Download as pdf or txt
Download as pdf or txt
You are on page 1of 30

Remote Sensing of Floods

Oxford Research Encyclopedia of Natural Hazard


Science
Remote Sensing of Floods  
Guy J.-P. Schumann
Subject: Floods Online Publication Date: Apr 2017
DOI: 10.1093/acrefore/9780199389407.013.265

Summary and Keywords

For about 40 years, with a proliferation over the last two decades, remote sensing data,
primarily in the form of satellite and airborne imagery and altimetry, have been used to
study floods, floodplain inundation, and river hydrodynamics. The sensors and data
processing techniques that exist to derive information about floods are numerous.
Instruments that record flood events may operate in the visible, thermal, and microwave
range of the electromagnetic spectrum. Due to the limitations posed by adverse weather
conditions during flood events, radar (microwave range) sensors are invaluable for
monitoring floods; however, if a visible image of flooding can be acquired, retrieving
useful information from this is often more straightforward. During recent years, scientific
contributions in the field of remote sensing of floods have increased considerably, and
science has presented innovative research and methods for retrieving information
content from multi-scale coverages of disastrous flood events all over the world. Progress
has been transformative, and the information obtained from remote sensing of floods is
becoming mature enough to not only be integrated with computer simulations of flooding
to allow better prediction, but also to assist flood response agencies in their operations.

Furthermore, this advancement has led to a number of recent and upcoming satellite
missions that are already transforming current procedures and operations in flood
modeling and monitoring, as well as our understanding of river and floodplain
hydrodynamics globally. Global initiatives that utilize remote sensing data to strengthen
support in managing and responding to flood disasters (e.g., The International Charter,
The Dartmouth Flood Observatory, CEOS, NASA’s Servir and the European Space
Agency’s Tiger-Net initiatives), primarily in developing nations, are becoming established
and also recognized by many nations that are in need of assistance because traditional
ground-based monitoring systems are sparse and in decline. The value remote sensing
can offer is growing rapidly, and the challenge now lies in ensuring sustainable and
interoperable use as well as optimized distribution of remote sensing products and
services for science as well as operational assistance.

Page 1 of 30

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, NATURAL HAZARD SCIENCE (oxfordre.com/
naturalhazardscience). (c) Oxford University Press USA, 2019. All Rights Reserved. Personal use only; commercial use is
strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 18 February 2019


Remote Sensing of Floods

Keywords: floods, remote sensing, satellite imagery, altimetry, flood forecasting, hydrodynamic modeling, disaster
assistance, interoperability

Introduction
Flooding is one of the costliest natural disasters. According to Munich RE, over the last
35 years, three of the five costliest natural disasters have been floods from storm surges
and tsunamis (the 2012 Sandy event, the 2011 Japan tsunami, and Katrina in 2005), and
this is the case for nearly every year in this decade (see Figure 1 as an example for 2015).
By 2050, costs of floods in coastal cities alone could reach $1 trillion annually (Hallegate,
Green, Nicholls, & Corfee-Morlot, 2013). Recently, all around the world, floods have been
of exceptionally high magnitude, with rainfalls exceeding record amounts and causing
unprecedented damage in many countries (e.g., U.S. East Coast, Malawi, Philippines,
India, U.S. Southwest, Northern England, U.S. Mississippi/Midwest, etc.).

Recent events have


covered spatial scales well
beyond past records and
are frequently surpassing
traditional regional
measurement and disaster
response coverage. In
essence, these large-scale
events highlight the need
Figure 1. The ten costliest natural disasters in 2015 for data and monitoring
by overall losses, insured losses, and fatalities. © coverage that can only be
Munich RE.
provided by remote
sensing platforms (i.e.,
satellite and airborne sensors).

Remote sensing, in its true definition, is the acquisition of information about an object or
phenomenon without making physical contact with the object, but is more commonly
referred to as the scanning of the Earth by satellite or high-flying aircraft and therefore
slowly becoming popularly synonymous with the science of Earth Observation.

Remote sensing of flood event processes and variables is of great value to many sectors
(flood risk mitigation planning, disaster relief services, global reinsurance markets, and
research) but the amount and quality of information available varies greatly with location,
spatial scales, and time. Remote sensing of floods can complement ground-based
observations and be integrated with computer models of flooding for event re-analysis
and forecasting to augment the amount of information available to end users, decision
makers, and scientists.

Page 2 of 30

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, NATURAL HAZARD SCIENCE (oxfordre.com/
naturalhazardscience). (c) Oxford University Press USA, 2019. All Rights Reserved. Personal use only; commercial use is
strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 18 February 2019


Remote Sensing of Floods

However, before using remote sensing data or products for flood monitoring and
management, it is important to consider end-user requirements and the appropriate
timeline as well as the spatial resolution of the delivered products. Figure 2 shows the
requirement in terms of spatial resolution and turnaround time for different flood
management sectors. For instance, flood mapping for emergency management could be
done at any spatial resolution, and accuracy may be less important, but data or products
should be made available within 48 hours, preferably in 12- to 24-hour intervals, whereas
for the insurance industry, the opposite situation would apply; spatial resolutions should
be finer than 10 m, but timeliness might be of lesser concern.

Although at first glance


this situation seems
favorable, as it is in line
with the fundamental
physical principle that by
increasing satellite swath
Figure 2. Temporal and spatial resolution coverage, thus decreasing
requirements by different sectors active in flood
monitoring and management. Composed after Blyth,
spatial resolution, the
1997. orbital revisit time is
increased, research
scientists and product developers are far from meeting needs, at least in the field of flood
monitoring. The main reason for this gap is two-fold. On the one hand, the science lacks
understanding of end-user needs, and on the other hand, decision makers are oftentimes
reluctant to use new data and methods in their operations or systems.

Nevertheless, over the past two decades, remote sensing has clearly been transformative
in the way we now understand flood processes at different scales, model and predict
floods, and assist flood disaster response. Success stories are numerous, but both the
science and end-user communities need to be aware of the fundamental limitations of
remote sensing of floods and manage expectations accordingly.

The Science of Remote Sensing of Floods


Within the electromagnetic spectrum, there are several ranges of wavelengths that are
typically used for remote sensing of floods. The atmosphere absorbs portions of the
signal, and there are only a few so called atmospheric windows that allow the satellite
signals to be transmitted (Figure 3). The most common ranges of wavelengths used for
floods, or surface water detection in general, extend from the visible (optical) to the near-
infrared and thermal infrared range, and the microwave (radar) range at the opposite
side of the spectrum.

Page 3 of 30

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, NATURAL HAZARD SCIENCE (oxfordre.com/
naturalhazardscience). (c) Oxford University Press USA, 2019. All Rights Reserved. Personal use only; commercial use is
strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 18 February 2019


Remote Sensing of Floods

Figure 3. Top panel: The electromagnetic spectrum.


Wavelength regions typically used in remote sensing
of floods are shown. Bottom panel: Graph showing
the parts of the spectrum that are referred to as
“atmospheric windows,” that is, where atmospheric
transmission of wavelengths is high. Modified after
Lubin and Massom (2007).

Imaging Sensors

Optical and Infrared Range


The information about floods that can be retrieved from a satellite or airborne instrument
operating in the visible-to-infrared ranges depends on daylight, weather conditions
(particularly clouds and rain), vegetation canopy density (especially emerging flooded
vegetation or overhanging vegetation), and the chemical composition as well as the
turbidity of the water (e.g., water sediment loads), all of which vary considerably during
flood events.

These limitations are considerable, and the situation is oftentimes further aggravated by
the fact that the satellite actually needs to be passing over the flood at the right time.
Unless satellites have a very frequent revisit time (such as NASA’s Aqua and Terra
satellites, or NOAA’s AVHRR series) or operate in constellation for a given mission (e.g.,
[the Italian space agency] ASI’s COSMO-SkyMed, ESA’s Sentinel-1, or the International
Charter), it is rather by chance to get a useful image of a flood, given that the average
satellite revisit time is about two weeks, which is more than the typical flood duration in
non-monsoonal regions.

If an image can be acquired, either by aircraft or satellite, in the visible to near-infrared


range, classifying water bodies and inundated area is often more straightforward than on
a radar image due to the fact that the human eye is more accustomed to this type of
image and thus interpretation is easier (Figure 4).

