Download as pdf or txt
Download as pdf or txt
You are on page 1of 18

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/301353707

Carbon Sinks and Tropical Forest Biomass Estimation: A Review on Role of


Remote Sensing In Aboveground-Biomass Modelling

Article in Geocarto International · April 2016


DOI: 10.1080/10106049.2016.1178814

CITATIONS READS

69 2,452

2 authors:

Nurul Ain Mohd Zaki Zulkiflee Abd Latif


Universiti Teknologi MARA Universiti Teknologi MARA
38 PUBLICATIONS 235 CITATIONS 118 PUBLICATIONS 1,329 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Nurul Ain Mohd Zaki on 21 December 2017.

The user has requested enhancement of the downloaded file.


Geocarto International

ISSN: 1010-6049 (Print) 1752-0762 (Online) Journal homepage: http://www.tandfonline.com/loi/tgei20

Carbon sinks and tropical forest biomass


estimation: a review on role of remote sensing in
aboveground-biomass modelling

Nurul Ain Mohd Zaki & Zulkiflee Abd Latif

To cite this article: Nurul Ain Mohd Zaki & Zulkiflee Abd Latif (2017) Carbon sinks and tropical
forest biomass estimation: a review on role of remote sensing in aboveground-biomass modelling,
Geocarto International, 32:7, 701-716, DOI: 10.1080/10106049.2016.1178814

To link to this article: http://dx.doi.org/10.1080/10106049.2016.1178814

Accepted author version posted online: 17


Apr 2016.
Published online: 10 May 2016.

Submit your article to this journal

Article views: 133

View related articles

View Crossmark data

Citing articles: 1 View citing articles

Full Terms & Conditions of access and use can be found at


http://www.tandfonline.com/action/journalInformation?journalCode=tgei20

Download by: [Universiti Teknologi Mara] Date: 18 May 2017, At: 04:52
Geocarto International, 2017
VOL. 32, NO. 7, 701–716
https://doi.org/10.1080/10106049.2016.1178814

REVIEWS

Carbon sinks and tropical forest biomass estimation: a review on


role of remote sensing in aboveground-biomass modelling
Nurul Ain Mohd Zaki and Zulkiflee Abd Latif
Applied Remote Sensing & Geospatial Research Group, Centre of Studies for Surveying Science & Geomatics, Faculty
of Architecture, Planning and Surveying, Universiti Teknologi MARA, Shah Alam, Malaysia

ABSTRACT ARTICLE HISTORY


Tropical forest embraces a large stock of carbon and contributes to the Received 12 November 2015
enormous amount of above- and below-ground biomass and the global Accepted 23 March 2016
carbon cycle. The carbon kept in the above-ground living biomass of trees
KEYWORDS
is typically the largest pool and the most directly impacted by deforestation Tropical forest; carbon pools;
and degradation. Hence, quantifying carbon stock and fluxes from tropical above-ground biomass;
forests by estimating the above-ground forest biomass is the critical step that remote sensing
will be investigated further in this paper. Remote sensing technology can
provide many advantages in quantifying and mapping forest structure and
monitoring and mapping above-ground biomass, and is both temporally and
spatially accurate. Therefore, a good data-set of biomass which comprises
canopy height and canopy structure can provide carbon sequestration
potential for forest reserves. This paper reviews a thorough research of
biomass estimation using remote sensing and geospatial technologies.

1. Introduction
The forest plays an important role for living things, as it provides habitats for the biodiversity and
livelihood for human beings. It provides a wide array of essential goods such as wood, timber, coal
and remedies, but the most significant is preserving the global carbon cycle by absorbing anthro-
pogenic emissions of carbon dioxide (CO2) into the atmosphere (Latif & Blackburn 2010, 2012).
Despite abundant species of trees, plants, animals, birds and insects, the forest also holds species of
living things that are rarely found in other places. Nowadays, forests are threatened by anthropogenic
factors due to the rapid increase of the population, which leads to various environmental problems
such as landslides, avalanches, increase in global temperature, water erosion and many more hazards
(FAO 2015). Therefore, the importance of forests as the main contributors that affect global climate
has been increasingly recognized due to its role within the carbon cycle.
A large-scale study of the twenty-first-century forest change by Hansen et al. (2013) demonstrated
that there has been an increase in forest loss in most tropical areas over the last decade, with south-east
Asia displaying one of the highest rates of deforestation. According to the Global Forest Resources
Assessment 2015, deforestation continues at an alarmingly high rate with 129 million lost hectares of
forest between 1990 and 2015, the largest loss occurring in tropical forests (FAO 2015). Referring to the

CONTACT Zulkiflee Abd Latif zulki721@salam.uitm.edu.my


© 2016 Informa UK Limited, trading as Taylor & Francis Group
702 N. A. MOHD ZAKI AND Z. ABD LATIF

