorphologists contribute to the
reconstruction of the frequency, extent, and intensity of past
events by studying their imprints on landforms and surface
materials. Geomorphologists play a key role in the analysis of
how these triggering events result in hazardous processes on
the Earth surface. They are specialized in analyzing the pos-
sible spatial extent of future hazardous processes related to a
set of environmental factors (geomorphology, topography,
geology, soils, land cover, etc.). Examples of this are the study
of potential landslide areas or the analysis of the effects of
surface materials and landforms in the amplification of seis-
mic waves or liquefaction. Geomorphologists are also good in
modeling the extent of the spreading of hazardous processes
over the topography (e.g., lahar flows, landslide runout, and
flood extent) based on the use of digital elevation models
(DEMs) and terrain analysis.
The study of hazards and risks has a very important spatial
component, as certain types of hazards are restricted to certain
geographical regions. Earthquakes occur along active tectonic-
plate margins and volcanos occur along subduction zones (e.g.,
around the margins of the Pacific plate, so-called ‘‘Ring of Fire’’).
Tsunamis occur in the neighborhood of active-plate margins,
but their effects can be felt at considerable distances from their
origin as the waves can travel long distances. Tropical cyclones
(in North America called ‘‘hurricanes’’ and in Asia called
‘‘typhoons’’) occur in particular coastal zones. Landslides occur
in hilly and mountainous regions. In the analysis of hazard and
risk, geospatial technologies and Earth observation via remote
sensing are increasingly playing important roles. Remote sens-
ing science and technology are now essential in monitoring
changes related to the Earth’s surface, oceans, and atmosphere,
and are increasingly used as the basis for early warning for
hazardous events. Furthermore, remote sensing enables the-
matic information extraction that represents the input data for
hazard modeling (e.g., topography, lithology, and land cover).
The use of spatio-temporal data and geographic infor-
mation systems (GIS) has now become a part of an integrated
approach to disaster risk management. New GIS algorithms
and analysis/modeling techniques are revolutionizing the
potential capacity to analyse hazards, vulnerability, and risks.
These spatial information systems are used for storage, situ-
ation analysis, modeling, and visualization (Twigg, 2004).
Disaster-risk management benefits greatly from the use of
geospatial technologies because spatial and temporal variation
can be accounted for, and new methodologies can be de-
veloped and fully explored. One of the key advantages of using
GIS-based software tools for the risk decision-making process
is the possibility to explore ‘what if’ analysis by varying par-
ameters and generating alternative scenarios in a spatial con-
text (Longley et al., 2005). Earlier publications on this topic
can be found in Wadge et al. (1993), Coppock (1995), Emani
(1996), and Kaiser et al. (2003).
266 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management

Table 1 Summary of definitions related to disasters, hazards, and vulnerability

Term Definition

Disaster A serious disruption of the functioning of a community or a society causing widespread human, material, economic, or
environmental losses that exceed the ability of the affected community or society to cope using its own resources.
Natural hazard A potentially damaging physical event, phenomenon, or human activity that may cause loss of life or injury, property
damage, social and economic disruption, or environmental degradation. This event has a probability of occurrence
within a specified period of time and within a given area, and has a given intensity.
Elements-at-risk Population, properties, economic activities, including public services, or any other defined values exposed to hazards in a
given area. Also referred to as ‘‘assets’’. The amount of elements-at-risk can be quantified either in numbers (of
buildings, people, etc.), monetary value (replacement costs, market costs, etc.), area, or perception (importance of
elements-at-risk).
Exposure Exposure indicates the degree to which the elements-at-risk are exposed to a particular hazard. The spatial interaction
between the elements-at-risk and the hazard footprints are depicted in a GIS by a simple map overlaying of the hazard
map with the elements-at-risk map.
Vulnerability The conditions determined by physical, social, economic, and environmental factors or processes, which increase the
susceptibility of a community to the impact of hazards. Can be subdivided in physical, social, economical, and
environmental vulnerability.
Capacity The positive managerial capabilities of individuals, households, and communities to confront the threat of disasters (e.g.,
through awareness raising, early warning, and preparedness planning).
Consequence The expected losses in a given area as a result of a given hazard scenario.
Risk The probability of harmful consequences or expected losses (deaths, injuries, property, livelihoods, economic activity
disrupted, or environment damage) resulting from interactions between (natural, human-induced, or man-made)
hazards and vulnerable conditions in a given area and time period.

Source: Based on UN-ISDR, 2004. Terminology of disaster risk reduction. United Nations, International Strategy for Disaster Reduction, Geneva, Switzerland. http://www.unisdr.org/
eng/library/lib-terminology-eng%20home.htm

The objective of this chapter is to give an overview of the

application of geo-information science and Earth observation
in the analysis of hazards and risk. The chapter starts with a
Vulnerable
background section discussing disaster trends and the frame- Hazards Risk
society
work for disaster–risk management. Subsequent sections dis-
cuss the use of GIS and remote sensing in the generation of
inventories of past hazardous events, the modelling of haz- (a)
ards, the generation of elements-at risk databases, and the
integration of these data in the modeling of potential losses.

Vulnerable
Hazards Disaster
3.10.2 Background society

Table 1 provides a summary of the various terms that are

relevant in the context of this chapter, relating to disasters,
disaster risk, and its various components (UN-ISDR,2004). It (b)
is important to distinguish between the terms disaster, hazard,
and risk. Risk results from the combination of hazards, con-
ditions of vulnerability, and insufficient capacity or measures
to reduce the potential negative consequences of risk (O’Keefe
et al., 1976). When the hazard or threat becomes a reality (i.e.,
Vulnerable
when it materializes), the risk becomes a disaster. For example, Hazards Risk
society
a certain river valley may be prone to flooding. There is risk if a
vulnerable society/community or property is located within
this flood-prone area. If the hazard materializes, that is, if the
flood actually occurs, it will cause losses to the vulnerable
(c)
society or property, thus creating a disaster (Figure 1).
Figure 1 Schematic representation of the relation between hazards,
vulnerable society, risk, and disasters. (a) Risk indicates the expected
3.10.2.1 Trends in Disaster Statistics losses to a vulnerable society as a result of hazards. (b) A disaster
occurs when the threat of a hazard becomes reality and impacts on a
Data on disaster occurrences, their effect on people, and vulnerable society. (c) Future trends of increasing hazards and
their cost to countries are very important for disaster-risk increasing vulnerability will lead to increasing risk.
 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management 267

management. There are now a number of organizations that
collect information on disasters at different scales and with
different objectives.
• Since 1988, the Centre for Research on the Epidemiology of
Disasters (CRED) has been maintaining an Emergency
Events Database (EM-DAT, 2009). Disasters have to fulfill
certain criteria in order to be included in the EM-DAT
database: They have to cause at least 10 casualties; 100 or
more people should be affected; it should result in a dec-
laration of emergency; or it should lead to a call for external
assistance.
• Data on disaster impacts are also collected by reinsurance
companies. For instance, the MunichRe database for natural
catastrophes (NatCatSERVICE) includes more than 28 000
entries on material and human-loss events worldwide
(MunichRe, 2010). A similar disaster-event database
(SIGMA) is maintained by SwissRe. These data, however, are
not publicly available.
• The Asian Disaster Reduction Center (ADRC) has initiated a
new disaster database, called Glidenumber (2010). The
specific feature of this database is that each disaster receives a
unique identifier and a number of relevant attributes.
• At a local level, disaster data have been collected by an ini-
tiative of NGOs, called LaRed, initially in Latin America, but
later on expanding also to other regions. They generated a
tool called DesInventar (2010), which allows local author-
ities, communities, and NGO’s to collect disaster infor-
mation. Recently, the DesInventar database has become
available online.
• There are also many disaster databases collected at the na-
tional level, or that are related to a specific type of hazard.
The Global Risk Identification Program (GRIP) and the
Centre for Research in Epidemiology of Disasters (CRED)
have initiated a service, called DisDAT, which brings together
all publicly available disaster databases from different
countries (GRIP, 2010). It contains 60 registered disaster
databases, of which 13 are global.
When examining the reported disasters in these databases,
there is a clear increase in hazardous events over the past
decades (Figure 2). The number of natural disasters in the last
decade has increased by a factor of 9 compared to the decade
1950–1959 (EM-DAT, 2009), which is mainly caused by an
increase in hydro-meteorological disasters. In terms of losses,
earthquakes resulted in the largest amount of losses (35% of
all losses), followed by floods (30%), windstorms (28%), and
others (7%). Earthquakes are also the main cause of fatalities,
which is estimated on the order of 1.4 million during the
period 1950–2000 (47%), followed by windstorms (45%),
floods (7%), and others (1%) (MunichRe, 2010; EM-DAT,
2009). It is interesting to note that human fatalities due to
natural disasters show a decreasing trend, which may be due
to better warning systems and improved disaster management,
but the number of people affected follows the increasing trend
of the number of events (see Figure 2).
Disaster information collected at the local level (e.g.,
DesInventar) is more complete, as it also includes small-
magnitude/high-frequency events, but the coverage of such
databases is limited worldwide. One of the major problems
with the use of disaster databases for natural hazard and risk
assessment is that they normally lack proper georeferencing of
the reported events (Verelst, 1999). A comparative study of the
EM-DAT, Sigma, and NATCAT databases carried out for four
countries showed that these databases differed significantly
(Guha-Sapir and Below, 2002).
The increase in the number of disasters, the losses, and
people affected cannot be explained only by better reporting
methods and media coverage of disasters, lack of which
probably made the number too low for the first part of the last
century. A number of factors influence the increase in the
number of disasters that can be subdivided as those leading to

500
Million
500 000 400 200
Nr. people affected (millions)
Nr. disasters reported

400 000
300 150
Nr. people killed

300 000
200 100
200 000
Nr. disasters reported
Nr. people killed 100 50
100 000
Nr. people affected

0 0 0
00

10

20

30

40

50

60

70

80

90

00

09
19

19

19

19

19

19

19

19

19

19

20

20

Year
Figure 2 Summary of natural disasters, showing the number of reported disasters, the number of people killed, and the number of people
affected over the period 1900–2009. Reproduced from EM-DAT, 2009. The OFDA/CRED International Disaster Database, Université Catholique de
Louvain, Brussels, Belgium. http://www.emdat.be
268 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management
a larger vulnerability, and those leading to a higher occurrence
of hazardous events.
The increased vulnerability is due to a number of reasons
including the rapid increase of the world population, which
has doubled in size from 3 billion in the 1960s to 7 billion in
2011 (World Bank, 2011). Depending on the expected growth
rates, world population is estimated to be between 7.9 and
11.0 billion by the year 2050 (UNPD, 2010a). The increase in
disaster impact, however, is higher than the increase in
population, which indicates that other important factors are
involved that increase the overall vulnerability of the world
population. One of the main factors is the large urbanization
rate. According to United Nation figures (UNPD, 2010b), the
worldwide urbanization percentage has increased from 29%
in 1950 to 50% in 2010, and is expected to rise to 69% in
2050. Another factor related to the population growth is that
areas become settled, that were previously avoided due to their
susceptibility to natural hazards. Many of the largest cities in
the world, the so-called ‘‘Megacities,’’ are located in hazardous
regions, either in coastal zones or in seismically active regions
(Smith and Petley, 2008; Kraas, 2008).
The increasing impact of natural disasters is also related
with the development of highly sensitive technologies, and the
growing susceptibility of modern industrial societies to
breakdowns in their infrastructure. Data from MunichRe
(2010) show that the economic losses have increased by a
factor of 8 over the past 50 years, and insured losses by a factor
of 15. There is a rapid increase in the insured losses, which are
mainly related to losses occurring in developed countries.
Windstorms clearly dominate the category of insured losses
(US $90 billion), followed by earthquakes (US $ 25 billion).
Insured losses to flooding are remarkably less (US $ 10 bil-
lion) due to the fact that they are most severe in developing
countries with lower insurance coverage (MunichRe, 2010).
It is not only the increased exposure of the population to
hazards, however, that can explain the increase in natural
disasters. The frequency of destructive events related to at-
mospheric extremes (such as floods, drought, cyclones, and
landslides) is also increasing (EM-DAT, 2009). During the past
10 years a total of 3 750 windstorms and floods were recorded,
accounting for two-thirds of all events. The number of catas-
trophes due to earthquakes and volcanic activity (approxi-
mately 100 per year) has remained constant (MunichRe,
2010). Although the time-span is still not long enough to in-
dicate it with certainty, these data suggest that climate change
is related to the increased occurrence of natural disasters
(IPCC, 2007).
An inverse relationship exists between the level of devel-
opment and loss of human lives in the case of disasters. Ap-
proximately 85 % of the disaster-related casualties occur in
less developed countries, where more than 4.7 billion people
live. The greater loss of lives is due to the lower quality of
buildings, lack of building codes or lack of enforcement,
construction of buildings in hazardous areas due to lack of
land-use planning, lower awareness and disaster preparedness,
less accurate or missing early-warning systems, lack of evacu-
ation planning, and lack of facilities for search-and-rescues
and medical attention. Although 65% of the overall losses
occur in high-income countries (with GNI US$ 412 000 per
capita) (World Bank, 2010), and only 3% in low-income
countries (GNI US$ o1000 per capita), the effect in the latter
group is devastating, as they may represent as much as 100%
of their gross national income (UN-ISDR, 2009). Economic
losses in absolute terms (billions of dollars) show an increase
with the level of development, as the absolute value of
elements-at-risk that might be damaged during a disaster in-
creases with increasing level of development. In relative terms,
however, the trend is reversed, showing a decrease in the losses
expressed as percentage of GDP with increasing level of
development (MunichRe, 2010).
3.10.2.2 Disaster-Risk Management Framework
Disaster Risk Management (DRM) is defined as ‘‘the system-
atic process of using administrative decisions, organization,
operational skills, and capacities to implement policies,
strategies, and coping capacities of the society and com-
munities to lessen the impacts of natural hazards and related
environmental and technological disasters’’. This comprises all
forms of activities, including structural and non-structural
measures to avoid (prevention) or limit (mitigation and pre-
paredness) adverse effects of hazards (UN-ISDR, 2004). Dis-
aster-risk management is aimed at disaster-risk reduction,
which refers to the conceptual framework of elements con-
sidered with the possibilities to minimize vulnerabilities and
disaster risks within the broad context of sustainable devel-
opment (UN-ISDR, 2004).
The past decades have witnessed a shift in focus from
disaster recovery and response to risk management and miti-
gation. This change was also from an approach that focused
primarily on the hazard as the main causal factor for risk,
and the reduction of the risk by physical-protection measures,
to a focus on vulnerability of communities and ways to
reduce those through preparedness and early warning. Later,
more focus was given to the strengthening of the capacity
of local communities and the development of community-
based coping strategies (Blaikie et al., 1994; Lavel, 2000;
Pelling, 2003). The Yokohama conference in 1994 put into
perspective the socio-economic aspects as a component of
effective disaster prevention. It was recognized that social
factors such as cultural tradition, religious values, economic
standing, and trust in political accountability are essential
in the determination of societal vulnerability. In order to
reduce societal vulnerability, and therewith decrease the
consequences of natural disasters, these factors need to be
addressed (Hillhorst, 2004). The ability to address socio-
economic factors requires knowledge and understanding of
local conditions, which can, in most cases, only be provided
by local participants.
The decade from 1990 to 2000 was declared by the United
Nations as the International Decade for Natural Disaster Re-
duction (IDNDR). As the impact of disasters increased dra-
matically during this decade, the international community
decided to continue this effort after 2000 in the form of an
International Strategy for Disaster Reduction (ISDR). The
ISDR stressed the need to move from a top-down manage-
ment of disasters and a cycle that focuses on reconstruction
and preparedness, toward a more comprehensive approach
that tries to avoid or mitigate the risk before disasters occur,
 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management 269

and at the same time fosters more awareness, public com-
mitment, knowledge sharing, and partnerships to implement
various risk reduction strategies at all levels (UN-ISDR,
2005b).
This more positive concept has been referred to as the
‘‘risk-management cycle’’, or better ‘‘spiral’’, in which learning
from a disaster can stimulate adaptation and modification in
development planning, rather than a simple reconstruction of
pre-existing social and physical conditions. This is illustrated
in Figure 3, by showing the disaster cycle and various com-
ponents (relief, recovery, reconstruction, prevention, and
preparedness), and how these changed through time. Initially,
most emphasis was given to disaster relief, recovery, and re-
construction, thereby getting into a cycle where the next dis-
aster was going to cause the same effects or worse (Figure 3A).
Later on, more attention was given to disaster preparedness
(Figure C), by developing warning systems and disaster
awareness programs (Figure 3B). Now, the efforts are focusing
on disaster prevention and preparedness Figure 3C, thus in-
creasing the time gap between individual disasters, and re-
ducing their effects, requiring less emphasis in relief, recovery,
and reconstruction. The eventual aim of disaster-risk man-
agement is to enlarge this cycle, and only reach the response
phase for extreme events with very low frequency.
Disaster prevention is achieved through risk management.
Figure 4 presents the general risk-management framework,
which is composed of a risk-assessment block and a block in
which risk-reduction strategies are defined. A summary of the
terminology used in risk management is given in Table 2.
Central to the procedure is risk analysis, in which the available
information is used to estimate the risk to individuals or
populations and property or the environment from various
hazards. Risk analysis generally contains the following steps:
1) hazard identification; 2) hazard assessment; 3) elements-at-
risk/exposure analysis; 4) vulnerability assessment; and 5) risk
estimation. Risk evaluation is the stage at which values and
judgments enter the decision process, explicitly or implicitly,
by including consideration of the importance of the estimated
risks and the associated social, environmental, and economic
consequences, in order to identify a range of alternatives for
reducing the risks (UN-ISDR, 2004).
Risk assessment is the combination of risk analysis and
risk evaluation. It is more than a purely scientific enterprise
and should be seen as a collaborative activity that brings
professionals, authorized disaster managers, local authorities,
and the people living in the exposed areas together (O’Brien,
2000; Plapp, 2001; Montague, 2004). Risk governance is
therefore an integral component. The final goal, reduction of

Prevention

Reconstruction
Preparedness
Recovery
B Relief
Disaster
Reconstruction event
Prevention ?

Recovery

Preparedness
Relief
A
Disaster
event
Recovery Reconstruction
e
m
Prevention Ti
Relief Preparedness

Disaster
event

Figure 3 Disaster cycle and its development through time.

270 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management

Risk management

Risk assessment

Risk analysis
Hazard
identification

(Spatial) Information management

Hazard
assessment

Risk governance/Risk communication

Elements at risk/
exposure

Vulnerability
assessment
Risk monitoring and updating

Risk estimation

Risk evaluation

Risk visualization

Non-structural Structural
Risk transfer Codes
Emergency planning Standard
Awareness/training Reinforcing
Early warning Protective measures
Landuse planning
EIA/SEA

Cost-benefit
assessment

Risk reduction

Figure 4 Risk-management framework.

Table 2 Summary of definitions related to risk management

Term Definition

Risk analysis The use of available information to estimate the risk to individuals or populations, property, or the environment from
hazards. Risk analysis generally contains the following steps: hazard identification, hazard assessment, elements-at-risk/
exposure analysis, vulnerability assessment, and risk estimation.
Risk evaluation The stage at which values and judgements enter the decision process, explicitly or implicitly, by including consideration of
the importance of the estimated risks and the associated social, environmental, and economic consequences, in order to
identify a range of alternatives for managing the risks.
Risk assessment The process of risk analysis and risk evaluation.
Risk control or risk The process of decision making for managing risks, and the implementation, or enforcement of risk-mitigation measures
treatment and the re-evaluation of its effectiveness from time to time, using the results of risk assessment as one input.
Risk management The complete process of risk assessment and risk control (or risk treatment).

Source: Based on UN-ISDR, 2004. Terminology of disaster risk reduction. United Nations, International Strategy for Disaster Reduction, Geneva, Switzerland. http://www.unisdr.org/
eng/library/lib-terminology-eng%20home.htm

disaster risk, should be achieved by combining structural and hazards are spatially distributed, as well as the vulnerable
non-structural measures that focuses on emergency pre- elements-at-risk.
paredness (e.g., awareness raising, early-warning systems, etc.),
inclusion of risk information in long term land-use planning,
3.10.2.3 Risk-Analysis Framework
and evaluation of the most cost-effective risk-reduction
measures (see Figure 4). In the entire risk-management As illustrated in Figure 5, there are three important com-
framework, spatial information plays a crucial role, as the ponents in risk analysis: 1) hazards; 2) vulnerability; and 3)
 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management 271

Risk = Probability of losses occurring

Risk = Hazard * Vulnerability * Amount

Temporal Consequences or
= probability * losses

Temporal Degree of loss to Quantification of

= *
probability elements at risk * elements at risk

Vulnerability
Hazard Elements at
risk
Temporal probability Vulnerability
function Type of elements
Duration
Type 1 at risk
Time of onset % Damage
Type of hazard Temporal variation
Type 2 quantification
Hazard intensity
• Number
Spatial extent • Economic value
• Initiation Harard
• Spreading/runout Location
intensity

Exposure:
Spatial overlay
of hazard and
elements at
risk

Figure 5 Risk analysis and its components.

elements-at-risk (Van Westen et al., 2008). They are charac-
terized by both spatial and non-spatial attributes. Hazards are
characterized by their temporal probability and intensity, de-
rived from frequency-magnitude analysis. Intensity expresses
the severity of the hazard, for example, water depth, flow
velocity, and duration in the case of flooding. The hazard
component in the equation actually refers to the probability of
occurrence of a hazardous phenomenon with a given intensity
within a specified period of time (e.g., annual probability).
Hazards also have an important spatial component, both re-
lated to the initiation of the hazard (e.g., a volcano) and the
spreading of the hazardous phenomena (e.g., the areas af-
fected by volcanic products such as lava flows) (Van Westen,
2009).
Elements-at-risk or ‘‘assets’’ are the population, properties,
economic activities, including public services, or any other
defined values exposed to hazards in a given area (UN-ISDR,
2004). Elements-at-risk also have spatial and non-spatial
characteristics. There are many different types of elements-at-
risk and they can be classified in various ways. The way in
which the amount of the elements-at-risk is characterized (e.g.,
as number of buildings, number of people, economic value, or
qualitative rating according to their importance) also defines
the way in which the risk is presented. The interaction of
elements-at-risk and hazard defines the exposure and the
vulnerability of the elements-at-risk. Exposure indicates the
degree to which the elements-at-risk are actually located in an
area affected by a particular hazard. The spatial interaction
between the elements-at-risk and the hazard footprints are
depicted in a GIS by spatial overlay of the hazard map with the
elements-at-risk map (Van Westen, 2009).
Vulnerability refers to the conditions determined by phys-
ical, social, economic, and environmental factors or processes,
which increase the susceptibility of a community to the impact
of hazards (UN-ISDR, 2004). The vulnerability of communities
and households can be analysed in a holistic qualitative man-
ner using a large number of criteria that characterize the
physical, social, economic, and environmental vulnerability.
The importance of each of these indicators is evaluated by as-
signing weights and combining them using spatial multi-
criteria evaluation. Physical vulnerability is evaluated as the
interaction between the intensity of the hazard and the type of
element-at-risk, making use of so-called vulnerability curves.
For further explanations on hazard and risk assessment, see
Alexander (1993), Okuyama and Chang (2004), Smith and
Petley (2008), and Alcantara-Ayala and Goudie (2010).
3.10.3 Hazard Assessment
Figure 6 presents a schematic overview of a set of
natural hazards and their cause–effect relationships. The
272 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management
Volcanic
Earthquakes
eruptions
Lava Fault
flows rupture
Pyroclastic
Tsunamis
flows
Ground
Ash fall
shaking
Lahars
movements
avalanches
Figure 6 Examples of multi-hazards and their interactions required for multi-hazard risk assessment. Partly based on CAPRA, 2009. Central
American Probabilistic Risk Assessment (CAPRA). World Bank. www.ecapra.org
upper row in the figure consists of the triggering events,
which are the endogenic or exogenic events that precipitate
other events. They may cause direct effects, such as ground
shaking resulting from an earthquake (Jimenez et al., 2000),
drought caused by deficiency in precipitation (Karnieli
and Dall’Olmo, 2003), pyroclastic flows and ash fall following
a volcanic eruption (Zuccaro et al., 2008), or wind speeds
caused by tropical cyclones (Holland, 1980; Emanuel et al.,
2006). The direct effects may trigger indirect effects, or
secondary hazards, such as landslides caused by ground
shaking in mountainous areas (Jibson et al., 1998), landslides
and floods occurring in recently burned areas (Cannon
et al., 2008), or tsunamis caused by earthquake-induced sur-
face displacement in the sea (Priest et al., 2001; Ioualalen
et al., 2007). Secondary hazards that are caused by other
hazards are also referred to as concatenated hazards or
cascading hazards. Figure 6 aims to depict the interrelation-
ships between the triggering factors, the primary hazards
and secondary hazards. These relationships can be very com-
plex, for instance, the occurrence of floods as a result of
the breaking of earthquake-induced landslide dams (Korup,
2002). Given this complexity, a multi-hazard assessment,
which forms the basis for subsequent risk assessment, should
always lead to some sort of simplification in terms of the
cause-effect relationships.
There are relatively few examples in literature on such
complete multi-hazard assessments, and most studies focus
on the evaluation of individual hazard types. Some of the
best examples of a multi-hazard assessment approach
and subsequent risk assessment will be discussed later in
Section 3.10.5.
Meteorological
extremes
Wind Rainfall Temperature
storms extremes extremes
Strong
Drought Cold wave
winds
Storm Precipitat
Heat wave
surge ion
Flooding Forest fires
Mass
Soil erosion
Snow Land
degradation
3.10.3.1 Scale and Hazard Assessment
Hazard assessment using GIS can be carried out at different
geographical scales. Although it is possible to use a range of
spatial resolutions of the input data for GIS analysis (meas-
urement scale), in practice the geographic scale determines the
size of the study area. This in turn restricts the scale of the
input data and the computational scale of analysis. The geo-
graphical scale also effectively determines the cartographic
scale at which information will be mapped. A number of
factors play a role in deciding what is the appropriate scale for
hazard and risk assessment (Fell et al., 2008; Van Westen et al.,
2008). These are related first of all to the aim of the hazard
assessment, the type of hazard, and the operational scale at
which these hazard processes are triggered and manifest
themselves. They also relate to the size and characteristics of
the study area, the available data and resources, and the re-
quired accuracy. Table 3 provides an overview of scales and
levels for different hazard types.
Hazard assessments that are carried out for the entire
Earth (global scale) focus on global problems, such as climate
change, or are aimed at displaying the distributions of a
particular hazard worldwide (e.g., land degradation). Risk
assessment at this scale is mainly intended to generate
risk indices for individual countries, to link them to indices
related to socio-economic development, and to make prior-
itizations for support by international organizations, such as
the World Bank, ADB, WHO, UNDP, FAO, etc. (Cardona,
2005; Peduzzi et al., 2009). The input data have a scale less
than 1:10 million and spatial resolutions on the order of
1–5 km. Under the umbrella of the ProVention Consortium
 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management 273

Table 3 Scales/levels for hazard assessment, with indication of basic mapping units and the optimal scale for displaying different types of
hazards (EQ¼Earthquakes, VO¼Volcanic hazards, DR¼Drought, WS¼Windstorms, FL¼Floods, CO ¼Coastal, LS ¼Landslides, WF ¼Wildfire).
Indicated is the applicability ( ¼highly applicable,  ¼ moderately applicable, and  ¼ Less applicable)

Scale Level Cartographic Spatial Area covered EQ VO DR WS FL CO LS WF

scale resolution (km2)
(million)

Global Global o1:5 1–5 km 148 million        

Very small Continental / large 1–5 1 5–20 million        
countries
Small National 0.1–1 0.1–1 km 30–600        
thousand
Regional Provincial 0.05–0.1 100 m 1000–10 000        
Medium Municipal 0.025–0.05 10 m 100        
Large Community 40.025 1–5 m 10        

staff from the Hazard Management Unit of the World Bank,
the Development Economics Research Group (DECRG) and
Columbia University carried out a global-scale multi-hazard
risk analysis that focused on identifying key ‘‘hotspots’’ where
the risks of natural disasters are particularly high (Dilley et al.,
2005). The project resulted in a series of global hazard and
risk maps, which can be downloaded from a website (CIESIN,
2005).
For individual continents or regions covering several
countries, hazard applications focus on understanding the
triggering mechanism(s) of hazards that cover vast areas (e.g.,
millions of km2) including tropical cyclones, earthquakes, or
drought. Applications also include evaluation of hazards
crossing national boundaries (e.g., flood hazard in large
catchments like the Rhine, Ganges, etc.), or those critical to
reduction policies at the international level (e.g., for the entire
European Union). The hazard maps are generated using
standardized methodologies, and used for risk assessment,
early warning (De Roo et al., 2007), and post-disaster damage
assessment. The geographic areas that are evaluated vary in
size, as some countries like China, India, or the USA are as
large as continents like Europe, but under one administrative
setup. The scale of the input maps can range between
1:100 000 and 1:5 million, and spatial resolutions may vary
from 90 m to 1 km, depending on the application.
Hazard and risk assessment at the national scale cover
areas ranging from tens to several hundred thousand km2,
depending on the size of the country. Hazard assessment
is carried out at a national scale for national planning pur-
poses, implementation of national disaster-risk reduction
policies, early-warning systems, disaster preparedness, and
insurance. The applications in planning become more con-
crete when zooming in on larger scales such as the provincial
level. For instance, hazard and risk assessment become
an integral component of regional development plans and
environmental impact assessments for infrastructure devel-
opments. At the municipal level, hazard and risk assessment
are carried out as a basis for land-use zoning and for the de-
sign of non-structural risk-reduction measures. At a com-
munity level, hazard and risk assessment are carried out in
participation with local communities and local authorities, as
a means to obtain commitment for disaster-risk reduction
programmes.
3.10.3.2 Spatial Data for Hazard Assessment
The assessment of multi-hazards and the subsequent risk as-
sessment is a very data intensive procedure. The availability of
certain types of spatio-temporal data can be one of the main
limitations for carrying out specific types of analyses. Table 4
provides a schematic overview of the main GIS data layers
required for hazard and risk assessment for different hazard
types. These can be subdivided into hazard inventory data and
triggering events, and environmental factors.
3.10.3.2.1 Hazard inventories and triggering events
The hazard-inventory data are by far the most important, as
they should provide insight into the distribution of past haz-
ardous phenomena, their types, mechanisms, causal factors,
frequency of occurrence, intensities, and the damage that has
been caused. The most straightforward way of generating
hazard inventories is through direct measurements of the
phenomena. Sensor-station networks are used to collect such
measurements (e.g., earthquake strong-motion data, flood-
discharge stations, meteorological stations, coastal-tide gau-
ging stations, or wave-measurement buoys).
Seismic networks have been formed globally (NERIES,
2009; ANSS, 2009; GSN, 2009), and the data can be managed
centrally and distributed using web-mapping applications
(e.g., the United States Geological Survey (USGS)). In the US,
a similar network has been established for recording stream-
discharge data for nearly 10 000 sites in a central database
linked with a web-mapping service (NWIS, 2010). Although a
tsunami warning system has been operational in the Pacific
Ocean for a number of decades, the 2004 Indian Ocean tsu-
nami has prompted the international community to imple-
ment such systems worldwide. For these sensor networks, the
spatial coverage is important so that potentially hazardous
areas are monitored. The density of observations required for
the monitoring networks differs strongly for various hazard
types. This is more problematic for flood-discharge stations, as
each potential hazardous river needs to be monitored,
whereas for seismic stations, the required density can be much
less. Also, the spacing between the individual stations is of
importance given the variability of specific parameters (e.g.,
rainfall varies significantly over relatively short distances in
mountain regions). The period for which measurements are
274 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management

Table 4 Overview of spatial data for hazard assessment, and their relevance for different types of hazards ( ¼highly relevant,
 ¼moderately relevant, and  ¼less relevant). (EQ ¼Earthquakes, VO ¼Volcanic hazards, DR¼Drought, WS¼Windstorms, FL¼Floods,
CO ¼Coastal, LS¼Landslides, WF ¼Wildfire)