Page 4 of 30

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, NATURAL HAZARD SCIENCE (oxfordre.com/
naturalhazardscience). (c) Oxford University Press USA, 2019. All Rights Reserved. Personal use only; commercial use is
strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 18 February 2019


Remote Sensing of Floods

Given the very high spatial


resolution of aerial
photography, flood extent
is often derived from color
or panchromatic aerial
photography by simply
digitizing the edges at the
contrasting land-water
interface. A very simple
and straightforward
approach to retrieving
useful flood information
from space is that of
Figure 4. Landsat-5 captures Missouri River flooding extracting a binary map
near Omaha, Nebraska, on July 6, 2011. The flood consisting of dry and
waters show up as very dark blue and, where the
water is shallow, medium blue. In the image, the
flooded pixels, most often
interstate is cut off by flood waters, just south of in a fully- or semi-
Missouri Valley, Iowa, and about 20 miles north of automated way. This
Omaha. © USGS/NASA.
procedure is applied
throughout the world by
many research teams and engineering and consulting companies as well as emergency
response services and governmental institutions. Probably the most common image
processing algorithm applied to an optical image for classifying water versus dry land is
the Normalized Difference Water Index (NDWI), or versions thereof, which uses the
reflected near-infrared radiation and visible green light to enhance the presence of open
water surfaces while eliminating the presence of soil and terrestrial vegetation features
(McFeeters, 1996).

The potential that (optical) satellite images can contribute to flood science and
applications has been known for over 40 years. Several studies in the early 1970s
demonstrated the value of optical satellite imagery to map the evolution of flooding from
space and indicated strong application potential for such maps for a number of sectors
(e.g., Currey, 1977; Deutsch & Ruggles, 1978; Robinove, 1978).

Presently, the Moderate Resolution Imaging Spectrometer (MODIS) onboard NASA’s


Terra and Aqua satellites acquires images twice daily all around the globe. This offers a
unique capability to monitor flood events and assist disaster response where and when
possible. Despite the relatively low spatial resolution (250 m), persistent cloud cover and
vegetation, as well as densely built areas that limit successful flood detection, NASA
processes these images in near-real time (NRT) and provides flood maps to a number of
flood relief agencies (for further reference, see NASA’s NRT Global Flood Mapping and
the Dartmouth Flood Observatory).

Page 5 of 30

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, NATURAL HAZARD SCIENCE (oxfordre.com/
naturalhazardscience). (c) Oxford University Press USA, 2019. All Rights Reserved. Personal use only; commercial use is
strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 18 February 2019


Remote Sensing of Floods

Notwithstanding the success of optical imagery for flood mapping (see Marcus &
Fonstad, 2008 for a detailed scientific review), as noted earlier, the systematic application
of such imagery is hampered by persistent cloud cover during floods, particularly in small
to medium-sized basins where floods often recede before weather conditions improve, as
is oftentimes the case in Europe for instance. Also, as already mentioned, the inability to
map flooding in urban areas or beneath vegetation canopies, limits the applicability of
optical sensors (Figure 5).

Microwave Range
Given the limitations of
sensors operating in the
visible and infrared
spectrum to acquire flood
information routinely,
microwave (radar) remote
sensing is often considered
an attractive alternative or
Figure 5. Photographs depicting flood disasters in complementary technology
environments that pose considerable challenges to
satellite remote sensing. for flood detection and
monitoring. Microwaves
penetrate cloud cover, fog, and light rain, and in commonly employed radar frequencies
(X or C, see Figure 3), active radar signals from synthetic aperture radar (SAR) are
reflected away from the sensor by smooth open water bodies, so, consequently, flooded
versus dry land is typically of high contrast.

Active radars transmit a signal and receive the backscatter characteristics of many
different surface features, which may be difficult to distinguish accurately. However, if
knowledgeable in backscatter characterization, valuable information can be retrieved
from a SAR image that is only possible with very high resolution commercial optical
imagery or aerial photography (Figure 6), such as emerging flooded vegetation (e.g.,
Hess, Melack, Filoso, & Wang, 1995; Hess, Melack, Novo, Barbosa, & Gastil, 2003), flood
conditions within urban areas (e.g., Mason, Speck, Devereux, Schumann, Neal, & Bates,
2010), or even road and other infrastructure damage (e.g., Mouratidis & Sarti, 2013).

Page 6 of 30

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, NATURAL HAZARD SCIENCE (oxfordre.com/
naturalhazardscience). (c) Oxford University Press USA, 2019. All Rights Reserved. Personal use only; commercial use is
strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 18 February 2019


Remote Sensing of Floods

However, the use of


passive microwave
systems over land
surfaces, meaning that
surface-emitted
wavelengths are measured
by the satellite, is difficult
given the large angular
beams of such systems
Figure 6. The port of Sendai after the tsunami in (Rees, 2001), resulting in
early 2011. This TerraSAR-X image, acquired on spatial resolutions as large
March 12, 2011, shows the devastation of the port of
the Japanese city of Sendai. The magenta-colored
as 20 to100 km.
areas reveal the extent of damage in the form of Interpretation of the wide
boulders and debris deposits; the blue areas denote
range of materials with
flooding. © DLR (CC-BY 3.0).
many different emissivities
is thus rendered nearly
impossible. Nevertheless, as the sensor is sensitive to changes in the electric field, very
large areas of water, for instance, can be detected but their uncertainties may be large
(Schumann, 2009).

Active microwave imagery from SAR seems to be, at the moment, the only reliable source
of information for routinely monitoring floods on rivers much smaller than 1 km in width.
The spatial resolution from current and planned spaceborne SAR sensors (i.e., typically 3–
30 m) should satisfy requirements for most applications. Satellite data with a ground
resolution of 100 m or even coarser would still be of value for rapid response
requirements for floods on large rivers, but this needs to be further investigated. Indeed,
the need for rapid dissemination of information is probably of greater importance in the
first instance than the production of a high-resolution product (Blyth, 1997). For example,
Di Baldassarre, Schumann, and Bates(2009A) demonstrate that inundation width derived
from a 75-m resolution SAR image in wide swath mode (delivered 24 h after an event on
the Po River, Italy, in early June 2008) can be used in near-real time to verify timely flood
inundation modeling. Obviously, near-real time availability of higher-resolution SAR data
is preferred, but the cost of such data may quickly become non-trivial.

More recently, advances in radar technology have led to several high resolution SAR
missions (e.g., TerraSAR-X, Radarsat-2, Cosmo-SkyMed constellation, Sentinel-1A) that
can complement optical imagery by allowing more successful detection of flooding, as
illustrated in Figure 7, and also demonstrated by Martinis, Schumann, and Bates(2013)
within a fully automated processing chain. This complementarity can be advantageous in
a variety of environments where optical imagery has clear limitations, such as in urban
areas, coasts, wetlands, forests, and during adverse weather conditions where optical
sensors are often limited (Schumann & Moller, 2015).

Page 7 of 30

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, NATURAL HAZARD SCIENCE (oxfordre.com/
naturalhazardscience). (c) Oxford University Press USA, 2019. All Rights Reserved. Personal use only; commercial use is
strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 18 February 2019


Remote Sensing of Floods

Many SAR image-


processing techniques
exist to more or less
successfully derive flood
area or extent (for a
review, refer to Schumann,
Bates, Horritt, Matgen, &
Pappenberger, 2009), but
classification inaccuracies
of flooded areas (i.e., dry
areas mapped as flooded
and vice versa) vary
greatly. Typically, detection
of flooding is from a single
image using Otsu’s method
Figure 7. Map showing rivers in flood and maximum
(Otsu, 1979) or a version
flood extent during the May–June 2015 flood disaster thereof, which
in Texas, Oklahoma, and Arkansas. automatically performs
image thresholding. The
algorithm maximizes the inter-class variance between image pixels assuming that the
image contains only two classes of pixels following a bi-modal distribution, which is often
an acceptable assumption in the case of a SAR flood image.

Noteworthy is that, in a number of cases, multitemporal SAR images have been used
successfully to monitor the evolution of a flood event or to map inundation dynamics (e.g.,
Bates, Wilson, Horritt, Mason, Holden, & Currie, 2006; Pulvirenti, Chini, Pierdicca,
Guerriero, & Ferrazzoli, 2011; Schumann, Neal, Mason, & Bates, 2011). In such cases,
rapid mapping and dissemination is preferable, of course; yet in urban areas, as well as in
wetlands and forests, detection of flooding from a satellite SAR image still poses
considerable challenges (Schumann & Moller, 2015).

Also, in some cases, SAR images acquired over the same area but at different times have
even been used to derive spatially distributed water levels through a complex but
powerful technique known as interferometry, or InSAR (refer to e.g., Alsdorf,
Melack,Dunne, Mertes, Hess, & Smith, 2000 for details on this technique), which will be
employed on the upcoming NASA/CNES Surface Water Ocean Topography (SWOT)
mission to measure water levels and map water surfaces of the world’s lakes and main
rivers (Fjørtoft et al., 2014).

Despite the many challenges, SAR systems are attractive for monitoring, mapping, and
analyzing floods, and much progress in addressing those challenges is expected in the
coming years given the growing number of current and planned satellite missions
carrying SAR sensors.