annual forest change from 2010 to 2015, it can be seen that there is a positive change in forest reduction
rates, despite the impact of increasing human populations, this being 8.8 million hectares per year, a
gain of 2.2 million hectares per year (FAO 2015). In South-east Asia generally, due to economic factors,
a lot of primary forest has historically been cleared for timber extraction and extensive conversion to
oil palm and rubber plantations. The recent forest cover research by Stibig et al. (2014) contends that
the main cause of forest loss in south-east Asia is land conversion from forest to cash crop plantation.
Increasing the rate of deforestation and conversion of the forest directly impacts ecosystem changes
and biodiversity function in relation to habitat provision, on both the local and the regional scales.
The 2007 Bali Climate Change Conference, under the auspices of the United Nations, adopted the
Kyoto protocol by setting binding obligations to reduce the emission of greenhouse gases (UNFCC
2015). At the conference, an important agreement was reached for developing countries to initiate
action to reduce emissions from the practices of deforestation and forest degradation (REDD). A
collective global target was set to reduce GHG emissions by about 5% of 1990 levels within the first
commitment period from 2008 to 2012. An amendment made to the protocol in Doha, Qatar, on 8
December 2012, known as the Doha Amendment to the Kyoto Protocol, aims to reduce GHG emissions
to at least 18% below 1990 levels from 2013 to 2020 during the second commitment period (UNFCC
2015). As a continuation of the persistent efforts to combat global warming and restrain the associ-
ated global risks, it is increasingly important to develop a mechanism for monitoring, reporting and
verifying (MRV) reducing emissions from deforestation and forest degradation (REDD+).
A destructive sampling method has been used in many forests in order to establish the allometric
equation to quantify the above-ground biomass (AGB) of specific tree species according to the type of
forest. For example, Ketterings et al. (2001) established an allometric equation for a mixed secondary
forest in Sumatra, Indonesia; Alvarez et al. (2012) explored the allometric equation developed for
the Colombian forest; Basuki et al. (2009) estimated the allometric equation to estimate the AGB
for Dipterocarp, Hopea, Palaqium and Shorea tree species in the Kalimantan forest, while Chave
et al. (2005, 2014) improved pantropical multi-species allometric equations for 58 countries all over
the world. Research on AGB has been enhanced by fusing a remote sensing technique with the field
measurement for better estimation of AGB.
Many previous researches have investigated potential methods to quantify AGB by including remote
sensing techniques. Remote sensing technology can provide many advantages to quantify and map
the forest structure and monitor and map the above-ground biomass; it is a promising tool that can
provide data that are both temporally and spatially accurate, and analyse them quantitatively. Various
remote sensing approaches for above-ground biomass estimation vary from low resolution optical
remote sensing imagery (e.g. Landsat, ASTER and MODIS) to very high spatial resolution imagery (e.g.
Quickbird, IKONOS, WorldView and GeoEye). The synergy of LiDAR (light detection and ranging)
and RADAR (radio detection and ranging) can improve the estimation of biomass and forest struc-
ture by providing a vertical reference for the canopy height model (CHM) of the tree. The objectives
of this paper are: (1) to review and assess the methodology of quantifying AGB from every aspect,
including field plot inventory, allometric equation and remote sensing technology; (2) to summarize
the recent advancements in optical and RS technologies for quantifying AGB through peer-reviewed
studies; and (3) to scrutinize the capabilities of airborne LiDAR to measure AGB for tropical forests
based on previous studies.

2. Above-ground biomass for tropical forest countries


Research on above-ground biomass for tropical forests makes wide use of destructive sampling and
conversion of volume into allometric equation (Ketterings et al. 2001; Alvarez et al. 2012; Fayolle et al.
2013; Hunter et al. 2013; Rutishauser et al. 2013; Aryal et al. 2014; Shimamoto et al. 2014); integration
of remote sensing and LiDAR technology (Asner et al. 2010; Gonzalez et al. 2010; Kronseder et al.,
2012; Baccini & Asner 2013; Rana et al. 2014) and the national forest inventory. Table 1 summarizes
the research that had been conducted all over the world across the tropic.
Geocarto International  703

Table 1. Research on above-ground biomass in the tropical forest.

Researcher/Criteria Method used Remarks Country


1 (Fayolle et al. 2013) Tree allomet- Destructive sampling and Single species model not Central Africa
ric in Central Africa: test the photogrammetry used measuring better than mul-
validity of pantropical mul- ti-species
ti-species allometric equations
for estimating carbon stocks
2 (Alvarez et al. 2012) Tree AGB Tree harvest and develop new Forest type and environment Colombia
allometries for carbon stocks equation factor are important to deter-
estimation in the natural mine AGB
forests of Colombia
3 (Rutishauser et al. 2013) Generic Destructive sampling and Comparing regional and generic Indonesia
allometric models including allometric equation allometric models to estimate
height, best estimate, forest biomass at tree level and plot
biomass and carbon stocks in level
Indonesia
4 (Basuki et al. 2009) Generic Destructive sampling and Generate site-specific equation Indonesia
allometric models including develop new equation for biomass estimation
height, best estimate, forest
biomass and carbon stocks
5 (Baccini & Asner 2013) Improving Random forest had been used To fine-tune a pantropical Colombia and
pantropical forest carbon maps (data mining algorithm) carbon map to specific Peru
with airborne LiDAR sampling sub-national regions
6 (Asner & Mascaro 2014) Mapping Using Lorey’s height Using discrete return data LIDAR Colombia, Ha-
tropical forest carbon waii, Mad-
agascar,
Panama,
Peru
7 (Asner et al. 2010) High-resolu- Using different scales and Regional mapping using mod- Amazon
tion forest carbon stocks and multi-temporal method to erate resolution, satellite data
emissions in the Amazon analyse AGB stock and LiDAR
8 (Ioki et al. 2014) Estimating AGB Using regression method from Used stepwise analysis Malaysia
of tropical rainforest of differ- LiDAR against field survey.
ent degradation levels using
airborne LiDAR
9 (Rana et al. 2014) Effect of field Sparse Bayesian method used Sample plot distance and Nepal, India
plot location on estimating to estimate biomass slope have effect on biomass
tropical forest AGB in Nepal estimation
using LIDAR
10 (Clark et al. 2011) Small footprint Using local minima, inverse First return Lidar sensor can be Costa Rica
LiDAR estimation of sub-cano- distance weighted (IDW) and used to predict sub-canopy
py elevation and tree height in ordinary kriging (OK) elevation in tropical rainforest
a tropical rain forest landscape

According to Baccini and Asner (2013), tropical forests have the largest carbon stock and har-
bour many of the world’s species. A tropical forest consists of different layer types: the emergent, the
sub-canopy, the canopy and the understory. Due to the complexity of the structure of tropical forest
trees, there are difficulties in measuring biophysical parameters through remote sensing technology
that varies in type and resolution (Saatchi et al. 2007). In short, measuring the tree height of a tropical
forest is quite challenging as it would include dense understory trees, tall canopies and closed-canopy
conditions which might obstruct the line of sight during field measurement (Hunter et al. 2013).