Group Data layer and types EQ VO DR WS FL CO LS WF

Hazard inventories Satellite-based monitoring        

Ground-based networks        
Archive studies        
Visual image interpretation        
Field mapping        
Participatory approaches        
Dating methods        
Topography Relief        
Altitude difference (in time)        
Slope steepness        
Slope direction        
Flow accumulation        
Geology Rock types        
Weathering        
Faults        
Structural geology        
Soils Soil types        
Soil depth        
Geotechnical properties        
Hydrological properties        
Hydrology Discharge        
Ground water tables        
Soil moisture        
Runoff        
Geomorphology Physiographic units        
Origin/genesis        
Landforms        
Active processes        
Land use Natural vegetation        
Land use        
Vegetation changes        
Land-use changes        
Linear infrastructures        
Built-up areas        
Triggering factors Rainfall        
Temperature        
Wind speed & direction        
Wave height        
Tides        
Earthquakes        
Volcanic eruptions        

available, and the spatio-temporal continuity of the meas-
urements, play an important role, as generally the period for
which measurements are available is not sufficiently large to
capture major events from the past. Catalogs from the meas-
urement networks should be carefully analysed before being
used in a hazard assessment. The monitoring networks located
on the ground or in the oceans are supported by a number of
satellite systems that are used for transmitting information
to data centres. There are also a large variety of satellite-based
monitoring systems that can measure characteristics of
hazards over larger areas on a regular basis, such as sea-
surface temperature, rainfall, altitude, clouds, and vegetation
indices.
For larger areas, lack of data from meteorological stations
can be substituted with rainfall estimates from satellite
imagery, such as from the Tropical Rainfall Measuring Mission
(TRMM) and Multi-Satellite Precipitation Analysis (TMPA),
which are used to issue landslide and flood warnings based on
a threshold value derived from earlier published intensity-
duration-frequency relationships for different countries (Hong
et al., 2007b). As another example, GEONETCast is a global
network of satellite-based data dissemination systems pro-
viding environmental data to the world-wide user community.
Products include meteorological satellites (Meteosat, GOES,
and FengYun), and vegetation monitoring using SPOT-Vege-
tation data. This information is made available to many users,
with low cost receiving stations and open-source software
(Mannaerts et al., 2009). Another example is the Sentinel Asia
program that is an initiative supported by JAXA and the
APRSAF (Asia-Pacific Regional Space Agency Forum) to share
 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management
disaster information in the Asia-Pacific region using a Digital
Asia (Web-GIS) platform, and to make the best use of Earth
observation satellites data for disaster management in the
Asia-Pacific region (Sentinel Asia, 2010).
An important initiative that is focused on the provision of
space-based information for disaster response is the inter-
national charter ‘‘Space and Major Disasters’’ (Disaster Char-
ter, 2010). A number of organizations and program are
involved in rapid mapping activities after major disasters in-
cluding UNOSAT (2010), DLR-ZKI (2010), SERTIT (2010),
GDACS (2010), and the Dartmouth Flood Observatory (2010).
In Europe, the Global Monitoring for Environment and Se-
curity (GMES) initiative of the European Commission and the
European Space Agency (ESA) is actively supporting the use of
satellite technology in disaster management, with projects
such as PREVIEW (Prevention, Information and Early Warning
pre-operational services to support the management of risks),
LIMES (Land and Sea Integrated Monitoring for Environment
and Security), GMOSS (Global Monitoring for Security and
Stability), SAFER (Services and Applications For Emergency
Response), and G-MOSAIC (GMES services for Management
of Operations, Situation Awareness and Intelligence for re-
gional Crises) (GMES, 2010). The United Nations Platform for
Space-based Information for Disaster Management and
Emergency Response (UN-SPIDER, 2010) has been established
by the UN to ensure that all countries have access to and
develop the capacity to use space-based information to sup-
port the disaster management cycle. They are working on a
space application matrix that will provide the satellite-based
approaches for each type of hazard and each phase of the
disaster management cycle. Overviews on the use of space-
based information in hazard inventory assessment can be
found in CEOS (2003), Tralli et al. (2005), IGOS (2007), and
Joyce et al. (2009).
For a number of hazards, satellite-based information is the
major source for generating hazard inventories and hazard
monitoring (e.g., tropical cyclones, forest fires, and drought).
For other hazards, such data support ground-based measure-
ments (e.g., earthquakes, volcanic eruptions, and coastal haz-
ards). Some hazard types cannot be recorded by a network of
measurement stations, as these do not have specific measurable
characteristics (e.g., landslides, forest fires, and snow ava-
lanches). There are also many areas where recorded infor-
mation is not available. Thus, the identification of hazardous
phenomena may require techniques such as automatic classi-
fication or expert visual interpretation of remotely-sensed data.
Automated classification methods make use of reflectance
variations in different parts of the electromagnetic spectrum,
and by data collected from active microwave and light de-
tection and ranging (LiDAR) sensors. For flooding appli-
cations, Earth-observation satellites can be used for mapping
historical events and sequential inundation phases, including
duration, depth of inundation, and direction of current
(Smith, 1997). Geomorphological information can be ob-
tained using optical (LANDSAT, SPOT, IRS, and ASTER) and
microwave (ERS, RADARSAT, ENVISAT, and PALSAR) data
(Marcus and Fonstad, 2008). The use of optical satellite data is
commonly hampered by the presence of clouds, and hazard
mapping is also hampered in areas with vegetation cover. The
use of synthetic-aperture radar (SAR) data are therefore better
275
for mapping hazard events such as floods (Schumann et al.,
2007).
Mapping of forest fires with satellite information is done
by mapping the fires themselves using thermal sensors (Giglio
and Kendall, 2001), or through the mapping of burnt areas
using MODIS or AVHRR, which have a high temporal reso-
lution (Trigg et al., 2005), or with SAR (Bourgeau-Chavez and
Kasischke, 2002).
To facilitate visual interpretation of hazard phenomena
that cannot be automatically obtained from satellite images
(e.g., landslides) and for many geomorphological interpret-
ations of hilly and mountainous areas, stereoscopic imagery
with a high to very-high resolution is required (Metternicht
et al., 2005). Very-high resolution imagery (QuickBird, IKO-
NOS, WorldView, GeoEye, SPOT-5, Resourcesat, Cartosat,
Formosat, and ALOS-PRISM) have currently become the best
option for visual mapping from satellite imagery, and the
number of operational sensors with similar characteristics is
growing year by year as more countries are launching Earth-
observation satellites with stereo capabilities and spatial
resolutions of 3 m or better. The high costs may still be a
limitation for obtaining high-resolution images for particular
study areas, especially for multiple dates after the occurrence
of main triggering events such as tropical storms or cyclones.
Automatic classification of landslides using digital airphotos
and very-high resolution satellite images has been applied
successfully by Hervas et al. (2003), Barlow et al. (2006), and
Martha et al. (2010).
Hazard inventory databases should contain information
for extended periods of time so that magnitude/frequency
relationships can be analysed. This requires the inclusion of
high-frequency/low-magnitude events for estimating hazards
with a high probability of occurrence, but should also contain
sufficient low-frequency/high-magnitude events to evaluate
the hazard for extreme events as well. Therefore, apart from
measuring, observing, and mapping recent hazard events, it is
very important to carry out extensive archive studies. For ex-
ample, one of the most comprehensive projects for landslide
and flood inventory mapping has been the AVI project in Italy
(Guzzetti et al., 1994). Another example is from China where
an analysis was made on extreme precipitation events based
on datasets derived from Chinese historical documents over
eastern China for the past 1500 years (Zheng et al., 2006).
Hazard inventories can also be produced using participatory
mapping and participatory GIS (PGIS). PGIS involves com-
munities in the production of spatial data and spatial de-
cision-making. For example, local people can interpret the
outputs from a GIS or contribute by integrating participatory
mapping of hazardous events to modify or update infor-
mation in a GIS database. Capturing local knowledge and
combining it with other spatial information is a central ob-
jective. This process may assist communities to look at their
environment and explore alternative scenarios based on
understanding of their own goals, constraints, and preferences
(McCall, 2003; Peters Guarin et al., 2005).
The data and techniques described above are intended to
support the generation of hazard inventory databases. Such
databases may have a very large degree of uncertainty, which
can be related to the incompleteness of historical information
with respect to the exact location, time of occurrence, and type
276 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management
Table 5 Global data sources for inventory of hazardous events and hazard assessment used in the PREVIEW project
Hazard type Historic events
Cyclones UNEP/GRID-Europe, based on various raw data sources
Cyclones storm surges UNEP/GRID-Europe, based on Cyclones – winds data
Droughts UNEP/GRID-Europe based on Climate Research Unit (CRU)
precipitation data
Earthquakes United States Geological Survey (USGS) ShakeMap Atlas
Fires European Space Agency (ESA-ESRIN) and World Fires Atlas Program
(ATSR).
Floods Dartmouth Flood Observatory (DFO).
Tsunamis National Geophysical Data Center (NGDC) Tsunami database, NOAA
Volcanoes Smithsonian Institution Volcanoes of the world
Landslides Not available
Source: Reproduced from UNEP/DEWA/GRID, 2010. PREVIEW Global Risk Data Platform. United Nations Environment Programme, Global Resource Information Database, Europe,
Geneva. http://www.grid.unep.ch/activities/earlywarning/preview/index.php
of hazard. Table 5 lists a number of sources for global-hazard
inventories that have been used in the PREVIEW project
(Peduzzi et al., 2009).
3.10.3.2.2 Environmental factors
The environmental factors are a collection of GIS data layers
that are expected to have an effect on the occurrence of the
hazardous phenomena, and can be utilized as causal factors in
the prediction of future events (given numerous assumptions).
The list of environmental factors in Table 4 is not exhaustive,
and it is important to select those factors that are related to a
specific type of hazard in each particular environment. They
represent, however, an example of data types related to
topography, geology, soils, hydrology, geomorphology, and
land use. The basic data can be subdivided into those that are
more or less static, and those that are dynamic and need to be
updated regularly. Examples of static data sets are related to
geology, soil types, geomorphology, and topography. The time
frame for the updating of dynamic data may range from hours
to days (e.g., meteorological data and its effect on hydrology),
to months and years for land-cover and land-use data. Land-
use information should be evaluated with care, as this is both
an environmental factor, which determines the occurrence of
new events (such as forest fires, landslides, and soil erosion),
as well as an element-at-risk, which may be affected by the
hazards. Table 4 provides an indication on the relevance of
these factors for hazard assessment for different types of haz-
ards (Van Westen, 2009).
As topography is one of the major factors in most types of
hazard analysis, the generation of a DEM and geomorpho-
metric analysis plays a critical role. Elevation data can be ob-
tained through a variety of techniques, such as digitizing
contours from existing topographic maps, topographic level-
ling, EDM (electronic distance measurement), differential
Global Positioning System (GPS) measurements, digital
photogrammetry, Interferometric Synthetic-Aperture Radar
(InSAR), and LiDAR. Data source selection depends on a
variety of factors that include data availability, price, and ap-
plication. Many topographic parameters can be produced
from DEMs using fairly simple GIS operations. For more
Hazards
UNEP/GRID-Europe
UNEP/GRID-Europe
International Research Institute for Climate
Prediction (IRI), Columbia University
UNEP/GRID-Europe, USGS, and GSHAP
(Global Seismic Hazard Assessment Project)
IONA Fire Atlas
UNEP/GRID-Europe
Norwegian Geotechnical Institute (NGI)
Hotspots project, International Centre for
Geohazards (ICG/NGI)
details regarding digital terrain modeling and geomorpho-
metry, see these chapters in this volume.
The main sources for global DEMs used in hazard and risk
analysis are GTOPO30 (USGS, 1997; Hastings and Dunbar,
1998) and the Shuttle Radar Topographic Mission (SRTM; Farr
and Kobrick, 2000; CGIAR-CSI, 2008). The relatively low
vertical accuracy of SRTM data (Falorni et al., 2005) does not
make it suitable for large-scale hazard assessments. It is ex-
tensively used, however, for many small-scale applications in
areas where other sources of DEM are not available, such as in
tsunami-hazard assessment (Blumberg et al., 2005). ASTER-
derived DEMs are also commonly used in hazard assessments
for (parts of) countries in the absence of more detailed data
(Fujisada et al., 2005). For smaller areas, the best option is to
generate DEMs from very-high resolution images (e.g.,
Quickbird, IKONOS, ALOS PRISM, and Cartosat).
Microwave (InSAR) data can also be used for the gener-
ation of DEMs, but in practice it is mostly used for detecting
changes in topographic heights related to different hazardous
geological processes, such as land subsidence, slow-moving
landslides, tectonic motions, ice movement, and volcanic ac-
tivity (Massonnet and Feigl, 1998; Ferretti et al., 2001; Hilley
et al., 2004; Salvi et al., 2004; Bürgmann et al., 2006). Multi-
temporal InSAR analyses using techniques such as the per-
manent scatterers (PSInSAR; Ferretti et al., 2001) can be used
to measure displacement of permanent scatterers (e.g., build-
ings) with millimetre accuracy, and allow the reconstruction of
the deformation history (Farina et al., 2008).
For detailed measurement of displacements, networks of
differential global positioning system (DGPS) receivers at
fixed points are used extensively for mapping strain rates,
tectonic plate movements (Vigni et al., 2005), volcanic
movements (Bonforte and Puglisi, 2003), and landslides (Gili
et al., 2000).
More detailed DEMs derived using LiDAR are used exten-
sively for geomorphologic mapping and terrain classification
(Asselen and Seijmonsbergen, 2006). Airborne LiDAR data
can be applied to glacial hazards (Favey et al., 2002), coastal
hazards (Miller et al., 2008), flood modeling (Cobby et al.,
2001; French, 2003), and landslide-hazard assessment
(Haugerud et al., 2003). Multi-temporal LiDAR can also be
 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management
used to model changes and quantify rates of active fluvial
processes such as river-bank erosion (Thoma et al., 2005).
Topographic information is used for heuristic hazard an-
alysis at small scales (e.g., hillshading images, physiographic
classification, internal relief, and drainage density), regional
scales (e.g., altitude zones, slope gradient, slope direction,
contributing area, plan curvature, profile curvature, and slope
length), and in physically-based modeling at local scales (e.g.,
local drainage direction, flow path, and slope gradient), and
spread modeling (Moore et al., 2001). The use of slope-gradient
maps in hazard assessment is greatly affected by the resolution
of the DEM (Zhou and Liu, 2004). As a general rule of thumb,
the use of slope-gradient maps is not advisable for small-scale
studies (Van Westen et al., 2008), although some have used
1 km resolution DEMs to calculate slope-angle distribution
(Hong et al., 2007a). In larger-scale studies, slope maps and
other DEM derivatives can be used as input factors for heuristic
or statistical analysis. In local-scale hazard assessment, DEMs
are used in hydrology modeling and slope maps are used for
physics-based modeling (Kuriakose et al., 2009a).
Geological maps represent a standard information com-
ponent in the hazard assessment of many hazard types (e.g.,
seismic, volcanic, landslides, and soil erosion). A geological
map of the world (CGMW) was developed in 2009 with maps
at the 1:5 million and 1:25 million scales. OneGeology is an
international initiative of the geological surveys of the world,
launched in 2007 as a contribution to the International Year of
Planet Earth, with the aim to create a web-based geological
map of the world (OneGeology, 2010). Digital geological
maps of chronostratigraphy, lithostratigraphy, faults, tectonic
lineaments, tectonic units, and other themes are available
online with scales ranging from 1:250 000 (for certain coun-
tries) to 1:50 million. For individual countries, geological in-
formation is generally digitally available at much larger scales.
For example, through the web-portal of the USGS, scanned
geological maps, as well as GIS data that can be downloaded
(USGS, 2010). The subdivision of geological formations into
meaningful units for hazard assessment is generally prob-
lematic at small scales of analysis. In detailed hazard studies,
specific engineering geological maps are collected and rock
types are characterized using field tests and laboratory meas-
urements. For detailed analysis, 3-D geological maps have also
been used, although the amount of outcrop and borehole
information collected will make it difficult to use this infor-
mation on a scale smaller than 1:5000 and its use is restricted
mostly to a site investigation level (e.g., Xie et al., 2003). Apart
from lithological information, structural information is very
important for hazard assessment (e.g., earthquakes, landslides,
and volcanic eruptions). At medium and large scales, attempts
have been made to generate maps indicating dip direction and
dip amount based on field measurements, but the success of
this depends very strongly on the amount of measurements
and the complexity of the geological structure (Günther,
2003).
Soil information is also required for hazard assessment.
This includes soil types with associated geotechnical and
hydrological properties, and soil sequences with depth infor-
mation. These data layers are essential components for any
physical numerical-modeling approach (e.g., earthquake
amplification studies, landslides, and soil erosion). Pedologic
277
soil maps usually portray soil classes based on the upper soil
horizons with rather complicated legends, and are relevant for
soil erosion, drought, and forest-fire hazard assessment. En-
gineering soil maps describe all loose materials on top of the
bedrock and portray classes according to geotechnical char-
acteristics. They are based on outcrops, borehole information,
and geophysical studies. The soil depth is very difficult to map
over large areas, as it may significantly vary over a relatively
small area. Soil thickness can be modeled using an inter-
polation technique that incorporates factors such as land use
and slope (Kuriakose et al., 2009b). Digital soil information is
available worldwide (FAO, 1981), and includes soil-type
classification, clay mineralogy, soil depth, soil-moisture cap-
acity, bulk density, soil compaction, and other variables. This
product is not based on satellite information directly, but
primarily on ground surveys and national databases.
Geomorphological maps are made at various scales to
show land units based on their shape, material, processes, and
genesis (e.g., Klimaszewski, 1982). No generally accepted le-
gend exists for geomorphological maps, and there may be
large variations in the information content based on the ex-
perience of the geomorphologist. An applied geomorpho-
logical mapping working group has been formed as part of the
International Association of Geomomorphologists (IAG) to
set guidelines for geomorphological mapping and develop a
digital atlas of geomorphological maps. Detailed geo-
morphological maps contain a wealth of information, but
require extensive field mapping and are very difficult to con-
vert into digital format (Gustavson et al., 2006). Unfortu-
nately, traditional geomorphological mapping seems to have
nearly disappeared with the developments of digital geospatial
techniques, and relatively few publications on hazard and risk
still focus on it (Carton et al., 2005; Castellanos and Van
Westen, 2007), or replace it by using morphometric infor-
mation. An important new field within geomorphology is the
quantitative analysis of the topography, called geomorpho-
metry or digital terrain analysis, which combines elements of
the Earth sciences, engineering, mathematics, statistics, and
computer science (Pike, 2000; Drăgut- and Blaschke, 2006).
Part of the work focuses on the segmentation of the topo-
graphy into land-surface objects or geomorphological land
units based on morphometric characteristics at multiple scales
(Giles and Franklin, 1998; Miliaresis, 2001), and on the ex-
traction of landform elements (Carrara et al., 1995). Digital
geomorphological maps are available for some parts of the
world including Germany (GMK, 2010), Austria, and New
Zealand (GNS, 2010).
Land cover can be considered as a static factor in some
hazard studies, although most types of hazard assessments
actually focus on the detection of land-cover changes in re-
lation to hazard phenomena. Changes in land cover and land
use resulting from human activities, such as deforestation,
forest logging, road construction, fire, drought, and cultivation
on steep slopes, can have an important impact on hazards.
An example is the evaluation of the effect of logging and
deforestation on landslides (e.g., Furbish and Rice, 1983).
Land-use maps are made on a routine basis from medium
resolution satellite imagery such as LANDSAT, SPOT, ASTER,
and IRS1-D. Other sources for land-cover data with higher
temporal and lower spatial resolution include MODIS
278 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management
(Moderate Resolution Imaging Spectroradiometer; Friedl et al.,
2002), MERIS (Medium Resolution Imaging Spectrometer),
NOAA-AVHRR, Global Imager (GLI), and SPOT-Vegetation
imagery with varying resolutions (250 m–1 km). They are
used on a routine basis for monitoring the global distribution
of land-cover types (e.g., 10-day basis; Cihlar, 2000). Specific
change-detection algorithms that can be used for land-cover
studies are reviewed by Coppin et al. (2004). Seasonal and
inter-annual variations in land cover that may be caused by
natural disasters and land-use changes can be detected using
high temporal frequency satellite data.
Several initiatives have produced global land-cover maps
for different time periods. For example, the CORINE Land
Cover 2000 dataset (CLC2000) has been produced using re-
motely sensed imagery to produce a land-cover database at a
scale of 1:100 000, a positional accuracy of 150 m, and a
minimum mapping unit of 25 ha in Europe, and a resolution
of 1 km globally. A more recent map is the ESA Globcover
global land-cover map based on MERIS data acquired between
mid-2005 and mid-2006 (Arino et al., 2007). For individual
continents, more detailed land-cover information is available
(e.g., Africover, 2010).
Hazard and risk assessments require a multitude of data
from different sources. Therefore, it is important to have a
strategy for data availability. Important information concerns
include data quality, metadata, and multi-user databases.
Many project-specific data sets can be used for various pur-
poses (e.g., resource management and risk assessment).
This requires that the potential users know what data exist and
have access. Spatial-risk information requires the use of
a spatial data infrastructure, where through the internet basic
GIS data can be shared among the different technical
and scientific organizations involved in hazard and risk as-
sessment. A spatial data infrastructure is the foundation
or basic framework (e.g., a system or organizational scheme)
with policies, resources, and structures to make spatial infor-
mation available to decision makers when they need it,
where they need it, and in a form where they can use it (al-
most) immediately. The website where the data are actually
exchanged is called a clearinghouse. A good example of that
is the European ORCHESTRA project (ORCHESTRA, 2009),
which designed and implemented the specifications for a
service-oriented spatial data infrastructure for improved
interoperability among risk-management authorities in
Europe. In the framework of the CAPRA project of the
World Bank (CAPRA, 2009), the GeoNode was developed
as an open-source platform that facilitates the creation, shar-
ing, and collaborative use of geospatial data for risk assess-
ment (GeoNode, 2010). Examples of initiatives that focus
on spatial-data infrastructure for disaster relief include
Reliefweb (2010), Alernet (2010), HEWSweb (2010), and
GDACS (2010).
3.10.3.3 Hazard Assessment Examples
As indicated in Table 3, hazard assessment is carried out at
various scales, and the methods for hazard assessment are
determined by the type of hazard and by the availability of
input data (Table 4). This section discusses the main
approaches for hazard assessment used at various geographic
scales.
3.10.3.3.1 Global hazard assessment
Hazardous events can encompass large areas such as wind-
storms, drought, earthquakes, and tsunamis. Therefore, global
or international hazard assessment is required. For example,
the Global Seismic Hazard Mapping Project (GSHAP, 1999), a
demonstration project of the UN/International Decade of
Natural Disaster Reduction, was conducted in the 1992–1998
period with the goal of improving global standards in seismic-
hazard assessment. The GSHAP produced regional seismic-
hazard maps for most parts of the world that display the
global seismic hazard as peak ground acceleration (PGA) with
a 10% chance of exceedance in 50 years, corresponding to a
return period of 475 years. The procedure involves the iden-
tification of seismo-tectonic zones in which earthquake char-
acteristics were analyzed from historic earthquake databases.
For each point, seismic hazard is then analyzed using modules
(Arnold, 1989).
For windstorms, international databases exist for tropical
cyclones in different parts of the world. For the North Atlantic
region, for example, the HURDAT database (Jarvinen et al.,
1984) contains all historic Hurricane tracks. Windstorm-haz-
ard models generate a set of stochastic events based on his-
torical and modeled windstorm tracks, with parameters on
intensity, size, and shape. For each simulated track, data are
calculated for wind velocity together with associated levels of
storm surge, and rainfall intensities using empirical relations
(Mouton and Nordbeck, 2003). Areas that may inundate due
to tidal changes are mapped using a DEM in coastal zones
(Lavelle et al., 2003). Drought-hazard assessment at an inter-
national level is carried out using monthly average precipi-
tation data (e.g., the Weighted Anomaly of Standardized
Precipitation (WASP)) developed by the International Re-
search Institute for Climate and Society (IRI), computed on a
2.5  2.5 1 grid (Lyon and Barnston, 2005).
For other hazards, such as floods and landslides, infor-
mation at international levels is too general for the estimation
of hazards, as the hazard events are too localized and require
more detailed information. Nadim et al. (2006, 2009) made an
attempt to generate a global landslide-hazard map, making use
of general spatial data sets with global coverage, such as an
SRTM-derived DEM with 1 km spatial resolution, the geological
map of the world at 1:25 million scale, a soil-moisture index,
monthly precipitation data, and the Global Seismic Hazard
Mapping Programme (GSHAP) results. Given the poor reso-
lution of the data as compared to the specific conditions in
which landslides occur, however, the results are only a general
indication of landslide susceptibility. Hong et al. (2007a) pre-
sented a qualitative method for a global landslide-susceptibility
map using GIS-based map overlay techniques, combining sev-
eral layers of different parameters (e.g., elevation, slope, land
use, etc.). Recently, an attempt to provide global-scale landslide
early warnings in near real time using stochastic models com-
bining a global landslide database, TMPA rainfall estimates,
SRTM DEM, and MODIS land-cover products was conducted at
Columbia University, the success of which was mainly limited
by the lack of completeness of the landslide database and the
quality of the rainfall estimates from Tropical Rainfall
 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management
Measuring Mission and Multi-satellite Precipitation Analysis
(TMPA) (Kirschbaum et al., 2009). Global flood-hazard studies
are difficult to carry out, as the DEMs available at global scale
are generally not of sufficient detail for flood modeling appli-
cations. One example of an approach used for flood-hazard
mapping over very large areas is based on an inventory of past
flood events (e.g., from Dartmouth Flood Observatory), cou-
pled with a very simple flood model based on the HYDRO1k
Elevation Derivative Database (USGS, 1996; Verdin and
Greenlee, 1996). HYDRO1k is a geographic database developed
to provide comprehensive and consistent global coverage of
topographically derived data sets, including streams, drainage
basins, and ancillary layers derived from the USGS 30 arc-
second DEM of the world.
At the global scale, few approaches have been carried out
for multi-hazard assessment that attempt to provide general
indicators or risk indices for countries, or for parts of coun-
tries, mainly for comparison of risk levels between countries.
Dilley et al. (2005) developed a methodology for global
hazard and risk assessment for the main hazard types in
Table 3. Peduzzi et al. (2009) presented a model designed for
the United Nations Development Programme as a component
of the Disaster Risk Index (DRI), which aims at monitoring
the evolution of risk. Four hazards (droughts, floods, cyclones,
and earthquakes) were modeled using a GIS based on the
datasets shown in Table 5.
3.10.3.3.2 (Inter)National hazard assessment
Hazard assessment is commonly carried out for individual
continents or countries, as they are related to the same
administrative area and controlled by national or inter-
national governments (e.g., EU, USA, and China). The meth-
odology for hazard and risk assessment is standardized, and
mostly follows established guidelines that are requested by
governments (e.g., the European Floods Directive). The ap-
plications at the (inter)national level are more refined
than those carried out globally, and require higher-resolution
data. For example, the European Flood Directive (EFD) indi-
cated that preliminary flood-risk assessments in Europe
should be completed by 2011, flood hazard and risk maps
should be available by 2013, and flood-management
plans should be completed by 2015 (EFD, 2007). In order to
accomplish these goals, standardized methods, datasets, and
GIS-based tools are to be used for the assessment and moni-
toring of flood risk for the whole of Europe. Flood-hazard
maps are generated based on DEMs with a resolution ranging
between 100 m and 1 km. The hazard factor is estimated by
using hydrological modeling (e.g., LISFLOOD) at different
scales and for many return periods (Barredo, 2007; van der
Knijff et al., 2010). Modeling of extreme precipitation and
resulting river discharge is calculated in real time, and flood
forecasts are made for the whole of Europe. In the USA, the
Federal Emergency Management Agency (FEMA) has estab-
lished a national flood-hazard mapping project with the
Federal Insurance and Mitigation Administration’s Hazard
Mapping Division, through their national flood insurance
program (FEMA, 2010).
Similar initiatives in Europe are in the field of forest fires.
The European Forest Fire Information System (EFFIS) makes a
rapid assessment of the burned areas through a series of daily
279
images from the MODIS instruments on board the TERRA and
AQUA satellites, and displays fires with burned area of ap-
proximately 40 ha or larger from a web-GIS (Ayanza et al.,
2003). A third example that is implemented at both the
European level as well as globally is the MARSOP-3 project on
crop yield forecasting, carried out by the Joint Research Centre
(JRC) of the EC, with other partners. This system includes the
management of a meteorological database, an agro-meteoro-
logical model and database, low-resolution satellite infor-
mation, statistical analyses of data, crop-yield forecasting, and
publishing of bulletins containing analysis, and forecasts and
thematic maps on crop-yield expectations using a Web-GIS
application (Reidsma et al., 2009). An overview on the use of
satellite data for drought monitoring and hazard assessment
can be found in Henricksen and Durkin (1986), Peters et al.
(2002), and White and Walcott (2009). The aforementioned
software tools are used for early warning as well as for hazard
assessment for the whole of Europe.
In the USA, a GIS-based tool for earthquake hazard assess-
ment, ShakeMaps, was developed by the USGS in cooperation
with regional seismic-network operators. ShakeMaps provides
near-real-time maps of ground motion and shaking intensity
after important earthquakes. It can also be used to generate
hazard maps using scenario earthquakes (Wald et al., 1999).
Later, a methodology was developed for modeling of seismic
site conditions using topographic slope as a proxy, using the
SRTM30 database, which provided the average shear-velocity
down to 30 m (Wald et al., 2004). Initiatives to incorporate
open-source software in seismic-hazard assessment have been
taken by OpenSHA (2010) and by the Global Earthquake
Model (GEM, 2010), an international initiative to develop
uniform and open standards and platforms for calculating
earthquake risk worldwide. The GEM brings together all major
players in the earthquake risk assessment field, including part-
ners from the insurance sector, international organizations,
public organizations, and research centers from all over
the world.
In terms of landslide-hazard assessment, this scale is still
too general to be able to map individual landslide phenom-
ena. The analysis of landslide hazards at this scale is still done
by weighting a number of input maps (e.g., Malet et al., 2009;
Castellanos and Van Westen, 2007).
3.10.3.3.3 Provincial and municipal level
At local and municipal scales, spatial information is generally
of sufficient quality to run more sophisticated models, which
can be either empirical (e.g., statistical) or deterministic
(physically-based). A flood-hazard assessment example fol-
lows. The first step is to transform catchment characteristics
like topography, relief, and land cover, complemented with
hydrological boundary conditions, into estimates of the dis-
charge at various locations along the river downstream. This
can be done with (distributed) 1-dimensional models. These
kinds of models are very useful to assess the response of the
river to extreme events and to changes in the topography and
land cover. Typical models to do this are HEC-HMS and
HEC-RAS of the US Army Corps of Engineers, MIKE-SHE
(Refsgaard and Storm, 1995), IHDM (Beven et al., 1987),
LISFLOOD (De Roo et al., 2000), and HEC-RAS (Brunner,
2002). They require the characterization of the terrain through
280 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management
a series of cross-sections perpendicular to the direction of flow
for which the average water depth and flow-velocity are cal-
culated. This type of modeling is typically applied for catch-
ment analysis, and the underlying assumption is that all flow
is parallel to a predefined river-network. In near-flat terrain
with complex topography, it cannot be assumed that all flow
will be parallel to the main river. Also in urban environments
and in areas with a dominant presence of man-made struc-
tures, models are required that calculate flow in both X- and
Y-directions. Such models like SOBEK (Stelling et al., 1998;
Hesselink et al., 2003), Telemac 2D (Hervouet and Van Haren,
1996), and MIKE21 can also be applied in the case of
diverging flow at a dike breach. They require high-quality
DEMs, which ideally are generated using LiDAR data (Dal Cin
et al., 2005; Alkema and Middelkoop, 2005). The flood
modeling is generally carried out at a municipal to provincial
scale for a selected stretch of the river. These models provide
information on how fast the water will flow and how it
propagates through the area. They are very suitable for as-
sessing the effects of the surface topography, like embanked
roads and different land-cover types on the flood behavior
(Stelling et al., 1998).
Furthermore, for landslide-hazard assessment, the provincial
and municipal scales offer many more possibilities, as sufficient
information can be collected on hazard inventories and
the factors that control the location of landslides (Dai et al.,
2002). Dai et al. (2002) differentiated between statistical
methods and physically based models. Guzetti et al. (2005)
provided an overview of the various statistical methods that can
be applied, focusing on the use of multivariate statistical
methods in which landslide inventories for different periods are
used in combination with environmental factors for predicting
landslide activity within slope units that are defined from a
DEM. Van Asch et al. (2007) provided an overview of the
physically based modeling approaches. Most of the physically
based landslide models make use of the infinite-slope model,
and are therefore only applicable to modeling shallow land-
slides. They can be subdivided into static models that do not
include a time component, and dynamic models, which use the
output of one time step as input for the next time step. Phys-
ically-based models for shallow landslides account for the
transient groundwater response of the slopes to rainfall, and the
effect of earthquake acceleration (Van Beek and van Asch,
2004).
The provincial and municipal scales are also the most ap-
propriate for volcanic hazard assessment, as a lot of this work
depends on the determination of the eruptive history on the
basis of geological investigation and age dating (Tilling,
1989). Given different volcanic eruption scenarios, several
modeling techniques can be carried out for the various vol-
canic hazards (ash fall, lava flow, pyroclastic flow, and lahars).
Most of these hazard-assessment methods require some sort of
spread modeling, where the volcanic products are distributed
over the terrain away from the vent. This requires the use of
dynamic models (Zuccaro et al., 2008). The evaluation of
volcanic hazards from tephra fallout is determined by utilizing
volcanic ash volumes, eruption height, and wind information
(Connor et al., 2001). Remote sensing also plays an important
role in volcanic-hazard assessment (e.g., Kerle and Oppen-
heimer, 2002).
3.10.3.3.4 Community level
Approaches based on local knowledge and experiences may be
useful in developing countries, where detailed information
required for conventional model-based risk analyses facili-
tated by GIS is generally not available. For instance, historical
records on river discharges and rainfall are commonly miss-
ing, whereas knowledge about hazardous events is generally
available within the local communities (Ferrier and Haque,
2003). A vast quantity exists of undocumented local
knowledge on disaster occurrences in the field, which usually
remains untapped because of the lack of funding, a format to
systematically collect it, and a low commitment to do so
(Hordijk and Baud, 2006). Anderson and Woodrow (1998)
stated that much of the information needed for risk assess-
ment and mitigation can be obtained from local people who
generally already know what the situation is, but do not al-
ways have the skills for understanding and organizing what
they know. Several organizations, such as the International
Federation of Red Cross and Red Crescent Societies (IFRC),
have developed community-based assessment instruments for
analyzing disaster situations at the grassroots level, and for
improving the community’s expertize in identifying and ar-
ticulating its needs and reducing its vulnerabilities. Some ex-
amples of these community-based methods are named
‘‘Capacity and Vulnerability Assessment (CVA)’’, ‘‘Hazards,
Vulnerability and Capacity Assessment (HVCA)’’, and ‘‘Dam-
age, Needs and Capacity Assessment methods (DNCA)’’
(Provention Consortium, 2010). These methods aim at elicit-
ing tacit local knowledge within communities on historic
disaster events, the perception of hazards, characterization of
elements-at-risk, identifying the main factors of vulnerability,
coping mechanisms, and disaster reduction scenarios. The
application of such collaborative approaches is not common
in many developing countries, and decision-making about risk
is commonly done in a top-down approach by local author-
ities where specialists diagnose problems, formulate alter-
natives, and determine options without a meaningful
consultation with communities (UN-ESCAP, 2003). Hazard
specialists commonly consider that community participation
is difficult to achieve, and the information perceived is
unscientific, not always easy to retrieve, difficult to be ex-
pressed in quantitative terms, or needs to be converted into
spatial formats (Peters Guarin, 2008).
The integration of geo-information systems and local-
community knowledge relevant to hazards, vulnerability, and
risk modeling is still in an initial stage (Maskrey, 1998; Ferrier
and Haque, 2003; Zerger and Smith, 2003). Very often the
sketches, paper maps, historical profiles, and other results
obtained through participatory mapping are not kept after a
risk project has finished, leading to a loss of valuable infor-
mation. As Cannon et al. (2003) advised, these products need
to be converted from raw data into useful spatial information
that allows the community and other participants to develop
analytical processes for risk analysis and exploration of man-
agement alternatives. Several authors have shown that local
communities are indeed the primary sources of information
for flood depths, time of occurrence, severity measured in
terms of damage, and the like (Whitehouse, 2001; Alcantara-
Ayala, 2004; Rautela, 2005). Systematic collection of data from
significant events using public participation can provide a very
 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management
useful component for the development of data-sets that can be
used as input for risk studies at the community level, and as a
basis for risk management and community planning (Ireland,
2001). Information from local communities can also be useful
in calibrating and verifying risk and disaster scenarios (Bassolé
et al., 2001; Peters Guarin, 2008).
3.10.4 Elements-At-Risk and Vulnerability
The next step in risk assessment, after analyzing the hazard, is
to evaluate the elements-at-risk. There are many different types
of elements-at-risk and they can be classified in various ways.
In this section, several types of elements-at-risk and their data
sources are evaluated, followed by a discussion on how these
are used in vulnerability assessment.
3.10.4.1 Elements-At-Risk Information
Elements-at-risk inventories can be carried out at various levels
depending on the requirement of the study. Table 6 provides
a more detailed description. Elements-at-risk data should
be collected for basic spatial units, which may be grid-cells
on a global scale (see Tables 3 and 6), administrative units
(countries, provinces, municipalities, neighbourhoods, and
census tracts), or so-called homogeneous units with similar
characteristics in terms of type and density of elements-at-risk.
Risk can also be analyzed for linear features (e.g., transporta-
tion lines) and specific sites (e.g., a dam site). The risk
assessment will be done for these spatial units of the
elements-at-risk, rather than those used in the hazard assess-
ment. In the HAZUS methodology (FEMA, 2004), the loss
estimation is done based on census tracts.
Digital information on coastlines, international bound-
aries, cities, airports, elevations, roads, railroads, water fea-
tures, and cultural landmarks are available from different
sources such as the Geonetwork established by FAO (2010),
with available data comprising base layers (e.g., boundaries,
roads, and rivers), thematic layers (e.g., protected areas), or a
backdrop image (e.g., World Forest, 2000).
One of the most important spatial attributes of the map-
ping units for an elements-at-risk inventory is land use. The
land use determines to a large extent the type of buildings that
can be expected in the unit, the economic activities that are
carried out, and the density of the population for different
time periods during the day. Land-cover and land-use maps
are prepared by image classification at small scales or through
visual interpretation at larger scales. Ebert et al. (2009) de-
veloped a method using object-oriented image classification
methods for the automatic characterization of land-use types
in urban areas.
3.10.4.1.1 Collaborative mapping and Mobile-GIS
Elements-at-risk information is collected from a wide variety
of sources. There are also many areas in the world for which
no detailed digital data are available on elements-at-risk. In
such situations, data should be digitized from analog maps, or
in case these also do not exist, be mapped in the field, for
instance, using mobile GIS. With the use of mobile GIS, it is
281
possible to directly collect the spatial information based on a
high-resolution image that can be uploaded into a palmtop
computer or smart phone, and link it with attribute infor-
mation that is collected in the field. Some of the most used
tools for mobile GIS in urban elements-at-risk mapping are
ArcPad (Montoya, 2003) and Cybertracker (McCall, 2008).
Several initiatives have been established for collaborative
mapping of topographic features, also referred to as ‘‘crowd-
sourcing’’. For example, OpenStreetMap is a free editable map
of the whole world, which is made using collaborative map-
ping by volunteers. It allows users to collect, view, edit, and
use geographical data in a collaborative way from anywhere
on Earth (OpenStreetMap, 2010). Another crowdsourcing ex-
ample is ‘‘Ushahidi’’ (Ushahidi, 2010), which means ‘‘testi-
mony’’ in Swahili, which was initially developed to map
reports of violence in Kenya in 2008, and which has been used
later in many disaster events to rapidly collect and visualize
spatial information. Other applications that are specifically
directed to post-disaster relief coordination are ‘‘Sahana’’
(Sahana, 2010) and ‘‘Virtual Disaster Viewer’’ (Virtual Disaster
Viewer, 2009). Sahana is a free web-based disaster manage-
ment system developed after the Indian Ocean tsunami as a
collaboration tool that addresses the common coordination
problems during a disaster. The Virtual Disaster Viewer is a
crowd-sourcing tool for collaborative disaster impact and
damage assessment, which has proven to be effective after the
Haiti earthquake in 2010. Hundreds of earthquake and re-
mote-sensing experts were assigned specific areas (tiles) of the
affected areas to review and provide their assessment by
comparing before and after high-resolution satellite images,
which became available on Google EarthTM immediately after
the disaster, and which served as the basis for the collaborative
mapping. Such collaborative-mapping applications might
become a very important tool in the future.
3.10.4.1.2 Population data
People are the most important elements-at-risk, and popu-
lation data exhibit a static and dynamic component. The static
component relates to the number of inhabitants per mapping
unit and their characteristics, whereas the dynamic com-
ponent refers to their activity patterns and their distribution in
space and time. Population distribution can be expressed as
either the absolute number of people per mapping unit or as
population density. The way population data are collected and
represented in a risk assessment depends on the scale of an-
alysis (see Table 3) and the availability of information (Rhind,
1991).
Census data are the obvious source for demographic data.
They are used as benchmark data for studying population
changes, and are key input for making projections concerning
population, households, labour force, and employment.
Census data are costly to collect, and updating of population
information is carried out on average every 10 years. Census
data are aggregated to census tracts, and normally data at an
individual household level are confidential. This is also the
reason why risk assessment is normally carried out at the
census tract level (FEMA, 2004). Census tracts are divisions of
land that are designed to contain 2500–8000 inhabitants with
relatively homogeneous population characteristics, economic
status, and living conditions. Census data may also contain
Table 6 Main elements-at-risk and how they can be spatially represented at various mapping scales