Page 8 of 30

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, NATURAL HAZARD SCIENCE (oxfordre.com/
naturalhazardscience). (c) Oxford University Press USA, 2019. All Rights Reserved. Personal use only; commercial use is
strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 18 February 2019


Remote Sensing of Floods

Altimeters

Altimeters (most common are radar altimeters) transmit signals to Earth, and receive the
echo from the surface. The satellite orbit has to be accurately tracked, and its position is
determined relative to an arbitrary reference surface (an ellipsoid). Altimetry instruments
determine the distance from the satellite to a target surface by measuring the satellite-to-
surface round-trip time of the signal pulse, thus retrieving precise surface height
measurements (Figure 8). The magnitude and shape of the echoes (or waveforms) also
contain information about the characteristics of the surface. Surfaces that are large and
spatially homogeneous, such as the ocean, or indeed flooding, are well suited for
altimetry and lead to high measurement accuracy. However, surfaces that contain
discontinuities or significant slopes, such as some ice, rivers, or land surfaces, make
accurate interpretation more difficult.

Satellite tracks passing


over rivers and lakes are
considered virtual
measurement stations
(VS), usually
complementing the
declining network of
gauging stations on the
ground (Alsdorf,
Rodriguez, & Lettenmaier,
2007). Altimetry data
series may provide
essential information for
characterizing a
hydrological regime,
Figure 8. A simple illustration of the measurement
especially in remote and
principle of altimetry. data-sparse regions. Some
recent studies have even
combined altimetry data with traditional information for the estimation of river discharge
in ungauged basins (see e.g., Bjerklie, Dingman, Vorosmarty, Bolster, & Congalton, 2003;
Durand et al., 2016; Tourian, Sneeuw, & Bárdossy, 2013). However, limitations related to
typical ground-track spacing (up to a few hundred km), wide footprint resolution (from a
few hundred meters to several km), and low revisit time of most altimetry missions (10–
35 days) make the use of altimetry data on rivers with small to medium width or with a
dynamic flow regime very difficult (Schumann & Domeneghetti, 2016).

Nevertheless, satellite altimetry has proven extremely valuable, and innovative


approaches are being developed to overcome some of the fundamental limitations, such
as spatial and temporal low resolution (e.g., Tourian et al., 2016). Altimetry is particularly
useful for the monitoring of water resources from reservoirs and lakes in countries like
Africa (Figure 9), where information is difficult to access and obtain with traditional

Page 9 of 30

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, NATURAL HAZARD SCIENCE (oxfordre.com/
naturalhazardscience). (c) Oxford University Press USA, 2019. All Rights Reserved. Personal use only; commercial use is
strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 18 February 2019


Remote Sensing of Floods

measurement techniques on the ground because of the high cost in equipment,


manpower, and communications, and because it is difficult to obtain these precious
hydrological data from many countries.

Over the years, long time


series (in some cases >15
years) of both radar and
laser altimetry data have
become available and have
stimulated innovative
research among
hydrologists, in particular
for flood modeling and
forecasting. For example,
Baugh, Bates, Schumann,
and Trigg, (2013) have
demonstrated the use of
ICESat-1 (NASA’s laser
altimetry mission)
vegetation canopy height
Figure 9. Example of altimetry-derived rivers and
lake levels across Africa. © ESA. measurements (Simard,
Pinto, Fisher, & Baccini,
2011) to correct floodplain bare-earth topography from the Shuttle Radar Topography
Mission (SRTM) digital elevation model (DEM), thereby improving the ability to simulate
floodplain flow patterns.

Altimetry data can also be used to understand first- and second-order hydrodynamics of
rivers (Figure 10), such as water surface slopes and backwater effects, as demonstrated
by O’Loughlin, Trigg, Schumann, and Bates(2013) for the Congo River, or other large
rivers that are now mostly ungauged (see also Hall, Schumann, Bamber, Bates, & Trigg,
2012 for a pioneer study using ICESat-1 water levels on the Amazon River). Also using
ICESat-1 altimeter data, Schumann et al. (2013B) successfully calibrated in-channel
water levels of a large scale flood inundation forecasting model over the Lower Zambezi
River in SE Africa.

Page 10 of 30

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, NATURAL HAZARD SCIENCE (oxfordre.com/
naturalhazardscience). (c) Oxford University Press USA, 2019. All Rights Reserved. Personal use only; commercial use is
strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 18 February 2019


Remote Sensing of Floods

One of the most notable


research studies in this
context has been
undertaken by Tourian et
al. (2016), who developed
an approach to densify the
time series of altimetry
data from many satellite
missions in both space and
Figure 10. ICESat-1 altimetry observations within
time by statistically and
160 m of a floodplain channel greater than 120 m in
width (cyan circles) across the middle reach of the hydrodynamically
Amazon River. From Hall (2013). connecting all the time
series after removing
inter-satellite biases. This produces an accurate water level time series at any location
along a river (~50 cm error on average).

With upcoming satellite radar and laser altimetry missions (e.g., Sentinel-3 constellation,
SWOT, ICESat-2), there will be a proliferation of measurements of the world’s major
rivers and lakes, and the challenge will be in making sense of all that data and what it
actually means to have that much information available (see Schumann & Domeneghetti,
2016 for a discussion on this topic).

Integration With Computer Simulations and


Forecasting of Floods
Integrating remotely sensed flood information with flood process models, such as one-
dimensional and two-dimensional hydrodynamic (hydraulic) models, can be done for
model evaluation (i.e., model calibration or validation) or assimilation with models.

Simply put, hydrodynamic models simulate water flow volumes and depths within channel
networks (commonly in 1-D) and in the adjacent floodplain lands when channel bank
overtopping occurs and water spreads across low-lying topography (in 2-D). Of course,
such models are needed for predicting flood events as well as for event re-analysis.
Although traditionally applied to relatively small sections or reaches of rivers, recent
advances in computational model code and computing power have enabled flood
simulations over spatial and temporal scales much larger than in the past; in fact, such
models can now be run at continental-to-global scales (Dottori, Salamon, Bianchi, Alfieri,
Hirpa, & Feyen, 2016; Sampson, Smith, Bates, Neal, Alfieri, & Freer, 2015).

Given the inherent nature of large spatial coverage and of pixel information being two-
dimensional, using imagery of floods, in particular from satellites, is an inviting
alternative to the typically few and far apart ground-based measurements traditionally
used to evaluate the performance and prediction skills of flood models.

Page 11 of 30

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, NATURAL HAZARD SCIENCE (oxfordre.com/
naturalhazardscience). (c) Oxford University Press USA, 2019. All Rights Reserved. Personal use only; commercial use is
strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 18 February 2019


Remote Sensing of Floods

As described in the previous sections, remote sensing can provide information about both
flood area and water levels; however, information about flow velocities and actual depth
of water is difficult or in many cases impossible to obtain, especially where and when
such information would be most valuable, such as within urban areas during high-
magnitude events. This is largely due to fundamental limitations related to the physical
principles of signal interactions with surface water; most wavelengths are typically either
fully absorbed when reaching certain depths or reflected away from calm, open water
surfaces. Only very few successful studies have been conducted that demonstrate the
ability of remote sensing to infer information about water depth or to retrieve flow
velocity fields, not merely enough to reach application credibility or acceptable readiness
levels.

Model Validation and Calibration

Valuable information can be retrieved from flood area and water level, either directly
from altimetry and SAR interferometry, or indirectly, by merging flood shorelines with a
digital elevation model (e.g., Schumann et al., 2007). A variety of methods exist to
integrate such information with flood models. As noted earlier, the most common use of
flood area or extent data is for model calibration or validation (see, e.g., Aronica, Bates, &
Horritt, 2002, as one of the classic studies on this topic). Calibration is defined as
adjusting model parameters (such as surface roughness or boundary conditions) to
improve the fit between model predictions and observations. Validation, or verification,
involves comparing model output with observations and using this to deduce model
performance (Figure 11). The processes of model calibration and validation with remotely
sensed data involve some common steps: (a) extraction of flood area, shorelines, or water
level from the remote sensing data, and (b) comparison with model predictions, using
some form of performance metric or statistical measure.

Page 12 of 30

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, NATURAL HAZARD SCIENCE (oxfordre.com/
naturalhazardscience). (c) Oxford University Press USA, 2019. All Rights Reserved. Personal use only; commercial use is
strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 18 February 2019


Remote Sensing of Floods

When using flood area, a


map overlay operation
with some sort of spatial
comparison metric is
typically used. This is
applied to find the best
model simulation within a
set of simulations run with
different model parameter
values, or to evaluate the
accuracy of a model. Early
Figure 11. Map showing fit between a global 2-D studies of this type have
flood inundation model and the U.K. Environment focused on simple wet/dry
Agency benchmark data for the Severn catchment
flood maps either in a
(~11,000 km2). Green shading represents predicted
correct flooding, blue shading represents flooded deterministic or
area unique to the global model, and red shading probabilistic model
represents flooded area unique to the benchmark
data. From Sampson et al. (2015). evaluation context (see
Schumann et al., 2009;
Yan, Di Baldassarre, Solomatine, & Schumann, 2015, for a review of existing literature).
More recently, however, probabilistic approaches have been employed to process a flood
image (e.g., Giustarini, Chini, Hostache, Pappenberger, & Matgen, 2015), which has been
shown to increase information content considerably, thereby achieving better model
calibration (Di Baldassarre, Schumann, & Bates, 2009B). The potential of radar and laser
altimetry for model calibration has also been examined by a number of studies (e.g.,
Domeneghetti, Castellarin, Tarpanelli, & Moramarco, 2015; Neal, Schumann, & Bates,
2012; Schumann et al., 2013B;).