3. Tropical forest responses to bioclimatic and environmental conditions


Variation in forest types also contributes to AGB changes, being closely related to bioclimatic and
environment conditions (Barbosa et al. 2014). Research has shown that the estimation of AGB using a
remote sensing method varies between different types of dry Dipterocarp forest, mixed deciduous and
tropical evergreen forest (Wang & Qi 2008).The geographical condition of the forest would affect the
AGB estimation: for example, in the study by Asner et al. (2009), it was shown that biomass decreases
704 N. A. MOHD ZAKI AND Z. ABD LATIF

by between 53 and 84% with the increase in height, while Rana et al. (2014) agree that the slope and
elevation would affect the AGB estimation. Moreover, (Jérôme Chave et al. 2014), who classified the
forest into dry forest, moist forest, moist mangrove and wet forest, state that the equation must take
into account the bioclimatic stress variable as one of the constraints. All these show that environmental
conditions affect the pattern of AGB estimation.
Research by Propastin (2012) investigated the effect of altitude on the spatial weighing matrices
of the geographically weighted regression (GWR) model in a tropical rainforest region. This research
shows that the relationship between AGB and the vegetation index (VI) has significantly improved
by use of the GAWR method that includes altitude, or elevation, as a spatial weight in GWR. Based
on this study, it is shown that the GAWR model is better than the traditional GWR model and that
horizontal and vertical (altitudinal) aspects of the spatial non-stationary are strongly correlated.
In addition, a study by Asner et al. (2009) stated that an increase in elevation resulted in a decrease
of mean AGB to 53–84% due to environmental and biotic factors. On the other hand, another research
found the opposite pattern for elevation gradient in the Brazilian tropical forest. According to a study
carried out by Van der Laan et al. (2014), there is a strong positive relationship between AGB against
the test variables, these being altitude, slope and soil type, as well as with anthropogenic variables.
Therefore, the relationship between forest AGB and environmental condition indicates the need to
study the varying patterns of those climatic variables to quantify AGB in certain places.

4. Quantifying above-ground forest biomass measurement


Several methods have been outlined to measure above-ground biomass, but primarily it can be divided
into direct method (destructive sampling) and indirect method (non-destructive method). Using the
destructive sampling method to develop a new technique would take a huge amount of resources and
would involve cutting down a lot of large trees (Ketterings et al. 2001; Chave et al. 2005; Basuki et al.
2009). On the other hand, utilizing the non-destructive method consists of developing the allometric
equation or regression model, calculating from forest inventory data. Application of remote sensing
and geographical information system (GIS) methods can also estimate the above-ground biomass
(Brown 1997; Lu et al. 2005). However, appropriate methods should be refined in order to estimate the
above-ground biomass using remote sensing integrated with field measurement: this concept would
reduce the labour-intensive data collection when it comes to a large area.

4.1. Destructive sampling method


Direct measurement also known as complete harvesting is a method in which a small sample of trees
within a plot is felled and oven-dried until the weight is constant; it is used globally in many researches
while harvesting the trees (Basuki et al. 2009; van Breugel et al. 2011; Alvarez et al. 2012; Feldpausch
et al. 2012; Fayolle et al. 2013). Felling, dissecting and weighing can use a lot of resources, and involve
a huge amount of work, even more if the root system is also excavated and included in the biomass of
tree (West 2009). Therefore, use of an allometric equation developed by other researchers is suggested
for certain areas that have the same characteristics and species composition, as it would take a huge
number of resources to develop a new allometric equation from destructive sampling.

4.2. Allometric equation method


Another method that is often used to estimate the above-ground biomass is by use of the method
of conversion of volume data to develop an allometric equation. Tree stand allometric equations are
developed by calculating relationships from field measurements of tree parameters such as the diameter
of the trunk, tree height, the diameter at breast height (DBH), tree species, age, crown density and
bioclimatic variables. Using the existing volume data, allometric equation of the above-ground biomass
can be estimated (Brown 1997). According to Basuki et al. (2009), Brown (1997) and Ketterings et al.
Geocarto International  705

(2001), common forms of allometric equation for biomass estimation use allometric form (y = axb),
exponential (y = aebx) and quadratic (y = a + bx + cx2). An abundance of allometric equations have
now been developed: for instance, Chave et al. (2005, 2014) have developed a pantropic multi-species
allometric equation to develop a huge data-set where there are over 400 trees (diameter more than
five metres) from 58 countries all over the world, excluding Africa, while Ketterings et al. (2001) have
established an allometric regression for the mixed secondary forest in Indonesia. However, these
forests are not considered as lowland Dipterocarp forest. However, the allometric equation developed
by Chave et al. (2005) is currently widely used for the approximation of sites where local equations
are not available (Alvarez et al. 2012; Fayolle et al. 2013). Different types of allometric regressions
concerning biomass and tree height are dependent on environmental variables, for example, climatic
condition, bioclimatic stress, and therefore Chave et al. (2014) improved the allometric equation based
on this problem. Selection of the allometric equation for AGB is dependent on the characteristics and
composition of the forest study area. The allometric equation in Equation (1) is used to estimate the
above-ground biomass from diameter at the breast height (DBH) of the tree.
In (AGB) = 𝛼 + 𝛽 In(𝜌 × D2 × H) (1)
According to the equation above, AGB is above-ground biomass, α and β are model coefficients
(derived from least square regression), D is trunk diameter (cm) and ρ (g cm3) is wood-specific gravity,
also known as oven-dry wood over green volume; H is total tree height (m), while ɛ is an error term
(Chave et al. 2014).
Ngo et al. (2013) assert that efforts have been devoted to quantifying carbon stock in primary and
secondary forests in Singapore utilizing the allometric equation developed by Chave et al. (2005). By
this method, the total estimated carbon stock in the primary forest was 337 Mg C ha−1, while in the
secondary forest, it was 274 Mg C ha −1. Hunter et al. (2013) improved the biomass estimation by
evaluating the accuracy of the diameter–height allometric in the Brazilian Amazon using LiDAR and
ground measurement. Therefore, the accuracy of individual tree height measurement varies from 3
to 20% of the total height, leading to imprecise outcomes with 5 or 6% uncertainty.