282
Scale

Global Continental National Provincial/Municipal Community

Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management
Basic unit 1 km grid or countries 90–1 km grid and countries 30–90 m and municipality Census tract Groups of buildings
Population Gridded population map Gridded population map By municipality By Census tract People per building
• Population density • Population density • Daytime/Night-time
• Daytime/Night-time • Gender
• Age
• Education, etc.
Buildings N.A. Gridded building density map By municipality By Census tract Building footprints
• Number of Buildings • Generalized use • Detailed use
• Height • Height
• Building types • Building types
• Construction type
• Quality/Age
• Foundation
Transportation N.A. Main roads, railroads, harbours, Road & railway networks, with All transportation networks with All transportation networks with
networks and airports general traffic density detailed classification, including detailed engineering works &
information viaducts etc. & traffic data detailed dynamic traffic data
Lifelines N.A. Main powerlines Only main networks Detailed networks: Detailed networks and related
• Water supply • Water supply facilities:
• Electricity • Waste water • Water supply
• Electricity • Waste water
• Communication • Electricity
• Gas • Communication
• Gas
Essential N.A. By Municipality As points Individual building footprints Individual building footprints
facilities • Number of essential facilities • General characterization • Normal characterization • Detailed characterization
• Buildings as groups • Buildings as groups • Each building separately
Agricultural data Gridded main land cover Gridded maps: By homogeneous unit By cadastral parcel By cadastral parcel, for a given period
types, crops • Crop types • Crop types • Crop types of the year
• Yield information • Yield information • Crop rotation • Crop types
• Yield information • Crop rotation and time
• Agricultural buildings • Yield information
Ecological data Main land cover types, Natural protected areas with Natural protected area with General flora and fauna data per Detailed flora and fauna data per
crops international approval national relevance cadastral parcel. cadastral parcel
Economic data GDP By region: Economic production, By Municipality By Mapping unit By household
import/export, type of economic • Economic production • Employment rate • Employment
activities • Import/export • Socio-economic level • Income
• Type of economic activities • Main income types plus larger • Type of business plus larger
scale data scale data

Source: Adapted from Van Westen, C.J., Castellanos Abella, E.A., Sekhar, L.K., 2008. Spatial data for landslide susceptibility, hazards and vulnerability assessment: an overview. Engineering Geology 102(3–4), 112–131.
 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management
other relevant characteristics that are used in risk assessment,
such as information on age, gender, income, education, and
migration.
For larger areas, census data may be aggregated into larger
administrative units. For large parts of the world, however,
census data are not available, out dated, or unreliable.
Therefore, other approaches have been used to model popu-
lation distribution with remote sensing and GIS, based on
a number of factors, such as land cover, roads, slopes, and
night-time illumination. The use of remote-sensing data in
combination with other data to redistribute population in-
formation over smaller areas based on general population
data for large administrative units is also referred to as
‘‘dasymetric mapping’’ (Balk et al., 2006). Global population
data are available from the LandScan Global Population
Database (Bhaduri et al., 2007; LandScan, 2010) that provide
the average population over 24 h in a 1 km resolution grid.
The Global Rural-Urban Mapping Project (GRUMP) is an-
other example of modeling human populations in a common
geo-referenced framework (GRUMP, 2004), as is the African
Population Database (APD, 2010). Higher-resolution popu-
lation databases have also been developed for specific areas,
especially in low-income countries where limited information
is available and there is a need to generate population infor-
mation using satellite data. Tatem et al. (2007) made a com-
parison between semi-automated population distribution
mapping for several countries in East Africa, based on 30 m
Landsat ETM data, and concluded that these produced more
accurate results than existing products at a cost of $0.01 per km2.
For risk assessment at municipal or community level,
population is required at a high spatial resolution for every
census tract or even for each building. In the absence of census
data, static-population information can be derived directly
using high-resolution satellite imagery (e.g., Harvey, 2002) or
through a building-footprint map, where the land-use type
and the floor-space are used to estimate the number of people
present in a particular building (Chen et al., 2004; Lwin and
Murayama, 2009).
3.10.4.1.3 Building data
After population, buildings are the second most important
group of elements-at-risk. They house the population, and the
‘‘behaviour’’ of a building under a hazard event determines
whether the people in the building might be injured or killed.
In order to assess the potential losses and degree of damage of
buildings, it is important to analyze the type of negative effects
that the event might have on the building exposed to it, and
the characteristics of the building. The negative effects of
hazardous events on buildings can be classified into a number
of groups depending on the type of hazard (Blong, 2003;
Hollenstein, 2005). Figure 7 depicts a schematic overview of
the various hazard processes that may occur and that have a
different effect on buildings. For instance, a building may
be impacted by a mass, and the damaging effects would be
determined by the volume of the mass, speed of impact, and
the medium, such as rocks, soil, debris, snow, water, or air.
Buildings are also affected by undercutting (erosion or
landslides), shaking (earthquakes), inundation, fires, loss-of-
support (subsidence), gasses, or loading (e.g., volcanic ashes).
In each of these situations, particular building characteristics
283
are important for evaluating the damaging effects, such as
structural type, construction materials, application of building
code, age, maintenance, roof type, height, floor space, volume,
shape, proximity to other buildings, proximity to hazard
source, proximity to vegetation, and openings (FEMA, 2004;
Jones et al., 2005; Grünthal et al., 2006; Douglas, 2007).
For risk maps that express losses in economic terms, an
estimation of building costs is also required. Several sources of
information can be used, such as data on house prices from
real-estate agencies, information from cadastres that indicate
the value used as the basis for taxation, engineering societies
that calculate the replacement costs, or insurance companies
(Grünthal et al., 2006). It is often difficult to get hold of the
building values used by the cadastres, whereas it is easier to
use the values from real-estate agencies. Samples are taken
from each type of building in the various land-use classes. In
some countries, building societies produce a monthly index
that permits an update of property prices. Cost estimation can
be based on using the replacement or market value. Apart
from building costs, content costs are also very relevant, es-
pecially for those hazards that have less structural damage
such as flooding.
Building information can be obtained in several ways.
Idealy, data are available on the number and types of buildings
per mapping unit, or even in the form of building-footprint
maps. If such data are not available, building-footprint maps
can be generated using screen digitizing from high-resolution
images (Van Westen et al., 2002). Automated building mapping
has also been carried out using high-resolution satellite images
(Fraser et al., 2002), InSAR (Stilla et al., 2003), and specifically
using LiDAR (Priestnall et al., 2000; Brenner, 2005; Oude
Elberink and Vosselman, 2009). LiDAR data also allow the
extraction of other relevant features and the calculation of
shapes, building height, and volumes that are needed in risk
assessment.
3.10.4.2 Vulnerability
Vulnerability is the most complicated component of risk
assessment, as indicated in Figure 5, because the concept
of vulnerability has a wide range of interpretations. The
concept originated from the social sciences in response to
the pure hazard-oriented perception of disaster risk in the
1970s. Since that time, different disciplines have developed
their own concepts. Multiple definitions and different con-
ceptual frameworks of vulnerability exist (e.g., Blaikie et al.,
1994; Pelling, 2003). For more details, see the overview
by Birkmann (2006). The definition of vulnerability (used in
Table 1) indicates that vulnerability can be viewed as multi-
dimensional (physical, social, economic, environmental, in-
stitutional, and human factors define vulnerability), dynamic
(it changes over time), scale-dependent (it can be expressed at
different scales from individuals to countries), and site-specific
(each location might need its own approach) (Bankoff
et al., 2003). Risk-assessment methods can be differentiated
based on quantitative and qualitative approaches. Figure 8
presents a framework for multi-hazard risk assessment that
will be subsequently explained. It is important to note that
quantitative methods focus mostly on physical vulnerability,
284 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management

Impact by mass
Snow
Rock fall avalanche Landslide

Airplane crash Lava flow Pyroclastic flow

impact
Wind

Tornado Cyclone Explosion

cutting

Shaking
Under

Erosion Landslide Earth quake

Inundation

Flooding Tsunami Volcanic lahar

Gasses
Fire

Fire Bush fire Chemical accident

support
Loss of

Covering

Liquefaction Landslide Ash fall

Figure 7 Examples of the types of hazardous processes to which buildings can be exposed. Each type of processes will have different effects.

whereas qualitative methods also incorporate the other def-
initional aspects.
Physical vulnerability is the potential for physical impact
on the built environment and population. It is defined as the
degree of loss to a given element-at-risk or set of elements-at-
risk resulting from the occurrence of a natural phenomenon of
a given magnitude, and expressed on a scale from 0 (no
damage) to 1 (total damage). As can be seen from Figures 3
and 7, vulnerability is related to the characteristics of the
elements-at-risk and to the hazard intensity. Physical vulner-
ability as such is therefore not a spatial component, but is
determined by the spatial overlay of exposed elements-at-risk
and hazard footprints (Van Westen et al., 2009). Economic
vulnerability is defined as the potential impact of hazards on
economic assets and processes (i.e., business interruption,
secondary effects such as increased poverty and job loss). So-
cial vulnerability is the potential impact of events on groups
within the society (such as the poor, single parent households,
pregnant or lactating women, the handicapped, children, and
elderly), and it considers public awareness of risk, ability of
groups to self-cope with catastrophes, and the status of insti-
tutional structures designed to help them cope. Environmental
 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management 285

Environmental factors Triggering factors Hazard inventory Elements at risk

Geology Land cover (t 0) Buildings

Earthquakes
Landslides
Geomorphology Roads
Land cover (t n)
Soils Windstorms Floods Land use
Topography Soil erosion Essential facilities
Etc. Hydrology (t 0) Rainfall Cadastral data
Forest fires
Census data
Hydrology (t n)
(a)

Initiation analysis Spatial probability

(Inventory, heuristic, Event-based
statistical, physical hazard inventories Magnitude - loss
Temporal probability
modeling) relationships
Magnitude probability • Vulnerability curve
• Vulnerability matrix
Run-out analysis • V=1
Hazard assessment
(Empirical, analytical, Pt, Ps and Pm for different return
numerical models) periods of triggering events and
hazard types
(b) (c) (d)

Hazard × Vulnerability × Amount

e
Specific ype)/tim
Loss (t
risk Σ All hazard types
Σ All hazard intensities
Σ All return periods
Σ All triggering events
e Σ All elements at risk
ype)/tim
Total risk Loss (t
(e)

Quantitative risk assessment Qualitative risk assessment

Combining specific risk curves

Total risk F-N Hazard index Vulnerability index

Probability

Probability

curves

Max Risk index

Min
Spatial multi criteria evaluation
Loss Number (N) of fatalities
(f) (g)

Cost-benefit analysis Early warning systems Spatial planning

Environmental
Mitigation measures Preparedness planning
impact assessment
(h)

Figure 8 Framework of the use of GIS for multi-hazard risk assessment. Based on Van Westen, C.J., Van Asch, T.W.J., Soeters, R., 2005.
Landslide hazard and risk zonation; why is it still so difficult? Bulletin of Engineering Geology and the Environment 650(2), 167–184, and Van
Westen, C.J., Castellanos Abella, E.A., Sekhar, L.K., 2008. Spatial data for landslide susceptibility, hazards and vulnerability assessment: an
overview. Engineering Geology 102(3-4), 112–131.
286 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management
vulnerability evaluates the potential impacts of events on the
environment (flora, fauna, ecosystems, and biodiversity)
(Birkmann, 2006).
Vulnerability can be expressed or presented in various ways
(Calvi et al., 2006). Vulnerability indices are based on indi-
cators of vulnerability and are mostly used for holistic
vulnerability and capacity and resilience assessment. Vulner-
ability tables show the relation between hazard intensity and
degree of damage in the form of a table. Vulnerability curves
display the relation between hazard intensity and degree of
damage for a group of elements-at-risk (e.g., a certain building
type) ranging from 0 to 1. Different types of elements-at-risk
will show different levels of damage given the same intensity
of hazard (see Figure 3). Vulnerability curves can be relative
(showing the percentage of property value damaged) or ab-
solute (showing the absolute amount of damage). Fragility
curves provide the probability for a particular group of
element-at-risk to be in or exceeding a certain damage state
(e.g., complete destruction, extensive damage, moderate
damage, and slight damage) under a given hazard intensity
(FEMA, 2004). A damage probability matrix (DPM) indicates
the probability that a given structural typology will be in a
given damage state for a given intensity.
Measuring physical vulnerability is a complicated process,
and can be done using either empirical or analytical methods
(Lang, 2002). Empirical methods are either based on damage
data from historical hazard events, or on expert opinion. For
events that are relatively frequent and widespread, it is pos-
sible to collect information on the degree of physical damage
to buildings or infrastructure after the event has occurred (e.g.,
Reese et al., 2007). This method is particularly suited for
flooding and for earthquakes, which normally affect many
buildings that are of the same type, and allow generating large
enough samples in order to make a correlation between the
hazard intensity (e.g., modified Mercalli intensity, ground ac-
celeration, water depth, etc.) and the degree of damage. The
result is either a DPM or a vulnerability curve. In many situ-
ations, expert opinion will be the most feasible option for
obtaining vulnerability information, either because there is no
prior damage information and not enough funding to apply
analytical methods, or because building classifications used
elsewhere do not reflect the local building stock (Douglas,
2007). This method involves the consultation of a group of
experts on vulnerability to give their opinion (e.g., on the
percentage damage they expect for the different structural
types with different intensities of hazard).
Analytical methods are used to study the behavior of
buildings and structures based on engineering design criteria,
analyzing the seismic load to derive the likelihood of failure,
using physical modeling tests (e.g., shake tables or wind tun-
nels), as well as computer simulation techniques. Analytical
methods are able to model the relation between the intensity
of the hazard and the level of damage of objects. They require,
however, detailed numerical input data. For instance, in the
case of earthquake vulnerability analysis of buildings, it is
important to have geotechnical reports to establish the value
of the effective peak acceleration coefficient, the value of the
effective peak-velocity-related acceleration coefficient, and the
soil-profile type. Spectral acceleration should also be obtained.
One of the common tests is using a shake table. This is a
device for shaking structural models or building components
with a wide range of simulated ground motions, including
reproductions of recorded earthquake time-histories (Calvi
et al., 2006).
Most of the work on the measurement of physical vul-
nerability is done for earthquakes, floods, and windstorms
(FEMA, 2004). Even though flood vulnerability has been de-
fined in a rather detailed manner (Moel et al., 2009), there are
still many uncertainties involved. For volcanic hazards, much
progress in defining vulnerability has been made in recent
years (Spence et al., 2004, 2005). For mass movement,
less work has been done on defining vulnerability (Glade,
2003), partly due to the large variation in mass-movement
processes, the difficulty in expressing landslide intensity versus
the degree of damage, and limited amount of landslide-
damage data. Some approaches exist for single landslide
types such as debris flows (e.g., Fuchs et al., 2007), but an
integrated methodology is still lacking. Hollenstein (2005)
developed an approach for multi-hazard vulnerability assess-
ment by definition of hazards using a common set of par-
ameters (e.g., acceleration, pressure, and temperature change)
and fragility functions, so that they were applicable to
all risks.
Population vulnerability can be subdivided into direct
physical population vulnerability (injury, casualties, and
homelessness) and indirect social vulnerability and capacity.
Physical population vulnerability is mostly carried out after a
building-vulnerability study by analyzing the effect of the
building damage on the population inside, using different
injury severity classes. Empirical relations exist for different
types of hazards, although most information is available for
earthquakes (Coburn and Spence, 2002; FEMA, 2004). For
example, such relations are available for volcanic hazards
(Spence et al., 2005), landslides (Glade et al., 2005), drought
(Wilhite, 2000), flooding, and windstorms (FEMA, 2004).
The methods described above aim at quantifying physical
vulnerability to natural hazards, and mostly follow an en-
gineering approach that is restricted to quantifying the phys-
ical effects of disasters on buildings and other infrastructure,
and secondary effects of these related to casualties and eco-
nomic losses. Other approaches also exist that look at vul-
nerability in a holistic way and try to incorporate all the
components of vulnerability using an indicator approach
(qualitative).
3.10.5 Multi-Hazard Risk Assessment
A framework of multi-hazard risk assessment is presented in
Figure 8 that identifies various components. The first com-
ponent (a) represents the input data, which can be subdivided
in datasets required for generating susceptibility maps, trig-
gering factors, multi-temporal inventories, and elements-
at-risk. The second component (b) focuses on susceptibility
assessment, and is divided into two parts. The first one dealing
with the modeling of areas where the hazard may initiate (e.g.,
earthquakes, landslide initiation, hydrological modelling, soil
erosion, and volcanic eruptions), which can make use of a
variety of different methods (inventory based, heuristic, stat-
istical, and physically-based models). The resulting maps form
 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management
the input as source areas in the modeling of potential
spreading of the phenomena (e.g., spreading of volcanic
deposits, landslide run-out, flood extent modelling, seismic
amplification, and forest-fire spreading).
The third component (c) deals with hazard assessment,
which heavily depends on the availability of magnitude-
frequency information. The susceptibility maps together with
the magnitude-frequency relations of the triggering events
are used to determine three components that are needed for
the hazard assessment: 1) the spatial probability (indicating
the probability that a given area will be affected by the hazard
of a given intensity); 2) the temporal probability (indicating
the probability of the event to happen in time); and 3) the
magnitude probability (indicating the probability that the
hazard event will have a given magnitude) (Corominas and
Moya, 2008).
The fourth component (d) focuses on vulnerability as-
sessment and indicates the various types of vulnerability as-
sessment approaches that can be used. Component (e) in
Figure 8 portrays the concept of risk assessment that integrates
the hazard, vulnerability, and amount of elements-at-risk. The
specific risk is calculated for many different situations related
to hazard type, hazard intensity, return period of the triggering
event, and type of element-at-risk. The integration of hazard,
vulnerability, and risk can be done in two ways: quantitative or
qualitative.
Component (f) presents the quantitative risk approach in
which the results are shown in risk curves plotting the ex-
pected losses against the probability of occurrence for each
hazard type individually, and also expressing the uncertainty,
by generating two loss curves expressing the minimum and
maximum losses for each return period of triggering events, or
associated annual probability. The individual risks curves can
be integrated into total risk curves for a particular area, and the
population loss can be expressed as F-N curves. The risk curves
can be made for different basic units (e.g., administrative units
such as individual slopes, road sections, census tracts, settle-
ments, municipalities, regions, or provinces). Component (g)
deals with methods for qualitative risk assessment, which are
mostly based on integrating a hazard index, and a vulner-
ability index, using spatial multi-criteria evaluation. The last
component (h) deals with the use of risk information in
various stages of disaster-risk management.
Hazards will impact different types of elements-at-risk, and
it is therefore important to calculate the risk for different
sectors/environments (e.g., housing, agriculture, transporta-
tion, education, health, tourism, protected areas, forests,
wetlands, etc.). Risk assessment should involve the relevant
stakeholders (i.e., individuals, businesses, organizations, and
authorities). The methodology for conducting risk assess-
ments can be broadly classified into qualitative and quanti-
tative approaches.
3.10.5.1 Qualitative Approaches
Qualitative methods for risk assessment are useful as an initial
screening process to identify hazards and risks. They are also
used when the assumed level of risk does not justify the time
and effort of collecting the vast amount of data needed for a
287
quantitative risk assessment, and where the possibility of ob-
taining numerical data is limited.
The simplest form of qualitative-risk analysis is to combine
hazard maps with elements-at-risk maps in GIS, using a sim-
ple-risk matrix in which the classes are qualitatively defined
(AGS, 2000). This method is widely applied, mostly at
(inter)national or provincial scales where the quantitative
variables are not available or they need to be generalized.
Qualitative approaches consider a number of factors that have
an influence on the risk. The approaches are mostly based on
the development of so-called risk indices, and on the use of
spatial multi- criteria evaluation. One of the first attempts to
develop global-risk indicators was done through the Hotspots
project (Dilley et al., 2005). In a report for the Inter-American
Development Bank, Cardona (2005) proposed different sets
of complex indicators for benchmarking countries in different
periods (e.g., from 1980 to 2000), and to make cross-national
comparisons. Four components or composite indicators reflect
the principal elements that represent vulnerability and show
the advances of different countries in risk management: 1)
Disaster Deficit Index (DDI); 2) Local Disaster Index (LDI); 3)
Prevalent Vulnerability Index (PVI); and 4) Risk Management
Index (RMI). Each composite index is generated on the basis
of a number of indicators. For instance, the DDI can be con-
sidered as an indicator of a country’s economic vulnerability
to disaster. The method has been applied thus far only in Latin
America and the Caribbean. Peduzzi et al. (2005, 2009) have
developed global indicators, not on the basis of administrative
units but based on gridded maps. The Disaster Risk Index
(DRI) (UN-ISDR, 2005b) combines both the total number
and the percentage of killed people per country in large- and
medium-scale disasters associated with droughts, floods, cyc-
lones, and earthquakes based on data from 1980 to 2000. In
the DRI, countries are indexed for each hazard type according
to their degree of physical exposure, their degree of relative
vulnerability, and their degree of risk.
At local scales, risk indices are also used, often in com-
bination with spatial multi-criteria evaluation (SMCE). Cas-
tellanos and Van Westen (2007) present an example of the use
of SMCE for the generation of a landslide-risk index for the
country of Cuba, generated by combining a hazard index and
a vulnerability index. The hazard index is computed using
indicator maps related to event-triggering factors (earthquakes
and rainfall) and environmental factors. The vulnerability
index was made using five key indicators including housing
condition and transportation (physical-vulnerability indi-
cators), population (social-vulnerability indicator), pro-
duction (economic-vulnerability indicator), and protected
areas (environmental-vulnerability indicator). The indicators
were based on polygons related to political-administrative
areas, which are mostly at the municipal level. Each indicator
was processed, analysed, and standardized according to its
contribution to hazard and vulnerability. The indicators were
weighted using direct pair-wise comparison and rank-ordering
weighting methods, and weights were combined to obtain the
final landslide risk-index map. The results were analyzed per
physiographic region and administrative units at provincial
and municipal levels. At the local level, Villagrán de Leon
(2006) incorporated 3 dimensions of vulnerability that in-
cluded the scale/level (from human being to national level),
288 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management
the various sectors of society, and six components of vulner-
ability. The method uses matrices to calculate a vulnerability
index that are grouped in qualitative classes (high, medium,
and low).
3.10.5.2 Quantitative Approaches
Quantitative approaches aim at expressing the risk in quanti-
tative terms, either as probabilities or expected losses. They can
be deterministic (scenario-based) or probabilistic (taking into
account the effect of all possible scenarios and uncertainties).
Quantitative risk assessment (QRA) follows an engineering
approach and focus on the evaluation of the direct physical
losses resulting directly from the impact of the hazard (e.g.,
flooded buildings and collapsed buildings). Some also
analyze indirect losses due to loss of function (e.g., disruption
of transport, business losses, or clean up costs). The focus is on
tangible losses that have a monetary (replacement) value.
Disasters also cause a large amount of intangible losses, for
example, lives and injuries, cultural heritage, environmental
quality, and biodiversity. Quantitative risk assessment aims at
quantifying the risk according to the equation given in
Figure 8. Several approaches, however, differ in how they
calculate hazards and vulnerability consequences. For a
number of different hazard scenarios, the consequences are
plotted against the temporal probability of occurrence of the
hazard events. Through these points a curve is fitted, the so-
called risk curve, and the area below the curve presents the
total risk. This procedure is carried out for all individual
hazard types, and care should be taken to evaluate inter-
relations between hazards. Since the risk is normalized into
annual risk, it is then possible to evaluate the multi-hazard
risk, and use the risk curves as the basis for disaster-risk re-
duction. The (epistemic and aleatory) uncertainties are in-
corporated in the modeling and used to calculate exceedance
probability curves, average annual losses (AAL), and probable
maximum losses (PML).
Loss estimation has been carried out initially from the early
days of insurance, and has evolved to computer-based catas-
trophe modeling since the late 1980s using advanced infor-
mation technology and GIS (Grossi et al., 2005). Since the end
of the 1980s, risk-modeling firms such as AIR Worldwide, Risk
Management Solutions (RMS), EQECAT, and others have led
the industry of probabilistic-risk modeling. Numerous inte-
grated numerical-models were developed for simulating cata-
strophic hazards that account for different types of hazards.
For instance, EQECAT developed a system (WORLDCA-
Tenterprise) that includes 181 natural-hazard models from 95
countries (EQECAT, 2010). These software tools are propri-
etary, however, and were used for the insurance market.
Software tools developed by the scientific community also
exist for disaster-risk management. One of the earliest publicly
available software methods for loss estimation was RADIUS
(Risk Assessment Tools for Diagnosis of Urban Areas against
Seismic Disasters). The very simple tool enabled users to
perform an aggregated loss estimation using a gridded mesh
and estimate the number of damaged buildings, length of
damaged lifelines, and the number of casualties and injured
people (RADIUS, 1999). The best publicly available software
tool for loss estimation to-date is HAZUS (which stands for
‘‘Hazards US’’), developed by FEMA together with the Na-
tional Institute of Building Sciences (NIBS). The goal of FEMA
was to create a methodology that was the standard national-
loss methodology for assessing losses from natural hazards
(FEMA, 2004). The first version of HAZUS was released in
1997 with a seismic loss estimation focus, and was extended
to multi-hazard losses in 2004, incorporating losses from
floods and windstorms. HAZUS was developed as a software
tool under ArcView and later ArcGIS. Although the HAZUS
methodology has been very well documented, the tool was
primarily developed for the US, and all data formats, building
types, fragility curves, and empirical relationships cannot be
exported easily to other countries.
Several other countries have adapted the HAZUS method-
ology to their own situation (e.g., in Taiwan (Yeh et al., 2006)
and Bangladesh (Sarkar et al., 2010)). The HAZUS method-
ology has also been the basis for the development of several
other open-source software tools for loss estimation. One is
called SELENA (SEimic Loss EstimatioN using a logic tree
Approach), and was developed by the International Centre for
Geohazards (ICG), NORSAR (Norway) and the University of
Alicante (Molina et al., 2010). Although most of the above
mentioned GIS-based loss-estimation tools focus on seismic
hazard, the Central American Probabilistic Risk Assessment
Initiative (CAPRA, 2009) has a true multi-hazard risk focus.
The aim of CAPRA is to develop a system that utilizes GIS,
Web-GIS, and catastrophe models in an open platform for
disaster-risk assessment, which allows users from the Central
American countries to analyze the risk in their areas, and be
able to make informed decisions on disaster-risk reduction.
The methodology focuses on the development and use of
probabilistic hazard-assessment modules for earthquakes,
hurricanes, extreme rainfall, and volcanic hazards, and the
hazards triggered by them, such as flooding, windstorms,
landslides, and tsunamis. These are based on event databases
with historical and simulated events. This information is
combined with elements-at-risk data focusing on buildings
and population. For the classes of elements-at-risk, vulner-
ability data can be generated using a vulnerability module. The
main product of CAPRA is a software tool, called CAPRA-SIG,
which combines the hazard scenarios, elements-at-risk, and
vulnerability data to calculate loss-exceedance curves.
In New Zealand, a comparable effort includes the RiskS-
cape methodology for multi-hazard risk assessment (Reese
et al., 2007; Schmidt et al., 2011). Another example of multi-
hazard risk assessment is the Cities project in Australia, which
is coordinated by Geoscience, Australia. Studies have been
made for six cities of which the Perth study is the latest
(Durham, 2003; Jones et al., 2005). Also in Europe, several
projects have developed multi-hazard loss estimations sys-
tems, such as the ARMAGEDOM system in France (Sedan and
Mirgon, 2003) and in Germany (Grünthal et al., 2006).
The aforementioned systems focus on the assessment of
losses before events, whereas other systems aim at providing
fast assessments of damage directly after the occurrence of
major events. For instance, the PAGER (Prompt Assessment of
Global Earthquakes for Response) system, developed by the
USGS is an automated system that rapidly assesses earthquake
impacts by comparing the population exposed to each level of
 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management 289

Table 7 Relationship between stakeholders in risk management and risk visualization options

Stakeholder Purpose Type of risk visualization

General public General information on risks
over large areas
Awareness raising
Community-based DRR projects
Businesses Investment policies and location
planning
Technical staff of (local) Land use regulation/zoning
authorities
Building codes
Spatial planning
Environmental Impact
Assessment
Disaster preparedness
Decision makers / local Decision making on risk
authorities reduction measures
Investments
Strategic Environmental
Assessment
NGO’s Influence political decisions in
favor of environment and
sustainable development
Scientists / technical staff of Hazard information exchange to
hazard data producers public and other agencies
Exchange of basic information
for hazard and risk
assessment
Insurance industry Development of insurance policy
Media Risk communication to public
Basic WebGIS applications in which they can overlay the location of major
hazard types with high-resolution imagery or topographic maps.
Animations (what if scenarios)
Simple maps of the neighborhood with risk class, buildings, evacuation
routes, and other features.
General information about hazards and risks in both graphical and map
format.
Map with simple legend including construction restricted, construction
allowed, further investigation required.
Maps indicating the types of building allowed (building type, number of
floors).
Hazard maps, with simple legends related to probabilities and possible
consequences.
Maps and possible loss figures for future scenarios.
Real time simple and concise Web-based information in both map and
graphical forms.
Statistical information, loss-exceedance curves, F-N curves, and maps.
Economic losses, projected economic losses for future scenarios.
General statistical information for administrative units.
This can vary from simple maps to Web-based applications, depending
on the objectives of the NGO.
WebGIS applications where they can access the basic information.
Spatial Data Infrastructure / Clearinghouse for exchanging information.
Loss-Exceedance Curves of economic losses, F-N curves.
Animations of hazard phenomena that clearly illustrate the problems.