Assimilation With Models

Assimilation, which solves the optimization problem by combining model and observation
while accounting for errors in both, may appear conceptually simple (Figure 12);
however, successful application of the many different mathematical variants that exist is
rather difficult and requires an expert skillset (see Evensen, 2006 for a popular
introduction to assimilation).

Page 13 of 30

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, NATURAL HAZARD SCIENCE (oxfordre.com/
naturalhazardscience). (c) Oxford University Press USA, 2019. All Rights Reserved. Personal use only; commercial use is
strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 18 February 2019


Remote Sensing of Floods

Water level and flood area


have been used in data
assimilation studies,
although assimilation of
flood area information is
still being studied, and
only a very limited number
of examples exist in the
literature to date
Figure 12. Abstract schematic of the process of (Andreadis & Schumann,
assimilation.
2014; Lai, Liang, Yesou, &
Daillet, 2014). In general,
the problem with assimilating inundated area in hydrodynamic models is that of
localization; that is, the impact the information content has is highly localized and, as a
result, may wear off rapidly in space and time. In addition, the inherent two-
dimensionality of flood area complicates this problem, and advanced local smoothing
techniques are needed (see e.g., Andreadis & Schumann, 2014). Assimilating water
levels, obtained either directly with altimetry or by intersecting flood shorelines with
accurate floodplain topography, however, is relatively straightforward and more
commonly reported in the scientific literature.

When carefully applied, assimilation can be used not only to correct or update a model
forecast but also to pinpoint areas in the floodplain and channel where important model
parameters (e.g., friction values) are homogeneous, thereby inferring localized flood flow
processes (see Hostache, Lai, Monnier, & Puech, 2010). In a similar context, assimilation
can help determine the impact of satellite observations on the prediction of a number of
flood parameters such as flood levels, discharge, and inundation area, as demonstrated
by Andreadis and Schumann (2014).

With an increasing number of satellite missions dedicated to hydrology and the


monitoring of floods, information on how satellite revisit time impacts forecast
performance is crucial to optimize satellite operations (i.e., acquisition of observations) as
well as models, and here assimilation can also help (García-Pintado, Neal, Mason, Dance,
& Bates, 2013).

In the plethora of methodologies and numerical techniques exploiting the potential of


satellite observations of flood parameters, data assimilation appears very promising,
especially for flood forecasting (Schumann & Domeneghetti, 2016). Assimilation studies
have begun to investigate the value of integrating remote sensing of floods with computer
models, but more research in this field is needed to fully grasp the potential that may be
offered by the current and future proliferation of satellite observations of floods. This is
especially true in the context of real-time forecasting during complex scenarios, or
indeed, when assimilating multiple observations in a coupled modeling system that can
simulate processes from “clouds to inundation.” Here, non-linearities and complex

Page 14 of 30

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, NATURAL HAZARD SCIENCE (oxfordre.com/
naturalhazardscience). (c) Oxford University Press USA, 2019. All Rights Reserved. Personal use only; commercial use is
strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 18 February 2019


Remote Sensing of Floods

feedbacks between different processes exist and need to be more fully understood
(Matgen et al., 2010).

End-User Applications and Operational


Assistance
It is clear that recent and upcoming satellite missions are transforming current
procedures and operations in flood mapping, monitoring, and modeling, as well as our
understanding of river and floodplain hydrodynamics globally (Bates, Neal, Alsdorf, &
Schumann, 2014). Indeed there is an increasing trend in initiatives that utilize remote
sensing data to strengthen support in global flood mapping and monitoring, and in
responding to flood disasters (e.g. the International Charter, the Dartmouth Flood
Observatory, UMD’s Global Flood Monitoring System, CEOS, NASA’s Servir, and ESA’s
Tiger-Net initiatives). Figure 13 illustrates three very successful end-user oriented
applications focusing on flood detection and forecasting.

Figure 13. News feature articles of successful end-


user-oriented applications based on satellite
observations.

Global Initiatives

The International Charter


The International Charter provides a unified system of space data acquisition and delivery
to those affected by disasters. Space agency members commit resources to support the
provisions of the Charter, which can be activated by pre-defined authorized users. Due to
its ongoing success over the last decades, the Charter is soon widening access to satellite
imagery for disaster response to any national disaster management authority without the
need for membership.

Tiger-Net
Tiger-Net is a major component of the European Space Agency (ESA) in supporting the
African continent with Earth Observation (EO) capacity for water resource monitoring,
including flood mapping and monitoring through ESA’s satellites, in particular SAR. This
is done in close collaboration with African water authorities and experts. Main

Page 15 of 30

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, NATURAL HAZARD SCIENCE (oxfordre.com/
naturalhazardscience). (c) Oxford University Press USA, 2019. All Rights Reserved. Personal use only; commercial use is
strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 18 February 2019


Remote Sensing of Floods

achievements of Tiger-Net include the development of an open-source Water Observation


and Information System (WOIS) for monitoring, assessing, and inventorying water
resources, and training of African water authorities and technical centers to fully exploit
the increasing observation capacity offered by current and upcoming generations of
satellites.

Servir
Servir is a joint venture between NASA and the U.S. Agency for International
Development (USAID). Similar to ESA’s Tiger-Net, it provides satellite-based monitoring,
imaging, and mapping, as well as GIS capabilities and predictive models to help improve
environmental decision-making among developing nations. Among other applications,
Servir helps end users to rapidly respond to and assess damage from natural disasters
such as floods.

With regard to remote sensing of floods, one precursor success story of Servir is without
doubt the implementation of an operational flood forecasting system based on radar
altimetry in Bangladesh (Figure 13). The operational forecasting system based on radar
altimetry provides water level forecasts with acceptable accuracies. The system is now
operated by Bangladesh’s Flood Forecasting and Warning Center and serves the entirety
of the Ganges-Brahmaputra-Meghna river basin complex as well as more than thirty
flood-prone nations in the region, currently deprived of real-time flow data from upstream
nations (Hossain et al., 2014).

The CEOS Flood Pilot


The working group on disasters of the Committee on Earth Observation Satellites (CEOS)
aims to deliver satellite data seamlessly to end users during natural disasters. The
working group’s Flood Pilot has been leveraging and coordinating a number of ongoing
projects, particularly in three regions (the Caribbean, southern Africa, and Southeast
Asia). One of the Flood Pilot’s main objectives is to create a Global Flood Dashboard to
serve as a “one-stop shop” for information from a number of existing systems for
monitoring and predicting floods in real-time, as also argued by Schumann and
Domeneghetti (2016) in an Invited Commentary on the proliferation of satellite data. Pilot
regions are encouraged to develop a basic capacity to access data and include them in
their decision-making process.

Global Flood Mapping, Detection, and Forecasting

The Dartmouth Flood Observatory (DFO) and NASA’s NRT Flood Mapping
The Dartmouth Flood Observatory (DFO) conducts global remote sensing-based flood
mapping and measurements in near-real time (NRT) and archives this information. The
DFO also performs global hydrological modeling, which it integrates with its global
surface water mapping. Collaborating and partnering with a number of humanitarian and
flood disaster emergency management agencies, such as the United Nations World Food
Programme (UN WFP), ensures maximum utility of the information. The DFO is most
known for its rapid flood mapping with the MODIS instrument onboard NASA’s Aqua and
Page 16 of 30

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, NATURAL HAZARD SCIENCE (oxfordre.com/
naturalhazardscience). (c) Oxford University Press USA, 2019. All Rights Reserved. Personal use only; commercial use is
strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 18 February 2019


Remote Sensing of Floods

Terra satellites (Figure 13). The observatory offers two map series accessible from the
global index: “Current Flood Conditions,” providing daily, satellite-based updates of
surface water extent, and the “Global Atlas of Floodplains,” a remote sensing record of
floods, 1993 to 2015. These systems have been sustained by grants and contracts, among
others, from NASA, the European Commission, the World Bank, and the Latin American
Development Bank.

NASA’s NRT flood mapping is similar to the DFO, and in fact, since 2012, feeds its maps
seamlessly to the DFO. The LANCE processing system at NASA Goddard provides such
products typically within a few hours of satellite overpass. As with the DFO, open water is
detected using a ratio of MODIS bands in the visible and near-infrared at 250 m spatial
resolution. The impact of clouds is minimized by compositing images typically over two or
more days. Flooding is classified as anomaly to a reference water layer denoting “normal”
water extent.