4.3. Remote sensing approaches: sensors for above-ground biomass estimation


There is today an abundance of research emerging about remote sensing technology for qualitative
and quantitative measurement of biophysical parameters and it has become a popular technique that
is used all over the world (Lu et al. 2005; Zheng et al. 2007; Gonzalez et al. 2010; Morel et al. 2011; Sun
et al. 2011; Hudak et al. 2012; Viana et al. 2012). Conventionally, obtaining biophysical information
about forest biomass was achieved by conducting a small research to measure a sample area of the
forest. There are three levels of forest biomass estimation using remote sensing in line with the spa-
tial resolution of sensors: (1) very high resolution satellite imagery; (2) moderate resolution satellite
imagery; and (3) coarser/low resolution satellite imagery. Multispectral imagery remote sensing has
a multiple band remote sensing imaging system which usually consists of two or more bands. A band
can be defined as a channel that stores the information gathered from a narrow wavelength range or
a given spectral width – for example, 20 nm.

4.3.1. Moderate/low resolution multispectral imagery (MRMI)


One of the first satellite imagery platforms in multispectral remote sensing was Landsat MSS, which
has been providing data since the 1970s and offers free open access to the public user. Various vege-
tation indices have been established by scholars with remotely sensed data due to the strong relation-
ship between vegetation and forest parameters – for example, the normalized difference vegetation
index (NDVI), green NDVI (gNDVI), principal components (PCs), simple ratio (SR), leaf area index
(LAI) and many more. Among the scholars who have used Landsat TM image processing are Lu et
al. (2002) Yan et al. (2015) and Zheng et al. (2007), who use vegetation indices including simple ratio
(SR) and the normalized difference vegetation index (NDVI) to show that vegetation indices have
706 N. A. MOHD ZAKI AND Z. ABD LATIF

a high correlation with biomass estimation. Roy and Ravan (1996) utilize Landsat TM for regional
biomass mapping using a multiple regression equation integrated with a physiographic map showing
a correlation coefficient of 0.77%. A study by Lu et al. (2002) concluded that the spectral bands for
Landsat TM and the vegetation index are both adequate to establish a good model for biomass using
multiple regression modelling. In addition, Frazier et al. (2014) and Gómez et al. (2014), also using
the Landsat data-set, studied the temporal spectral trajectory metrics to investigate the relationship
between AGB and the spectral properties of the forest attributes of boreal forest and Mediterranean
pines. A valuable aspect of their research was the use of the tasselled cap (TC) component to estimate
the AGB, showing that the TC component has a good relationship with forest density.
In a further multisensory study, Fernández-Manso et al. (2014) explored the relationship between
AGB and spectral bands using ASTER imagery and forest inventory data collection. In this study,
the authors estimated the AGB on fraction images from linear spectral mixture analysis (LSMA) to
overcome the deficiencies of the medium resolution sensor used, this being ASTER (15–90 m spatial
resolution). Pearson correlation and stepwise multiple regression were applied to the images to select
the best predictors from the fraction images, NDVI and TC components. The study found that using
ASTER fraction images in regression modelling increases the AGB estimation in the Mediterranean
pine forest. Research by Mitchard et al. (2014) investigated the accuracy of pantropical carbon maps
using a huge inventory plot located in nine countries in the Amazon tropical forest. Using MODIS
integrated with a geoscience laser altimeter system (ICES at GLAS) sensor, extrapolation of AGB was
established using multi-variable non-linear models and random forest calculation. The authors further
noted the importance of measuring tree height with the species at least at the genus level, as the tropical
forest holds a huge diversity of species which needs to be matched according to their wood density.

4.3.2. Very high spatial resolution multispectral imagery (VHRM)


Several types of very high resolution multispectral (VHRM) imagers such as GeoEye, WV2, WV3
IKONOS and QB can be used for quantifying biomass at different resolutions and scales (Gonzalez
et al. 2010; Eckert 2012; Karna et al. 2015; Hussin et al. 2014). These optical sensors have a variety of
spatial, radiometric, temporal and spectral resolutions depending on their type and capabilities. In
order to quantify and model AGB using high resolution spatial resolution imagery, suitable methods
have to be chosen. Based on several researches utilizing VHRM by Gibbs et al. (2007), Hussin et al.
(2014) and Lu, Chen, et al. (2014), variables for tree forest parameters and biomass modelling could
be extracted from multispectral imagers, RADAR and LiDAR. Despite being more costly and small
scale compared to moderate multispectral imagers, VHRM could be considered for use because more
data could be discerned for forest tree extraction.
Researchers Hussin et al. (2014) developed a biomass/carbon stock model using segmentation of
an object-based image analysis (OBIA) classification technique by correlating the crown projection
area (CPA) of GeoEye-1 satellite imagery with a species of the tree Shorea Robusta. They concluded
that a regional growing approach showed better results compared to valley-following techniques for
a mixed forest type. Further to this research, Mbaabu et al. (2014) explored the capabilities of OBIA
using GeoEye-1 satellite imagery with airborne LiDAR data with an accuracy output of 70–82% for
classification of results. Meanwhile, Sousa et al. (2015) used a multi-resolution segmentation method
that included the NDVI formula to generate the CPA and vegetation mask for the Mediterranean for-
est in Portugal. However, Jing et al. (2012) tested the integration of aerial photographs with intensity
components as an input for tree biomass delineation using watershed segmentation methods. Based
on this research, the proposed method resulted in a high-quality tree crown map which was useful for
biomass calculation. In short, VHRM is a promising approach to separate the crowns of individual trees
but is quite complicated for the complex structure of the tropical rainforest that has different layers of
tree canopy. Eckert (2012) tested the WV2 images by performing a stepwise multiple linear regression.
He concluded that texture measure correlates more with biomass than with vegetation index.
Geocarto International  707