shaking intensity with models of economic and fatality losses
based on past earthquakes in each country or region of the
world (PAGER, 2010).
3.10.5.3 Spatial Risk Visualization
Risk management cannot take place without proper risk gov-
ernance. Risk governance has been promoted in the ISDR,
Hyogo framework for action to: ‘‘Promote and improve dia-
logue and cooperation among scientific communities and
practitioners working on disaster-risk reduction, and en-
courage partnerships among stakeholders, including those
working on the socio-economic dimensions of disaster-risk
reduction’’ (UN-ISDR, 2005a). Governance depends on the
level of political commitment and strong institutions. Good
governance is identified in the ISDR Framework for disaster
reduction as a key area for the success of effective and sus-
tained disaster-risk reduction (IRGC, 2005). One of the im-
portant processes in risk governance is risk communication,
which is the interactive exchange of information about risks
among risk assessors, managers, news media, interested
groups, and the general public. An important component of
that is the visualization of risk. Since risk is a spatially varying
phenomenon, GIS technology is now the standard approach
for the production and presentation of risk information. Risk
can be presented in the form of statistical information per
administrative unit, such as a risk index value resulting from
qualitative risk assessment, the probable maximum loss or the
average annual loss, loss-exceedance curve for economic risk,
or F-N curves for societal population risk. Risk can also be
visualized in map form that depicts the spatial variation
of risk.
The type of risk (qualitative/quantitative, direct/indirect,
societal risk/individual risk, etc.) and the visualization tech-
nique used depends on the stakeholder to which the risk in-
formation is communicated. Table 7 provides an overview of
the relation between stakeholders and the type of risk
visualization.
Internet-based GIS systems have been developed in which
all the individual layers are separated (multi-tier approach),
thus allowing many clients to access and visualize the geo-data
at the same time. A WebGIS is a special GIS tool that uses the
Internet as a means to access and transmit remote data, con-
duct analysis, and present GIS results. WebGIS applications for
risk visualization have been developed for different purposes.
At the global level, the PREVIEW Global Risk Data Platform is
the result of efforts of UNEP, UNISDR, UNDP, and World
Bank, to share spatial data and information on global risk
from natural hazards through the internet. Users can visualize,
download, or extract data on past hazardous events, human
and economical hazard exposure, and risk from natural
290 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management
Figure 9 Global Risk Data Platform, PREVIEW. Reproduced from UNEP/DEWA/GRID, 2010. PREVIEW Global Risk Data Platform. United Nations
Environment Programme, Global Resource Information Database, Europe, Geneva. http://www.grid.unep.ch/activities/earlywarning/preview/
index.php
hazards on a platform compliant with OGC Web Services
(OWS). It covers tropical cyclones and related storm surges,
drought, earthquakes, biomass fires, floods, landslides, tsu-
namis, and volcanic eruptions (see Figure 9). The collection of
data is made via a wide range of partners (UNEP/DEWA/
GRID, 2010).
An example of risk visualization at the international level is
the multi-hazard risk atlas for the Andean region (Commu-
nidad Andina, 2009) that is available in paper atlas and Web-
based versions. This atlas provides a comprehensive overview
of the elements-at-risk in the region (population, production,
and infrastructure), the hazard phenomena (earthquakes,
tsunami, volcanic eruptions, landslides, flooding, cold waves,
and drought), and the risks in a very well designed manner.
Examples of different approaches for visualizing flood hazard
and risk maps from 19 European countries, USA, and Japan
are presented in EXCIMAP (2007). Many countries are also
developing their own Web-based risk maps. For example, the
CEDIM Risk Explorer, Germany, is a web-based map viewer
that interactively presents the results of the CEDIM project
"Riskmap Germany" (Müller et al., 2006). A more compli-
cated Web-GIS system has been developed in the Netherlands,
which can be accessed by the general public as part of the
national risk communication strategy. A secured section of the
same system can be accessed by professionals involved in risk
management, allowing them to get more detailed information
required for emergency response planning. National-scale risk
mapping in the Netherlands was carried out after the occur-
rence of major technical and flood disasters in the past
decades. The Web-GIS application (see Figure 10) shows in-
formation on natural hazards (flooding, natural fires, and
earthquakes), technological hazards (transportation accidents,
hazardous substances, and nuclear), and vulnerable objects
(Risicokaart, 2008). The flood-prone areas are defined by
more than 1 m flooding depth with a frequency larger than
1/4000 per year.
3.10.6 Conclusions
This chapter provides a framework for understanding hazards
assessment and disaster-risk management. Spatial data re-
quirements and techniques for multi-hazard risk assessment
have been addressed. It should be emphasized that data col-
lection, analysis/modeling, and information production as
part of this process is a complex task, because risk is a dynamic
concept and has many facets. This is illustrated in Figure 11. It
is evident that the world undergoes rapid changes in terms of
population growth, urbanization, economic development,
and socio-political structures. Furthermore, there is convincing
evidence that greenhouse gas forcing may be causing changes
in the Earth’s climate that are expected to lead to an increase in
 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management 291

Figure 10 The national risk atlas of the Netherlands is publicly available on the internet (www.risicokaart.nl).

hazardous events due to a hydro-meteorological trigger (IPCC,
2007).
The difficulty in predicting the magnitude of these changes
and the frequency of occurrence of extreme events reiterates
the need for a thorough change in our adaptation manage-
ment of hydro-meteorological risks (EEA, 2004). According to
recent European studies, the projected impact of flooding in
Europe will increase dramatically in the coming decades. By
2080, it is estimated that between 250 000 and 400 000
people will be affected each year by flooding, and the total
annual expected flood damage will range between 7.7 and 15
billion Euros. These values are more than double of those in
the period 1961–1990 (Ciscar, 2009). Very limited work has
been carried out up to now to include the cascading or con-
joint (also called domino) effects in the analysis of future
impacts of environmental changes to hydro-meteorological
hazards. The exposure of elements-at-risk also increases and
therefore the risk of natural hazards is constantly growing.
Land-use changes will occur as a result of technological, socio-
economic, and political developments, as well as global en-
vironmental change. The nature and rate of change will
strongly depend on policy decisions. Many environmental
problems are caused by unplanned urban expansion. By 2050,
approximately 70% of all people will be living in urban areas,
whereas in several countries the proportion will be 90% or
more. Some of the drivers of change to the urban environment
292 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management
Change analysis Analysis of uncertainty
Climate change Changing
hazards
Land use
Changing
change
elements at risk
Socio-economic Changing
change vulnerability
Scenario
development
Scenario A
Scenario B
Scenario C
Scenario ..
Figure 11 Framework of the implementation of environmental change scenarios in risk management.
are the global economy, cross-border transport networks,
large-scale societal, economic and demographic changes, and
differences in national planning laws. As the level of un-
certainty of the components used in the risk equation (hazard,
vulnerability, and quantification of the exposed elements-at-
risk) is very high, the analysis of the changes in future risk
should incorporate these uncertainties in a probabilistic
manner.
Impacts of natural hazards on the environment and on the
society are still tackled by mono-disciplinary approaches. The
focus is reflected in the domains of scientific research (single
approach and tools for each type of hazard), in the existing
management tools, and in the legislative basis of these activ-
ities. Management tools, models, and local-to-regional tech-
nical solutions have been proposed by numerous projects for
single hazards. Only a few of them have tackled the issue of
risk assessment and management, however, from a multi-
hazard perspective, especially including possible combined
and domino effects. Probabilistic tools for multi-hazard risk
assessment are not available to stakeholders at the local level.
Insurance companies and specialized risk-assessment con-
sultants have developed models, but these are not open for
public use. The implementation of risk-management measures
such as disaster-preparedness program, land-use planning,
regulatory zoning and early warning systems are considered
Risk management
Risk assessment
Multi-hazard
Risk governance/Risk communication/Risk visualization
assessment
Elements at risk
/exposure
(Spatial) information management
Vulnerability
assessment
Risk analysis
Risk evaluation
Risk reduction measures
Emergency
preparedness
Land use planning
Cost-benefit
assessment
essential. Fleischauer et al. (2006) concluded that spatial
planning is only one of many aspects in risk management and
that it is, in general, not properly implemented. Furthermore,
multi-risk assessment approaches are not used in planning
practice: Risk indicators are hardly used and vulnerability in-
dicators are not at all.
Therefore, approaches are needed for integrating disaster-
risk assessment in long-term resource allocation and land-use
planning at all levels of administration. Additionally, scientific
advances in hazard and risk assessment and demands of
stakeholders/end-users are still not well connected. In many
cases, the scientific outcomes remain rooted solely within the
scientific community, or new knowledge is not fabricated
enough to be implemented by stakeholders and end-users
(IRGC, 2005). A key cause of the gap between the science
community and stakeholders/end-users is in the complexity of
human-environment interactions. This has led to the devel-
opment of a diversity of approaches, commonly not easy to
implement by the end-user community. A need exists for the
development of a harmonized decision-making structure for
applying hazard and risk mitigation through spatial planning
in risk-prone areas. A need also exists for capacity building in
the field of multi-hazard risk assessment, and the transfer of
the knowledge from developed countries to developing
countries using open-source software tools and methods
 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management
adapted to the data availabilities in these countries (Van
Westen et al., 2009). The Hyogo framework of action
2005–2015 of the UN-ISDR indicates risk assessment and
education as two of the key areas for the development of ac-
tion in the coming years.
In conclusion, it is clear that geomorphology, geographic
information science, and Earth observation (remote sensing)
have made significant contributions to the understanding of
natural hazard processes, and the way these can be monitored
and assessed/modeled at various scales using a range of
geospatial technologies. Such capabilities have also been used
successfully in analyzing the risk to vulnerable societies, and
the results have been communicated to stakeholders that have
often used them in appropriate disaster-risk reduction strat-
egies. Nevertheless, as extreme events and the number of
vulnerable people are expected to increase in the future, more
work is required to improve our ability to accurately estimate
future hazards and risks. Geomorphologists should further
develop their scientific work into practical applications that
can be used to save lives and reduce economic losses due to
natural disasters.
Acknowledgements
This research was supported by the United Nations University
– ITC School for Disaster Geo-Information Management. The
author would like to thank Sekhar Lukose Kuriakose, Norman
Kerle and Michael Bishop for their comments on the draft.
References
Africover, 2010. Africover Land cover database for Africa. Food and Agricultural
Organization of the United Nations. FAO.
AGS, 2000. Landslide risk management concepts and guidelines, Australian
Geomechanics Society (AGS), Sub-committee on landslide risk management.
Alcantara-Ayala, I., 2004. Flowing Mountains in Mexico, Incorporating Local
Knowledge and Initiatives to Confront Disaster and Promote Prevention’. In
Mountain Research and Development 24(1), 10–43.
Alcantara-Ayala, I., Goudie, A.S., 2010. Geomorphological Hazards and Disaster
Prevention. Cambridge University Press, Cambridge, 291 pp.
Alernet, 2010. Interactive map. AlertNet, Alerting humanitarians to emergencies.
Thomas Reuters Foundation, http://www.alertnet.org/map/index.htm
Alexander, D., 1993. Natural disasters. UCL Press Ltd., University College, London.
Alkema, D., Middelkoop, H., 2005. The influence of floodplain compartmentalization
on flood risk within the Rhine - Meuse delta. Natural hazards: journal of
the international society for the prevention and mitigation of natural hazards
36(1-2), 125–145.
Anderson, M.B., Woodrow, P.J., 1998. Rising from the Ashes: Development
Strategies in Times of Disaster. London, IT Publications.
ANSS, 2009. Advanced National Seismic System. http://earthquake.usgs.gov/
monitoring/anss/
APD, 2010. African Population Database. United Nations Environmental Programme
(UNEP). http://www.na.unep.net/globalpop/africa/
Arino, O., Gross, D., Ranera, F., et al., 2007. GlobCover: ESA service for global
land cover from MERIS. Geoscience and Remote Sensing Symposium, 2007.
IGARSS 2007. IEEE International : 2412–2415
Arnold, E., 1989. Program SEISRISK III Adapted to Personal computers. Open-File
Report, 89–557.
Asselen, S.V., Seijmonsbergen, A.C., 2006. Expert-driven semi-automated
geomorphological mapping for a mountainous area using a laser DTM.
Geomorphology 78(3–4), 309–320.
293
Ayanza, J.S.M., Barbosa, P.M., Schmuck, G., Liberta, G., (2003). The European
Forest Fire Information System (EFFIS). European Commission. JIR Institute for
Environment and sustainability.
Balk, D.L., Deichmann, U., Yetman, G., Pozzi, F., Hay, S.I., 2006. Determining
global population distribution: methods, applications and data. Advances in
Parasitology 62, 119–156.
Bankoff, G., Frerks, G., Hillhorst, D. (Eds.), 2003. Mapping Vulnerability: Disasters,
Development and People. Earthscan, London.
Barlow, J., Franklin, S., Martin, Y., 2006. High spatial resolution satellite imagery,
DEM derivatives, and image segmentation for the detection of mass wasting
processes. Photogrammetric Engineering and Remote Sensing 72(6), 687–692.
Barredo, J.I., 2007. Major flood disasters in Europe: 1950–2005. Natural Hazards
42(1), 125–148.
Bassolé, A., Brunner, J., Tunstall, D., 2001. GIS: supporting environmental planning
and management in West Africa. A report of the joint USAID/World Resources
Institute Information Group for Africa.
Beven, K., Calver, A. and Morris, E.M., 1987. The Institute of Hydrology distributed
model. Institute of Hydrology report 98, Wallingford UK.
Bhaduri, B., Bright, E., Coleman, P., Urban, M., 2007. LandScan USA: A High
Resolution Geospatial and Temporal Modeling Approach for Population
Distribution and Dynamics. GeoJournal 69, 103–117.
Birkmann, J. (Ed.), 2006. United Nations University Press, Tokyo, 524pp524pp.
Blaikie, P., Cannon, T., Davis., I., Wisner, B., 1994. At risk: natural hazards, people’s
vulnerability and disasters. Routledge, London.
Blong, R., 2003. A new damage index, Nat. Hazards 30(1), 1–23.
Blumberg, D., Bach, D., Weissel, J., Gorokhovich, Y., Small, C., & Balk, D. 2005.
The 2004 Sumatra tsunami event: Contribution of SRTM data to the analysis of
devastation. The Shuttle Radar Topography Mission – Data Validation and
Applications, Workshop, June 14–16, 2005, Reston, Virginia.
Bonforte, A., Puglisi, G., 2003. Magma uprising and flank dynamics on
Mount Etna volcano, studied using GPS data (1994–1995). J. Geophys. Res
108(B3), 2153.
Bourgeau-Chavez, L., Kasischke, E., 2002. Mapping fire scars in global boreal
forest using imaging radar data. International Journal of Remote Sensing 23,
4211–4234.
Brenner, C., 2005. Building reconstruction from images and laser scanning.
International Journal of Applied Earth Observation and Geoinformation 6(3–4),
187–198.
Brunner, G.W., 2002. HEC-RAS River Analysis System; Hydraulic Reference Manual.
US Army Corps of Engineers, Davis, USA.
Bürgmann, R., Hilley, G., Ferretti, A., Novali, F., 2006. Resolving vertical tectonics
in the San Francisco Bay Area from Permanent Scatterer InSAR and GPS
Analysis. Geology 34, 221–224.
Calvi, G.M., Pinho, R., Magenes, G., Bommer, J.J., Restrepo-Vélez, L.F., Crowley,
H., 2006. Development of Seismic Vulnerability Assessment Methodologies Over
The Past 30 Years. ISET Journal of Earthquake Technology, Paper No. 472
43(3), 75–104.
Cannon, S.H., Gartner, J.E., Wilson, R.C., Bowers, J.C., Laber, J.L., 2008. Storm
rainfall conditions for floods and debris flows from recently burned areas in
southwestern Colorado and southern California. Geomorphology 96(3-4),
250–269.
Cannon, T., Twigg, J., Rowell, J., 2003. Social vulnerability, sustainable Livelihoods
and disasters. Greenwich, DFID, 63 p.
CAPRA, 2009. Central American Probabilistic Risk Assessment (CAPRA). World
Bank. www.ecapra.org
Cardona, O.D., 2005. Indicators for Disaster Risk and Risk Management. Instituto
de Estudios Ambientales, Universidad Nacional de Columbia, Manizales,
Columbia, Program for Latin America and the Caribbean: Summary Report.
Carrara, A., Cardinali, M., Guzzetti, F., Reichenbach, P., 1995. GIS-based techniques
for mapping landslide hazard. In: Carrara, A., Guzzetti, F. (Eds.), Geographical
Information Systems in Assessing Natural Hazards. Kluwer Academic
Publications, Dordrecht, The Netherlands, 135-176135-176.
Carton, A., Coratza, P. and Marchetti, M., 2005. Guidelines for geomorphological
sites mapping: examples from Italy , Géomorphologie : relief, processus,
environnement, 3/2005.
Castellanos, E., Van Westen, C.J., 2007. Qualitative landslide susceptibility
assessment by multicriteria analysis; a case study from San Antonio del Sur,
Guantanamo, Cuba. Geomorphology 94(3-4), 453–466.
CEOS, 2003. Geohazards Team Report. Committee on Earth Observation Satellites
(CEOS). Integrated Global Observing Strategy (IGOS) http://www.igospartners.
org/Geohazards.htm
CGIAR-CSI, 2008. SRTM 90m Digital Elevation Data. The CGIAR Consortium for
Spatial Information (CGIAR-CSI) http://srtm.csi.cgiar.org/
294 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management
Chen, K., McAneney, J., Blong, R., Leigh, R., Hunter, L., Magill, C., 2004. Defining
area at risk and its effect in catastrophe loss estimation: a dasymetric mapping
approach. Applied Geography 24, 97–117.
CIESIN, 2005. Global Multihazard Proportional Economic Loss Risk Deciles. Center
for Hazards and Risk Research (CHRR); Center for International Earth Science
Information Network (CIESIN). Columbia University, Palisades, NY. http://
www.ldeo.columbia.edu/chrr/research/hotspots/
Cihlar, J., 2000. Land cover mapping of large areas from satellites: status and
research priorities. International Journal of Remote Sensing, 1366–5901 21(6),
1093–1114.
Ciscar, J.C. (Ed), 2009. Climate change impacts in Europe. Final report of the
research project. European Commission Joint Research Centre. EUR 24093 EN
Cobby, D.M., Mason, D.C., Davenport, I.J., 2001. Image processing of airborne
scanning laser altimetry data for improved river flood modeling. ISPRS Journal
of Photogrammetry and Remote Sensing 56(2), pp. 121–138.
Coburn, A.W., Spence, R.J.S., 2002. Earthquake Protection, 2nd ed. John Wiley &
Sons, Chichester, UK.
Communidad Andina 2009. Atlas de las dinamicas del Territorio Andino.
Poblacion y bienes expuestos a amenazas naturales. Communidad Andina,
Comité Andino para la Prevención y Atención de Desastres, CAPRAD.
http://www.comunidadandina.org/predecan/atlasweb/index.html
Connor, C.B., Hill, B.E., Winfrey, B., Franklin, N.M., La Femina, P.C., 2001. Estimation
of volcanic hazards from tephra fallout. Natural Hazards Review 2, 33–42.
Coppin, P., Jonckheere, I., Nackaerts, K., Muys, B., Lambin, E., 2004. Digital
change detection methods in ecosystem monitoring: a review. International
Journal of Remote Sensing 25(9), 1565–1596.
Coppock, J.T., 1995. GIS and natural hazards: an overview from a GIS perspective.
In: Carrara, A., Guzzetti, F. (Eds.), Geographical information systems in
assessing natural hazards. Kluwer, Dordrecht, pp. 21–34.
Corominas, J., Moya, J., 2008. A review of assessing landslide frequency for
hazard zoning purposes. Engineering geology 102, 193–213.
Dai, F.C., Lee, C.F., Ngai, Y.Y., 2002. Landslide risk assessment and management:
an overview. Engineering Geology 64, 65–87.
Dal Cin, C., Moens, L., Dierickx, P., Bastin, G., Zech, Y., 2005. An integrated
approach for realtime floodmap forecasting on the Belgian Meuse river. Natural
Hazards 36(1–2), 237–256.
Dartmouth Flood Observatory, 2010. Space based measurement of surface water.
http://floodobservatory.colorado.edu/
De Roo, A., Barredo, J., Lavalle, C., Bodis, K., Bonk, R., 2007. Potential Flood
Hazard and Risk Mapping at Pan-European Scale. Digital Terrain Modelling.
Lecture Notes in Geoinformation and Cartography 2007, 183–202.
De Roo, A.P.J., Wesseling, C.G., Van Deursen, W.P.A., 2000. Physically based river
basin modelling within a GIS: the LISFLOOD model. Hydrological Processes 14,
1981–1992.
DesInventar, 2010. Inventory system of the effects of disasters. http://
online.desinventar.org
Dilley, M., Chen, R.S., Deichmann, U., et al., 2005. Natural Disaster Hotspots: A
Global Risk Analysis. The World Bank, Hazard, Management Unit, Washington,
D.C. Disaster Risk Management Series No. 5, 132 pp.
Disaster Charter (2010). The International Charter on Space and Major Disasters.
http://www.disasterscharter.org/home
DLR-ZKI, 2010. Center for Satellite Based Crisis Information (ZKI) , German Remote
Sensing Data Center (DFD) of DLR. http://www.zki.dlr.de/mission
Douglas, J., 2007. Physical vulnerability modelling in natural hazard risk
assessment. Nat. Hazards Earth Syst. Sci 7, 283–288.
Drăgut- , L., Blaschke, T., 2006. Automated classification of landform elements using
object-based image analysis. Geomorphology 81(3-4), 330–344.
Durham, K., 2003. Treating the Risks in Cairns. Natural Hazards 30(2), 251–261.
Ebert, A., Kerle, N., Stein, A., 2009. Urban social vulnerability assessment with
physical proxies and spatial metrics derived from air- and spaceborne imagery
and GIS data. Natural hazards 48(2), 275–294.
EEA, 2004. "Mapping the impacts of recent natural disasters and technological
accidents in Europe", Environmental Issue Report No. 35, European Environment
Agency (EEA), Copenhagen, Denmark.
EFD, 2007. European Floods Directive. European Commission. http://ec.europa.eu/
environment/water/flood_risk/index.htm
EM-DAT, 2009. The OFDA/CRED International Disaster Database, Université
Catholique de Louvain, Brussels, Belgium. http://www.emdat.be
Emani, S., 1996. Applications in hazard assessment and management. Explorations
in Geographic Information Systems Technology, Vol. 6. Worcester (USA), United
Nations Institute for Training and Research (UNITAR).
Emanuel, K., Ravela, S., Vivant, E., Risi, C., 2006. A statistical deterministic
approach to hurricane risk assessment. Bull. Am. Met. Soc 87, 299–314.
EQECAT, 2010. WORLDCAT enterpriseTM, EQECAT’s catastrophe risk modeling
software platform. http://www.eqecat.com
EXCIMAP, 2007. Atlas of flood maps. European Commission.
Falorni, G., Teles, V., Vivoni, E.R., Bras, R.L., Amaratunga, K.S., 2005. Analysis and
characterization of the vertical accuracy of digital elevation models from the
shuttle radar topography mission. Jounal of Geophysical Research 110.
FAO, 1981. Soil Map of the World, United Nations Food and Agriculture
Organization, Vols. 1-10 (1:5M scale maps and accompanying texts). UNESCO,
Paris.
FAO, 2010. Geonetwork. Food and Agriculture Organisation of the United Nationa
(FAO). http://www.fao.org/geonetwork/srv/en/main.home
Farina, P., Casagli, N., Ferretti, A., 2008. Radar-interpretation of InSAR
measurements for landslide investigations in civil protection practices. Proc. 1st
North American Landslide Conference. Vail, Colorado, 272–283.
Farr, T.G., Kobrick, M., 2000. Shuttle Radar Topography Mission produces a wealth
of data. EOS Trans 81, 583–585.
Favey, E., Wehr, A., Geiger, A., Kahle, H.-G., 2002. Some examples of European
activities in airborne laser techniques and an application in glaciology. Journal
of Geodynamics 34, 347–355.
Fell, R., Corominas, J., Bonnard, C., Cascini, L., Leroi, E., Savage, W.Z., 2008.
Guidelines for landslide susceptibility, hazard and risk zoning for land use
planning. Engineering Geology 102, 85–98.
FEMA, 2004. HAZUS-MH. FEMA’s Methodology for Estimating Potential Losses
from Disasters. US Federal Emergency Management Agency. http://
www.fema.gov/plan/prevent/hazus/index.shtm
FEMA, 2010. National Flood Insurance Programme. US Federal Ergency
Management Agency. http://www.fema.gov/business/nfip/
Ferretti, A., Prati, C., Rocca, F., 2001. Permanent Scatterers in SAR interferometry.
IEEE Trans. Geosci. Remote Sens 39(1), 8–20.
Ferrier, N., Haque, C.E., 2003. Hazards Risk Assessment Methodology for
Emergency Managers: A Standardized Framework for Application. Natural
Hazards 28(2-3), 271–290.
Fleischhauer, M., Greiving, S., Wanczura, S. (Eds.), 2006. Natural Hazards and
Spatial Planning in Europe. Dortmund 2006, 206 pp.
Fraser, C.S., Baltsavias, E., Gruñe, A., 2002. Processing of IKONOS imagery for
submetre 3D positioning and building extraction. ISPRS Journal of
Photogrammetry and Remote Sensing 56(3), 177–194.
French, J.R., 2003. Airborne LiDAR in support of geomorphological and hydraulic
modeling. Earth Surface Processes and Landforms vol. 28, 3.
Friedl, M.A., McIver, D.K., Hodges, J.C.F., et al., 2002. Global land cover mapping
from MODIS: algorithms and early results. Remote Sens Environ 83(1–2),
287–302.
Fuchs, S., Heiss, K., Hübl, J., 2007. Towards an empirical vulnerability function for
use in debris flow risk assessment. Natural Hazards and Earth System Sciences
7, 495–506.
Fujisada, H., Bailey, G.B., Kelly, G.G., Hara, S., Abrams, M.J., 2005. ASTER DEM
performance. IEEE Transactions on Geoscience and Remote Sensing 43(12),
2707–2714.
Furbish, D.J., Rice, R.M., 1983. Predicting landslides related to clearcut logging,
Northwestern California, USA. Mountain Research and Development 3, 253–259.
GDACS, 2010. Global Disaster Alert and Coordination System. Joint Research
Centre of the European Commission, and United Nations Office for Coordination
of Humanitarian Affairs. http://www.gdacs.org/
GEM, 2010. Global Earthquake Model. http://www.globalquakemodel.org/
GeoNode, 2010. GeoNode. Initiative of OpenGeo and the World Bank. http://
www.geonode.org/
Giglio, L., Kendall, J.D., 2001. Application of the Dozier retrieval to wildfire
characterization: a sensitivity analysis. Remote Sensing of Environment 77,
34–49.
Giles, P.T., Franklin, S.E., 1998. An automated approach to the classification of the
slope units using digital data. Geomorphology 21, 251–264.
Gili, J.A., Corominas, L., Rius, J., 2000. Using Global Positioning System
techniques in landslide monitoring. Engineering Geology 55(3), 167–192.
Glade, T., 2003. Vulnerability assessment in landslide risk analysis. Die Erde 134,
121–138.
Glade, T., Anderson, M., Crozier, M.J., 2005. Landslide Hazard and Risk. John
Wiley & Sons, Ltd., Chichester, England, 802 pp.
Glidenumber, 2010. Disaster database. ADRC. See www.glidenumber.net
GMES, 2010. Global Monitoring for Environment and Security. European
Commission.
GMK, 2010. Geomorphologische Karten im Netz. Bonn University.
GNS, 2010. Central South Island Glacial Geomorphology. Institute of Geological
and Nuclear Sciences. http://maps.gns.cri.nz/website/csigg/
 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management
GRIP, 2010. DISDAT the disaster data collection initiative portal. Global Risk
Identification Program (GRIP). http://www.gripweb.org/
Grossi, P., Kunreuther, H., Patel, C.C., 2005. Catastrophe Modeling: A New
Approach to Managing Risk. Springer, New York.
GRUMP, 2004. Center for International Earth Science Information Network (CIESIN),
Columbia University. Global Rural-Urban Mapping Project (GRUMP). CIESIN,
Columbia University, Palisades, NY. http://sedac.ciesin.columbia.edu/gpw
Grünthal, G., Thieken, A.H., Schwarz, J., Radtke, K.S., Smolka, A., Merz, B., 2006.
Comparative Risk Assessments for the City of Cologne – Storms, Floods,
Earthquakes. Natural Hazards 38(1-2), 21–44.
GSHAP, 1999. The Global Seismic Hazard Assessment Project. International
Lithosphere Program (ILP) with the support of the International Council of
Scientific Unions (ICSU). http://www.seismo.ethz.ch/static/GSHAP/
GSN, 2009. Global Seismographic Network. http://earthquake.usgs.gov/monitoring/
gsn/
Guha-Sapir, D. and Below, R., 2002. The quality and accuracy of disaster data: a
comparative analyses of three global data sets, technical document. The
ProVention Consortium and the Disaster Management Facility, The World Bank,
WHO Centre for Research on the Epidemiology of Disasters, University of
Louvain School of Medicine.
Günther, A., 2003. SLOPEMAP: programs for automated mapping of geometrical
and kinematical properties of hard rock hill slopes. Computers and Geoscience,
865–875.
Gustavsson, M., Kolstrup, E., Seijmonsbergen, A.C., 2006. A new symbol and GIS
based detailed geomorphological mapping system: renewal of a scientific
discipline for understanding landscape development. Geomorphology 77,
90–111.
Guzzetti, F., Cardinali, M., Reichenbach, P., 1994. The AVI project: a bibliographical
and archive inventory of landslides and floods in Italy. Environmental
Management 18(4), 623–633.
Guzzetti, F., Reichenbach, P., Cardinali, M., Galli, M., Ardizzone, F., 2005.
Probabilistic landslide hazard assessment at the basin scale. Geomorphology 72,
272–299.
Harvey, J.T., 2002. Estimating census district populations from satellite imagery:
Some approaches and limitations. International Journal of Remote Sensing 23,
2071–2095.
Hastings, D.A., Dunbar, P.K., 1998. Development and assessment of the global land
1 km base elevation digital elevation model (GLOBE). International Archives of
Photogrammetry and Remote Sensing 32(4), 218–221.
Haugerud, R.A., Harding, D.J., Johnson, S.Y., Harless, J.L., Weaver, C.S., Sherrod,
B.L., 2003. High-resolution LiDAR topography of the Puget Lowland,
Washington - a bonanza for earth science. GSA Today 13, 4–10.
Henricksen, B.L., Durkin, J.W., 1986. Growing period and drought early warning in
Africa using satellite data. International Journal of Remote Sensing 7(11),
1583–1608.
Hervas, J., Barredo, J.I., Rosin, P.L., Pasuto, A., Mantovani, F., Silvano, S., 2003.
Monitoring landslides from optical remotely sensed imagery: the case history of
Tessina landslide, Italy. Geomorphology 54, 63–75.
Hervouet, J.M., Van Haren, L., 1996. Recent advances in numerical methods for
fluid flows. In: Anderson, M.G., Walling, D.E., Bates, P.D. (Eds.), Floodplain
Processes. John Wiley & Sons Ltd., England, pp. 183–214.
Hesselink, A.W., Stelling, G.S., Kwadijk, J.C.J., Middelkoop, H., 2003. Inundation of
a Dutch river polder, sensitivity analysis of a physically based inundation model
using historic data. Water Resources Research 39, 9, art. no. 1234.
HEWSweb, 2010. Humanitarian Early Warning Service. Inter Agency Standing
Committee, World Food Programme. http://www.hewsweb.org
Hilhorst, D., 2004. Unlocking domains of disaster response. In: Bankoff, G., Frerks,
G., Hilhorst, D. (Eds.), Mapping Vulnerability: Disasters, Development and
People. Earthscan Publications, London, pp. 52–66.
Hilley, G.E., Burgmann, R., Ferretti, A., Novali, F., Rocca, F., 2004. Dynamics of
slow-moving landslides from Permanent Scatterer analysis. Science 304(5679),
1952–1955.
Holland, G.J., 1980. An analytic model of the wind and pressure profiles in
hurricanes, Mon. Weather Rev 108, 1212–1218.
Hollenstein, K., 2005. Reconsidering the risk assessment concept: Standardizing the
impact description as a building block for vulnerability assessment. Nat. Hazards
Earth Syst. Sci., 5, 301–307.
Hong, Y., Adler, R., Huffman, G., 2007a. Use of satellite remote sensing
data in the mapping of global landslide susceptibility. Natural Hazards 43,
245–256.
Hong, Y., Adler, R.F., Negri, A., Huffman, G.J., 2007b. Flood and Landslide
Applications of Near Real-time Satellite Rainfall Estimation. Natural Hazards
43(2), 285–294.
295
Hordijk, M.A., Baud, I.S.A., 2006. The role of research and knowledge generation in
collective action and urban governance: How can researchers act as catalysts?’
to Habitat International. Habitat International 30, 668–689.
IGOS 2007. Geohazards Earth Observation Requirements. Integrated Global
Observing Strategy (IGOS). Geohazards initiative. BRGM/RP 55719-FR. http://
igosg.brgm.fr/documents.asp
Ioualalen, M., Asavanant, J., Kaewbanjak, N., Grilli, S.T., Kirby, J.T., Watts, P., 2007.
Modeling the 26 December 2004 Indian Ocean tsunami: Case study of impact in
Thailand. Journal of Geophysical Research 112(c07024), 21 pp.
IPCC, 2007. Impacts, Adaptation and Vulnerability. Intergovernmental Panel on
Climate Change, Fourth Assessment Report, Working Group II. http://
www.ipcc.ch/ipccreports/ar4-wg2.htm
Ireland, D., 2001. Disaster Risk Management from a Remote Shire’s Perspective.
Centre for Disasters Studies, JCU, Australia, 57 p.
IRGC, 2005. Risk Governance: Towards an Integrative Approach. International Risk
Governance Council, White Paper 1 Geneva, http://www.irgc.org/
Publications.html
Jarvinen, B.R., Neumann, C.J., Davis, M.A.S., 1984. A tropical cyclone data tape for
the North Atlantic basin, 1886–1983, contents, limitations and uses. Miami,
Florida, NOAA Tech Memo, NWS NHC 22.
Jones, T., Middelmann, M. and Corby, N. 2005. Natural Hazard Risk in Perth,
Western Australia. The Cities Project, Geoscience Australia. http://www.ga.gov.au/
hazards/reports/perth/
Joyce, K.E., Belliss, S.E., Samsonov, S.V., McNeill, S.J., Glassey, P.J., 2009. A
review of the status of satellite remote sensing and image processing techniques
for mapping natural hazards and disasters. Progress in Physical Geography
33(2), 183–207.
Jibson, R.W., Harp, E.L. and Michael, J.A., 1998, A Method for Producing Digital
Probabilistic Seismic Landslide Hazard Maps: An Example from the Los
Angeles, California, Area, U.S. Geological Survey Open File Report 98-113, 16.
Jimenez, M.J., Garcia-Fernandez, M., Zonno, G., Cella, F., 2000. Mapping soil
effects in Barcelona, Spain, through an integrated GIS environment. Soil
Dynamics and Earthquake Engineering 19(4), 289–301.
Kaiser, R., Spiegel, P.B., Henderson, A.K., Gerber, M.L., 2003. The application of
Geographic Information Systems and Global Positioning Systems in
humanitarian emergencies: lessons learned, programme implications and future
research. Disasters 27, 127–140.
Karnieli, A., Dall’Olmo, G., 2003. Remote sensing monitoring of desertification,
phenology, and droughts. Management of Environmental Quality 14(1), 22–38.
Kerle, N., Oppenheimer, C., 2002. Satellite remote sensing as a tool in lahar
disaster management. Disasters 26, 140–160.
Kirschbaum, D.B., Adler, R., Hong, Y., Lerner-Lam, A., 2009. Evaluation of a
preliminary satellite-based landslide hazard algorithm using global landslide
inventories. Natural Hazards and Earth System Sciences 9, 673–686.
Klimaszewski, M., 1982. Detailed geomorphological maps. ITC Journal 3, 265–271.
Korup, O., 2002. Recent research on landslide dams-a literature review with special
attention to New Zealand. Progress in Physical Geography 26, 206–235.
Kraas, F., 2008. Megacities as Global Risk Areas. Urban Ecology 2008 V, 583–596.
Kuriakose, S.L., van Beek, L.P.H., Van Westen, C.J., 2009a. Parameterizing a
physically based shallow landslide model in a data poor region. Earth Surface
Processes and Landforms 34(6), 867–881.
Kuriakose, S.L., Devkota, S., Rossiter, D.G., Jetten, V.G., 2009b. Prediction of soil
depth using environmental variables in an anthropogenic landscape, a case
study in the Western Ghats of Kerala, India. Catena 79(1), 27–38.
Lang, K. 2002. Seismic Vulnerability of Existing Buildings, IBK Report No. 273,
Swiss Federal Institute of Technology (ETH), Zurich, Switzerland.
LandScan, 2010. LandScanTM Global Population Database. Oak Ridge National
Laboratory, Oak Ridge, TN. http://www.ornl.gov/landscan
Lavell, A., 2000. An approach to concepts and definitions in risk management
terminology and practice. ERD-UNDP, Geneva, 32 pp.
Lavelle, F., Vickery, P.J., Schauer, B., Twisdale, L.A., Laatsch, E. 2003. HAZUS-MH
Hurricane Model, Proceedings, 11th International Conference on Wind
Engineering, Texas Tech University, June 2003, 1015–1022.
Longley, M.F., Goodchild, D.J., Maguire, Rhind, D.W., 2005. Geographical
Information Systems: Principles, Techniques, Management and Applications.
John Wiley & Sons, Hoboken, NJ.
Lwin, K., Murayama, Y., 2009. A GIS Approach to Estimation of Building Population
for Micro-spatial Analysis. Transactions in GIS 13, 401–414.
Lyon, B., Barnston, A.G., 2005. ENSO and the Spatial Extent of Interannual
Precipitation Extremes in Tropical Land Areas. Journal of Climate 18,
5095–5109.
Malet, J.-P., Thiery, Y., Puissant, A., Hervás, J., Günther, A., Grandjean, G., 2009.
Landslide susceptibility mapping at 1:1M scale over France: exploratory results
296 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management
with a heuristic model. In: Malet, J.-P., Remaı̂tre, A., Boogard, T. (Eds.), Proc.
International Conference on Landslide Processes: from Geomorphologic
Mapping to Dynamic Modelling, 6–7 February 2009. Strasbourg, France, CERG
Editions, Strasbourg, pp. 315–320CERG Editions, Strasbourg.
Mannaerts, C.M., Maathuis, B.H.P., Molenaar, M. and Lemmens, R.L.G. 2009. The
ITC GEONETCast toolbox : a geo capacity building component for education
and training in global earth observation and geo information provision to
society. In: IGARSS 2009: Proceedings of the 2009 IEEE international
geoscience and remote sensing symposium : Earth observation, origins and
applications, July 12–17, 2009, Cape Town, South Africa, 385–388.
Marcus, W.A., Fonstad, M.A., 2008. Optical remote mapping of rivers at
sub-meter resolutions and watershed extents. Earth Surf. Process. Landf. 33(1),
4–24.
Martha, T.R., Kerle, N., Jetten, V., Van Westen, C.J., Kumar, K.V., 2010.
Characterising spectral, spatial and morphometric properties of landslides for
semi - automatic detection using object - oriented methods. Geomorphology
116(1-2), 24–36.
Maskrey, A., 1998. Navegando entre brumas. La aplicación de los sistemas de
información geográfica al análisis del riesgo en América Latina. ITDG-Peru. LA
RED.
Massonnet, D., Feigl, K.L., 1998. Radar interferometry and its application to
changes in the Earth’s surface. Rev. Geophys 36, 441–500.
McCall, M.K., 2003. Seeking good governance in participatory-GIS. Habitat
International 27, 549–573.
McCall, M.K., 2008. Participatory Mapping and Participatory GIS (PGIS) for CRA,
Community DRR and Hazard Assessment. Provention Consortium. http://
www.proventionconsortium.org/themes/default/pdfs/CRA/PGIS_Sept08.pdf
Metternicht, G., Hurni, L., Gogu, R., 2005. Remote sensing of landslides: An
analysis of the potential contribution to geo-spatial systems for hazard
assessment in mountainous environments. Remote Sensing of Environment
98(23), 284–303.
Miller, P., Mills, J., Edwards, S., Bryan, P., Marsh, S., Mitchell, H., Hobb, P., 2008.
A robust surface matching technique for coastal geohazard assessment and
management. ISPRS Journal of Photogrammetry & Remote Sensing 63,
529–542.
Miliaresis, G.C., 2001. Geomorphometric mapping of Zagros Ranges at regional
scale. Computers & Geosciences 27, 775–786.
Moel, H., de, van Alphen, J., Aerts, J.C.J.H., 2009. Flood maps in Europe –
methods, availability and use. Nat. Hazards Earth Syst. Sci. 9, 289–301.
Molina, S., Lang, D.H., Lindholm, C.D., 2010. SELENA – An open-source tool for
seismic risk and loss assessment using a logic tree computation procedure.
Computers & Geosciences 36(3), 257–269.
Montague, P., 2004. Reducing the harms associated with risk assessments.
Environmental Impact Assessment Review 24, 733–748.
Montoya, L., 2003. Geo-data acquisition through mobile GIS and digital video: an
urban disaster management perspective. Environmental Modelling & software
18(10), 869–876.
Moore, I.D., Grayson, R.B., Ladson, A.R., 2001. Digital terrain modelling: a review
of hydrological, geomorphological, and biological applications. Hydrological
Processes 5(1), 3–30.
Mouton, F., Nordbeck, O., 2003. Cyclone database manager: a tool for converting
point data from cyclone observations into tracks and wind speed profiles in a
GIS. University of Geneva.
Müller, M., Vorogushyn, S., Maier, P., et al., 2006. CEDIM Risk Explorer - a map
server solution in the project "Risk Map Germany". Natural Hazards and Earth
System Sciences 6, 711–720.
MunichRe, 2010. NATCATSERVICE. Downloadcenter for statistics on natural
catastrophes. Münchener Rückversicherungs-Gesellschaft, Geo Risks Research,
Munich, Germany. http://www.munichre.com/en/reinsurance/business/non-life/
georisks/natcatservice/default.aspx
Nadim, F., Kjekstad, O., 2009. Assessment of Global High-Risk Landslide Disaster
Hotspots. In: Sassa, K., Canuti, P. (Eds.), Landslides - Diaster Risk Reduction.
Springer, pp. 213–221.
Nadim, F., Kjekstad, O., Peduzzi, P., Herold, C., Jaedicke, C., 2006. Global landslide
and avalanche hotspots. Landslides 3(2), 159–174.
NERIES, 2009. Network of Research Infrastructures for European Seismology. http://
neries.knmi.nl
NWIS, 2010. National Water Information system. United States Geological Survey.
http://nwis.waterdata.usgs.gov/nwis
O’Brien, M., 2000. Making better Environmental Decisions; An Alternative to Risk
Assessment. MIT Press, Cambridge, Massachusets, USA.
O’Keefe, P., Westgate, K., Wisner, B., 1976. Taking the naturalness out of natural
disasters. Nature 260, 566–567.
Okuyama, Y., Chang, S.E. (Eds.), 2004. Modeling spatial and economic impacts of
disasters. Springer, Advances in spatial science, 329 pp329 pp.
OneGeology, 2010. OneGeology Portal with worldwide geological data. http://
portal.onegeology.org/
OpenSHA, 2010. OpenSHA open source platform. Southern California Earthuake
Center (SCEC) at the University of Southern California (USC) and the United
States Geological Survey (USGS). http://www.opensha.org/
OpenStreetMap, 2010. Open Street Map. http://www.openstreetmap.org/
ORCHESTRA, 2009. Reference Model–ORCHESTRA Architecture (RM-OA). EC
ORCHESTRA project. http://www.eu-orchestra.org/documents.shtml
Oude Elberink, S., vossleman, G., 2009. Building Reconstruction by Target Based
Graph Matching on Incomplete Laser Data: Analysis and Limitations. Sensors
2009 9, 6101–6118.
PAGER, 2010. PAGER - Prompt Assessment of Global Earthquakes for Response.
United States Geological Survey. http://earthquake.usgs.gov/earthquakes/pager/
Peduzzi, P., Dao, H., Herold, C., 2005. Mapping Disastrous. Natural Hazards Using
Global Datasets, Natural Hazards 35(2), 265–289.
Peduzzi, P., Dao, H., Herold, C., Mouton, F., 2009. Assessing global exposure and
vulnerability towards natural hazards: the Disaster Risk Index. Nat. Hazards Earth
Syst. Sci. 9, 1149–1159.
Pelling, M., 2003. The Vulnerability of Cities. Natural disasters and Social
Resilience. Earthscan Publications, London, 256 p.
Peters, A.J., Walter Shea, E.A., JI, L., Vliia, A., Hayes, M., Svoboda, M.D., 2002.
Drought Monitoring with NDVI-Based Standardized Vegetation Index.
Photogrammetric Engineering & Remote Sensing 68(1), 71–75.
Peters Guarin, G., 2008 Integrating local knowledge into GIS based flood risk
assessment, Naga city, The Philippines. Wageningen, Enschede, Wageningen
University, ITC, 2008. ITC Dissertation 157, 352 p.
Peters Guarin, G., Van Westen, C.J., Montoya, A.L., 2005. Community-based flood
risk assessment using GIS for the town of San Sebastian, Guatemala. J. of
Human Security and Development 1, 29–49.
Pike, R.J., 2000. Geomorphometry – diversity in quantitative surface analysis.
Progress in Physical Geography 24(1), 1–20.
Plapp, T., 2001. Understanding risk perception from natural hazards. In: Amman,
Dannenmann, Vulliet (Eds.), RISK 21 – Coping with Risks due to Natural
Hazards in the 21st Century. Taylor & Francis Group, London, pp. 101–108.
Priest, G., A. Baptista, E. Myers III, R. Kamphaus 2001. Tsunami hazard assessment
in Oregon. ITS 2001 Proceedings, NTHMP Review Session, R-3, 55–65.
Priestnall, G., Jaafar, J., Duncan, A., 2000. Extracting urban features from LiDAR
digital surface models. Computers, Environment and Urban Systems 24(2),
65–78.
Provention Consortium, 2010. Community Risk Assessment methodologies and
case studies. http://www.proventionconsortium.org/pageid=43
RADIUS, 1999. RADIUS method (Risk Assessment Tools for Diagnosis of Urban
Areas against Seismic Disasters). http://www.geohaz.org/news/images/
publications/RADIUS_RiskAssessment.pdf
Rautela, P., 2005. Indigenous Technical Knowledge Inputs for Effective Disaster
Management in the Fragile Himalayan Ecosystem. In Disaster Prevention and
Management 14(2), 233–241.
Reese, S., Bell, R.G., King, A.B., 2007. RiskScape: a new tool for comparing risk
from natural hazards. Water & Atmosphere 15(3), 24–25.
Reese, S., Cousins, W.J., Power, W.L., Palmer, N.G., Tejakusuma, I.G., Nugrahadi,
S., 2007. Tsunami vulnerability of buildings and people in South Java – field
observations after the July 2006 Java tsunami. Nat. Hazards Earth Syst. Sci., 7,
573–589.
Refsgaard, J.C., Storm, B., 1995. MIKE-SHE. In: Singh, V.J. (Ed.), Computer
models in watershed hydrology. Water Resources Publications.
Reidsma, P., Ewert, F., Boogaard, H., Van Diepen, K., 2009. Regional crop
modelling in Europe: The impact of climatic conditions and farm characteristics
on maize yields. Agricultural Systems 100(1-3), 51–60.
Reliefweb, 2010. ReliefWeb, serving the needs of the humanitarian relief community.
http://www.reliefweb.int
Rhind, D.W., 1991. Counting the people: The role of GIS. In: Maguire, D.J.,
Goodchild, M.F., Rhind, D.W. (Eds.), Geographical Information Systems.
Longman, London, pp. 127–137.
Risicokaart, 2008. National hazard and risk map of the Netherlands. Landelijke
Beheer Organisatie Risicokaart, Interprovinciaal Overleg (IPO). http://
www.risicokaart.nl
Sahana, 2010. Sahana Free and Open Source Disaster Management System. http://
sahanafoundation.org/
Salvi, S., Atzori, S., Tolomei, C., et al., 2004. Inflation rate of the Colli Albani
volcanic complex retrieved by the permanent scatterers SAR interferometry
technique. Geophys. Res. Lett. 31, 21.
 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management
Sarker, J.K., Ansary, M.A., Rahman, M.S., Safiullah, M.M., 2010. Seismic hazard
assessment for Mymensingh, Bangladesh. Environ Earth Sci., 60, 643–653.
Schmidt, J., Matcham, I., Reese, S., et al., 2011. Quantitative multi-risk analysis for
natural hazards: a framework for multi-risk modeling. Natural Hazards 58,
1169–1192.
Schumann, G., Matgen, P., Pappenberger, F., Hostache, R., Puech, C., Hoffmann, L.,
Pfister, L., 2007. High-resolution 3-D flood information from radar for effective
flood hazard management. IEEE Trans. Geosci. Remote Sens 45(6), 1715–1725.
Sedan, O. and Mirgon, C. (2003). Application ARMAGEDOM Notice
utilisateur,BRGM open file BRGM/RP-52759-FR.
Sentinel Asia, 2010. Sentinel Asia initiative for space-based information and
disaster management in the Asia-Pacific region. Web site: http://
dmss.tksc.jaxa.jp/sentinel/
SERTIT, 2010. SErvice Régional de Traitement d’Image et de Télédétection.
University of Strasbourg, http://sertit.u-strasbg.fr
Smith, L.C., 1997. Satellite remote sensing of river inundation area, stage, and
discharge: a review. Hydrological Processes 11, 1427–1439.
Smith, K., Petley, D.N., 2008. Environmental hazards. Assessing risk and reducing
disaster. Taylor & Francis, London.
Spence, R.J.S., Baxter, P.J., Zuccaro, G., 2004. Building vulnerability and human
casualty estimation for a pyroclastic flow: a model and its application to
Vesuvius. J. Volcanol 133, 321–343.
Spence, R.J.S., Kelman, I., Calogero, E., Toyos, G., Baxter, P.J., Komorowski, J.C.,
2005. Modelling expected physical impacts and human casualties from
explosive volcanic eruptions. Natural Hazards and Earth System Sciences 5,
1003–1015.
Stelling, G.S., Kernkamp, H.W.J., Laguzzi, M.M., 1998. Delft Flooding System, a
powerful tool for inundation assessment based upon a positive flow simulation.
In: Babovic, Larsen (Eds.), Hydro-informatics 1998. Balkema, Rotterdam the
Netherlands, pp. 449–456.
Stilla, U., Soergel, U., Thoennessen, U., 2003. Potential and limits of InSAR data
for building reconstruction in built-up areas. ISPRS Journal of Photogrammetry
and Remote Sensing 58(1-2), 113–123.
Tatem, A.J., Noor, A.M., von Hagen, C., Di Gregorio, A., Hay, S.I., 2007. High
Resolution Population Maps for Low Income Nations: Combining Land Cover
and Census in East Africa. PLoS ONE 2(12), e1298. http://dx.doi.org/10.1371/
journal.pone.0001298.
Thoma, D.P., Gupta, S.C., Bauer, M.E., Kirchoff, C.E., 2005. Airborne laser scanning
for riverbank erosion assessment. Remote Sensing of Environment 95, 493–501.
Tilling, R.I., 1989. Volcanic Hazards, Amer. Geophys. Union, Washington DC, 123 pp.
Tralli, D.M., Blom, R.G., Zlotnicki, V., Donnellan, A., Evans, D.L., 2005. Satellite
remote sensing of earthquake, volcano, flood, landslide and coastal
inundation hazards. ISPRS Journal of Photogrammetry and Remote Sensing 59,
185–198.
Trigg, S.N., Roy, D.P., Flasse, S., 2005. An in situ study of the effects of surface
anisotropy on the remote sensing of burned savannah. International Journal of
Remote Sensing 26, 4869–4876.
Twigg, J., 2004. Disaster Risk Reduction: mitigation and preparedness in
development and emergency programming. Overseas Development Institute,
Humanitarian Practice Network, London, Good Practice Review no. 9.
UNEP/DEWA/GRID, 2010. PREVIEW Global Risk Data Platform. United Nations
Environment Programme, Global Resource Information Database, Europe,
Geneva. http://www.grid.unep.ch/activities/earlywarning/preview/index.php
UN-ESCAP, 2003. Guidelines on Participatory Planning and Management for Flood
Mitigation and Preparedness, United Nations- Economic and Social Commission
for Asia and the Pacific (ESCAP). Water Resources Series No. 82. UN.
New York.
UN-ISDR, 2004. Terminology of disaster risk reduction. United Nations, International
Strategy for Disaster Reduction, Geneva, Switzerland. http://www.unisdr.org/eng/
library/lib-terminology-eng%20home.htm
UN-ISDR, 2005a. Hyogo framework for Action 2005-2015. Building the Resilience
of Nations and Communities to Disasters. United Nations, International Strategy
for Disaster Reduction, Geneva, Switzerland. http://www.unisdr.org/eng/hfa/
hfa.htm
UN-ISDR, 2005b. Reducing Disaster Risk: A Challenge for Development. United
Nations, International Strategy for Disaster Reduction, Geneva, Switzerland http://
www.undp.org/bcpr/disred/rdr.htm
UN-ISDR, 2009. Economic damages: share of GDP, by natural disaster and country.
United Nations, International Strategy for Disaster Reduction, Geneva,
Switzerland. http://www.unisdr.org/disaster-statistics/top50.htm
UNPD, 2010a. World Population Prospects: the 2008 Revision. United Nations
Population Division. United Nations, Department of Economic and Social Affairs,
New York. http://esa.un.org/unpd/wpp2008/index.htm
297
UNPD, 2010b. World Urbanization Prospects:the 2009 Revision. United Nations
Population Division. United Nations, Department of Economic and Social Affairs,
New York, http://esa.un.org/unpd/wup/index.htm
UNOSAT, 2010. Institute for Training and Research (UNITAR) Operational Satellite
Applications Programme. http://unosat.web.cern.ch/unosat/
UN-SPIDER, 2010. United Nations Platform for Space-based Information for
Disaster Management and Emergency Response - UN-SPIDER. http://www.un-
spider.org/
USGS, 1996. HYDRO1k Elevation Derivative Database. U.S. Geological Survey,
Sioux Falls, South Dakota, USA. http://eros.usgs.gov/#/Find_Data/
Products_and_Data_Available/gtopo30/hydro
USGS, 1997. GTOPO30. Global 30-Arc-Second Elevation Data Set. U.S. Geological
Survey, Sioux Falls, South Dakota, USA. http://gcmd.nasa.gov/records/
GCMD_GTOPO30.html
USGS, 2010. National Geological Map Database (NGMB). United States Geological
Survey. http://ngmdb.usgs.gov/
Ushahidi, 2010. Ushahidi http://www.ushahidi.com/
Van Asch, T.W.J., Malet, J.P., van Beek, L.P.H., Amittrano, D., 2007. Techniques,
issues and advances in numerical modelling of landslide hazard. Bulletin de la
Societé Geologique de France 178, 65–88.
Van Beek, L.P.H., van Asch, T.W.J., 2004. Regional assessment of the effects of
land-use change and landslide hazard by means of physically based modeling.
Natural Hazards 30(3), 289–304.
Van der Knijff, J.M., Younis, J., De Roo, A.P.J., 2010. LISFLOOD: a GIS-based
distributed model for river-basin scale water balance and flood simulation.
International Journal of Geographical Information Science 24(2), 189–212.
Van Westen, C.J. (Ed.), 2009. Distance Education course on the use of spatial
information in Multi-hazard risk assessment.
Van Westen, C.J., Castellanos Abella, E.A., Sekhar, L.K., 2008. Spatial data for
landslide susceptibility, hazards and vulnerability assessment: an overview.
Engineering Geology 102(3–4), 112–131.
Van Westen, C.J., Montoya, L., and Boerboom, L. 2002. Multi-Hazard Risk
Assessment using GIS in urban areas: A case study for the city of Turrialba,
Costa-Rica, In: Proc. Regional workshop on Best Practise in Disaster Mitigation,
Bali, 120–136.
Verdin, K.L., and Greenlee, S.K., 1996. Development of continental scale digital
elevation models and extraction of hydrographic features. In: Proceedings, Third
International Conference/Workshop on Integrating GIS and Environmental
Modeling, Santa Fe, New Mexico, January 21-26, 1996. National Center for
Geographic Information and Analysis, Santa Barbara, California.
Verelst, L., 1999. Recommendations for Geo-referencing EM-DAT Data Base,
Consultancy Report prepared for the Asia FIVIMS Project and the CRED, Rome.
Vigni, C., et al., 2005. Insight into the 2004 Sumatra–Andaman earthquake from
GPS measurements in southeast Asia. Nature 436, 201–206.
Villagrán de Leon, J.C., 2006. Vulnerability: A Conceptual and Methodological
Review. UNU-EHS. UNU. No 4/2006.
Virtual Disaster Viewer (2009). Virtual disaster viewer. ImageCat. http://
virtualdisasterviewer.com
Wadge, G., Wislocki, A.P., Pearson, E.J., 1993. Spatial analysis in GIS for natural
hazard assessment. In: Goodchild, M.F., Parks, B.O., Steyaert, L.T. (Eds.),
Environmental Modeling with GIS. Oxford University Press, New York, pp.
332–338.
Wald, D.J., Earle, P.S., Quitoriano, V., 2004. Topographic Slope as a Proxy
for Seismic Site Correction and Amplification. EOS. Trans. AGU 85(47),
F1424.
Wald, D.J., Quitoriano, V., Heaton, T.H., Kanamori, H., Scrivner, C.W., Worden, B.C.,
1999. TriNet "ShakeMaps": Rapid generation of peak ground-motion and
intensity maps for earthquakes in southern California. Earthquake Spectra 15(3),
537–556.
White, D.H., Walcott, J.J., 2009. The role of seasonal indices in monitoring and
assessing agricultural and other droughts: a review. Crop and Pasture Science
60, 599–616.
Whitehouse, G., 2001. Community involvement in flood and floodplain
management: the Australian Scene. Best practice guidelines for floodplain
management in Australia. Draft report. Canberra. 17 p.
Wilhite, D.A., 2000. Drought: A Global Assessment. Routledge Publishers, U.K,
Natural Hazards and Disasters Series. World Bank, 2010. GDP per capita. http://
data.worldbank.org/indicator/NY.GDP.PCAP.CD
World Bank, 2011. Population data. http://data.worldbank.org/indicator/SP.POP.TOTL
Xie, M., Tetsuro, E., Zhou, G., Mitani, Y., 2003. Geographic Information Systems
based three-dimensional critical slope stability analysis and landslide hazard
assessment. Journal of Geotechnical and Geoenvironmental Engineering
129(12), 1109–1118.
298 Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management