Global Radar-Based Flood Mapping on ESA’s G-POD System

Also in a global context, ESA hosts a SAR-based mapping tool on their Grid-Processing on
Demand (G-POD) system, which is freely available to end users, who can query the ESA
SAR database for a flood image and retrieve an automatically generated flood map. The
mapping algorithm is based on a region-growing algorithm refined by change detection
and works on different SAR image modes and resolutions (Matgen, Hostache, Schumann,
Pfister,Hoffmann, & Savenije,2011).

Global Flood Detection System (GFDS)


The Global Disaster Alert and Coordination System’s (GDACS) experimental Global Flood
Detection System (GFDS) monitors floods worldwide using near-real time passive
microwave remote sensing from the Advanced Microwave Scanning Radiometer-Earth
Observing System (AMSR-E). Dynamics over a large inundated area can be observed
when surface water increases significantly and affects the emitted microwave signal,
which allows the system to flag the area as flooded. Time series are calculated in more
than 10,000 monitoring areas, along with flood maps and animations.

Global Flood Monitoring System (GFMS)


Real-time quasi-global hydrological calculations at 1/8th degree and 1 km resolution
inundation simulations are performed with the University of Maryland’s Global Flood
Monitoring System (GFMS), which is a NASA-funded experimental system that uses real-
time TRMM Multi-satellite Precipitation Analysis (TMPA) data and now the iMERG
product from the Global Precipitation Measurement (GPM) mission. This system issues
flood forecasts with 4–5 days lead time and mapping of inundation at a 3-hour time step.
Although at a relatively low resolution, the timeliness of the system and its forecast
capability make it very attractive to flood relief services and flood disaster response
organizations worldwide (Figure 13).

Page 17 of 30

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, NATURAL HAZARD SCIENCE (oxfordre.com/
naturalhazardscience). (c) Oxford University Press USA, 2019. All Rights Reserved. Personal use only; commercial use is
strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 18 February 2019


Remote Sensing of Floods

Primarily in the developing world, global initiatives and systems such as presented in this
article are now becoming established and recognized by many nations that are in need of
assistance because traditional ground-based monitoring systems are sparse and in
decline. Satellite data only realize their full value if those who can benefit from these data
know where and how to access the many products and services, and know how to use
them. Achieving this is still extremely challenging, however, especially in the area of
natural disasters such as floods, where the data are needed only infrequently and many
satellite acquisitions of floods are still inherently “opportunistic,” with the exception of
the International Charter.

Nevertheless, the value satellites can offer during floods is growing rapidly, and the
challenge now lies in ensuring sustainable and interoperable use as well as optimized
distribution of remote sensing products and services for science as well as operational
assistance (see discussions by e.g., Hossain et al., 2016 as well as Schumann &
Domeneghetti, 2016).

The Era of Data Proliferation and


Interoperability
For about 40 years, with a proliferation over the last two decades, remote sensing data,
primarily in the form of satellite and airborne imagery and altimetry, have been used to
monitor, map, and study floods (Figure 14). The sensors and data processing techniques
that exist to derive information about floods are numerous. During recent years, scientific
contributions in the field of remote sensing of floods have increased considerably, and
science has presented innovative research and methods for retrieving information
content from multi-scale coverage of disastrous flood events all over the world. Progress
has been transformative, and the information obtained from remote sensing of floods is
becoming mature enough not only to be integrated with computer simulations of flooding
to allow better prediction, but also to assist flood response agencies in their operations.

This said, numerous


challenges remain and
need to be resolved if the
current trend of data
proliferation continues,
and even more
importantly, if remote
Figure 14. Graphical illustration of past, current,
and future satellite missions relevant to the sensing of floods is to
monitoring of surface water, floods, and reservoirs. assist flood disaster
A: Imagery-sensor missions. B: Altimetry missions.
Modified from Schumann and Domeneghetti (2016).
response worldwide in a
timely manner, so that
relief teams and decision
makers can act quickly, and organizations all around the world can deal swiftly with

Page 18 of 30

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, NATURAL HAZARD SCIENCE (oxfordre.com/
naturalhazardscience). (c) Oxford University Press USA, 2019. All Rights Reserved. Personal use only; commercial use is
strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 18 February 2019


Remote Sensing of Floods

successive large events and deliver relevant geospatial data and flood imagery when and
where needed.

Most major challenges related to remote sensing of floods have been alluded to in the
previous sections. Generally speaking, the most pressing challenge is related to
widespread confusion about what data and products there are, where and when they can
be accessed, and what information they actually contain that is relevant to decision
making at hand. This is further complicated by the fact that this “confusion” may now be
as common among scientists and product developers as it is to the end-user communities.

Central to solving current challenges is the notion of data interoperability.


Interoperability of data or a system means that data and system interfaces are completely
understood, work with other products or systems, present or future, in either
implementation or access, without any restrictions (Figure 15). Interoperability is
certainly a welcomed concept in the era of data, products, and system multiplicity.
Achieving this, however, is easier said than done and could take many years or even
decades.

One of the most notable


examples that integrates
remote sensing of floods,
big data processing
technology, and
interoperability is the
Water Observations from
Space (WOfS) dataset
provided by Geoscience
Australia (Mueller et al.,
2016). WOfS is a web
service, available as an
online viewer and as a Web
Mapping Service (WMS),
displaying historical
Figure 15. Illustration of data interoperability in the
surface water observations
context of the development and dissemination of derived from satellite
remote sensing–based flood products. A new imagery for all of Australia
“community of practice” (CoP) needs to form in order
to engage the different communities involved. A data from 1987 to 2014 (Figure
interoperability platform connects traditionally 16).
disparate and separated data streams and
establishes a seamless connectivity across all parts
The purpose of this big
of the new CoP network. Interoperability increases
access to information and timeliness of distribution, dataset is to map from
thereby leveraging information in more valuable continent-wide satellite
ways, all while maintaining the integrity and
intellectual property of the data and products imagery (acquired by
exchanged. Landsat-5 and -7) where
water is frequently

Page 19 of 30

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, NATURAL HAZARD SCIENCE (oxfordre.com/
naturalhazardscience). (c) Oxford University Press USA, 2019. All Rights Reserved. Personal use only; commercial use is
strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 18 February 2019


Remote Sensing of Floods

present, where it is rarely observed, and where inundation of the surface (i.e., flooding)
has been occurring. In addition to the frequency of water detection, the service provides
a confidence level that a water observation in a given location is correct. Over- and
under-estimation of surface water can occur in the presence of clouds, shadow, steep
slopes, tall and dense vegetation, and urban areas, as well as in occasional snow cover.

Given the relatively long repeat cycle of the Landsat satellite series of 16 days, not all
historical floods may have been observed. Also, in places like dams, where changes in
water drainage and infrastructure take place over time, past water observations may no
longer be adequate to infer future probabilities of water observations.

The WOfS project began in 2011, and is now complete, but data will continue to be
updated every three months. For storing, organizing, and processing the huge data
volumes, the project uses a high performance compute structure known as the Australian
Geoscience Data Cube (AGDC), designed for large-scale parallel processing. For a
complete scientific account of WOfS, the reader should refer to Mueller et al. (2016).

Big and potentially


globally available and free
datasets from satellites
like the WOfS are expected
to change ways in which
we understand processes
such as floods at global
scales. It is clear that the
Figure 16. A snapshot of the online WOfS viewer. © wealth of remote sensing
Geoscience Australia.
information that can be
produced and
disseminated in general and also during a flood disaster is much under-utilized by end
users. As an example, the reader should refer to the commentary article on the Texas
flood disaster of 2015 by Schumann et al. (2016).

This underutilization is due to a number of reasons, most of which relate to the relative
novelty of these data: (a) limited time and personnel capacity to understand, process, and
handle new types of geospatial datasets; (b) limited near-real time data accessibility,
bandwidth, and sharing capacity; (c) incompatibility between user platforms and
geospatial data formats; (d) data availability may be simply unknown and/or data latency
may be inadequate; and (e) limited understanding by scientists and engineers about end-
user information product and timing needs.

To address this frequently encountered mismatch between data availability and end-user
needs, the scientific community should seek to collaborate closer with end users, as
advocated by Hossain et al. (2016). A step in the right direction would be to build a “one-
stop-shop” (i.e., data portal) dedicated to the remote sensing of floods and, more broadly

Page 20 of 30

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, NATURAL HAZARD SCIENCE (oxfordre.com/
naturalhazardscience). (c) Oxford University Press USA, 2019. All Rights Reserved. Personal use only; commercial use is
strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 18 February 2019


Remote Sensing of Floods

speaking, river-related variables (Schumann & Domeneghetti, 2016). In a first instance,


this could be data and products on water level and flood area.