4.4. Remote sensing approaches: LiDAR for quantifying AGB in forests


LiDAR is an acronym derived from light detection and ranging. Historically, airborne laser scanning
started in the early 1970s, proving that an airborne laser system can penetrate distance with precision
of under one metre (Vosselman & Maas 2010). Over the past few decades, interest has grown among
researchers in the use of LiDAR in forestry and ecosystem studies (Drake et al. 2002; Patenaude et al.
2004; Clark et al. 2011; Tonolli et al. 2011; Li et al. 2014; Vincent et al. 2014). LiDAR is a promising
technology that provides accuracy of biophysical parameters while encompassing height in three
dimensions (Lefsky et al. 2002). Recent studies show increasing research interest in using LiDAR
metrics for forest biomass studies (Hudak et al. 2012; Vincent et al. 2012; Fassnacht et al. 2014; Vega
et al. 2014; Badreldin & Sanchez-Azofeifa 2015). Figure 1 summarizes the overall quantifying of AGB
using LiDAR for forest studies according to researchers.
Typically, studies of LiDAR for forest biomass measurement can be classified according to several
methods of LiDAR metrics which rely on the type of platform used (space-borne, airborne or ground
platform) (Gleason & Im 2012; Baccini & Asner 2013); the scanning design (profiling and scanning)
(Chen & Qi 2013); the type of signal return recorded (discrete or full waveform) (Blackburn et al.
2014); and the footprint size (single tree, small footprint, moderate footprint and large footprint LiDAR
profile) (Drake et al. 2002). Widening ALS technology is pioneered by several airborne LiDAR sensor
providers: for example, Riegl (LMS), Optech (ALTM) and Leica (ALS). The ground-based techniques
platform involves terrestrial laser scanning (TLS) that measures the dimension of objects in 3D and
provides data point clouds for forest biomass estimation. TLS developers available include Faro, Leica,
Topcon and many more.
In terms of LiDAR captured by space-borne satellite, the types of LiDAR that are widely used for
biomass research are LVIS and GLAS. LiDAR data can be categorized based on output signal returns
recorded either full-waveform or discrete (Lefsky et al. 2002; Pirotti 2011). Previous biomass for
forestry research shows the variety of LiDAR scanning footprint sizes depends on research interest.
For instance, small footprint is to within approximately one metre (Vincent et al. 2012), moderate
footprint approximately10–30 metres (Meyer et al. 2013) and large footprint is 50 metres or more

Figure 1. Summarization of LiDAR for AGB estimation.


708 N. A. MOHD ZAKI AND Z. ABD LATIF

(Wulder et al. 2012). This review will scrutinize the application of LiDAR for forestry purposes and
narrow it down to the tropical forest characteristics.

4.4.1. Quantifying AGB using airborne small footprint LiDAR discrete return
Forest mensuration using small footprint airborne LiDAR discrete return is widely used by research-
ers to quantify the biomass of the forest structure for the area base (Patenaude et al. 2004; Popescu
et al. 2011); to estimate biomass change (Hudak et al. 2012); for tree species classification (Dalponte
et al. 2014); for tree crown delineation (Jing et al. 2012): and to quantify individual trees biomass for
the plot scale (Vega et al. 2014).The key components of airborne laser scanning are a laser scanner
system and a GPS/IMU combination (Vosselman & Maas 2010). The numerous earlier AGB studies
utilizing small airborne discrete returns can generally be separated into two methods, these being the
area-based method and single tree extraction (Chen & Qi 2013).
The individual tree-based or 3D individual tree method of AGB is the extraction of a single tree’s
attributes such as tree crown, tree top, stem diameter, tree height and other parameters that relate to
the biomass of the tree. Previously, many researches extracted single tree forest characteristics such
as tree tops (Chen et al. 2006); 3D single tree crown (Mongus & Žalik 2015); 3D multilayer single
tree and crown radius (Popescu et al. 2003); and many more. This type of method is quite tough as
it requires the use of high-point density data (approximately 10 points per square metre or more),
especially for a complex forest structure like a tropical rain forest (Lu et al., 2014). Nowadays, the
advent of LiDAR technology has improved: a lot of sensors that have higher density laser repetition
rates have been developed and higher density LiDAR can be collected. However, the most challenging
aspects of individual tree-based AGB estimation are: (i) the irregular size of crown cover for certain
types of forest (e.g. tropical rain forest and sub-tropical forest); (ii) the multilayer understory level
of the tree (e.g. tropical rain forest and temperate broadleaved forest). However, the introduction of
LiDAR waveform technology offers a good solution to this problem, though it needs an intellectual
algorithm to solve the problem.

4.4.2. Quantifying AGB using airborne large/small footprint waveform LiDAR


Small footprint waveform LiDAR could retrieve better characteristics of the forest tree stands than a
discrete LiDAR data-set, as the data are denser and capture more energy signals redirected from the
tree (Pirotti 2011). A number of studies have utilized space-borne large footprint waveform LiDAR
to characterize the forest structure (e.g. Popescu et al. 2011; Baccini et al. 2012; Baccini & Asner 2013;
Huang et al. 2013; Gwenzi & Lefsky 2014), but only a few studies use small footprint waveform LiDAR
(Li et al. 2014; Lu et al., 2014) as the waveform LiDAR data are quite challenging, especially for very
dense forests (Chen & Qi 2013).
Reitberger et al. (2009) compared the segmentation single tree extraction of small waveform LiDAR
using a watershed segmentation procedure, stem detection method and normalized cut segmentation
for alpine spruce forest. As a result, higher density of more than 10 points per square metre does
not increase the detection rate but would be beneficial for quantifying timber volume. However, the
waveform LiDAR still could not segment all the lower layers of the single tree if the forest was too
dense. Other scholars tested seven models using small waveform LiDAR, these being ordinary least
square, cubist, bagging, random forest, boosted regression trees, the generalized additive model and
support vector regression (SVR) to quantify AGB for plot level in a mixed-hardwood conifer forest
(Li et al. 2014). Based on their results, LiDAR waveform strongly relates to AGB when using the
higher percentile height, which is the 80th and 90th percentile, and also suggests the use of kurtosis
distribution of intensity data to improve the estimation.
A large footprint waveform LiDAR data-set that has commonly been used by previous researchers is
GLAS, which was aboard the NASA ice, cloud and land elevation (ICEsat) satellite that operated from
2003 to 2009 (Abdalati et al. 2010). Usually, researchers would use this type of data-set to quantify AGB
and carbon stocks at a regional level (Popescu et al. 2011; Baccini et al. 2012). Popescu et al. (2011)
compared the AGB estimation of forests using GLAS and small footprint LiDAR and found that GLAS
Geocarto International  709