Yeh, C.H., Loh, C.H., Tsai, K.C., 2006. Overview of Taiwan Earthquake Loss
Estimation System. Natural Hazards 37(1-2), 23–37.
Zerger, A., Smith, D., 2003. Impediments to using GIS for real-time disaster
support. Computers, Environmental and Urban Systems 27, 123–141.
Zheng, J., Wang, W.C., Ge, Q., Man, Z., Zhang, P., 2006. Precipitation Variability
and Extreme Events in Eastern China during the Past 1500 Years. Terr. Atmos.
Ocean. Sci. 17(3), 579–592.
Zhou, Q., Liu, X., 2004. Analysis of errors of derived slope and aspect related to
DEM data properties. Computers & Geosciences 30(4), 369–378.
Zuccaro, G., Cacace, F., Spence, R., Baxter, P., 2008. Impact of explosive eruption
scenarios at Vesuvius. Journal of Volcanological and Geothermal Research
178(3), 416–453.

Biographical Sketch

Dr. Cees Van Westen is an associate professor in the Department of Earth Systems Analysis, and on the Faculty of
Geo-Information Science and Earth Observation (ITC), University of Twente, the Netherlands. He obtained his
M.S. in Physical Geography from the University of Amsterdam (1988) and his Ph.D. in Engineering Geology from
the Technical University of Delft (1993), with emphasis on geographic information systems for landslide hazard
zonation. Since 2005, he is the Director of the United Nations University - ITC School on Geo-information for
Disaster Risk Management. He received the ITC research award in 1993 and the Richard Wolters Prize of the
International Association of Engineering geology (IAEG) in 1996. He is specialized on topics related with the use
of spatial information for landslide hazard and risk assessment, participatory GIS for flood-risk assessment,
volcanic-hazard assessment, seismic hazard and risk assessment, technological risk assessment, and multi-hazard
risk assessment. Most of his research is in the field of landslides, dealing with topics including generation of event-
based landslide inventories using remote sensing, historical records and field mapping, combination of heuristic
and statistical models for landslide susceptibility analysis, dynamic modelling of landslide initiation, landslide
run-out analysis, and different approaches for landslide-risk assessment. He has published more than 45 papers in
ISI journals, 10 book chapters, and numerous papers in conference proceedings. He has been involved in many
projects funded by the EU (FP6, FP7), World Bank, ADB, Dutch government, and US-AID. He is currently the
project coordinator of the CHANGES project, an EU FP7 Marie Curie International Training Network.
 3.11 Geovisualization
MJ Smith, Kingston University, Surrey, UK
JK Hillier, Loughborough University, Loughborough, UK
J-C Otto and M Geilhausen, University of Salzburg, Salzburg, Austria
r 2013 Elsevier Inc. All rights reserved.

3.11.1 Introduction 300

3.11.2 Background 301
3.11.2.1 Historical Context 301
3.11.2.2 Geomorphology and Geovisualization 303
3.11.2.3 Applications and Emergent Technologies 304
3.11.3 Visual Processing 305
3.11.3.1 Detection of Landforms 306
3.11.3.2 Image Enhancement 307
3.11.3.3 Enhancement of DEMs 309
3.11.3.3.1 Regional–residual separation 309
3.11.3.3.2 Land-surface parameters 310
3.11.3.4 Recommendations for Terrain Visualization 312
3.11.4 Visual Interaction 313
3.11.4.1 Display 313
3.11.4.2 Digitization and Overlay 313
3.11.4.3 2D to 2.5D in Space 314
3.11.4.4 3D in Space 315
3.11.4.5 Virtual Globes 315
3.11.5 Visual Outputs 316
3.11.5.1 Geomorphological Maps 316
3.11.5.1.1 Legend systems 316
3.11.5.1.2 Map design 316
3.11.5.2 Digital Mapping 318
3.11.5.2.1 Open standards 319
3.11.5.2.2 GeoPDF 319
3.11.5.2.3 Principles of web mapping and WebGIS 319
3.11.6 Conclusions 321
References 322

Glossary Detectability How readily a landform may be

ALS Airborne laser scanning is used to collect high-
resolution data for generation of a digital surface or digital
terrain model.
Bathymetry Depth of the Earth’s surface in water-covered
areas.
Decluttering Regional–residual separation performed on
a digital surface model with the intention of removing
superficial clutter to leave ‘bare earth’ heights for the
generation of a digital terrain model.
DEM A digital elevation model refers to any digital
representation of the topography that characterizes the
altitude field. See also DTM and DSM.
distinguished or correctly identified by an interpreter in an
image that is displayed. The concept may be extended to
automated approaches for identification and mapping of
landforms.
DSM A digital surface model represents the highest, upper
surface of all visible objects on the landscape, as recorded
using a particular measurement technique.
DTM A digital terrain model is a representation of the
Earth’s surface free from anthropogenic features such as
buildings and vegetation.
Elevation Elevation is the height above a reference datum,
historically sea level.

Smith, M.J., Hillier, J.K., Otto, J.-C., Geilhausen, M. 2013. Geovisualization.

In: Shroder, J. (Editor in Chief), Bishop, M.P. (Ed.), Treatise on
Geomorphology. Academic Press, San Diego, CA, vol. 3, Remote Sensing
and GIScience in Geomorphology, pp. 299–325.

Treatise on Geomorphology, Volume 3 http://dx.doi.org/10.1016/B978-0-12-374739-6.00054-3 299

300 Geovisualization

Geomorphology The study of the form and processes that
shape the Earth or other planetary surfaces, and how they
have evolved through time.
Geovisualization This generically refers to the depiction
of spatial data. In geomorphology, it facilitates geospatial
analysis and processing. See the International Cartographic
Association (ICA) Commission on Geovisualization
(http://geoanalytics.net/ica/).
Heuristics Using experience-based rules to solve a
problem.
Hyperspectral imagery Imagery produced from an
instrument with sensors sufficient to create observations
across a continuous region of the spectrum for each pixel.
Kernel In image processing, the kernel refers to the width
of a spatial filter and is synonymous with the window size.
Landform A shape or morphology of a particular part of
the landscape that is identified as a distinct spatial entity.
Multispectral imagery Imagery created from an
instrument with sensors sensitive to different regions of the
electromagnetic spectrum from a relatively small number of
spectral bands that exhibit a relatively large bandwidth.
Panchromatic (image) A single image generated from a
sensor that records the entire region of the visible spectrum.
Point cloud Clustering of point data in a 3D coordinate
space. Data consist of irregularly spaced (x,y,z) triplets that
can be used to generate a DSM or DTM.
Raster A data format where attributes are represented in
tessellations that usually form a regularized spatial grid or
layer. For satellite imagery and DEMs, reflectance and
surface elevation, values are stored in each grid cell.
Raw data Data as measured from sensor systems,
although significant processing has usually been applied
before presentation to the user community.
Regional–residual (relief) separation (RRS) Numerical
operation to isolate a layer of spatially varying vertical
thickness related to features of interest from a DEM.
Signal-to-noise ratio Ratio of information about the
landforms of interest (i.e., topographic signal) to all other
features in an image or DEM. The ratio may be increased by
filtering the data before viewing, or by how the data are
displayed.
Sounding Measurement of depth. See bathymetry.
TLS Terrestrial laser scanning is used to produce high-
resolution digital elevation models. It represents a field-
based approach.
Topographic signal The presence of landforms of interest
in a particular study as expressed in datasets such as height.
Topography The shape of the Earth’s surface. This can
include landscape features such as vegetation and man-
made features.
Visual processing The act of visual exploration of the
terrain or landscape for the purpose of landform
identification, or in order to generate other derived
observational data sets.
VNIR The visible and near infra-red regions of the
electromagnetic spectrum.

Abstract

Geovisualization involves the depiction of spatial data in an attempt to facilitate the interpretation of observational and
simulated datasets through which Earth’s surface and solid Earth processes may be understood. Numerous techniques can
be applied to imagery, digital elevation models, and other geographic information system data layers to explore for patterns
and depict landscape characteristics. Given the rapid proliferation of remotely sensed data and high-resolution digital
elevation models, the focus is on the visualization of satellite imagery and terrain morphology, where manual human
interpretation plays a fundamental role in the study of geomorphic processes and the mapping of landforms. A treatment of
some techniques is provided that can be used to enhance satellite imagery and the visualization of the topography to
improve landform identification as part of geomorphological mapping. Visual interaction with spatial data is an important
part of exploring and understanding geomorphological datasets, and a variety of methods exist ranging across simple
overlay, panning and zooming, 2.5D, 3D, and temporal analyses. Specific visualization outputs are also covered that focus
on static and interactive methods of dissemination. Geomorphological mapping legends and the cartographic principles for
map design are discussed, followed by details of dynamic web-based mapping systems that allow for greater immersive use
by end users and the effective dissemination of data.