On the one hand, scientists and product developers could collect and synthesize
knowledge on this platform as well as data from past, present, and for future EO missions
while sharing algorithms, accuracy assessments, and documentation. On the other hand,
decision makers could pull data and products from this portal and request tailored
information layers as needed for their operations. Other flood-related information that
could be made available alongside remote sensing data are output layers from models
such as those produced by the GFMS for flood wave lead times using satellite rainfall, by
Sampson et al. (2015) for flood hazard return period layers, or indeed as proposed by
Dottori et al. (2016) for the Global Flood Awareness System (GloFAS).

Figure 17 illustrates a great example of the appealing complementarity between flood


hazard model simulations and remote sensing information on flooding. Although this
exists at the moment only for the whole of Australia, recent advances in high performance
cloud computing, large-scale flood hazard modeling, and online high-end data analytics
platforms geared towards interoperability should enable such efforts to be extended to
the global scale.

Certainly, another
desirable add-on feature
would be to use model
forecasts, complemented
and verified by social
media, to pinpoint target
regions for satellite image
acquisition and delivery of
flood products. Product
formats should allow high
interoperability and thus
integrate seamlessly with
Figure 17. Number of times water was detected
end-user operation
between 1987 and 2014 by Landsat-5 and -7.
Frequently observed water (such as permanent lakes systems and platforms
and dams) is shown in purple and blue, down and, ideally, should also
through green to infrequently observed water (such
as floods) in yellow, and finally to very low
provide relevant, simple,
percentages in red. © Geoscience Australia. Overlay and timely information, at
map of ~17,000 km2 showing maximum inundation
low bandwidth, that can be
depth over a 40-year simulation period (Schumann,
Andreadis, & Castillo, 2013A) downscaled onto the accessed by anyone from
90 m SRTM-DEM on top of the 28-year Landsat anywhere at any time on
observations (predicted correct [flooding]: 89.6%;
area in error: 10.9%).
any device.

Such a clearing house


would not only organize and structure data availability better, thereby clarifying existing
confusion over data and products; it would also create innovation and improve added

Page 21 of 30

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, NATURAL HAZARD SCIENCE (oxfordre.com/
naturalhazardscience). (c) Oxford University Press USA, 2019. All Rights Reserved. Personal use only; commercial use is
strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 18 February 2019


Remote Sensing of Floods

value of the different data products. In addition, the platform could be used for algorithm
development and testing since EO data can be more easily exploited. Ultimately,
methodological frameworks and best practice guidelines can be defined and established.
At the same time, the end-user community should have the opportunity to leave feedback
on data and products, which should in turn be used to improve the different types of
information disseminated.

Despite the notable progress in recent years in remote sensing technology, data
processing, and interoperability, the lack of a clear framework to date for a systematic
selection and extraction of actionable information, as well as the lack of a coordinated
dissemination process of such information, seriously limits the use of remote sensing in
practical and operational applications, especially during flood disasters. Consequently,
much effort from the scientific community is needed to change the current situation and
make products not only more accessible but also more credible and thus more valuable to
the many potential end users worldwide.

Outlook
There is now a general consensus among space agencies to strengthen the support that
satellites can offer in relation to floods. Especially during flood disasters and for response
operations, there is a strong need for near-real time (NRT) acquisition and processing of
satellite data, so derived products can assist field organizations in a timely manner. This
is particularly true when other information is scarce or not easily accessible.

However, remotely sensed data often require calibration, pre- and post-processing, and
some form of validation, all with the common end goal to increase information content
and credibility. In addition, the many ancillary and auxiliary geospatial data that may be
available, such as field data collected or flood hazard forecasts, make timely
dissemination of information challenging. Consequently, of the many products that are
produced from remote sensing, most are either not in the desirable format, or end users
such as emergency managers may simply be unaware of their existence.

Despite the many challenges and non-trivial issues, remote sensing has clearly shifted
flood science and applications from a data-poor to a data-rich environment in the past 15
years. This shift is embodied in the upcoming NASA/CNES Surface Water Ocean
Topography (SWOT) mission, which is the first satellite mission dedicated to hydrology.
The satellite carries a Ka-band radar planned for continuous operation, and as such it will
be of significant advantage over most radar satellites currently operating. The proposed
revisit time would be 21 days, with more frequent sampling at mid-latitudes
(Biancamaria, Lettenmaier, &Pavelsky,2015). The mission, with an expected launch date
of 2021, would allow invaluable data gathering for hydrology at the global scale.
Information about water bodies (lakes and rivers), including floods, would be collected at
an average pixel spacing of 50 m at each satellite overpass.

Page 22 of 30

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, NATURAL HAZARD SCIENCE (oxfordre.com/
naturalhazardscience). (c) Oxford University Press USA, 2019. All Rights Reserved. Personal use only; commercial use is
strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 18 February 2019


Remote Sensing of Floods

SWOT as well as the many current and planned remote sensing platforms, onboard
aircrafts, and satellites, provide datasets with great potential for enhanced monitoring,
measuring, and mapping of floods, improving flood models through new data assimilation
techniques, and scaling of processes within models (Schumann & Moller, 2015). This
situation will lead to innovative ways in which new data may not only advance science
(e.g., by providing more accurate flood forecasts), but also better assist end users in their
decision making.

Having said that, both the science and end-user communities need to collaborate closely
to address the many challenges that lie ahead. Of particular importance is the
development of more computationally efficient and robust operational image mapping
algorithm for floods and flood damage assessment, which would ideally be independent of
image properties such as spatial resolution, spectral signature, or viewing angle, and can
be applied in a variety of environments. Also in this context, more research and applied
case studies are needed to ultimately demonstrate to decision makers that utilizing
satellite data should be an integral part of flood disaster management and relief
operations. Here, a “one-stop-shop,” dedicated to remote sensing of surface water
measurements including floods, as outlined earlier, is on the top of many people’s wish
list.

Last but not least, as noted by Schumann (2015) in an editorial on remote sensing of
floods, with a proliferation of free Earth Observation data now and in the near future,
there is a need not only to understand the limitations and errors of the data and methods
but also to develop more sophisticated data processing algorithms, as well as robust
frameworks for handling the many heterogeneous geospatial data sets and for effective
information management and transfer across networks. This becomes even more pressing
as communities will see more data and products coming from emerging technologies in
fast-growing sectors, such as micro- and nano-satellites, and unmanned aerial vehicles
(UAVs), which are already becoming a widespread reality.

Conclusion
In recent years, there has been a significant increase in the number and types of satellite
instruments that can be used to map floods and infer information about flooding. Of
course, the ability to monitor floods with sensors mounted on aircraft and satellites has
been known for decades. The availability of aerial photography and early launches of
satellites allowed investigation of the potential value to map and monitor floods, and to
support flood management applications. Over the years there has been much stimulating
research in this area, and significant progress has been achieved in fostering our
understanding of the ways in which remote sensing can support and advance flood
modeling, even flood forecasting, and assist flood disaster response operations.

This article reviewed the utility of remote sensing, mostly from satellites, to map and
monitor floods. Examples of applications in different landscape settings and at various
spatial and temporal scales have been illustrated. Many current and upcoming satellite
Page 23 of 30

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, NATURAL HAZARD SCIENCE (oxfordre.com/
naturalhazardscience). (c) Oxford University Press USA, 2019. All Rights Reserved. Personal use only; commercial use is
strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 18 February 2019


Remote Sensing of Floods

missions are collecting data that can inform directly or indirectly about water bodies and
flood inundation processes. This data proliferation has shifted the research and
application fields in the area of remote sensing of (flood) hydrology from a data-poor
(pre-2000) to a data-rich (post-2000) environment. Consequently, innovative methods and
products from these data have been developed over the years, which have led not only to
better understanding of flood processes at various spatial and temporal scales and better
flood forecasts, but also to global initiatives and applications that utilize and promote
remote sensing for improved decision-making activities, particularly in developing
nations.

Global-scale initiatives and end-user oriented applications are now becoming established
and also recognized by many nations that are in need of assistance because traditional
ground-based monitoring systems are sparse and in decline. The value remote sensing
can offer is growing rapidly, and many challenges lie ahead. New sensor technologies, for
instance light-weight, small satellites and drones, now add many terabytes of new data
every day, and as a result, innovative and powerful online data analytics platforms are
being offered to retrieve actionable information from these data.

For remote sensing of floods, the grand challenge now lies in ensuring sustainable and
interoperable use as well as optimized distribution of remote sensing products and
services for science and end-user applications, as well as for operational flood disaster
assistance.

Suggested Readings
Alsdorf, D. E., Rodriguez, E., & Lettenmaier, D. P. (2007). Measuring surface water
from space. Reviews of Geophysics, 45(2).

Bates, P. D., Horritt, M. S., Smith, C. N., & Mason, D. (1997). Integrating remote sensing
observations of flood hydrology and hydraulic modelling. Hydrological Processes, 11,
1777–1795.