elevations correlated with airborne LiDAR (R = 0.995) and RMSE (0.78 m), which is well and highly
correlated. However, this research was done in a temperate forest, and would be more complicated
for other types of forests with a complex structure, such as tropical rainforest.

4.4.3. Quantifying AGB using LiDAR for tropical rain forest


A number of studies have been conducted to quantify AGB and carbon stock for tropical forest all over
the world, including Indonesian forest (Kronseder et al. 2012), Columbian forest (Baccini & Asner
2013) and Malaysian forest (Ioki et al. 2014). Several used the random forest algorithm and some of
the researchers used regression (stepwise multiple regression, power law function and local minima).
Usually, the type of LiDAR data used by the researcher is the canopy height model (CHM) generated
from the LiDAR data, which shows the difference between last return and first return, and computes
the tree height (see Figure 2). Previous researchers have generated the canopy height model derived
from the LiDAR data-set and found that canopy height and mean canopy height are important pre-
dictors to quantify AGB in sub-tropical forest and tropical forest (De Sy et al. 2012; Cao et al. 2014).
Regression using the logarithmic equation of the power model is used by a large number of studies to
estimate AGB using LiDAR: Latifi et al. (2012) used a linear model (stepwise linear model) and Ene
et al. (2012) used a non-linear transformation model to construct the relationship between LiDAR
and forest biomass.
The most common method of relating the LiDAR forest characteristics is using an area-based
statistical model at the stand or plot level and then applying the model to the whole forest (Chen &
Qi 2013). The area-based statistical model to quantify AGB/carbon stocks has been applied to various
forest types such as at tropical rainforest (Asner et al. 2010; Kronseder et al. 2012; Baccini & Asner 2013;
Asner & Mascaro 2014), sub-tropical (Cao et al. 2014), savanna forest (Gwenzi & Lefsky 2014), boreal
forest (Sun et al. 2011; Gobakken et al. 2012; Neigh et al. 2013), temperate forest (Jeanne Anderson
et al. 2006) and coniferous forest (He et al. 2013). Overall, these studies are at the landscape scale
and the number of plots used is quite small (approximately 20 to 100), but the study conducted by
Anderson et al. (2008) used a large field plot sample where 400 field plots were identified in the study
area. Table 2 summarizes the research overview and the methodology of utilizing LiDAR data in the
tropical forest.

Figure 2. Canopy height model of tropical rain forest. Figure 2(a): The top view of the CHM. Figure 2(b): The 3D view from the top of
CHM, Figure 2(c): The interpolation of height of the CHM and finally Figure 2(d): The perspective view of the CHM.
710 N. A. MOHD ZAKI AND Z. ABD LATIF

5. Techniques and algorithms used for quantifying the biomass model


Many techniques have been established by researchers to quantify biomass. Based on Lu et al. (2014),
there are two wide categories of algorithms for determining biomass estimation: the parametric and
the non-parametric approaches. Parametric algorithm has a fixed number of variables, while non-par-
ametric algorithm uses a flexible number of variables. In the parametric algorithm, the hypothesis
relationship between the dependent and independent variables has explicit model functioning that
can be derived from the parameters (Lu et al., 2014). For instance, the parametric algorithm may be
a linear regression model or multiple linear regression models. On the other hand, the non-paramet-
ric algorithm accomplishes the model structure in a data-driven way (Lu et al., 2014). For example,
the non-parametric algorithms that have commonly been used are the random forest (RF), nearest
neighbours (NN), Euclidean distance (ED) and others.

5.1. Parametric-based algorithms


Among researchers, many have used regression techniques and remote sensing technology to quantify
biomass (Gonzalez et al. 2010; Latifi et al. 2012; Li et al. 2014). Overall, to quantify biomass using a
parametric-based algorithm, the dependent variable would be the biomass, while the independent
variables vary: they may be textural image, height, canopy height model or many others. Eckert (2012)
performed a stepwise multiple linear regression model to model the biomass and carbon stock models
and obtained a strong correlation for the texture measures. Furthermore, research carried out by Roy
and Ravan (1996) and Zheng et al. (2007) used the multiple regression and stepwise region regres-
sion methods to model the AGB estimation. Based on their results, the empirical model with satellite
measures and spectral response has a good correlation with biomass. Kumar et al. (2013) also used
regression in many forms of modelling, such as linear, logarithm and power exponential, to obtain
the correlation AGB estimation. Based on these findings, the study showed significant correlation
(R2 = 0.53) to predict biomass using simple regression. Zhu and Liu (2015) compared the parametric
and non-parametric algorithms using simple linear regression, random forest and artificial neural
network to improve forest AGB estimation using the Landsat NDVI time series. From their research,
each of the algorithms has its own particular strengths and it is necessary to choose a modelling
method according to the specific research needs.