3.11.1 Introduction concerned with geomorphological mapping of various as-

Earth scientists aim to study and understand the Earth’s
dynamics systems. The interaction of atmospheric, geologic,
and ecological processes greatly influences surface pro-
cesses and topographic evolution. Geomorphological stud-
ies focus on understanding various forcing factors, process
mechanics, and the role of surface processes in landscape
evolution. Fundamental to geomorphology is the linking
of process to form and the identification of landforms
that have genetic, morphological, and chronological
characteristics. Therefore, geomorphology is fundamentally
pects of the system including process domains, materials
and structure, morphology, landforms, and chronology.
Visualization of the landscape is a primary means to study
geomorphological conditions, and this is accomplished
through the use of field studies and Earth observation that
focus on collecting spatio-temporal land-surface information.
To study landforms, however, it is necessary to be able to
characterize the topography and specifically identify various
landforms and features/conditions that are the result of spe-
cific processes, feedback mechanisms, or system couplings.
Consequently, it is essential to collect such data and analyze
 Geovisualization 301

patterns and trends so that new insights, relationships, and

understandings advance the discipline. Quantitative analysis
of data can be used and numerous analytical spatial and
temporal techniques exist (e.g., geostatistics and Fourier an-
alysis). The power of the human visualization system must
also be considered, as accurate, detailed and reliable visual-
ization can significantly augment and facilitate the interpret-
ation of quantitative analysis results, and can also be of
primary importance to the geomorphological study of pro-
cesses shaping the Earth and other planetary landscapes, given
that completely automated spatial analysis and mapping
models are still being developed and evaluated for a wide
range of geomorphological applications.
In geomorphology, the shape of the land is so funda-
mental, that although various forms of spatial data exist to
characterize different aspects of the land surface, topographic
information is of central importance to geovisualization.
Other forms of landscape information such as imagery, bio-
physical surface properties, land cover, and land use, as well as
other thematic information can easily be integrated with
topographic information to facilitate effective geovisualization
of the landscape. Consequently, many techniques for the ex-
traction of information from spatial data sets also represent
visualization techniques. This includes the extensive and
complex range of image processing and spatial analysis,
techniques that can be applied to a variety of datasets, al-
though they may not be visually displayed in an effective
manner.
Therefore, the scope of this chapter focuses on ‘visual pro-
cessing,’ ‘visual interaction’ and ‘visual outputs,’ all of which
form part of the geovisualization of terrain. This chapter does
not cover all aspects or techniques of scientific visualization for Figure 1 Examples of common methods of visualization, Bowridge,
a variety of different datasets, but instead focuses on the use of Scotland. (a) Geomorphological map of glacial landforms, utilizing
spatial data (e.g., imagery and topography) and geovisualiza- Ordnance Survey 1:25 000 base mapping with 25-ft contour lines.
Reproduced from Google Maps. (b) Aerial photo (‘A’ notes the location
tion for geomorphological studies and mapping. It provides
of Bowrdige drumlin depicted in (a) and (c). Reproduced from Rose,
an historical perspective, formal definitions of geovisualization
J., Smith, M.J., 2008. Glacial geomorphological maps of the Glasgow
and an outline of emergent technologies and applications. The region, western central Scotland. Journal of Maps v2008, 399–416,
chapter then addresses visual processing and the optimization with permission from JOM. Note the photo is as composite as
of imagery for landform visualization. The mechanics of vari- presented in Google EarthTM). (c) Oblique photo (J. Rose). The arrow
ous methods are then examined by which a user can interact indicates the approximate photo orientation.
with data, given that it has been prepared for viewing. Finally,
a discussion follows on the dissemination of data in both (Figure 1(b)) or oblique (Figure 1(c)) aerial photos. Alongside
traditional paper and electronic formats, highlighting various maps, alternative portrayals of spatial data have been routinely
visualization products. used (e.g., Kraak and Ormeling, 2010; Bonham-Carter, 1994)
including field sketches (Figure 2), conceptual diagrams
(Figure 3), vertical/oblique aerial photos (Figure 1(b) or 1(c)),
satellite imagery, digital elevation models (DEMs), and aug-
3.11.2 Background
mented reality (e.g., Reitmayr et al., 2005).
Early geomorphological maps were generally produced for
3.11.2.1 Historical Context
military and engineering purposes and designed for use in the
The visual interpretation of data and numerous conceptual field (Klimaszewski, 1982). With the onset of a morphological
ideas have been a key aspect of human understanding view of landscapes and the description of their physiography
(e.g., Kraak and Ormeling, 2010). Historically, maps have at the turn of the twentieth century, landform analysis
commonly been created to assist in interpretation and under- and morphological description became new objectives for
standing, as attribute information is linked to spatial coordin- geomorphological mapping (e.g., Passarge, 1912, 1914). From
ates or location, depicting spatial patterns. Figure 1(a) this standpoint, maps were more than just a medium for
illustrates the utility of creating maps for the investigation of visualization but represented a research tool for landscape
geomorphic processes by displaying the distribution of glacial analysis providing a generalized inventory of landforms, sur-
landforms and their composition (indicated by the symbol fill face structures, geomorphological processes, surface and sub-
color), something that is difficult to convey using vertical surface materials, and genetic information. The applications
302 Geovisualization

Narrow
rock
outcrop

10 m high, 45°
bare soil slope

Level vegetated
rock shelf, 1.5 m
deep

2 m high rock face

B Back and side
slopes of basin
Appr all about 45°
ox. 8
m

s,
Rocky step
ai ns t be dd ing plane
ag

2m Distance A−B approx. 18 m

A
Figure 2 Field sketch of a slope profile highlighting morphological features on the eastern slope of Sg. Belalong, northwest Borneo. Reproduced
from Dykes, A.P., 1995. Hydrological controls on shallow mass movements and characteristic slope forms in the tropical rainforest of
Temburong District, Brunei, Department of Geography. Unpublished PhD thesis, King’s College, University of London.

of geomorphological maps range from simple descriptions of
a field site (e.g., accompanying a journal publication or con-
struction site report), to land system analyses (Bennett et al.,
2010), land surveys, land management, or natural hazard
assessment (Brunsden et al., 1975; Seijmonsbergen and de
Graaff, 2006).
Maps remained the fundamental geomorphological output
through to the 1980s, as they provided both 2D visualization
and an effective data storage paradigm for spatial data. Geo-
morphological mapping subsequently declined due to a pre-
occupation with cartographic symbolization and a move to
field-scale experimentation. Since the 1990s, widely available
remotely sensed data, progress in computing power, and
rapid improvements of geographic information systems (GIS;
e.g., Wessel and Smith, 1998; ESRI, 2003) have permitted the
combination of field-scale and regional approaches that have
led to a resurgence in mapping. In particular, the emergence of
GIS through the wider field of geographic information science
or geomatics has provided software tools that can be used to
store, manipulate, analyze, and display spatial data. The latest
developments are also permitting more temporal analysis
capabilities. Such information technology provides a powerful
methodology to combine diverse datasets including total-
station data, global positioning system (GPS) measurements,
satellite imagery, postcodes, and historic data forms. Specific-
ally, GIS can be used to prevent data overload, facilitating a
clear and carefully constructed approach for analysis, inter-
pretation, and higher level use of the data and information.
Historically, geovisualization has been a paper-based ana-
log technique; however, it now involves the routine collection
of large digital datasets, detailed analytical analysis, and
computer-based geovisualization. The largest datasets are
generated by sensors aboard mobile platforms, and numerous
datasets can be used for geomorphological applications
(Smith and Pain, 2009). In terms of data volume, satellite
imagery remains the single most important product, although
the issues associated with spectral, spatial, and temporal
resolution can be problematic for specific applications. In
terms of impact upon geomorphology, however, the gener-
ation of DEMs has arguably had the largest influence. This is
profound because a variety of elevation data can be syn-
thesized or fused into a DEM and, once in raster format,
 Geovisualization 303

High altitude glacial

and periglacial
Degraded debris slide
Rock face with rockfall 1
Ancient high level terrace
Boulder field or erosion surface
Degraded sideslope Talus/scree
of former river valley Incised gully
Taluvium cover to Taluvium
Active lower slopes
weathered bedrock
Colluvium 2 2
Colluvium
Rotational slide
Tributary
3
valley
Principal
4 3 valley

4
5
4

Floodplain deposits

Residual soil or
weathered rock Debris slide
5
Braided channel Rockslope
with bar deposits
Rapids
Fan
Alluvium
Young river
terrace deposits
Figure 3 Conceptual block diagram of high-mountain environments showing landforms and processes of five major zones: 1 – high-altitude
glacial and periglacial; 2 – free rock faces and associated slopes; 3 – degraded middle slopes and ancient valley floors; 4 – active lower slopes;
and 5 – valley floors. Reproduced from Fookes, P.G., Sweeney, M., Manby, C.N.D., Martin, R.P., 1985. Geological and geotechnical engineering
aspects of low-cost roads in mountainous terrain. Engineering Geology 21, 1–152.

image-processing or spatial analysis techniques, common in
geovisualization, can be applied. Contemporary mapping is
therefore computer based, reliant upon the input of digital
datasets, with analysis and output performed using GIS.
3.11.2.2 Geomorphology and Geovisualization
Geovisualization is widely utilized in geomorphology for the
exploration and analysis of spatio-temporal data. Although a
spatial framework is not a requirement for geomorphological
study (e.g., Smith and McClung, 1997), it is natural in many
studies to use ‘space’ as the organizing paradigm (e.g., Benetti
et al., 2010). A better understanding of many geomorphic
phenomena can therefore be gained through the recording
(i.e., ‘mapping’) and analysis of their spatial distribution. In
the past, the observed distribution and form (i.e., morph-
ology) would typically have been communicated using a
geomorphological map (e.g., Rose and Smith, 2008).
The term ‘geovisualization’ is a contraction of geographic
visualization. This was first mooted by MacEachren et al.
(2004), who defined geovisualization as a ‘‘process for lever-
aging y data resources to meet scientific and societal needs
and a research field that develops visual methods and tools
to support a wide array of geospatial data applications.’’
This definition extends geovisualization beyond simply
communication using the presentation of an image or images.
The definition proposes that the term ‘visualization’ include
four primary functions (Figure 4): (1) ‘exploration’ of datasets
in order to find patterns and information; (2) ‘analysis’ of
patterns and relationships between data; (3) ‘synthesis’, in-
volving the generation of an overview and understanding of
the inherent information; and (4) ‘presentation’ of the find-
ings. Within this definition, geovisualization incorporates a
wide gamut of activities and capabilities. In contrast, Kraak
(2008) suggested that the term geovisualization was being
used increasingly widely and indiscriminately, becoming
synonymous with ‘mapping,’ and therefore increasingly less
useful as a term. Consequently, he favors the use of the term
‘geovisual analytics’ (Thomas and Cook, 2005; Andrienko
et al., 2007) for the range of activities in and around the
geovisualization cube, implicitly retaining a more focused
definition of ‘geovisualization.’ Discussion exists about the
terminology surrounding geographic visualization; yet, at its
simplest, geovisualization is simply a synthesis of the long-
developed visual communication of cartography with current
digital analytical technologies, principally GIS. Indeed, it
could be argued that in the past cartography was geovisuali-
zation; that is, cartography embodied the sum of geovisual
techniques that were possible before practical, pervasive
desktop computing. The introduction of computer technolo-
gies in the 1960s, however, saw a split in the discipline of
304 Geovisualization
- Explore - analyze - synthesize - present -
Info
. actio
sha ..task
ring ...kn
con owle
stru dge
ctio
n leased. The UK government has released significant quantities
Figure 4 Four functions of geovisualization modified from
MacEachren, A.M., Gahegan, M., Pike, W., Brewer, I., Cai, G.,
Lengerich, E., Hardistry, F., 2004. Geovisualization for knowledge
construction and decision support. Computer Graphics and
Applications 24(1, 24), 13–17, that include exploration, analysis,
synthesis, and presentation. Geovisualization feature-space is defined
by the types of task, user, and interactivity.
geographic visualization with users either focused princi-
pally upon design and communication, or data handling.
The former we now think of as cartography, whereas the
latter became geographic information science. The meaning
of the term geovisualization is therefore debated, but in its
widest, intuitive, sense it is ‘the visual depiction of spatial data
and information.’ It is a convenient term to employ within
geomorphology and is used here with this latter broad
meaning.
3.11.2.3 Applications and Emergent Technologies
Some applications of geovisualization are depicted in
Figure 4, with the axes of the cube illustrating the task being
performed, the user doing the task, and the degree of inter-
action with the data being visualized. Within this, the four
functions proposed by MacEachren et al. (2004) move se-
quentially from highly interactive exploration of the data by
specialists to generate basic observational knowledge, to pre-
senting synthesized information to the public involving
little interaction with the data. ‘Knowledge construction’ at the
start of this sequence requires a specialist with a high degree
of interaction, and can be considered a research-intensive
application. At the opposite end of the spectrum, ‘information
sharing’ generally requires lower levels of interaction by
non-specialists, and is an application for the public and de-
cision makers. This latter area is now an important aspect of
research, as funding bodies are aware of their accountability,
and therefore desire further downstream application, as well
as interaction with the general public or, more generally,
‘public relations.’
Two trends not well represented by the four functions
proposed within the geovisualization cube are the growth in
Publi
both public interaction with data and data distribution to the
widest possible audience. The latter facilitates the former, and
c ... u
with both specialists and the public exploring data, the face of
sers
the cube representing high levels of interaction is becoming
increasingly occupied. Specifically, there are three require-
... sp
ments for this occupation to be achieved:
eciali
• freely or easily available data;
•
s
nonproprietary or ‘open’ formats for geospatial data; and
ts
• free or easily available software to download, visualize, and
analyze data.
low The public release of data is not new (e.g., SYNBAPS; Van
n ...
Wyckhouse, 1973), but there has been increasing political
ter pressure to do so. In the USA, the National Geophysical Data
... in
igh Center (http://www.ngdc.noaa.gov/) makes data publically
H
available, and it is a requirement of funding that data be re-
of data (http://data.gov.uk), and Research Councils UK also
require scientists to place their data in repositories. Data
sharing and archival requires a move away from proprietary
data formats, and this has been achieved through the devel-
opment of industry-standard, open, formats (e.g., Keyhole
Markup Language (KML); see Section 3.11.5.2.1) and openly
specified proprietary formats (e.g., the shapefile format; ESRI,
2003).
The ability to select base data and easily overlay specialist
datasets is fundamental across a spectrum of activities, ranging
from the development to the use of software applications. For
example, Generic Mapping Tools (Wessel and Smith, 1998),
GRASS and QGIS allow scripting to enable advanced users to
create wider dynamic access to data. The Seamount Catalog
(http://earthref.org/SBN/; Koppers et al., 2010) exemplifies a
scripted front end to geospatial data. For both technical and
nonspecialist users, GeoMapApp (http://www.geomapapp.org/;
Carbotte et al., 2004) is a free, easy-to-use software package that
contains datasets, displays (e.g., maps and profiles), and per-
mits importing and overlays. Google Earth performs similar
tasks in an accessible way, encouraging map-making by the
public at large. The public use of geovisualization itself is no
better displayed than in the culture of ‘mashups’ (e.g., Wood
et al., 2007), and by the ease of use of a site that allows anyone
to make maps (e.g., https://arcgis.com).
Emergent technologies are driving research capabilities in
geomorphology and are being developed to meet new chal-
lenges. Innovations in on-screen visualization and public en-
gagement have led to the development of technologies that
allow existing forms of display to be added to, or ‘augmented.’
By enhancing visualization, augmentation presents new op-
portunities for researchers to interact with, and demonstrate,
their work. This is particularly useful for field trips, interaction
with the general public, and consultancy work. Augmentation
takes a base information ‘layer,’ such as a real-time view of a
landscape, and allows users to overlay specific information on
top. This information could be pre-identified landforms,
outlines of historic spatial extents, or information on the
process that formed them and the material they consist of.

The base layer is commonly a live video stream, for
which the location and orientation of the camera must be
known. The technology is now well developed and imple-
mented on many smartphones. For example, Layar (http://
www.layar.com/) allows the development of data layers that
can be made available to end users. This technology is known
as ‘augmented reality’.
In addition to augmented reality, geomorphologists have
new research opportunities by using high-resolution satellite
imagery. For example, the development of low-altitude im-
aging systems (e.g., Hardin and Jensen, 2011) allow for the
capture of submeter resolution data (o10 cm). A variety of
innovative platforms are available (e.g., Smith et al., 2009;
Laliberte and Rango, 2011) that enable geomorphologists to
examine the landscape at high spatial and temporal frequen-
cies. Likewise, terrestrial-laser scanning (TLS; Heritage and
Large, 2009) can be used to dramatically increase the data
volume and precision of topographic data. Examples of new
approaches based upon such data include Williams et al.
(2011) who outline methods for monitoring channel changes
to gravel-bed rivers before and after flood events, and Pesci
et al. (2011) who monitor landslide deformation using TLS
data over three epochs. Perhaps the greatest excitement has
been generated around real-time sensors that relay measure-
ments of physical processes as they occur. For example, Hart
et al. (2006) utilized subglacial sensors to monitor changes to
till through time. Linked clusters of sensors, or ‘sensor net-
works,’ are a powerful methodology to remotely monitor
processes at very high spatial and temporal resolutions, par-
ticularly in areas that are hazardous or remote (http://glacs-
web.org/). Importantly they have mission-critical capabilities,
where communication and data storage are replicated across
probes so that sensor failure does not cause network failure.
Overall, the increasing amounts of data, both spatially and
temporally, present new challenges for geovisualization where
high data storage and multidimensional datasets require in-
novative solutions.
3.11.3 Visual Processing
Geomorphologists are interested in the geovisualization of
spatial data in order to study and understand the Earth’s sur-
face. Given their objectives, they generate an observational
data set that provides them with information that they can
analyze, and one that facilitates the identification of land-
forms. The observational data are representations of the lo-
cation and characteristics of the landscape, and representation
of the topography may be termed digital terrain modeling
(DTM).
An intermediate output product of terrain analysis (i.e.,
geomorphometry) may be an enhanced data layer for further
visual inspection, analysis, and/or numerical modeling. Ul-
timate output products are typically maps for distribution,
analysis, and archival. Although researchers may conduct
fieldwork for geomorphological mapping, we only consider
terrain analysis using VNIR (visible and near infra-red) im-
agery and topographic information in this section. Specific-
ally, the concepts of ‘topographic signal’ (Hillier and
Smith, 2008) and ‘detectability’ (Smith and Wise, 2007) are
Geovisualization 305
introduced. These concepts can be used as a basis for evalu-
ation of image processing and terrain analysis techniques
that enhance visibility of specific landforms by increasing the
ratio of (topographic) signal to noise of an image. The sec-
tion concludes with recommendations for enhancement and
cautions about limitations in geomorphological mapping
using digital data.
The process of landform identification can be performed
using manual or automated/semi-automated techniques.
Manual mapping techniques require a skilled interpreter to
identify and outline landforms of interest. Skill, how well
landforms are identified, is based upon many factors and
includes the basic expertise and experience of the analyst
(Colwell, 1983). The process is qualitative and relies upon the
development of ordered relationships between features in an
image using complex visual heuristics in order to identify
object types. Although many have focused on using pan-
chromatic aerial photos, geomorphologists now have access to
multispectral, thermal, and radar imagery. Such data permit
better interpretation using physics (matter–energy inter-
actions) as a basis for understanding reflectance, emission,
and backscatter variations. The basic image elements remain
the same: that is, the identification of landforms via the
assessment of tone, texture, shape, size, shadow, pattern, lo-
cation, and association. For further details on the image
elements and basic image-interpretation strategies, see general
textbooks on remote sensing (e.g., Gibson and Power, 2000;
Campbell, 2007; Lillesand et al., 2008).
Automated and semi-automated mapping techniques use a
variety of algorithmic approaches to identify landforms (e.g.,
Hillier and Watts, 2004; Van Asselen and Seijmonsbergen,
2006; Seijmonsbergen et al., 2011). Such approaches have the
benefit of being numerical, repeatable, and not reliant on an
individual interpreter or systematic variation during process-
ing. Any inherent biases tend to originate from the ultimately
subjective calibration underpinning the method. A significant
number of these approaches are primarily statistical, having
been developed from clustering methods used more generally
in remote sensing. The increasing availability of DEMs, how-
ever, has allowed the incorporation of elevation and derived
land-surface parameters, so that morphological information
can be used in the classification process. The wealth of
remotely sensed data now available has opened considerable
opportunities for developing the processing of extensive
datasets, as well as integrating them into new types of ana-
lyses. For example, Camargo et al. (2009) integrated reflect-
ance data into a geomorphometric object-based analysis, and
this approach is likely to become more common in an attempt
to evaluate surface materials from surface spectra. Further in-
sights can be gained through the integration of subsurface data
from active and passive sensors. At sea, the subsurface struc-
ture has been commonly probed using sound (e.g., seismic
reflection), electrical (e.g., resistivity), magnetic and gravity
fields. See Jones (1999) for further details. On land, passive
airborne systems, such as aeromagnetics (detecting subsurface
magnetic features), and active systems, such as airborne elec-
tromagnetics (3D conductivity), provide subsurface data
(Smith and Pain, 2009). The use of automated and semi-
automated techniques that integrate a variety of datasets is
therefore expected to become increasingly prominent,
306 Geovisualization

however, for some mapping applications, manual mapping
remains the most prevalent technique for recording the
shape and distribution of landforms. Therefore, visualization
techniques that can be used to facilitate this objective are
presented.
however, the detectability of a landform will be primarily
dependent on sensor system characteristics. The following
three object characteristics in relation to sensor system char-
acteristics combine to affect landform detectability and should
be considered by an interpreter prior to selecting data:

1. Relative size. The minimum resolvable size of a landform

3.11.3.1 Detection of Landforms
Manual mapping requires the visual detection of individual
landforms and the recording of their spatial distribution onto
some kind of a base map. Detection may be accomplished
using aerial photography or satellite imagery, or may require
additional topographic information to delineate the bound-
aries of landform features. The use of specific regions of
the electromagnetic spectrum may make detection easier, as
variations in reflectance between the landform and the
surrounding landscape depends on landform composition
versus landscape composition. Detection may also relate
to surface structure variations that may be characterized by
image texture. Smith and Wise (2007) outline the concept of
detectability; that is, the ability for an interpreter to accurately
identify the true dimensions of a feature of interest. This is
dependent on the data source, image pre-processing, and en-
hancement, and the skill of the interpreter. Interpreter skill is a
difficult capability to quantify and can be highly variable
between individuals, although there are procedures that can
be implemented to minimize this influence. Fundamentally,
that is visible in an image is a function of the spatial
resolution (Figure 5). The spatial resolution must be high
enough, or the landform large enough, that it contains a
sufficient number of pixels to be recognizable. Small fea-
tures may be under-represented in any population, and
where spatial clustering of small landforms exists, system-
atic errors in interpretation may consequently result.
2. Azimuth biasing. The orientation angle of an elongated
landform (e.g., drumlin); this causes their appearance to
differ in an image depending on the relationship between
the orientation angle and the solar geometry (Figure 6).
This becomes more pronounced with greater elongation
(Smith and Clark, 2005).
3. Landform signal strength. The amount of tonal and textural
information that is available to visually distinguish indi-
vidual landforms; also termed the ‘topographic signal’
(Figure 7; Hillier and Smith, 2008). For DEMs, the signal
strength is related to altitude variation with respect to the
landforms of interest and the surrounding terrain. For
imagery, topography is inferred from associations between
landforms and land cover (e.g., Punkari, 1982) and tonal

160 000 165 000 170 000 175 000 180 000 160 000 165 000 170 000 175 000
308 000

308 000

308 000
308 000
302 000

302 000
302 000

302 000
296 000

296 000
296 000

296 000
290 000

290 000
290 000

290 000
284 000

284 000
284 000

284 000

(a) 160 000 165 000 170 000 175 000 180 000 (b) 160 000 165 000 170 000 175 000
Figure 5 Illustration of the effects of resolution on the detectability of drumlins. (a) 50 m DEM. (b) 150 m DEM. Note that small landforms in
the north of the area would be difficult to detect from the 150 m DEM. Irish Grid coordinates in meters, relief shaded with a solar elevation of
451 and a solar azimuth of 451.
 Geovisualization 307

160 000 165 000 170 000 175 000 160 000 165 000 170 000 175 000

308 000

308 000
308 000

308 000
302 000

302 000

302 000
302 000
296 000

296 000

296 000
296 000

290 000
290 000

290 000
290 000
284 000

284 000

284 000
284 000

(a) 160 000 165 000 170 000 175 000 (b) 160 000 165 000 170 000 175 000
Figure 6 Illustration of the effects of solar azimuth angle on the detectability of drumlins from a relief-shaded DEM. (a) Azimuth angle parallel
to the dominant drumlin orientation. (b) Azimuth angle orthogonal to the principal drumlin orientation. Arrows indicate azimuth angle. Irish Grid
coordinates in meters.

and textural variations from shadows (Slaney, 1981), par-
ticularly where solar elevation angles are relatively low.
Furthermore, synthetic-aperture radar (SAR) systems use an
oblique (rather than nadir) viewing geometry (Ford, 1984;
Vencatasawmy et al., 1998), which enhances the topo-
graphic expression in the imagery.
The minimum-resolvable plan form size for detecting
landforms can only be reduced by increasing the spatial
resolution of the data (Smith et al., 2006). Although such data
may be available, there will almost certainly be a trade-off
between financial cost and spatial coverage, and project ob-
jectives must be tailored to available resources (e.g., Punkari,
1982). For satellite imagery, routine repeat coverage from
moderate-resolution sensors (e.g., Landsat) mitigates this
somewhat, as images at optimal illumination angles and
azimuths may be available in existing archives.
Azimuth biasing is perhaps the most significant problem
for satellite imagery, although other issues associated with
landform mapping from imagery still exist. Bias is minimized
through the acquisition of imagery with relatively high solar
elevation (i.e., illumination is basically overhead); however,
this minimizes the landform signal strength so that
features are not as clearly observable (Figure 7). Smith and
Wise (2007) recommend solar elevation angles o201; how-
ever, polar-orbiting satellites operate with fixed overpass times,
which means that it is not possible to specify acquisition time
at a site. With a reduced ability to obtain ideal satellite
imagery, ephemeral landforms are commensurately more
difficult to identify and monitor. If solar geometry conditions
cannot be adequately controlled, interpreters should be aware
of the possibility of introducing systematic biases through the
exclusion of landforms suites. Azimuthal bias does not affect
DEMs unless relief shading is generated using low-angle illu-
mination. Using both satellite imagery and DEMs, it is pos-
sible to enhance the landform signal strength to make
landforms more detectable, thereby increasing the accuracy of
manual mapping.
3.11.3.2 Image Enhancement
With the acquisition of satellite imagery and appropriate pre-
processing, it is sometimes necessary to enhance the imagery
to provide the best possible visualization of landforms. En-
hancement is largely based on a subset of the standard image-
processing techniques that include spectral and spatial
enhancements that are commonly used in remote sensing
(e.g., Mather, 2004; Lillesand et al., 2008), and have proven
useful for geomorphological visualization. Clark (1997) noted
that brightness/color variations are most effectively detected
by humans, and therefore recommends the use of images
rendered in gray scale. For multispectral imagery, it has been
common practice to generate image ratio images that effect-
ively portray spectral differences that are associated with sur-
face-compositional variations. The geological community
308 Geovisualization

155 000 160 000 165 000 170 000 175 000 180 000

311 000
311 000

308 000
308 000

305 000
305 000

302 000
302 000

299 000
299 000

296 000
296 000

(a) 155 000 160 000 165 000 170 000 175 000 180 000

155 000 160 000 165 000 170 000 175 000 180 000

311 000
311 000

308 000
308 000

305 000
305 000

302 000
302 000

299 000
299 000

296 000
296 000

(b) 155 000 160 000 165 000 170 000 175 000 180 000
Figure 7 Illustration of the effects of landform signal strength using Landsat TM imagery acquired on contrasting dates (B1801 solar azimuth).
(a) Imagery with low solar elevation (111). (b) Imagery with high solar elevation (481). The images overlap with the top half of Figures 4, 6, 8, 9,
10, and 11. Irish Grid coordinates in meters.

commonly uses image ratios and examines false-color com-
posites of ratio images. Standard contrast enhancements such
as linear, piece-wise, and other histogram-modifying techni-
ques should be utilized to maximize the tonal contrast for
original imagery and spectral features such as ratios. Principal
component analysis (PCA) is another standard technique for
spectral-feature extraction that isolates the fundamental linear
variation from the entire multispectral dataset, thereby pro-
ducing spectral features (principal component images) that
highlight different aspects of the landscape. The first principal-
component image contains the majority of the information
from the original dataset. It is possible that vegetation–
landform, topography–landform, and moisture– landform
relationships can be ascertained by interpreting these images.
 Geovisualization 309

Finally, spatial-feature enhancement techniques can also be 160 000 165 000 170 000 175 000 180 000
applied. Examples include texture enhancement, spatial con-

308 000

308 000
volution (kernel) filtering, and more advanced techniques
including fractal analysis and Fourier analysis. Edge-detection
techniques and high-pass filters can be used to detect subtle
variations in reflectance or topography that may help to de-

302 000
302 000
lineate landforms of interest. They are commonly used for
mapping lithological units and faulting, and play an import-
ant role in tectonic geomorphology.
The work of Punkari (1982) illustrates how multi-
spectral imagery is important for identifying landforms. In

296 000
296 000
his work, inter-drumlin regions exhibited greater surface
moisture conditions. These areas were enhanced for mapping
by using those regions of the spectrum that were most sen-
sitive to moisture conditions (i.e., NIR, SWIR), thereby per-

290 000
290 000
mitting the identification of drumlins. Other techniques
could also be used including the use of the wetness index and
the principal component image that enhances surface mois-
ture conditions. In a second example, Jansson and Glasser

284 000

284 000
(2005) used a simple false-color composite to effectively
integrate near infra-red, thermal infra-red, and topographic
information (shaded relief) to permit landform identifi-
cation. With respect to topographic information, it is im-
portant to note that many image enhancement algorithms
160 000 165 000 170 000 175 000 180 000
can also be applied to terrain parameter images generated
from terrain analysis. Figure 8 A DEM gray-scale image. Irish Grid coordinates in meters.

3.11.3.3 Enhancement of DEMs
DEMs represent the altitude field, and they can be used to
derive morphological information and be displayed in a variety
of ways. Numerous spatial-analysis algorithms exist that char-
acterize different aspects of the topography, and this infor-
mation can be easily integrated and draped over the altitude
field. By default, most image-processing software can be used to
display DEMs as a gray-scale image with a palette of gray tones
representing altitude that linearly varies between the maximum
and minimum values in the dataset (Figure 8). Displaying the
topography in this way has its disadvantages because subtle
variations in the altitude field cannot be easily seen, and low
amplitude landform features cannot be identified against larger
scale higher amplitude relief variations in the landscape. It is
therefore necessary to focus on specific areas of the landscape
containing features of interest and enhancing the ‘landform
signal strength’ to facilitate identification.
3.11.3.3.1 Regional–residual separation
As an aid to better understanding of surface processes, land-
forms can be divided into classes or categories, based on some
similarity of a topographic characteristic or a priori knowledge
that they relate to a physical process. These classes of features
can constitute a ‘component’ of a landscape. If H is elevation,
the decomposition of a DEM into components may be
expressed as HDEM ¼ H1 þ H2 þ yHn, where n is the number of
components, and HDEM the observed relief (e.g., Hillier, 2011).
Hn is the thickness of a component, a layer that varies spatially
in thickness; i.e., the values of H1, H2 and so on to Hn vary for
each location (x,y) of the DEM. Thicknesses are always greater
than or equal to zero. Summing the components recreates the
terrain. Isolating landforms into a component of the landscape
so that they may be studied independently is termed regional-
residual (relief) separation (RRS). Wessel (1998) coined the
term as applied to DEMs, and Hillier and Smith (2008)
introduced it to subaerial geomorphology. The approach is
more difficult to apply where classes of landform are mor-
phologically similar to each other or to the underlying
topography.
Figure 9 illustrates one RRS method that was used to
isolate drumlins and hills from one another. Other appli-
cation examples of RRS include quantifying the large-scale
subsidence trend across the oceans to understand how tec-
tonic plates evolve (e.g., Sclater et al., 1975; Calcagno and
Cazenave, 1994), isolating underwater volcanoes (White,
1993; Hillier and Watts, 2007), removing trees and other
surface features to ‘declutter’ digital surface models and create
DTMs for tasks such as flood modeling (e.g., Sithole and
Vosselman, 2004; Mason et al., 2007), or archeological pro-
specting (Hillier et al., 2007; Hesse, 2010). It is important to
note that the spatial components are not terrain units that
exhibit similar properties such as those portrayed in a con-
ventional geomorphological map.
By convention, the regional component contains features
of a larger width-scale than the residual components because,
in older methods (e.g., Watts, 1976; McKenzie et al., 1980;
Hillier and Smith, 2008), the altitude trend representative of a
region was calculated first and then subtracted from the ele-
vation to obtain the residual. Alternatively, it is possible to
directly isolate (i.e., identifying and determining spatial limits
for) all individual features of a certain type within landscapes
(e.g., Hillier and Watts, 2004; Hillier, 2008) leaving the re-
gional trend. But, the same naming convention is still used.
310 Geovisualization

0 10 000 20 000

High
North South
Elevation (m) 150

100

30 0000

Relative elevation
(a) 50
Elevation (m)

20
10
0

290 000
–10
(b) –20
elevation
Relative

Low
0 10 000 20 000
160 000 170 000 180 000
(c) Distance along profile (m) (d)

Figure 9 Regional–residual separation (RSS) applied to a profile across the DEM (Hillier and Smith, 2008). (a) 1  1 km median sliding-window
filter is used to approximate regional relief (hills; regions indicated by the bold line). (b) Subtraction of regional relief from the DEM to leave
topographic signal (drumlins; thin line). Envelope (gray shade) is based upon filters returning maximum and minimum residual values within a
1  1 km kernel. (c) Normalization of heights, equivalent to a local contrast stretch, so that landforms of differing amplitude may be effectively
displayed using a single color palette. (d) Display of relative elevations using gray scale. Irish Grid coordinates in meters. Profile is along
white line.