Bates, P. D., Neal, J. C., Alsdorf, D., & Schumann, G. J.-P. (2014). Observing global surface
water flood dynamics. Surveys in Geophysics, 35(3), 839–852.

Carbonneau, P. E., & Piégay, H. (2012). Fluvial remote sensing for science and
management. Chichester, U.K.: Wiley-Blackwell.

Di Baldassarre, G., & Schumann, G. (2011). Recent advances in mapping and modelling
flood processes in lowland areas. Physics and Chemistry of the Earth, Parts A/B/C, 36(7–
8), 221–334.

Hossain, F. (Ed.). (2016). Earth science satellite applications: Current and future
prospects. Cham, Switzerland: Springer International Publishing.

Page 24 of 30

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, NATURAL HAZARD SCIENCE (oxfordre.com/
naturalhazardscience). (c) Oxford University Press USA, 2019. All Rights Reserved. Personal use only; commercial use is
strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 18 February 2019


Remote Sensing of Floods

Hossain, F., Serrat-Capdevila, A., Granger, S., Thomas, A., Saah, D., Ganz, D., et al.
(2016). A global capacity building vision for societal applications of Earth observing
systems and data: Key questions and recommendations. Bulletin of the American
Meteorological Society, 97(7), 1295–1299.

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, 4–24.

Paron, P., Di Baldassarre, G., & Shroder, J. F., (2015). Hydro-meteorological hazards, risks
and disasters. Boston: Elsevier.

Schumann, G. J.-P. (2015). Remote sensing in flood monitoring and management


[Special issue]. Remote Sensing, 7(12).

Schumann, G. J.-P., Bates, P. D., Horritt, M. S., Matgen, P., & Pappenberger, F. (2009).
Progress in integration of remote sensing derived flood extent and stage data
and hydraulic models. Reviews of Geophysics, 47.

Schumann, G. J.-P., Frye, S., Wells, G., Adler, R., Brakenridge, R., Bolten, J., et al. (2016).
Unlocking the full potential of Earth Observation during the 2015 Texas flood disaster.
Water Resources Research, 52(5), 3288–3293.

Schumann, G. J.-P., & Moller, D. K. (2015). Microwave remote sensing of flood inundation.
Physics and Chemistry of the Earth, Parts A/B/C, 83–84, 84–95.

Smith, L. C. (1997). Satellite remote sensing of river inundation area, stage, and
discharge: A review. Hydrological Processes, 11(10), 1427–1439.

Yan, K., Di Baldassarre, G., Solomatine, D. P., & Schumann, G. J.-P. (2015). A review of
low-cost space-borne data for flood modelling: Topography, flood extent, and water level.
Hydrological Processes, 29(15), 3368–3387.

References
Alsdorf, D. E., Melack, J. M., Dunne, T., Mertes, L. A. K., Hess, L. L., & Smith, L. C. (2000).
Interferometric radar measurements of water level changes on the Amazon flood plain.
Nature, 404, 174–177.

Alsdorf, D. E., Rodriguez, E., & Lettenmaier, D. P. (2007). Measuring surface water
from space. Reviews of Geophysics, 45.

Andreadis, K. M., & Schumann, G. J.-P. (2014). Estimating the impact of satellite
observations on the predictability of large-scale hydraulic models. Advances in Water
Resources, 73, 44–54.

Aronica, G., Bates, P. D., & Horritt, M. S. (2002). Assessing the uncertainty in distributed
model predictions using observed binary pattern information within GLUE. Hydrological
Processes, 16, 2001–2016.

Page 25 of 30

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, NATURAL HAZARD SCIENCE (oxfordre.com/
naturalhazardscience). (c) Oxford University Press USA, 2019. All Rights Reserved. Personal use only; commercial use is
strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 18 February 2019


Remote Sensing of Floods

Bates, P. D., Neal, J. C., Alsdorf, D., & Schumann, G. J.-P. (2014). Observing global surface
water flood dynamics. Surveys in Geophysics, 35(3), 839–852.

Bates, P. D., Wilson, M. D., Horritt, M. S., Mason, D., Holden, N., & Currie, A. (2006).
Reach scale foodplain inundation dynamics observed using airborne synthetic aperture
radar imagery: Data analysis and modelling. Journal of Hydrology, 328, 306–318.

Baugh, C. A., Bates, P. D., Schumann, G., & Trigg, M. A. (2013). SRTM vegetation removal
and hydrodynamic modeling accuracy. Water Resources Research, 49(9), 5276–5289.

Biancamaria, S., Lettenmaier, D. P., & Pavelsky, T. M. (2015). The SWOT mission and its
capabilities for land hydrology, Surveys in Geophysics, 37(2), 307–337.

Bjerklie, D. M., Dingman, S. L., Vorosmarty, C. J., Bolster, C. H., & Congalton, R. G. (2003).
Evaluating the potential for measuring river discharge from space. Journal of Hydrology,
278, 17–38.

Blyth, K. (1997). Floodnet: A telenetwork for acquisition, processing, and dissemination of


Earth observation data for monitoring and emergency management of floods.
Hydrological Processes, 11, 1359–1375.

Currey, D. T. (1977). Identifying flood water movement. Remote Sensing of Environment,


6(1), 51–61.

Deutsch, M., & Ruggles, F. H. (1978). Hydrological applications of Landsat imagery used
in study of 1973 Indus river flood, Pakistan. Water Resources Bulletin, 14(2), 261–274.

Di Baldassarre, G., Schumann, G., & Bates, P. (2009a). Near real time satellite imagery to
support and verify timely flood modelling. Hydrological Processes, 23, 799–803.

Di Baldassarre, G., Schumann, G., & Bates, P. D. (2009b). A technique for the calibration
of hydraulic models using uncertain satellite observations of flood extent. Journal of
Hydrology, 367, 276–282.

Domeneghetti, A., Castellarin, A., Tarpanelli, A., & Moramarco, T. (2015). Investigating
the uncertainty of satellite altimetry products for hydrodynamic modelling. Hydrological
Processes, 29, 4908–4918.

Dottori, F., Salamon, P., Bianchi, A., Alfieri, L., Hirpa, F. A., & Feyen, L. (2016).
Development and evaluation of a framework for global flood hazard mapping. Advances in
Water Resources, 94, 87–102.

Durand, M., Gleason, C. J., Garambois, P. A., Bjerklie, D., Smith, L. C., Roux, H., et al.
(2016). An intercomparison of remote sensing river discharge estimation algorithms from
measurements of river height, width, and slope. Water Resources Research, 52(6), 4527–
4549.

Page 26 of 30

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, NATURAL HAZARD SCIENCE (oxfordre.com/
naturalhazardscience). (c) Oxford University Press USA, 2019. All Rights Reserved. Personal use only; commercial use is
strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 18 February 2019


Remote Sensing of Floods

Evensen, G. (2006). Data assimilation: The Ensemble Kalman Filter. Secaucus, NJ:
Springer-Verlag.

Fjortoft, R., Gaudin, J.-M., Pourthie, N., Lalaurie, J.-C., Mallet, A., Nouvel, J.-F., et al.
(2014). KaRIn on SWOT: Characteristics of near-nadir Ka-band interferometric SAR
imagery. IEEE Transactions on Geoscience and Remote Sensing, 52(4), 2172–2185.

Garcia-Pintado, J., Neal, J. C., Mason, D. C., Dance, S. L., & Bates, P. D. (2013). Scheduling
satellite-based SAR acquisition for sequential assimilation of water level observations into
flood modelling. Journal of Hydrology, 495, 252–266.

Giustarini, L., Chini, M., Hostache, R., Pappenberger, F., & Matgen, P. (2015). Flood
hazard mapping combining hydrodynamic modeling and multiannual remote sensing
data. Remote Sensing, 7(10), 14200–14226.

Hall, A. C. (2013). Observing water level dynamics in the Amazon using satellite
altimetry. PhD diss. University of Bristol, Bristol, UK.

Hall, A. C., Schumann, G. J.-P., Bamber, J. L., Bates, P. D., & Trigg, M. A. (2012). Geodetic
corrections to Amazon River water level gauges using ICESat altimetry. Water
Resources Research, 48(6).

Hallegatte, S., Green, C., Nicholls, R. J., & Corfee-Morlot, J. (2013). Future flood losses in
major coastal cities. Nature Climate Change, 3, 802–806.

Hess, L. L., Melack, J. M., Filoso, S., & Wang, Y. (1995). Delineation of inundated area and
vegetation along the Amazon floodplain with the SIR-C Synthetic Aperture Radar. IEEE
Transactions on Geoscience and Remote Sensing, 33(4), 896–904.

Hess, L. L., Melack, J. M., Novo, E. M. L. M., Barbosa, C. C. F., & Gastil, M. (2003). Dual-
season mapping of wetland inundation and vegetation for the central Amazon basin.
Remote Sensing of Environment, 87, 404–428.