5.2. Non-parametric-based algorithms


Non-parametric methods have described the non-linear and complex relationship between biomass
variables and remote sensing data (Zhu & Liu 2015). Many studies have used non-parametric tests
for their research in order to estimate the biomass of the forest (Fazakas et al. 1999; Latifi et al. 2010;
Lumbres & Lee 2014). Based on previous researches, one of the non-parametric techniques, k-nearest
neighbour (k-NN), is widely used to map biomass, land use and land cover (McRoberts et al. 2007;
Latifi et al. 2012; Lumbres & Lee 2014); to retrieve the forest attributes (Fazakas et al. 1999; McRoberts
et al. 2007); and also for a forest’s species-specific prediction (Breidenbach et al. 2010). Practically,
k-NN algorithm works to represent the observation where response and predictor parameters are
recognized as references: if k is greater than 1, the assigned value calculates the values of k-NN for
each response parameter (Breidenbach et al. 2010).
Random forest is one of the non-parametric tree-based models that constructs a large number of
regression trees by choosing a random sample from the data-set, whether discrete or continuously
(Lu et al., 2014; Mascaro et al. 2014). Breidenbach et al. (2010) tested two non-parametric algorithms
– random forest (RF) and most similar neighbours’ inference (MSN) – to overcome the segmentation
result of individual tree crown (ITC) by calculating the distance between the segments. The researchers
concluded that the results show a significant underestimation, especially for the rare deciduous trees.
However, the biases using RF are larger when compared to using MSN (Breidenbach et al. 2010).
Geocarto International  711

Table 2. Overview of research utilizing LiDAR for tropical rainforest from the entire world.

Researcher Criteria Method Used Remarks Result Scale


(Baccini & Fine-tune a pantrop- Random forest algo- Carnegie Airborne Total mass carbon 15.7 €
Asner 2013) ical carbon map rithm Observatory (CAO) and 17.6 MgC ha−1
(LiDAR) to specific used to calibrate
sub-national regions MODIS data to im-
(MODIS) prove AGB mapping
(Baccini et al. Use multisensory RS Random forest algo- Use multisensory Total mass of carbon ¥
2012) data and field test rithm satellite data to in America, Africa
to estimate AGB/C estimate AGB and and Asia to be 117.7
stock carbon (±8.4), 64.5(±8.4)
and 46.5(±3.0) Pg C
(Asner & Mas- Using field inventory Lorey’s height Evaluate the RF in Regional input : £
caro 2014) plots with LiDAR top up-scaling airborne R2=0.85 and RM-
of canopy height LiDAR SE=24.7MgCha−1
to estimate AGB/C Universal input : R2
stocks =0.83 and RMSE=
26.3 Mg C ha−1
(Vincent et al. Used small footprint Linear regression used Direct multiple regres- RMSE of Prediction €
2012) LiDAR to predict to predict BA sion was the most less than 10% for BA
basal area (BA) and accurate
its components
(Asner et al. Used multi-scale and CLASlite algorithms to Regional mapping of Total mass carbon €
2010) multi-temporal map forest land use AGB/C stock (secondary forest)
RS data to analyse is 30.6 ± 16.7 Mg
AGB/C stock C ha−1
(Kronseder AGB estimation across Stepwise multiple LiDAR 3D to quantify AGB values using #
et al. 2012) different types of regression method AGB at plot level allometric model
forests degradation used 15–547 Mg ha−1
levels
(Mascaro et al. Error in LiDAR and Tested the plot edge LiDAR error range RMSE=10.7 Mg C ha−1 €
2014) predictions of AGB discrimination errors from 17 to 40 Mg
C ha−1
(Ioki et al. Estimate AGB at dif- Used multidimension- Using laser penetra- The average AGB = #
2014) ferent degradation al scaling (nMDS) to tion rate (LP) to 382 Mg/ha
levels compare different estimate AGB The height percentile
degradation levels models correlated
with AGB, with R2
> 0.60
(Meyer et al. Estimate AGB/C stocks Model developed Compare difference of Total mass carbon (R2 μ, €
2013) dynamics using using power law two biomass using = 0.7 and RMSEmean
temporal LiDAR data function two models using = 27.6Mgha−1)
relative height and
mean canopy height
(MCH)
(Drake et al. Estimate tropical for- Quadratic mean stem Agreed the capabil- R2 value: QMSD: 0.93, μ
2002) est structure using diameter (QMSD), ities of LIDAR for basal area: 0.72, and
LiDAR BA and AGB esti- estimate biomass AGB 0.93
mated in dense tropical
forests

Notes: € = Discrete return LiDAR in plot level, £ = Discrete return LiDAR in large scale, ¥ = GLAS footprint in large plot, # = Small
footprint full-waveform LiDAR in plot level and μ = Land, Vegetation and Ice Sensor – LVIS data.

An artificial neural network (ANN) model is widely used for forest biomass estimation with
remote sensing technology (Foody et al. 2001; Cutler et al. 2012; Deng et al. 2014; Zhu & Liu 2015).
When handling a very complex relationship data-set, ANN is the best algorithm that can be used
– for instance, in dealing with non-linearity and non-normality issues (Zhu & Liu 2015). Based on
Cutler et al. (2012), the ANN algorithm is used to test the texture measures derived from SAR and
Landsat TM imagers in four training scenarios. The results from the study are that the combination
of SAR texture and Landsat TM data from three regions showed the strongest correlation with AGB
at all sites.
712 N. A. MOHD ZAKI AND Z. ABD LATIF

6. Linking measurement of carbon stocks and options from geospatial technology


In this review, we emphasize the benefits and drawbacks of estimating above-ground forest biomass
from remote sensing. Overall, it is seen that the most critical procedure is to obtain an accurate and pre-
cise estimation of above-ground biomass. Furthermore, an evaluation of studies shows the importance
of using multi-temporal data to monitor the above-ground biomass, as the value of biomass change
might differ across several years. The synoptic use of remote sensing will provide more detailed analysis
with the integration of geospatial technology, which could improve forest management, evaluation of
carbon sink, environmental monitoring and climate change mitigation focus on forest structure, and
would help decision-makers such as REDD+ and many related organizations. Predictive models can
also be developed with the existing biomass estimation available for each year.