The skill in performing a successful RRS is always determining
a property that makes the features you wish to isolate dis-
tinctively different.
The earliest RRSs were manual (e.g., Menard, 1973; Sclater
et al., 1975), and the quantitative differences among the
components were not known. Computationally, early meth-
ods used efficient frequency-domain filters (e.g., Watts and
Daly, 1981; Cazenave et al., 1986) to obtain the regional
component (e.g., 400–4000 km wavelength topography), and
high-pass filters to emphasize the small-scale residual. This
approach depends upon nonoverlapping spatial frequencies.
This so-called spectral overlap in the frequency domain is
therefore a problem (Wessel, 1998).
Alternatively, a valid regional elevation can be esti-
mated using low-pass spatial filtering (e.g., Watts, 1976;
McKenzie et al., 1980; Wessel and Smith, 1998). If the DEM
contains many statistical outliers (e.g., anomalously high
points), bias can occur in the regional (Smith, 1990). Use of
other, ‘robust’ statistics such as the median and mode can
mitigate this substantially (Smith, 1990; Crosby et al., 2006;
Kim and Wessel, 2008), as can iterative statistical approaches
(Marty and Cazenave, 1989; Wang et al., 2001).
These older techniques are good first approximations, and
remain useful in some situations (e.g., Hillier and Smith,
2008). The main difficulties include:
• Spectral overlap between classes of landform (e.g., Wessel,
1998).
• Spatial overlap of landforms.
• High spatial density of smaller landforms obscuring the
regional trend (e.g., Smith, 1990).
• Complex high-amplitude regional trends.
• Widely ranging shapes and sizes within a class of feature
(e.g., seamounts, drumlins).
A number of complex algorithms have focused on over-
coming these problems to create a regional digital surface model
when de-cluttering LiDAR data (e.g., Sithole and Vosselman,
2004). For geomorphic features, several algorithms also exist
that identify the smaller features directly (Hillier and Watts,
2004; Hillier, 2008). These seek to avoid the problems experi-
enced by the older approaches above, and have the advantage
that features are implicitly mapped during the separation but,
like the de-cluttering algorithms, are somewhat task specific.
Lastly, if the locations of features are known a priori, this in-
formation may be used (e.g., Smith et al., 2009). Once a com-
ponent is isolated through RRS, the image can be enhanced and
displayed to facilitate landform mapping.
3.11.3.3.2 Land-surface parameters
To facilitate landform mapping, land-surface parameters
(LSPs) (Hengl and Reuter, 2008) are commonly used, as they
characterize unique morphometric properties of the topo-
graphy. These may then be displayed separately, or in unique
combinations to identify landforms and delineate their
boundaries. Such parameters may represent physical attributes
of the topography, components of biophysical variables, or
empirical proxies related to potential or position.
Relief shading is the most common visualization technique
that can be used to enhance the topography (Kraak and
Ormeling, 2010). The landscape is illuminated from a specific
solar azimuth and elevation angle, and one can assume

perfectly specular (i.e., mirror-like) reflection from the surface.
Relief shading, however, highlights a break-in-slope as seen
from a particular direction, which does not necessarily char-
acterize the surface morphology, and suffers from azimuth
biasing. Figure 6 illustrates the problem with a DEM that is
illuminated parallel and orthogonal to the principal landform
orientation. It is better illustrated by using an animation,
where the solar azimuth angle is rotated through 3601 at 51
intervals. Such an animation reveals landforms appearing,
disappearing, and changing shape. It is therefore common
to utilize relief-shaded images generated from multiple
solar-azimuth angles. An alternative solution is to use PCA to
produce optimal combinations of relief-shaded images (Smith
and Clark, 2005).
Supplementary material related to this article can be found
online at doi:10.1016/B978-0-12-374739-6.00054-3.
The above discussion illustrates that even if a DEM accur-
ately represents the topography, the method used to display
topographic information can introduce significant bias. It is
therefore preferable to use LSPs that are designed to numer-
ically (rather than for display) characterize terrain properties,
as they provide the basis for progress toward reproducible
visualization and mapping methodologies.
The slope of the terrain is a fundamental parameter
(Evans, 1972; Olaya, 2008) as it controls the component of
gravitational force available for geomorphic processes and is
defined at a point as a plane tangential to the land surface.
That plane has a steepness (gradient) and orientation (slope
azimuth). The slope gradient, or slope angle, is commonly
160 000 165 000 170 000 175 000 180 000
308 000
308 000
302 000
302 000
296 000
296 000
290 000
290 000
284 000
284 000
(a) 160 000 165 000 170 000 175 000 180 000
Figure 10 DEM-derived land-surface parameters. (a) Slope gradient image. (b) Profile curvature image.
Geovisualization 311
used as an LSP for mapping, as many landforms have steep or
shallow slopes, although users generally find it difficult to
interpret and must become familiar with the topography. For
visualization purposes, gray-scale display of a slope image is
best when flat areas are represented using light gray tones,
giving an image a shaded-relief quality that does not exhibit
azimuth biasing (Figure 10(a)). Terrain curvature is a meas-
ure of the rate of change of gradient, and may be quantified
as planform, profile, and tangential curvature (Schmidt
et al., 2003). Profile curvature measures downslope curvature
(the derivative of gradient) and highlights breaks in slope
(Figure 10(b)). It is therefore most appropriate for land-
form mapping, but like gradient, some find it difficult to
interpret.
Another parameter useful for mapping is surface roughness.
It represents the variability of elevation of a topographic surface
at a given scale (Figure 11(a)). For example, McKean and
Roering (2004) separated landslide debris from different
time periods based on decreasing surface roughness due to
subaerial erosion. There are numerous metrics to characterize
surface roughness and these are reviewed by Grohmann et al.
(2010), who classified them as area ratio, vector dispersion, and
standard deviation methods. The first approach calculates the
ratio between the area of a dipping surface and its area when
projected onto a horizontal plane (Hobson, 1972), with values
approaching one indicating flat surfaces. The second approach
calculates vectors normal to local tangents to the land surface,
and then characterizes how concentrated or dispersed these
vectors are (Hobson, 1972; Guth, 2001; McKean and Roering,
160 000 165 000 170 000 175 000 180 000
308 000
308 000
302 000
302 000
296 000
296 000
290 000
290 000
284 000
284 000
(b) 160 000 165 000 170 000 175 000 180 000
312 Geovisualization
160 000 165 000 170 000 175 000
308 000
308 000
302 000
302 000
296 000
296 000
290 000
290 000
284 000
284 000
(a) 160 000 165 000 170 000 175 000
Figure 11 DEM-derived land-surface parameter and enhancement. (a) Surface roughness image. (b) Local contrast-stretch enhanced image.
2004). For flat terrain, vectors are near vertical and dispersion is
low. Finally, standard deviation calculations of elevation, slope,
and residual relief are commonly used.
Landforms can also be highlighted using the local-relative
elevation (e.g., Pike and Wilson, 1971; Hillier et al., 2007).
This approach highlights local altitudinal variations. Smith
and Clark (2005) refer to this as a local contrast stretch, where
a linear stretch is applied across a user-specified window or
kernel of appropriate width (Figure 11(b)), and a value is
assigned to the pixel according to the relative position be-
tween the lowest and highest point in the locality.
Yokoyama et al. (2002) proposed a terrain parameter
called ‘openness.’ Openness (or the opposite, ‘enclosure’) is
calculated using line-of-sight methods that emphasize con-
vexity and concavity in terrain, capturing the degree of geo-
metric ‘dominance.’ It utilizes standard viewshed algorithms,
but calculates these at nadir and zenith for eight compass
directions. The parameter is sensitive to changes in local relief,
where multiscalar expressions of openness (i.e., distance) will
have an impact. Conceptually it is related to measures of
both roughness and curvature, although the interpretation is
different.
3.11.3.4 Recommendations for Terrain Visualization
A sample of techniques designed to enhance satellite imagery
and DEMs for landform detection have been presented,
although many other techniques and approaches exist. The
selection of the most appropriate visualization technique(s)
160 000 165 000 170 000 175 000
308 000
308 000
302 000
302 000
296 000
296 000
290 000
290 000
284 000
284 000
(b) 160 000 165 000 170 000 175 000
will depend on the objectives of the study, the type of analysis
that is required, and the information that must be produced.
There are, however, some approaches that can be generally
recommended as useful.
Simple image display of DEM data is rarely useful as a
visualization technique for studying the landscape or land-
form mapping. It is therefore necessary to utilize techniques
that are designed to highlight landforms and features of
interest. Although appropriate RRS methods are not neces-
sarily trivial to design, simple methods can be effective in
highlighting individual landforms of interest, relegating all
other aspects of terrain to ‘noise’ (e.g., Hillier and Smith,
2008). An appropriate RRS to isolate the landforms of interest
will always tend to improve the output of any subsequently
applied technique.
The technique of relief shading highlights subtle topo-
graphic features well; it is widely implemented in software, fast
to compute, and very useful when appropriate care is taken to
allow for azimuthal bias. Azimuthal bias from the artificial
illumination source alters the position of breaks-in-slope,
thereby giving the impression of a landform changing shape
or disappearing. The use of multiple relief-shaded images at
least partially mitigates this problem. Nevertheless, techniques
that do not involve illumination are preferable. In general, the
use of LSPs can greatly facilitate the visualization of the
landscape, as a multitude of topographic characteristics can be
examined. Numerous enhancement techniques can be used
on multispectral and terrain images to facilitate mapping ef-
forts. Smith and Clark (2005) and Hillier and Smith (2008)
reviewed many methods, concluding that no single technique

is ideally suited to terrain visualization. Therefore, the best
practice for landform visualization and mapping is likely to
involve the integration of terrain information using visual-
ization techniques applicable to specific objectives.
3.11.4 Visual Interaction
Section 3.11.3 presented a variety of techniques that could be
used for static visualization of terrain. It is frequently neces-
sary, however, to combine and interact dynamically with
data to explore for patterns in order to make associations.
This section introduces common techniques to dynamically
interact with spatial data, and then explores increasingly so-
phisticated methods involving 2D planes and 2.5D surfaces,
before outlining how surface data can complement true 3D
volumetric data.
3.11.4.1 Display
Raster outputs of terrain are usually displayed and inspected
on a video display unit (VDU) using the red, green, and blue
(RGB) additive model of mixing colors (Figure 12). Given the
sensitivity of the human eye, the color cube allows the mixing
of the three primary colors at different intensities to provide
the full gamut of possible viewable colors. It is therefore
possible to display up to three separate images of the same
area at any one time, by encoding the pixel values as a 3D
coordinate position within the color cube. This is known as a
‘false-color composite’ (unless it represents the actual object
color as perceived by the human eye in which case it is a
‘true-color composite’). The technique is widely used in re-
mote sensing and, within geovisualization, allows the inter-
active inspection of multiple terrain datasets.
Figure 13 presents an example of how a false-color com-
posite can be used to interpret the landscape, as it combines
slope angle (red: 3  3 kernel), topographic openness (green:
51  51 kernel), and topographic openness (blue: 501  501
DN (red)
Yellow
Magenta White
Gray line
Black DN (green)
Cyan
DN (blue)
Figure 12 A red, green, and blue (RGB) color cube. The cube
defines the intensity of each primary color, allowing the 3D plotting
of all colors within the cube. The gray line shows how colors are
plotted for a gray scale each where each RGB component has equal
intensity.
Geovisualization 313
kernel). Reddish colors represent high slope and low openness
(e.g., at steep and enclosed hillslopes), whereas greenish/
bluish colors depict low slopes and relatively high openness
(e.g., in flat and open areas).
3.11.4.2 Digitization and Overlay
The main output of visual processing is a raster dataset, or
datasets, generated using a variety of techniques. These can
then be interpreted by geomorphologists for the identification
of individual landforms and the recording of their position
and extent. The process is generically known as ‘digitization’
and can form a body of work for the production of geo-
morphological maps (e.g., Hughes and Clark, 2010), or serve
as input layers for further quantitative processing. Digitization
is the electronic creation of geographic coordinates, usually
through a mouse click at the position identified on screen.
When the coordinates are used individually (points) or com-
bined together (lines or polygons), they are known as ‘feature
types,’ with each feature type forming a separate layer.
Layers can be stacked (e.g., overlain) and reordered,
allowing the interpreter to interact with different data layers
and visually inspect their interaction, as well as removing or
adding layers. This simple organizational framework provides a
powerful paradigm through which geomorphologists can work
digitally, and is conceptually the same as overlaying tracing
paper on an aerial photo and tracing the outlines of landforms.
The process in nondestructive and can involve modifying other
interactive features such as layer symbolization. Raster layers
(such as DEMs or satellite images) are typically rendered as
opaque, obscuring any data they are overlaid onto, although a
simple solution is to change the opacity of a specific data layer.
Interactive methods include switching layers on and off so that
Figure 13 False-color composite generated using topographic
openness (green: 51  51 kernel), topographic openness (blue: 501 
501 kernel) and slope (red: 3  3 kernel). Reddish colors represent
high slope and low openness and greenish/bluish colors depict low
slopes and relatively high openness (e.g., in flat and open areas).
314 Geovisualization
Figure 14 Integrated topographic map, geomorphological map, and aerial photo for Bruchi and Massa River valleys, Switzerland. Reproduced from
Otto, J.-C., Gustavsson, M., Geilhausen, M., 2011. Cartography: design, symbolisation and visualisation of geomorphological maps. In: Smith, M.J.,
Paron, P., Griffiths, J. (Eds.), Geomorphological Mapping: A Handbook of Techniques and Applications. Elsevier, London.
the interpreter can rapidly move between them, either manu-
ally or automatically at a prescribed frequency. Alternatively,
users can operate a slider to move a new layer horizontally or
vertically across the data viewer. Simultaneous viewing of a
raster data grid and vector data is simpler, as the vector data can
simply be overlain. The ability to manage large datasets and
interact with them, allowing detailed inspection is an import-
ant, although simple, feature of a digital workflow. Figure 14
depicts how multiple datasets, including a topographic map,
geomorphological map, and aerial photo, can be integrated
and viewed simultaneously.
The challenges in digital workflows come not through the
use of technology, but in planning and organizing the digit-
ization process. It is important to predetermine the features that
are of interest (e.g., Sahlin and Glasser, 2008), and group them
thematically (e.g., fluvial, glacial, periglacial, and mass move-
ments). The choice of feature type used to represent the land-
form will depend on the landform being digitized and the scale
of mapping, although it is important to remember that feature
types are simplified categorizations (or abstractions) of reality,
and that definitions change over time. Polygons approximate
outlines of 2D features, or area features, such as drumlins or
landslides. Over larger areas, however, lines and points can also
be used to represent some area features. For example, regional-
scale mapping may represent drumlins as lines and landslides
as points. Further detail on the overarching rational for geo-
morphological mapping is discussed by Dramis et al. (2011),
with Smith (2011) detailing the digital workflows involved.
3.11.4.3 2D to 2.5D in Space
A simple way to interact with data is through activities such as
zooming, panning, rotating, and performing simple contrast
enhancements. There is a risk that these techniques become to
seem trivial, as a result of the ease with which the user is able to
perform them. Rudimentary visual exploration, however, is an
essential task. Google EarthTM, for example, provides an elegant
interface for the navigation of a single set of imagery for any-
where on Earth, and this is a remarkable achievement. Indeed
the scale of their data licensing means that Google EarthTM is
generally not only the first data source queried, but also the
most up-to-date and highest spatial resolution (e.g., Dykes,
2008). Most commercial image processing and GIS software
supports basic interactive visualization, commonly augmented
with ‘aspatial’ statistical output limited to summarizing the
whole dataset. Although somewhat restricted, these ‘global’
measures can be used to explore features of individual datasets.
For example, histograms of DEMs can highlight artifacts re-
sulting from the conversion of contours (e.g., Smith, 1993;
Wise, 2000). Wood and Fisher (1993) introduced spatial
measures, but highlighted the benefits of visualization in
identifying error in DEMs (Gousie and Smith, 2010).
Animations provide a powerful methodology for data
visualization by depicting attribute change to vector or raster
data. Figure 6 demonstrates that effect of changing illumin-
ation azimuth when relief shading is used to display a DEM,
and in this instance, each frame of the animation is rendered
using a different solar azimuth angle. Perhaps more com-
monly, temporal data can be animated where each frame
represents a successive time ‘slice.’ Frames can progress at a
fixed rate, or be manipulated manually using a ‘time slider’
where users can step through a historical archive of satellite
imagery and aerial photography. Given various landscape
evolution models, the DEM at each time-stage of the simu-
lation could be similarly displayed. This type of functionality
is now incorporated into many desktop GISs. For quantitative
work, change detection results could be used based on a time-
series of raster data using simple raster differencing (e.g.,
Etzellmüller, 2000; Chen et al., 2004). TIMESAT (Jonsson and
Eklundh, 2004) extends imagery analysis to multiepoch data,
and similar results are achievable in geomorphology.

Figure 15 Perspective view of the Lough Gara region, Ireland. The
surface has been color-coded according to elevation, relief-shaded,
and then draped over a 3D rendition of the DEM. Generated using a
solar azimuth of 3451, with a vertical exaggeration of 10x. Long edge
is 30 km. GeoEye Satellite image.
The discussion so far has focused on the 2D manipulation
of spatial data. DEMs form a special case, as they are a 2D
array representing the upper surface of a 3D volume, but are
not 3D as they do not formally define a volume. They are also
not 2D in the sense of a plane or map, so are sometimes
referred to as 2.5D. A variety of GIS applications allow a user
to navigate a 3D perspective, examining the DEM surface
vertically and horizontally through rotation and panning
(Figure 15; e.g., Jordan et al., 2009). Data can not only be
viewed on traditional 2D screens, but also on 3D display
systems allowing the user to interact with perspective views of
the terrain, overlay aerial photos, and satellite imagery from
different time periods, and add digitized vector line-work that
can subsequently be edited. This enables virtual field re-
connaissance and is particularly important for remote or in-
accessible sites. The penetration of Google EarthTM into the
geomorphological community has enabled many users to
become familiar with this method of navigation and the cre-
ation of ‘flight paths’ that can be automatically navigated or
animated as ‘fly-throughs.’
3.11.4.4 3D in Space
It is natural to move from 2.5D to full 3D visual interaction
with data, either in terms of estimating volumes, or being dis-
played in conjunction with true 3D volumetric data (e.g., seis-
mic reflection data). As noted in Section 3.11.4.3, DEMs do not
provide volumetric data, however, where it is possible to define
a lower boundary through an understanding of a geological or
geomorphological basal surface (e.g., Sclater et al., 1975; White,
1993; Wessel, 1998; Hillier and Watts, 2004; Smith et al., 2009),
volume can be estimated. In this context, volume is computed
as the height or thickness between the lower and upper
bounding surfaces of a landform. These calculations may be the
only estimate of volume possible in a study, but uncertainty
exists, as the basal surface is frequently based on inference.
Where subsurface and additional surface data can be col-
lected, a DEM can be seen as just one dataset contributing to
an understanding of feature or landscape volumes. Airborne
radiometric, magnetic, and electromagnetic measurements of
the subsurface, for instance, could be integrated. Further
datasets derived from laser scanning can give information
about supra-surface features such as vegetation canopy
Geovisualization 315
structure (e.g., ECHIDNA) to, for example, estimate biomass
(Lovell et al., 2003). Integrating such geophysical data with
DEMs to understand geomorphic processes has been the focus
of much recent work (e.g., Johnson et al., 2006; Hillier et al.,
2008; Jordan et al., 2009; Otto et al., 2009). Processing and
display to achieve insightful interrogation of 3D data are ex-
emplified in tools for the oil industry for processing seismic
reflection data and integrating this with other geospatial in-
formation such as well-log measurements (e.g., GeoVisionary,
GeoWall, ProMAX, GeoProbe, and KINGDOM). Such 3D
visualization packages integrate data such as outcrop-
sedimentary logs, strike-and-dip measurements, and horizon
interpretations with photography draped over LiDAR DEMs to
characterize subaerial conditions (e.g., Fabuel Perez et al.,
2010). These packages typically display one or many planar
slices through a 3D volume, or render the 3D volume onto a
screen. Visualization of such data can be enhanced by true 3D
display, usually through the application of stereoscopy. A
variety of systems are currently available, with the general
popularity of 3D video films leading to the development of
low-cost home systems that utilize polarized or shutter-based
techniques. In polarized systems, a single monitor can display
polarized images that are interlaced on the screen. Through
the use of polarized glasses, the human visual system as-
sembles the display into a 3D image. The use of interlacing
effectively halves the resolution of the image, although some
manufacturers use a twin monitor set-up to mitigate this ef-
fect. Shutter-based systems use glasses that are actively con-
trolled by the host computer that synchronizes the use of a
blocking shutter on each lens of the glasses with the display of
left and right stereoscopic images and can utilize the full
resolution of the display. Projection systems (e.g., GeoWall)
are better developed as twin projectors, using polarized filters
and screen to allow very large images to be displayed at full
resolution. We therefore see integration of surface, surface
structure, and subsurface information as being integral to the
future building of detailed models that can be used for a wide
range of Earth-science applications.
3.11.4.5 Virtual Globes
Virtual globes and online maps have become an important
aspect of the Internet, since the first release of NASA’s World
Wind (http://worldwind.arc.nasa.gov). The seamless inte-
gration of satellite imagery and DEMs has changed our per-
ception of the Earth within a few years, placing more public
emphasis on geomorphology (Tooth, 2006). The geoscience
community has embraced the use of virtual globes with
emerging applications including simple terrain inspection and
feature mapping (Sato and Harp, 2009; Welsh and Davies,
2010), data visualization, and geo-data exchange (http://
www.usgs.gov). Many national research agencies visualize
their data using virtual globes. For example, the United States
Geological Survey (USGS) provides stream-flow, watershed, or
earthquake data for display in Google EarthTM. The
National Snow and Ice Data Centre (NSIDC) also offers in-
formation about ICEsat data, NASA’s Ice, Cloud and Land
Elevation satellite (http://nsidc.org/data/virtual_globes/glas/
anchorage.kml). The application localizes each laser footprint
316 Geovisualization
acquired by the satellite and provides further information on
physical parameters (e.g., waveforms of laser reflection) of the
data that help researchers choose the right data set for their
objectives (Ballagh et al., 2011). Land cover and landform
changes are ideal information for display on virtual globes, as
overlays can be used to depict different multitemporal im-
agery. Terrascope, for example, allows the rapid comparison of
Landsat imagery at different acquisition times that document
land-cover changes in the tropics (http://www.ambiotek.com/
terrascope). Landform and hazardous-process monitoring and
inventories make use of virtual globes to visualize their data,
such as the database of glacier and permafrost disasters com-
piled by the Glacier and Permafrost Hazards Group (GAPHAZ,
http://www.geo.uio.no/remotesensing/gaphaz/home.html).
Virtual globes not only enhance spatial thinking, which is
regarded as an ‘educational necessity,’ but also augment sci-
ence in general, far beyond simple data visualization (Ballagh
et al., 2011). Geomorphologists use virtual globes like Google
EarthTM or NASA World Wind for field-work planning, and the
identification of study sites in presentations. It should be
noted, however, that virtual globes are increasingly applied to
convey scientific data and research results. Tooth (2006) stated
that virtual globes help to address key geomorphological
questions such as scale-dependency of form, simplified by
techniques like zooming. For example, tectonic geomorph-
ology and planetary remote sensing studies strongly benefit
from the large-scale visualization of a planet surface.
Virtual globes make use of data overlays using an existing
database. Thus, visualization via virtual globes is restricted to
the existing data, and does not allow for more complex data
analysis provided by GIS applications. The standard protocol
for data overlay is the KML. This (and KMZ, the compressed
version of KML) is a data format that displays geo-registered
place-marks, annotations, geometries, imagery, and 3D surface
objects on virtual globes, comparable to Hypertext Markup
Language (HTML) that displays text and imagery in a web
browser. The data format is human-readable and can be
written using a simple text editor (Werneke, 2008).
3.11.5 Visual Outputs
A variety of visual-output products can be produced for re-
search or public consumption. Although visualization work-
flows remain almost entirely digital and interactive, specific
outputs need to be targeted toward the intended audience and
publication medium. For spatial information dissemination,
static maps can be made available in printed or digital form.
Interactive spatial-output products are increasingly be-
coming common as part of research outcomes, and as a
method designed to engage stakeholders to make research
results more widely available for downstream applications.
Interactive maps have become a standard component on
many websites, and can currently be considered a state-of-the-
art method for the dissemination of geomorphological data
and information. Virtual globes provide the 3D visualization
of satellite imagery and topographic data of the Earth and are
used to locate the sites where photographs and scientific data
have been collected in the field. A wide range of GIS appli-
cations and tools have been developed to generate Internet
maps for the visualization of research results, and to enhance
the search for, or effective distribution of, scientific data. This
section summarizes static and interactive visualization prod-
ucts used in geomorphology. It provides a review of the legend
systems commonly used and basic cartographic principles in
map design. Details regarding digital-map production, web
mapping, and WebGIS are then covered.
3.11.5.1 Geomorphological Maps
3.11.5.1.1 Legend systems
Traditional geomorphological maps differ from other the-
matic maps in that qualitative information usually prevails
over quantitative data. Consequently, most geomorphological
maps are compiled using descriptive symbols to represent
landforms and processes. Many different symbol sets and
mapping systems have evolved during the twentieth century in
different countries with different thematic emphases and
varying usage of symbols and color (Figure 16). Even though
attempts were made to create a general legend for geo-
morphological maps by the International Geographical Union
(IGU) in the 1960s (Demek et al., 1972), no universally ap-
plicable legend system has been established. Two main styles
of geomorphological maps include: (1) multilayer maps de-
picting morphology, landform genesis, current processes,
surface and subsurface material, and chronology and (2)
simpler maps focusing on landform relationships and
morphology (Evans, 2012). The former style results in multi-
colored maps with several stacked layers of data that tend to
be overloaded with information and may be difficult to in-
terpret (e.g., the German system; Barsch and Liedtke, 1980),
whilst the latter is usually presented in black and white and
represents a reduction of information for practical purposes
(e.g., the British system; Evans, 1990). Furthermore, in add-
ition to more universal legend systems that provide symbols
for the entire breadth of geomorphological landscapes
(Demek et al., 1972; Barsch and Liedtke, 1980; Cooke and
Doornkamp, 1990), specialized legends for high mountains
(Kneisel et al., 1998; De Graaff et al., 1987) or hazardous
processes exist (Kienholz, 1978; Kienholz and Krummenacher,
1995). See Otto et al. (2011) for an overview of selected geo-
morphological legend/mapping systems.
3.11.5.1.2 Map design
Geomorphological maps are complex thematic maps that
place demands on cartographic visualization techniques to
provide a comprehensible and readable map. It is therefore
beneficial to consider some underlying principles of cartog-
raphy and map design. The basic representations of objects
on maps are the symbol primitives: point, line, and area
(Robinson et al., 1995). Whether a linear feature in nature is
represented by a line symbol on a map is primarily a question
of scale. For example, a blue line can be used to depict a river,
whereas on larger scale maps a river can be depicted using an
area symbol, covering the space between the riverbanks. The
map scale also determines if a landform is depicted by a point
symbol, or if it is split up into its morphological components.
Rock glaciers, for example, could be represented by a single
point symbol on small-scale maps, or by the assemblage of
line and area symbols that differentiate the step height of the

Landform example
Legend system
Moraine ridge
IGU unified key
(Demek et al., 1972)
ITC (NL)
(Verstappen & Zuidam, 1968)
GMK 25/100 (GER)
(Barsch, 1978)
British Geomorphological Maps
(Evans, 1990)
ARAG (NL)
(DeGraaf et al., 1987)
IGUL (CH)
(Schoeneich et al., 1998)
Mapping system by Gustavsson
(Gustavsson et al., 2006)
BUWAL/BAFU (CH)
(Kienholz, 1976)
Figure 16 Symbols for moraine ridge and fluvial terrace from different geomorphological legend systems. Reproduced from Otto, J.-C.,
Gustavsson, M., Geilhausen, M., 2011. Cartography: design, symbolisation and visualisation of geomorphological maps. In: Smith, M.J., Paron,
P., Griffiths, J. (Eds.), Geomorphological Mapping: A Handbook of Techniques and Applications. Elsevier, London.
rock-glacier front, furrows, and ridges, and the accumulation
of boulders and blocks on top of the rock glacier, if the
map scale increases. Note that this is different and in-
dependent of the feature types used when digitizing data. For
specific projects, it is often the case that landforms may be
well aligned (e.g., drumlins digitized as lines are presented
on a geomorphological map as lines); however, it may be
necessary to use the same base data for the production of
different scale maps. This would require the generalization of
feature types.
A differentiation of these basic representations, to express
relationships among or differences between the data, can be
achieved by variations of the basic visual variables: color,
texture, size, shape, or orientation (Robinson et al., 1995;
Kraak and Ormeling, 2010). For example, different classes of
steps or breaks in slopes, depicting various levels of river ter-
races, could be visualized using different sizes or variations of
the same line symbol (Figure 17). Variable symbol shapes
demonstrate qualitative differences, and this is the most
commonly applied visual variable in geomorphological maps,
because of the great number of different symbols for different
landforms and processes.
Color is an important visual variable, mainly used to depict
qualitative differences. Geomorphological maps, however, are
commonly produced in black and white especially when they
are part of a journal publication to keep production costs
low. If color is used, variation of color characteristics, hue
(color variation), value (lightness), and chroma (saturation)
can be used to emphasize certain aspects of the map. Within
cartography, some color conventions exist that should be
Geovisualization 317
Emphasis
Fluvial terrace
Morphogenesis
Process/genesis
Genesis
Form/genesis
Genesis/
surface material
Morphogenesis/
landforms
Morphogenesis
Process/landform
acknowledged to avoid confusion. For example, on topo-
graphic maps, blue is used for objects related to water, like
rivers, springs, or lakes, and green generally represents
areas covered by vegetation. Color decisions can be facilitated
using the online software tool ‘Colorbrewer’ (http://color-
brewer2.org/). The software assists users in choosing the right
composition of colors by displaying different color schemes.
Color combinations can be tested on a complex map sample
that enables the designer to experience the differentiation and
perception of the colors used. In many geomorphological le-
gend systems, colors are applied to represent variations in
landform genesis (Barsch and Liedtke, 1980), process do-
mains (Gustavsson et al., 2006), or lithology (Pasuto et al.,
1999).
Map layout consists of the arrangement of the map com-
ponents into a functional composition and a meaningful and
aesthetically pleasing design to facilitate visual communi-
cation (GITTA, 2006). Geomorphological maps commonly
include the following elements surrounding the main map:
title, legend, scale, directional indicator (north arrow), co-
ordinate grid, or border, information on coordinate system
and map projection, and author credits. Commonly inset
maps are included that show the location of the mapped
area (essential for large-scale maps), an overview of the
geological situation, or other additional information on the
study area (e.g., a slope map). These items need to be arranged
carefully to guide the viewer’s eyes toward the focus of
the map.
Most geomorphological maps are now produced using
graphics or GIS software that provide tools to facilitate map
318 Geovisualization
0 25 50 Meters
1850
40
18
Figure 17 Mapping different levels of fluvial terraces using classified line symbols.
creation. The underlying cartographic principles outlined
above, however, remain important to produce maps that are
fit for purpose. On a map, all information is spatially related
and needs to be considered holistically. The composition of
map items decides if and how the reader understands the
message, with perception and understanding occurring sub-
consciously. To engage map-users and enable them to de-
velop an understanding of the meaning of the map, a visual
sense of the symbols and their attributes that correspond to
the intention of the cartographer is required (Robinson et al.,
1995).
3.11.5.2 Digital Mapping
Digital map creation is performed using either vector graphics
software (e.g., Adobe Illustrator) or GIS software. The main
advantage of graphical software with respect to the generation
of geomorphological maps is the great number of tools that
are available for the creation and modification of graphic
objects, as well as advanced printing options. Generally, these
can be adjusted and customized to the user’s needs and the
specific requirements of the legend system applied. These
functions still exceed the cartographic capabilities of GIS
software.
The primary advantage of a GIS is that it enables the
geographical reference system of the map to be retained, with
data stored in geographical databases for later analysis. Thus,
a GIS offers the ability to combine basic information on
N
18
50
184
0
Legend
Step height >5 m
Step height 2–5 m
Step height < 2 m
Creek
Contours 5 m
landforms such as process/material type and geometry with
secondary data, for example, from physical sampling, la-
boratory analyses, geophysical investigations, or the results
from geospatial analyses within the GIS (Minár et al., 2005;
Gustavsson et al., 2008). The results of GIS analyses are gen-
erally compiled into maps and consequently, GIS software
includes mapping facilities and some graphic-design capabil-
ities. Among these are automatic tools to generate the legend,
scale bar, north arrow, or coordinate grid, and functions
managing text labels and symbols. These map elements are
automatically and self-consistently updated when changes
are made in scale or symbol type. One challenge in the
process of map production is the generation of reusable,
standardized digital symbols. Although many symbols for
various objects or thematic purposes exist, only a few spe-
cialized geomorphological symbol sets are available for GIS
software (Bundesamt für Wasser und Geologie, 2002; Otto
and Dikau, 2004; Otto, 2008; IGUL, 2010). Given the existing
cartographic restrictions of many GIS, however, geomorpho-
logical legends and symbols need to be simplified for GIS-
based map production (Gustavsson et al., 2006; Gustavsson
and Kolstrup, 2009). Special symbol editors are provided to
compose and define the symbol set for the map (e.g., in
ArcGIS). As with graphic software, GIS software offers tools to
digitize vectors (points, line, polygons) with a high degree of
accuracy and provides the ability to modify single vector
nodes. An optimized graphical output is reached by transfer-
ring the GIS map to the graphics software to finalize the
geomorphological map.

Generating geomorphological maps using a GIS enables
numerous possibilities for the dissemination of research
outputs extending beyond simple paper products. Internet
technologies can contribute to both the dissemination of
geomorphological maps and access to geomorphologic data
and help to make geomorphological knowledge available
to the general public. In contrast to static digital maps
(i.e., simple images of maps, for example (http: http://gidimap.
giub.uni-bonn.de/gmk.digital/home_en.htm)), dynamic-web
maps are characterized by interactive capabilities, as the user
can interact with the map by zooming, panning, querying, or
adding thematic layers (Mitchell, 2005). Geo-registered map
data can be transferred and published in several digital ways
including GeoPDF, dynamic-web maps (e.g., WebGIS), and
virtual globes (e.g., Google EarthTM). These techniques are
outlined in the following sections.
3.11.5.2.1 Open standards
Data distribution and access in distributed web-based geos-
patial infrastructures need to be specified to achieve inter-
operability in a way that different applications (e.g., spatial
databases, Internet web servers, or web browsers) on various
platforms (e.g., Linux and Microsoft Windows) can interact
and communicate with each other. The specific needs for
interoperable geospatial technologies are implemented in
specifications or standards describing the basic data models to
represent different geographical features. They are important
because they contribute to both the interoperability for users
‘mashing’ up different applications and allowing software/
technology developers to make complex spatial information
and services universally accessible.
The standards are specified by the Open Geospatial Con-
sortium (OGC), a non-profit international standards organ-
ization, with members from commercial, governmental, and
research organizations (http://www.opengeospatial.org). It
leads the development of standards to establish interoperability
and ensures platform- and software-independent usability of
geospatial services and data sharing. The standards or specifi-
cations are the main outcomes of the OGC and appear as
openly published technical documents that detail interfaces or
encodings. These enable software developers to build support
for the interfaces or encodings into their products and services.
There are currently more than 30 standards defined, the most
prominent of which are web services, specifically:
• Web Map Service (WMS), which provides maps as raster
images;
• Web Feature Service (WFS), which allows the retrieval of
feature types (i.e., point, line, and polygon vector features);
• Web Coverage Service (WCS), which allows the retrieval of
tessellated surfaces, most commonly raster images. This
could be satellite imagery, DEMs, or aerial photography.
For data description and storage, Extensible Markup Lan-
guage (XML)-based languages such as GML (Geography
Markup Language), GeoSciML (GeoScience Markup Lan-
guage), or KML have been developed. XML defines a set of
rules for document encoding, comparable to HTML, used for
creating web pages. The OGC specifications and standards
have greatly influenced the direction of web-based GIS
Geovisualization 319
developments making it much easier to publish, visualize, and
exchange any geospatial data over the Internet (Mooney and
Winstanley, 2009). Basic functionality, advantages, and limi-
tations of WMS and KML are discussed in Section 3.11.5.2.3
and exemplified by case studies. In addition to specifications
and standards, the OGC publishes several White and Dis-
cussion Papers or Best Practice Guides (e.g., GeoPDF Encoding
Best Practice Version 2.2; Graves and Carl, 2009). Sec-
tion 3.11.5.2.2 introduces the GeoPDF, a merging of geospatial
data with the portable document format (PDF) file format.
3.11.5.2.2 GeoPDF
A GeoPDF, an OGC standard, includes one or multiple map
frames within a PDF page associated with a coordinate refer-
ence system (Graves and Carl, 2009). It enables the sharing of
geospatially referenced maps and data in PDF documents.
Multiple, independent map frames with individual spatial
reference systems are possible within a GeoPDF for map
overlays or insets. Geospatial functionality includes scalable
map display, layer visibility control, access to attribute data,
coordinate queries, and spatial measurements. This allows
similar functionality to simple interactive queries that would
be performed in a GIS. Adobe Reader (starting with Version
9.0) supports the geospatial functions of GeoPDFs; however,
full GeoPDF functionality requires the free TerraGo plug-in for
Adobe Reader (http://www.terrago.com). GeoPDFs can be
created either directly from a GIS (e.g., ESRI ArcGIS 9.3) or
using bespoke software (e.g., TerraGo Publisher or Map2PDF).
A GeoPDF enables fundamental GIS functionality outside
specialized GIS documents, turning the formerly static PDF
maps into interactive, portable, geo-referenced maps. It is an
interesting and valuable way of dissemination for geo-
morphological maps. Some geospatial data providers, such as
the USGS or the Australian Hydrographic Service (AHS), have
already started publishing interactive maps using the GeoPDF
format (http://store.usgs.gov).
3.11.5.2.3 Principles of web mapping and WebGIS
Web mapping is a common way of presenting dynamic
maps online. It links the Internet with GIS technology and
enables visualization, localization, and dissemination of geo-
registered data. Web-mapping applications such as Google
Maps or Bing Maps, which provide access to street data or
aerial imagery, are very popular and widespread and have in-
creased the interest and access to mapping. Mooney and
Winstanley (2009) point out that web mapping and WebGIS
applications are key components in the distribution of geos-
patial data and information.
Depending on the software system components, advanced
symbols, map overlays from different applications, and their
integration into a Desktop GIS are possible. Interoperability is
achieved through the use of OGC web services that include
mechanisms for the integration and visualization of infor-
mation from multiple sources. The terms ‘web mapping’
and ‘WebGIS’ are generally used synonymously, although they
do not necessarily refer to the same technologies. When
analytical GIS functionality is provided, the term WebGIS is
generally used (Mitchell, 2005; Mooney and Winstanley,
2009). GIS processing is performed online using a GIS server
and maps are visualized in interactive web viewers (e.g.,
320 Geovisualization
(a) WebGIS
Map request
Browser/
Web server
web client
Map
(b) Web service
Map request
Local client
(e.g. Google earth/
desktop GIS)
Map
Web server Mapserver
KML file KML file
Figure 18 Simplified scheme of information and data transfer of a WebGIS and Web-Service application.
OpenLayers – www.openlayers.org, MapBender – www.
mapbender.org, ka-Map! – www.ka-map.maptools.org, Map-
Builder – www.mapbuilder.net, Google Maps – maps.google.
com, or Bing Maps – www.bing.com/maps). Although there are
many ways to develop a WebGIS or to access web maps, de-
pending on the software components used, most applications
are based on the same principles (Figure 18).
The principles are as follows. The user operates a web
viewer inside a web browser and this provides selected
GIS functionality (e.g., zooming or panning, data query, layer
selection). The software compiles the user requests and for-
wards them to the application server (Figure 18(a)). The
server passes the map requests to the map-server, where the
central software performs the GIS processing. The map-server,
having access to the spatial data, executes the map requests
and returns the maps as images to the web server, which
finally serves them back to the user’s web viewer. The appli-
cation acts as a web-based information system. One popular
package is Maptool’s ‘MapServer for Windows’ (http://
www.maptools. org/ms4w/), which uses open-source com-
ponents to provide a map-server environment including lib-
raries for data input and output. MapServer is GIS software
running on a web server that enables interaction with GIS data
over the Internet and generates cartographic output of geo-
graphic content. For more information on common WebGIS
tools, see Mitchell (2005).
Most map-servers provide standardized web services like
WMS for accessing maps online. The WMS contains the map
request and parameters specifying GIS processing for the map-
server (e.g., choice of layers or spatial extent). Desktop GIS
Server
Forwarding
Access Geodata
Mapserver
Map
Geodata
Data
Server server
(optional)
Geodata
Geodata
Data
server
(optional)
(e.g., ArcGIS, MapInfo, and Global Mapper) as well as Internet
map viewers compile WMS data providing direct access to map
data from Internet servers. WMS technology permits users to
visualize entire GIS projects, such as a geomorphological map
on the Internet. Thematic layers of the map can be provided as
well as full analysis functionality, depending on the server-side
GIS software (e.g., Mapserver and GeoServer). Figure 19 shows
a WebGIS that visualizes the results of a geomorphological field
mapping campaign in the Turtmann valley, Switzerland (http://
www.geomorphology.at). The application employs MapServer
generating the maps as WMS, the spatial database management
system PostgreSQL (http://www.postgresql.org) maintaining
the geometries, and the web-mapping client Mapbender (http://
www.mapbender.org). Aerial images and a relief shaded map
are provided as base layers and several thematic layers present
information on process domains, surface materials, landforms,
and single processes. Due to MapServer’s powerful cartographic
engine, complex geomorphological symbols can be designed
and displayed. Symbols based on the legend for high-mountain
geomorphological systems established by Kneisel et al. (1998)
have been implemented. The WebGIS map thus uses the same
symbols as the printed map of the same area (Otto and Dikau,
2004). MapServer uses one symbol file that defines the com-
position of symbols for all types of vector geometries. Point
information, such as individual landforms, is displayed using a
geomorphological True Type font (Otto and Dikau, 2004) and
the spatial orientation of each character is achieved by pro-
viding the rotation angle as attribute data. Line features such as
crests and ridges are constructed using multilevel symbols, and
advanced polygon symbols are supported by hatchuring or