Hossain, F., Maswood, M., Siddique-E-Akbor, A. H., Yigzaw, W., Mazumdar, L. C., Ahmed,
T., et al. (2014). A promising radar altimetry satellite system for operational flood
forecasting in flood-prone Bangladesh. IEEE Geoscience and Remote Sensing Magazine,
2(3), 27–36.

Hossain, F., Serrat-Capdevila, A., Granger, S., Thomas, A., Saah, D., Ganz, D., et al.
(2016). A global capacity building vision for societal applications of Earth observing
systems and data: Key questions and recommendations. Bulletin of the American
Meteorological Society, 97(7), 1295–1299.

Hostache, R., Lai, X., Monnier, J., & Puech, C. (2010). Assimilation of spatially distributed
water levels into a shallow-water flood model. Part II: Use of a remote sensing image of
Mosel River. Journal of Hydrology, 390, 257–268.

Page 27 of 30

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, NATURAL HAZARD SCIENCE (oxfordre.com/
naturalhazardscience). (c) Oxford University Press USA, 2019. All Rights Reserved. Personal use only; commercial use is
strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 18 February 2019


Remote Sensing of Floods

Lai, X., Liang, Q., Yesou, H., & Daillet, S. (2014). Variational assimilation of remotely
sensed flood extents using a 2-D flood model. Hydrology and Earth System Sciences,
18(11), 4325–4339.

Lubin, D., & Massom, R. (2007). Remote sensing of Earth’s polar regions: Opportunities
for computational science. Computing in Science & Engineering, 9(1), 58–71.

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, 4–24.

Martinis, S., Twele, A., Strobl, C., Kersten, J., & Stein, E. (2013). A multi-scale flood
monitoring system based on fully automatic MODIS and TerraSAR-X processing chains.
Remote Sensing, 5, 5598–5619.

Mason, D. C., Speck, R., Devereux, B., Schumann, G. J.-P., Neal, J. C., & Bates, P. D.
(2010). Flood detection in urban areas using TerraSAR-X. IEEE Transactions on
Geoscience and Remote Sensing, 48, 882–894.

Matgen, P., Hostache, R., Schumann, G., Pfister, L., Hoffmann, L., & Savenije, H. (2011).
Towards an automated SAR-based flood monitoring system: Lessons learned from two
case studies. Physics and Chemistry of the Earth, 36, 241–252.

Matgen, P., Montanari, M., Hostache, R., Pfister, L., Hoffmann, L., Plaza, D., et al. (2010).
Towards the sequential assimilation of SAR-derived water stages into hydraulic models
using the particle filter: Proof of concept. Hydrology and Earth System Sciences, 14(9),
1773–1785.

McFeeters, S. K. (1996). The use of the normalized difference water index (NDWI) in the
delineation of open water features. International Journal of Remote Sensing, 17, 1425–
1432.

Mouratidis, A., & Sarti, F. (2013). Flash-flood monitoring and damage assessment with
SAR data: Issues and future challenges for Earth observation from space sustained by
case studies from the Balkans and Eastern Europe. In J. M. Krisp (Ed.), Earth observation
of global changes (pp. 125–136). Berlin: Springer Berlin.

Mueller, N., Lewis, A., Roberts, D., Ring, S., Melrose, R., Sixsmith, J., et al. (2016). Water
observations from space: Mapping surface water from 25 years of Landsat imagery
across Australia. Remote Sensing of Environment, 174, 341–352.

Neal, J. C., Schumann, G. J.-P., & Bates, P. D. (2012). A sub-grid channel model for
simulating river hydraulics and floodplain inundation over large and data sparse
areas. Water Resources Research, 48.

O’ Loughlin, F., Trigg, M., Schumann, G., & Bates, P. D. (2013). Hydraulic characterization
of the middle reach of the Congo River. Water Resources Research, 49(8), 5059–5070.

Page 28 of 30

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, NATURAL HAZARD SCIENCE (oxfordre.com/
naturalhazardscience). (c) Oxford University Press USA, 2019. All Rights Reserved. Personal use only; commercial use is
strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 18 February 2019


Remote Sensing of Floods

Otsu, N. (1979). A threshold selection method from gray-level histograms. IEEE


Transactions on Systems, Man, and Cybernetics, 9(1), 62–66.

Pulvirenti, L., Chini, M., Pierdicca, N., Guerriero, L., & Ferrazzoli, P. (2011). Flood
monitoring using multitemporal COSMO-SkyMed data: Image segmentation and
signature interpretation. Remote Sensing of Environment, 115, 990–1002.

Rees, W. G. (2001). Physical principles of remote sensing. Cambridge, U.K.: Cambridge


University Press.

Robinove, C. J. (1978). Interpretation of a Landsat image of an unusual flood phenomenon


in Australia. Remote Sensing of Environment, 7(3), 219–225.

Sampson, C. C., Smith, A. M., Bates, P. B., Neal, J. C., Alfieri, L., & Freer, J. E. (2015). A
high-resolution global flood hazard model. Water Resources Research, 51(9), 7358–7381.

Schumann, G. (2009). Spatially distributed water stages from remotely sensed imagery:
Prospects for improved flood inundation modelling. Saarbrücken, Germany: VDM.

Schumann, G., Andreadis, K., & Castillo, C. J. (2013a). A satellite and model based
flood inundation climatology of Australia. American Geophysical Union Fall Meeting
Abstracts, #H51P-05.

Schumann, G., Matgen, P., Pappenberger, F., Hostache, R., Puech, C., Hoffmann, L., et al.
(2007). High-resolution 3D flood information from radar for effective flood hazard
management. IEEE Transactions on Geoscience and Remote Sensing, 45, 1715–1725.

Schumann, G. J.-P. (2015). Preface: Remote sensing in flood monitoring and management.
Remote Sensing, 7(12), 17013–17015.

Schumann, G. J.-P., Bates, P. D., Horritt, M. S., Matgen, P., & Pappenberger, F. (2009).
Progress in integration of remote sensing derived flood extent and stage data
and hydraulic models. Reviews of Geophysics, 47.

Schumann, G. J.-P., & Domeneghetti, A. (2016). Exploiting the proliferation of current and
future satellite observations of rivers. Hydrological Processes, 30(16), 2891–2896.

Schumann, G. J.-P., Frye, S., Wells, G., Adler, R., Brakenridge, R., Bolten, J., et al. (2016).
Unlocking the full potential of Earth Observation during the 2015 Texas flood disaster.
Water Resources Research, 52(5), 3288–3293.

Schumann, G. J.-P., & Moller, D. K. (2015). Microwave remote sensing of flood inundation.
Physics and Chemistry of the Earth, Parts A/B/C, 83–84, 84–95.

Schumann, G. J.-P., Neal, J. C., Mason, D. C., & Bates, P. D. (2011). The accuracy of
sequential aerial photography and SAR data for observing urban flood dynamics: A case
study of the UK summer 2007 floods. Remote Sensing of Environment, 115, 2536–2546.

Page 29 of 30

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, NATURAL HAZARD SCIENCE (oxfordre.com/
naturalhazardscience). (c) Oxford University Press USA, 2019. All Rights Reserved. Personal use only; commercial use is
strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 18 February 2019


Remote Sensing of Floods

Schumann, G. J.-P., Neal, J. C., Voisin, N., Andreadis, K. M., Pappenberger, F.,
Phanthuwongpakdee, N., et al. (2013b). A first large scale flood inundation forecasting
model. Water Resources Research, 49, 6248–6257.

Simard, M., Pinto, N., Fisher, J., & Baccini, A. (2011). Mapping forest canopy height
globally with spaceborne LiDAR. Journal of Geophysical Research, 116(G4).

Tourian, M. J., Sneeuw, N., & Bárdossy, A. (2013). A quantile function approach to
discharge estimation from satellite altimetry (ENVISAT). Water Resources Research,
49(7), 4174–4186.

Tourian, M. J., Tarpanelli, A., Elmi, O., Qin, T., Brocca, L., Moramarco, T., et al. (2016).
Spatiotemporal densification of river water level time series by multimission satellite
altimetry. Water Resources Research, 52(2), 1140–1159.

Yan, K., Di Baldassarre, G., Solomatine, D. P., & Schumann, G. J.-P. (2015). A review of
low-cost space-borne data for flood modelling: Topography, flood extent, and water level.
Hydrological Processes, 29, 3368–3387.

Guy J.-P. Schumann

School of Geographical Sciences, University of Bristol, Bristol, UK; Remote Sensing


Solutions Inc., Monrovia, CA, USA

Page 30 of 30

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, NATURAL HAZARD SCIENCE (oxfordre.com/
naturalhazardscience). (c) Oxford University Press USA, 2019. All Rights Reserved. Personal use only; commercial use is
strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 18 February 2019

You might also like