Figure 19 The graphical user interface (GUI) of the geomorphological WebGIS application Turtmanntal (Universities of Salzburg and Bonn,
available at http://www.geomorphology.at).
image fills. The Turtmanntal WebGIS offers simple functionality
of a Desktop GIS such as spatial navigation, coordinate queries,
length and area calculations, as well as the selection of single
layers of information. The composed image of the map frame
can be exported as a high-resolution PDF (300 dpi) in A4 and
A3 landscape or portrait orientation. For educational purposes,
a glossary defines geomorphological terms.
Usually no restrictions exist concerning the number of
WMS services included within a WebGIS application. Thus,
WebGIS applications are powerful tools to disseminate geo-
spatial information to users from different organizations (e.g.,
local authorities and environmental agencies).
It is important to note that Desktop-GIS software benefits
from the processing power of the local computer, web-based
applications perform all operations online in real time, and
factors such as bandwidth capacity, network latency, browser
type, and system performance need to be considered. In add-
ition, users expect rapid applications and instantaneous re-
sponses to their spatial queries (Mooney and Winstanley, 2009).
3.11.6 Conclusions
Geovisualization has garnered considerable interest, as a term
covering a wide swath of activities ranging from exploration,
through to analysis, synthesis, and presentation. An intuitive,
useful definition is ‘the visual depiction of spatial data,’ and
Geovisualization 321
this chapter has focused more on the use of geovisualization
in geomorphology under this definition. There is considerable
overlap between some of the techniques outlined here with
analysis techniques covered in other chapters in this volume,
and this demonstrates the important role that visualization
plays in the analysis of spatial data for geomorphological
applications. It also clearly demonstrates the very important
role that visualization plays in the exploration of patterns, as
we seek to formalize concepts and theories, and better
understand landscape evolution.
Advances in geospatial technologies have led to the wide-
spread availability of high-resolution terrain data that have fa-
cilitated advances in geomorphometry and geovisualization,
collectively revolutionizing geomorphology. The generation of
new observational geospatial data sets, through the use of
geovisualization and mapping, may be functionally termed
visual processing. In geomorphology, researchers are specifically
interested in attempting to visualize the topography in new
ways through quantitative characterization and qualitative
visualization. Numerous techniques can be used to describe
various characteristics of the topography that can be visualized,
including location, extent, variation, height, volume, and many
other parameters. This is still predominantly performed by
manual mapping efforts using image interpretation, although
new automated approaches also exist for the generation of
land-surface parameters and mapping, although they must be
carefully evaluated and ultimately formalized.
322 Geovisualization
For manual mapping, interpreters should be aware of
issues pertaining to the identification of landforms and spe-
cifically the minimum resolvable landform size, the effect of
azimuth biasing, and the concept of landform signal strength.
Satellite imagery rarely provides ideal datasets for accurate
landform mapping, and care should be taken in the selection
of imagery to account for those regions of the electromagnetic
spectrum that best differentiate landforms versus the sur-
rounding landscape. Image-processing techniques facilitate
visualization and range from simple to complex algorithms
and approaches. Many produce similar results, so it is critical
to utilize the best algorithms for enhancement.
Similarly, there are a variety of terrain visualization techni-
ques. Examples that were presented include relief shaded,
regional–residual separation to isolate landforms of interest,
and the use of multiscale geomorphometric parameters that
highlight or enhance specific topographic characteristics.
Combining terrain information in innovative ways that facili-
tate geomorphological mapping and visualization of the
topography are required to standardize visualization products
that characterize specific aspects of geomorphological systems.
Geovisualization also extends to interacting with spatio-
temporal data, a process that is now far more common across
disciplines. This can incorporate simple methods for spatial
overlay, panning, and zooming of 2D data, as well as ex-
tending to 2.5D visualization and exploration of surface data,
as well as true 3D and temporal analysis. Although the de-
velopment and evaluation of many of these techniques and
approaches are research intensive and require specialist
knowledge, simple functions such as panning, zooming, and
rotation illustrate the real power of interacting with, and
understanding, the nature of spatial data. There is no better
demonstration of this than in web-based interactive mapping
applications, which essentially represent a ‘stripped down’ GIS
interface for users to explore the characteristics of their data
within a managed setting. This straddles analysis and product
generation, and visualization products that also include tra-
ditional paper-based cartographic outputs. Although such
interactive maps might seem to be radically different products,
the collective power of human visualization and spatial
overlay for the organization and management of geospatial
data lies in the end-to-end digital workflows. That is, satellite
imagery and DEMs represent the primary data inputs and,
after subsequent processing, are used as the basis for gener-
ating geomorphological maps. The system can be used for the
generation of a variety of output that is appropriate for the
intended end-user, be that specific data layers, intermediate
processing results, or a printed map. Regardless of the medium
employed, care should be taken during the design phase in
order to synthesize a comprehensible and readable map.
Given that geomorphological maps are complex thematic
maps, there are high demands on cartographic visualization
techniques. Although effective geospatial communication may
not be the primary objective compared with information
production or science objectives, stakeholder engagement has
become a significant part of the research process. Researchers
are now required to engage the general public in terms of
outcomes, and relevance, of their work, whilst also providing
products that maximize downstream application and use.
Web-mapping systems provide the flexibility to target different
end-user groups and geomorphologists should make full use
of the facilities available.
It is also pertinent to consider the potential for new de-
velopments in geovisualization. It is very likely that more
automated approaches to landform mapping will dominate in
the future, as new analysis and visualization techniques permit
better characterization of topographic features. Dynamic
visualization techniques, and particularly in the areas of 3D
and temporal analysis, will gradually improve, and we can
expect to see new output products that better engage people
and make use of augmented reality. Such technologies already
exist in limited ways, and new developments for assessing
subsurface and material variations in the near-surface
environment are sorely needed.
References
Andrienko, G., Andrienko, N., Jankowski, P., Keim, D., Kraak, M.J., MacEachren,
A.M., Wrobel, S., 2007. Visual analytics for spatial decision support: setting the
research agenda. International Journal of Geographic Information Science 8,
839–857.
Ballagh, L.M., Raup, B.H., Duerr, R.E., Khalsa, S.J.S., Helm, C., Fowler, D., Gupte,
A., 2011. Representing scientific data sets in KML: methods and challenges.
Computers & Geosciences 37(1), 57–64.
Barsch, D., Liedtke, H., 1980. Principles, scientific value and practical applicability
of the geomorphological map of the Federal Republic of Germany at the scale of
1:25,000 (GMK 25) and 1:100,000 (GMK 100). Zeitschrift fur Geomorphologie
Suppl 36, 296–313.
Benetti, S., Dunlop, P., Cofaigh, C.O., 2010. Glacial and glacially-related features on
the continental margin of northwest Ireland mapped from marine geophysical
data. Journal of Maps v2010, 14–29.
Bennett, G.L., Evans, D.J.A., Carbonneau, P., Twigg, D.R., 2010. Evolution of a
debris-charged glacier landsystem, Kviarjokull, Iceland. Journal of Maps v2010,
40–67.
Bonham-Carter, G.F., 1994. GIS for Geoscientists. Elsevier, Oxford.
Brunsden, D., Doornkamp, J.C., Fookes, P.G., Jones, D.K.C., Kelly, J.M.H., 1975.
Large scale geomorphological mapping and highway engineering design.
Quarterly Journal of Engineering Geology 8, 227–253.
Bundesamt für Wasser und Geologie, B., 2002. Naturgefahren Symbolbaukasten zur
Kartierung der Phänomene – EDV-Legende für die digitale Karthographie.
Bundesamt für Umwelt, Switzerland, Bern.
Calcagno, P., Cazenave, A., 1994. Subsidence of the sea-floor in the Atlantic and
Pacific Oceans – regional and large-scale variations. Earth and Planetary
Science Letters 126, 473–492.
Camargo, F.F., Florenzano, T.G., de Almeida, C.M., de Oliveira, C.G., 2009.
Geomorphological mapping using object-based analysis and ASTER DEM in the
Paraı́ba do Sul Valley, Brazil. International Journal of Remote Sensing 30,
6613–6620.
Campbell, J.B., 2007. Introduction to Remote Sensing. Taylor & Francis, London.
Carbotte, S.M., Arko, R., Chayes, D.N., et al., 2004. New integrated data
management system for Ridge2000 and MARGINS research. EOS Transactions
85, 553.
Cazenave, A., Dominh, K., Allègre, C.J., Marsh, J.G., 1986. Global relationship
between oceanic geoid and topography. Journal of Geophysical Research
91(B11), 11,439–11,450.
Chen, J., Jönsson, P., Tamura, M., Gu, Z., Matsushita, B., Eklundh, L., 2004. A
simple method for reconstructing a high-quality NDVI time-series data set based
on the Savitzky–Golay filter. Remote Sensing of Environment 91, 332–344.
Clark, C.D., 1997. Reconstructing the evolutionary dynamics of palaeo-ice sheets
using multi-temporal evidence, remote sensing and GIS. Quaternary Science
Reviews 16, 1067–1092.
Colwell, R.N., 1983. Manual of Remote Sensing. American Society of
Photogrammetry, Falls Church, VA.
Cooke, R.U., Doornkamp, J.C., 1990. Geomorphology in environmental
management. A New Introduction, 2nd ed. Clarendon Press, Oxford.
Crosby, A.G., McKenzie, D., Sclater, J.G., 2006. The relationship between depth,
age and gravity in the oceans. Geophysical Journal International 166, 553–573.

De Graaff, L.W.S., De Jong, M.G.G., Rupke, J., Verhofstad, J., 1987. A
geomorphological mapping system at scale 1:10 000 for mountainous areas
(Austria). Zeitschrift fur Geomorphologie 31, 229–242.
Demek, J., Embleton, C., Gellert, J.F., Verstappen, H.T., 1972. Manual of detailed
geomorphological mapping. International Geographical Union Commission on
Geomorphological Survey and Mapping. Academia, Prague.
Dramis, F., Guida, D., Cestari, A., 2011. Nature and aims of geomorphological
mapping. In: Smith, M.J., Paron, P., Griffiths, J. (Eds.), Geomorphological
Mapping: Methods and Applications. Elsevier, London, chap. 3.
Dykes, A.P., 2008. Geomorphological maps of Irish peat landslides created using
hand-held GPS. Journal of Maps 2008, 258–276.
ESRI, 2003. Spatial Data Standards and GIS Interoperability. ESRI white paper.
http://www.esri.com/library/whitepapers/pdfs/spatial-data-standards.pdf (accessed
12/07/2010).
Etzellmüller, B., 2000. On the quantification of surface changes using grid-based
digital elevation models (DEMs). Transactions in Geographical Information
Systems 4, 129–143.
Evans, I.S., 1972. General Geomorphometry, derivatives of altitude and descriptive
statistics. In: Chorley, R.J. (Ed.), Spatial Analysis in Geomorphology. Harper and
Row, New York, pp. 17–90.
Evans, I.S., 1990. Cartographic techniques in geomorphology. In: Goudie, A. (Ed.),
Geomorphological Techniques. Unwin Hyman, London, pp. 97–108.
Evans, I.S., 2012. Geomorphometry and landform mapping: What is a landform?
Geomorphology 137, 94–106.
Fabuel Perez, I., Hodgetts, D., Redfern, J., 2010. Integration of digital outcrop
models (DOMs) and high resolution sedimentology – workflow and implications
for geological modelling. Petroleum Geoscience 16, 1–23.
Ford, J.P., 1984. Mapping of glacial landforms from Seasat radar images.
Quaternary Research 22, 314–327.
Gibson, P., Power, C., 2000. Introductory Remote Sensing Digital Image Processing
and Applications. Routledge, London.
GITTA, 2006. Layout Design Settings/Graphical Semiology. Geographic Information
Technology Training Alliance. http://www.gitta.info/LayoutDesign/en/html/
index.html
Gousie, M.B., Smith, M.J., 2010. DEMView: 3D visualization of DEM error,
Accuracy 2010, Leicester, July 2010, pp. 165–168.
Graves, P., Carl, R., 2009. GeoPDF Encoding Best Practice Version 2.2, OGC
project document: OGC 08-139r2, Version 0.5.1. Open Geospatial Consortium
Inc.
Grohmann, C.H., Smith, M.J., Riccomini, C., 2010. Multi-scale analysis of
surface roughness. IEEE Transactions on Geoscience and Remote Sensing 99,
1–14.
Gustavsson, M., Kolstrup, E., 2009. New geomorphological mapping system used
at different scales in a Swedish glaciated area. Geomorphology 110, 37–44.
Gustavsson, M., Kolstrup, E., Seijmonsbergen, A.C., 2006. A new symbol-and-GIS
based detailed geomorphological mapping system: renewal of a scientific
discipline for understanding landscape development. Geomorphology 77, 90–111.
Gustavsson, M., Seijmonsbergen, A.C., Kolstrup, E., 2008. Structure and contents
of a new geomorphological GIS database linked to a geomorphological map -
With an example from Liden, central Sweden. Geomorphology 95, 335–349.
Guth, P.L., 2001. Quantifying terrain fabric in digital elevation models. In: Ehlen, J.,
Harmon, R. (Eds.), The Environmental Legacy of Military Operations. Geological
Society of America Reviews in Engineering Geology, Boulder, CO, pp. 13–25.
Hardin, P.J., Jensen, R.R., 2011. Introduction – small-scale unmanned aerial
systems for environmental remote sensing. GIScience and Remote Sensing
48(1), 1–3.
Hart, J.K., Martinez, K., Ong, R., Riddoch, A., Rose, K.C., Padhay, P., 2006. A
wireless multi-sensor subglacial probe: design and preliminary results. Journal
of Glaciology 52, 389–397.
Hengl, T., Reuter, H.I., 2008. Geomorphometry: Concepts, Software, Applications,
Elsevier.
Heritage, G., Large, A., 2009. Laser Scanning for the Environmental Sciences.
Wiley-Blackwell, Chichester.
Hesse, R., 2010. LiDAR-derived local relief models - a new tool for archaeological
prospection. Archaeological Prospection 17, 67–72.
Hillier, J., Bunbury, J., Graham, A., 2007. Monuments on a Migrating Nile. Journal
of Archaeological Science 34, 1011–1015.
Hillier, J., Smith, M.J., 2008. Residual relief separation: DEM enhancement for
geomorphological mapping. Earth Surface Processes and Landforms 33, 2266–2276.
Hillier, J., Tilmann, F., Hovius, N., 2008. Submarine geomorphology: new views on
an ’unseen’ landscape. Basin Research 20, 467–472.
Hillier, J.K., 2008. Seamount detection and isolation with a modified wavelet
transform. Basin Research 20, 555–573.
Geovisualization 323
Hillier, J.K., 2011. Submarine geomorphology: quantitative methods illustrated with the
Hawaiian volcanoes. In: Smith, M.J., Paron, P., Griffiths, J. (Eds.),
Geomorphological Mapping Methods and Applications. Elsevier, London, chap. 12.
Hillier, J.K., Watts, A.B., 2004. ‘‘Plate-like’’ subsidence of the east Pacific rise –
south Pacific superswell system. Journal of Geophysical Research 109. http://d
x.doi.org/10.1029/2004JB003041.
Hillier, J.K., Watts, A.B., 2007. Global distribution of seamounts from ship-track
bathymetry data. Geophysical Research Letters 34(13), L13304. http://dx.doi.org/
10.1029/2007GL029874.
Hobson, R.D., 1972. Surface roughness in topography: quantitative approach.
In: Chorley, R.J. (Ed.), Spatial analysis in geomorphology. Methuen, London,
pp. 225–245.
Hughes, A., Clark, C.D., 2010. Subglacial bedform map of Britain. Journal of Maps
v2010, 543–563.
IGUL, 2010. La légende géomorphologique de l’IGUL. http://www.unil.ch/igul/
page19238.html
Jansson, K.N., Glasser, N.F., 2005. Using landsat 7 ETM þ imagery and Digital
Terrain Models for mapping lineations on former ice sheet beds. International
Journal of Remote Sensing 26, 3931–3941.
Johnson, A., Leigh, J., Morin, P., Van Keken, P., 2006. GeoWall: stereoscopic
visualization for geoscience research and education. IEEE Computer Graphics
and Applications 26, 10–14.
Jones, E.J.W., 1999. Marine Geophysics. Wiley, Chichester.
Jonsson, P., Eklundh, L., 2004. TIMESAT – a program for analysing time-series of
satellite sensor data. Computers and Geosciences 30, 833–845.
Jordan, C., Bateson, L., Bow, J., Newell, A., Napier, B., Sabine, R.J., 2009.
Geovisionary Software for 3D Visualisation and Petroleum Exploration in
Southern Tajikistan, RSPSoc 2009: New Dimensions in Earth Observation.
Remote Sensing Society, Leicester, UK, pp. 637-642.
Kienholz, H., 1978. Maps of geomorphology and natural hazards of Grindelwald,
Switzerland: scale 1:10000. Arctic and Alpine Research 10, 169–184.
Kienholz, H., Krummenacher, B., 1995. Symbolbaukasten zur Kartierung der
Phänomene. In: Bundesamt für Umwelt, W.u.L.B. (Ed.), Mitteilungen des
Bundesamtes für Wasser und Geologie Nr. 6. Bundesamt für Umwelt, Wald und
Landschaft (BUWAL), Bern.
Kim, S., Wessel, P., 2008. Directional median filtering for the regional–residual
separation of bathymetry. Geochemistry Geophysics Geosystems 9, Q03005.
Klimaszewski, M., 1982. Detailed geomorphological maps. ITC Journal 3,
265–271.
Kneisel, C., Lehmkuhl, F., Winkler, S., 1998. Legende für geomorphologische
Kartierungen in Hochgebirgen (GMK Hochgebirge). Trierer Geographische
Studien 18, 12.24.
Koppers, A.A.P., Staudigel, H., Minnett, R., 2010. Seamount catalog: seamount
morphology, maps, and data files. Oceanography 23, 36–37.
Kraak, M.-J., 2008. Editorial: from geovisualisation toward geovisual analytics.
Cartographic Journal 45, 163–164.
Kraak, M.-J., Ormeling, F., 2010. Cartography: Visualisation of Spatial Data, Third
ed. Prentice Hall, Harlow, Essex.
Laliberte, A.S., Rango, A., 2011. Image processing and classification procedures for
analysis of sub-decimeter imagery acquired with an unmanned aircraft over arid
rangelands. GIScience and Remote Sensing 48(1), 4–23.
Lillesand, T.M., Kiefer, R.W., Chipman, J.W., 2008. Remote Sensing and Image
Interpretation, Sixth ed. Wiley, New York.
Lovell, J.L., Jupp, D.L.B., Culvenor, D.S., Coops, N.C., 2003. Using airborne and
ground based ranging Lidar to measure canopy structure in Australian forests.
Canadian Journal of Remote Sensing 29, 607–622.
MacEachren, A.M., Gahegan, M., Pike, W., Brewer, I., Cai, G., Lengerich, E.,
Hardistry, F., 2004. Geovisualization for knowledge construction and decision
support. Computer Graphics and Applications 24(1, 24), 13–17.
Marty, J.C., Cazenave, A., 1989. Regional variations in subsidence rate of oceanic
plates: a global analysis. Earth and Planetary Science Letters 94, 301–315.
Mason, D.C., Horritt, M.S., Hunter, N.M., Bates, P.D., 2007. Use of fused airborne
scanning laser altimetry and digital map data for urban flood modeling.
Hydrological Processes 21, 1436–1447.
Mather, P.M., 2004. Computer Processing of Remotely-Sensed Images. Wiley
Blackwell, London.
McKean, J., Roering, J.J., 2004. Objective landslide detection and surface
morphology mapping using high-resolution airborne laser altimetry.
Geomorphology 57, 331–351.
McKenzie, D.P., Watts, A.B., Parsons, B., Roufosse, M., 1980. Planform of mantle
convection beneath the Pacific Ocean. Nature 288, 442–446.
Menard, H.W., 1973. Depth anomalies and the bobbing motion of drifting islands.
Journal of Geophysical Research 78, 5128–5137.
324 Geovisualization
Minár, J., Mentlı́k, P., Jedlička, K., Barka, I., 2005. Geomorphological information
system – idea and options of practical implementation. Geografický Časopis
57(3), 247–266.
Mitchell, T., 2005. Web Mapping Illustrated. O’Reilly, Sebastopol.
Mooney, P., Winstanley, A.C., 2009. Representation of environmental data in web-
based GIS. In: Mount, N., Harvey, G., Aplin, P., Priestnall, G. (Eds.),
Representing, Modelling, and Visualising the Natural Environment. CRC Press,
Boca Raton, London, New York.
Olaya, V., 2008. Basic Land-Surface Parameters. In: Hengl, T., Reuter, H.I. (Eds.),
Geomorphometry: Concepts, Software, Applications. Elsevier, Amsterdam, pp.
3–30.
Otto, J.-C., 2008. Symbols for geomorphologic mapping in high mountains for
ArcGIS. http://www.geomorphology.at/index.phpoption=com_content&task=
view&id=133&Itemid=176.
Otto, J.-C., Dikau, R., 2004. Geomorphic system analysis of a high mountain valley
in the Swiss Alps. Zeitschrift für Geomorphologie N.F 48, 323–341.
Otto, J.-C., Gustavsson, M., Geilhausen, M., 2011. Cartography: design,
symbolisation and visualisation of geomorphological maps. In: Smith, M.J.,
Paron, P., Griffiths, J. (Eds.), Geomorphological Mapping: A Handbook of
Techniques and Applications. Elsevier, London.
Otto, J.C., Schrott, L., Jaboyedoff, M., Dikau, R., 2009. Quantifying sediment
storage in a high alpine valley (Turtmanntal, Switzerland). Earth Surface
Processes and Landforms 34, 1726–1742.
Passarge, S., 1912. Physiologische morphologie. Mitt. Geogr. Ges. Hamburg 26,
135–337.
Passarge, S., 1914. Morphologie des MeXtischblattes Stadtremda. Mittheilungen der
Geographischen Gesellschaft in Hamburg 28, 1–221.
Pasuto, A., Soldati, M., Siorpaes, C., 1999. Geomorphological Map of the
Surroundings of Cortina d’Ampezzo (Dolomites, Italy): Scale 1: 20,000. Dipartimento
di Scienze della Terra, Università di Modena e Reggio Emilia, Modena.
Pesci, A., Teza, G., Casula, G., et al., 2011. Multitemporal laser scanner-based
observation of the Mt. Vesuvius crater: characterization of overall geometry and
recognition of landslide events. ISPRS Journal of Photogrammetry and Remote
Sensing 66, 327–336.
Pike, R.J., Wilson, S.E., 1971. Elevation-relief ratio, hypsometric integral, and
geomorphic area-altitude analysis. GSA Bulletin 82(4), 1079–1084.
Punkari, M., 1982. Glacial geomorphology and dynamics in the eastern parts of the
Baltic Shield interpreted using Landsat imagery. Photogrammetric Journal
Finland 9, 77–93.
Reitmayr, G., Eade, E., Drummond, T., 2005. Localisation and interaction for
augmented maps. Proceedings of the 4th IEEE/ACM International Symposium on
Mixed and Augmented Reality. IEEE Computer Society, Washington DC, USA,
pp. 120–129.
Robinson, A.H., Morrison, J.L., Muehrcke, P.C., Kimerling, A.J., Guptill, S.C., 1995.
Elements of Cartography, Sixth ed. Wiley, Chichester.
Rose, J., Smith, M.J., 2008. Glacial geomorphological maps of the Glasgow region,
western central Scotland. Journal of Maps v2008, 399–416.
Sahlin, E.A.U., Glasser, N.F., 2008. Geomorphological map of Cadair Idris,
Wales. Journal of Maps 2008, 299–314. http://dx.doi.org/10.4113/
jom.2008.1011.
Sato, H.P., Harp, E.L., 2009. Interpretation of earthquake-induced landslides
triggered by the 12 May 2008, M7.9 Wenchuan earthquake in the Beichuan
area, Sichuan Province, China using satellite imagery and Google Earth.
Landslides 6, 153–159.
Schmidt, J., Evans, I.S., Brinkmann, J., 2003. Comparison of polynomial models
for land surface curvature calculation. International Journal of Geographic
Information Science 17, 797–814.
Sclater, J.G., Lawver, L.A., Parsons, B., 1975. Comparison of long-wavelength
residual elevation and free air gravity anomalies in the north Atlantic and
possible implications for the thickness of the lithospheric plate. Journal of
Geophysical Research 80, 1031–1042.
Seijmonsbergen, A.C., de Graaff, L.W.S., 2006. Geomorphological mapping and
geophysical profiling for the evaluation of natural hazards in an alpine
catchment. Natural Hazards and Earth System Science 6, 185–193.
Seijmonsbergen, A.C., Hengl, T., Anders, N.S., 2011. Semi-automated identification
and extraction of geomorphological features using digital elevation data.
In: Smith, M.J., Paron, P., Griffiths, J. (Eds.), Geomorphological Mapping:
Methods and Applications. Elsevier, London, chap. 11.
Sithole, G., Vosselman, G., 2004. Experimental comparison of filter algorithms for
bare-Earth extraction from airborne laser scanning point clouds. ISPRS Journal
of Photogrammetry and Remote Sensing 59, 85–101.
Slaney, V.R., 1981. Landsat images of Canada – a geological appraisal. Geological
Survey of Canada Paper 80-5, p. 102.
Smith, M.J., 2011. Digital Mapping: visualisation, interpretation and quantification
of landforms. In: Smith, M.J., Paron, P., Griffiths, J. (Eds.), Geomorphological
Mapping: Methods and Applications. Elsevier, London, chap. 8.
Smith, M.J., Chandler, J., Rose, J., 2009. High spatial resolution data acquisition
for the geosciences: kite aerial photography. Earth Surface Processes and
Landforms 34, 155–161.
Smith, M.J., Clark, C.D., 2005. Methods for the visualisation of digital elevation
models for landform mapping. Earth Surface Processes and Landforms 30,
885–900.
Smith, M.J., McClung, D.M., 1997. Avalanche frequency and terrain
characteristics at Rogers’ Pass, British Columbia, Canada. Journal of Glaciology
43, 165–171.
Smith, M.J., Pain, C., 2009. Applications of remote sensing in geomorphology.
Progress in Physical Geography 33, 568–582.
Smith, M.J., Rose, J., Booth, S., 2006. Geomorphological mapping of glacial
landforms from remotely sensed data: an evaluation of the principal data
sources and an assessment of their quality. Geomorphology 76, 148–165.
Smith, M.J., Rose, J., Gousie, M.B., 2009. The cookie cutter: a method for
obtaining a quantitative 3D description of glacial bedforms. Geomorphology
108, 209–218.
Smith, M.J., Wise, S.M., 2007. Problems of bias in mapping linear landforms from
satellite imagery. International Journal of Applied Earth Observation and
Geoinformation 9, 65–78.
Smith, W.H.F., 1990. Marine geophysical studies of seamounts in the Pacific Ocean
Basin. Unpublished PhD thesis, Columbia University.
Smith, W.H.F., 1993. On the accuracy of digital bathymetric data. Journal of
Geophysical Research 98, 9591–9603.
Thomas, J.J., Cook, C.A., 2005. Illuminating the Path: The Research and
Development Agenda for Visual Analytics. IEEE Press, Washington.
Tooth, S., 2006. Virtual globes: a catalyst for the re-enchantment of
geomorphology? Earth Surface Processes and Landforms 31, 1192–1194.
Van Asselen, S., Seijmonsbergen, A.C., 2006. Expert-driven semi-automated
geomorphological mapping using a laser DTM. Geomorphology 78,
309–320.
Van Wyckhouse, R., 1973. SYNBAPS. Technical Report TR-233, National
Oceanographic Office, Washington D.C.
Vencatasawmy, C.P., Clark, C.D., Martin, R.J., 1998. Landform and lineament
mapping using radar remote sensing. In: Lane, S.N., Richards, K.S., Chandler,
J.H. (Eds.), Landform Monitoring and Analysis. Wiley, New York.
Wang, Y., Mercer, B., Tao, V.C., Sharma, J., Crawford, S., 2001. Automatic
generation of bald earth digital elevation models from digital surface models
created using airborne IFSAR, Proceedings of 2001 ASPRS Annual Conference,
St. Louis, Missouri, USA.
Watts, A.B., 1976. Gravity and bathymetry in the central Pacific Ocean. Journal of
Geophysical Research 81, 1533–1548.
Watts, A.B., Daly, S.F., 1981. Long wavelength gravity and topography anomalies.
Annual Review of Earth and Planetary Sciences 9, 415–448.
Welsh, A., Davies, T., 2010. Identification of alluvial fans susceptible to debris-flow
hazards. Landslides 8(2), 183–194.
Werneke, J., 2008. The KML Handbook: Geographic Visualization for the Web.
Addison-Wesley Professional, San Francisco.
Wessel, P., 1998. An empirical method for optimal robust regional–residual
separation of geophysical data. Mathematical Geology 30, 391–408.
Wessel, P., Smith, W.H.F., 1998. New, improved version of generic mapping tools
released. EOS Transactions of the American Geophysical Union 79, 579.
White, R.S., 1993. Melt production-rates in mantle plumes. Philosophical
Transactions of the Royal Society of London Series A-Mathematical Physical and
Engineering Sciences 342, 137–153.
Williams, R., Brassington, J., Vericat, D., Hicks, M., Labrosse, F., Neal, M., 2011.
Monitoring braided river change using terrestrial laser scanning and optical
bathymetric mapping. In: Smith, M.J., Paron, P., Griffiths, J. (Eds.),
Geomorphological Mapping: Methods and Applications. Elsevier, London,
chap. 20.
Wise, S., 2000. Assessing the quality for hydrological applications of digital
elevation models derived from contours. Hydrological Processes 14, 1909–1929.
Wood, J., Dykes, J., Slingsby, A., Clarke, K., 2007. Interactive visual exploration of
a large spatio-temporal dataset: reflections on a geovisualization mashup. IEEE
Transactions on Visualization and Computer Graphics 13, 1176–1183.
Wood, J.D., Fisher, P.F., 1993. Assessing interpolation accuracy in elevation
models. IEEE Computer Graphics and Applications 13, 48–56.
Yokoyama, R., Shirasawa, M., Pike, R.J., 2002. Visualizing topography by
Openness: a new application of image processing to digital elevation models.
Photogrammetric Engineering and Remote Sensing 68, 257–265.
 Geovisualization 325

Biographical Sketch

Mike J. Smith received a BS (Hons.) in geography from the University of Wales, Aberystwyth, an MS degree in
geography from the University of British Columbia, and a PhD degree in paleo-glaciology from the University of
Sheffield. He is currently a senior lecturer in GIS in the School of Geography, Geology and the Environment at
Kingston University, where he contributes to bachelor and masters programs on the application of remote sensing
in the geosciences. His research interests are based on the application of digital elevation models in geo-
morphology and specifically focused upon the visualization and geomorphometric modeling of glacial land-
scapes. Recent interests also include field spectroscopy of loess. He is the founder and editor of the Journal
of Maps.

John K. Hillier received BA and MS degrees in Natural Sciences (Geology) from the University of Cambridge, and
a PhD degree in Marine Geophysics from the University of Oxford. He was awarded a research fellowship at St.
Catharine’s College Cambridge and has worked for Zurich Insurance as a catastrophe modeler. He is currently a
lecturer at Loughborough University, with research interests in geohazards and quantitative geomorphology.
Recent interests include drumlin formation, volcano formation, evolution of oceanic tectonic plates, and the
correlation between natural perils.

Jan-Christoph Otto has studied Geography, Geology, and Soil Science at the Universities of Bonn and Grenoble.
He received his Diploma and PhD in Geography from the University of Bonn.
He is currently a postdoctoral researcher in the Department of Geography and Geology, University of Salzburg,
within the Geomorphology and Environmental Systems Research Group. His main research interests include
landscape change in high-mountain environments with special focus on sediment budgets and the dynamics and
consequences of permafrost changes in mountain areas. His recent interests include the creation of digital geo-
morphological maps and the combination of high-resolution surface and subsurface data using GIS technology.

Martin Geilhausen studied Geography at the University of Bonn and received a diploma investigating fluvial and
glacial deposits using geophysics. After working for a geoinformation company, where he developed webGIS and
web-mapping applications, he enrolled as a PhD student in the Department of Geography and Geology, Uni-
versity of Salzburg (Geomorphology and Environmental Systems Research Group). His research involves the
quantification of recent and postglacial sediment dynamics in glacier fore-fields, Austrian Alps. Martin is applying
a suite of field techniques ranging from fluvial erosion measurement and geophysics (ground penetrating radar,
seismic refraction, and resistive tomography) to terrestrial laser scanning.