7. Conclusion
Previous work on above-ground forest biomass estimation has been reviewed from the perspectives
of field inventory methods, destructive sampling, allometric equations and remote sensing methods.
To access the biomass estimation for the tropical rain forest, with its complex structure of understory
trees, vast diversity of species and dense canopies, was really challenging. Destructive sampling uses
a lot of resources and is labour intensive, but it has a high degree of accuracy for biomass estimation.
However, to regenerate the number of trees that are cut would require hundreds of years, assuming
minimal damage to the forest floor during selective logging. This study has highlighted the limita-
tions of the use of destructive sampling and the evolution of the allometric equation from destructive
sampling methods to quantify AGB.
Remote sensing is an advanced method for biomass estimation with the various scales of data source.
However, biomass estimation using remote sensing is an ample procedure with no exact algorithm
agreed as a standard procedure. Therefore, reducing uncertainty in the biomass estimation is the best
method for reliable AGB estimation. LiDAR is the most promising technique for AGB estimation:
there is less uncertainty with the canopy height model extracted from the high density LiDAR, with
its accurate displays of up to centimetre level (Vosselman & Maas 2010). With the growth of the
technologies, the integration of LiDAR and VHR satellite imaging for AGB is a good combination
for better biomass mapping with spatial accuracy. With the availability of various scales of remote
sensing imagery, from coarser to higher resolution, integration of this multisensory technology could
improve the AGB estimation from local to regional scales’ estimation (Baccini et al. 2012). Overall,
the reliability estimation of AGB using a remote sensing technique is influenced by the type of remote
sensing data source, sensor type, algorithm used, processing technique, bioclimatic conditions and
type of forest. Hence, the development of modelling regression to acquire the best estimation of above-
ground biomass would be a new direction, especially in understanding the integration of allometric
equation and remote sensing modelling.

Acknowledgements
The authors would like to express their gratitude for the project funding through the Fundamental Research Grant Scheme
(FRGS 144/2015) awarded by the Ministry of Higher Education (MOHE), Malaysia, and the Research Management
Centre (RMC), Universiti Teknologi MARA, Malaysia, for its administrative support.

Disclosure statement
No potential conflict of interest was reported by the authors.

Funding
This work was supported by the Fundamental Research Grant Scheme (FRGS) [600-RMI/FRGS 5/3 (144/2015)].
Geocarto International  713

References
Abdalati W, Zwally HJ, Bindschadler R, Csatho B, Farrell SL, Fricker HA, Harding D, Kwok R, Lefsky M, Markus
T, et al. 2010. The ICESat-2 laser altimetry mission. Proc IEEE. 98:735–751. doi:http://dx.doi.org/10.1109/
JPROC.2009.2034765. .
Alvarez E, Duque A, Saldarriaga J, Cabrera K, de las Salas G, del Valle I, Lema A, Moreno F, Orrego S, Rodríguez L.
2012. Tree above-ground biomass allometries for carbon stocks estimation in the natural forests of Colombia. For
Ecol Manage. 267:297–308.
Anderson J, Martin ME, Smith M-L, Dubayah RO, Hofton MA, Hyde P, Petterson BE, Blair JB, Knox RG. 2006. The use
of waveform lidar to measure northern temperate mixed conifer and deciduous forest structure in New Hampshire.
Remote Sens Environ. 105:248–261.
Anderson J, Plourde L, Martin M, Braswell B, Smith M, Dubayah R, Hofton M, Blair J. 2008. Integrating waveform lidar
with hyperspectral imagery for inventory of a northern temperate forest. Remote Sens Environ. 112:1856–1870.
Aryal DR, De Jong BHJ, Ochoa-Gaona S, Esparza-Olguin L, Mendoza-Vega J. 2014. Carbon stocks and changes in
tropical secondary forests of southern Mexico. Agric Ecosyst Environ. 195:220–230.
Asner GP, Mascaro J. 2014. Mapping tropical forest carbon: calibrating plot estimates to a simple LiDAR metric. Remote
Sens Environ. 140:614–624.
Asner GP, Flint Hughes R, Varga TA, Knapp DE, & Kennedy-Bowdoin T. 2009. Environmental and biotic controls over
aboveground biomass throughout a tropical rain forest. Ecosystems. 12:261–278.
Asner GP, Powell GVN, Mascaro J, Knapp DE, Clark JK, Jacobson J, Kennedy BT, Balaji A, Paez-Acosta G, Victoria E, et al.
2010. High-resolution forest carbon stocks and emissions in the Amazon. Proc Nat Acad Sci USA. 107:16738–16742.
Baccini A, Asner GP. 2013. Improving pantropical forest carbon maps with airborne LiDAR sampling. Carbon Manag.
4:591–600.
Baccini A, Goetz SJ, Walker WS, Laporte NT, Sun M, Sulla-Menashe D, Hackler J, Beck PSA, Dubayah R, Friedl MA,
et al. 2012. Estimated carbon dioxide emissions from tropical deforestation improved by carbon-density maps. Nat
Clim Change. 2:182–185.
Badreldin N, Sanchez-Azofeifa A. 2015. Estimating forest biomass dynamics by integrating multi-temporal landsat
satellite images with ground and airborne LiDAR data in the Coal Valley Mine, Alberta, Canada. Remote Sens.
7:2832–2849.
Barbosa JM, Broadbent EN, Bitencourt MD. 2014. Remote Sensing of Aboveground Biomass in Tropical Secondary
Forests : A Review. Int J Forestry Res. 1–14.
Basuki TM, van Laake PE, Skidmore AK, Hussin Ya. 2009. Allometric equations for estimating the above-ground biomass
in tropical lowland Dipterocarp forests. For Ecol Manage. 257:1684–1694.
Blackburn GA, Abdul Latif Z, Boyd DS. 2014. Forest disturbance and regeneration: a mosaic of discrete gap dynamics
and open matrix regimes? J Veg Sci. 25:1341–1354.
Breidenbach J, Næsset E, Lien V, Gobakken T, Solberg S. 2010. Prediction of species specific forest inventory attributes
using a nonparametric semi-individual tree crown approach based on fused airborne laser scanning and multispectral
data. Remote Sens Environ. 114:911–924.
van Breugel M, Ransijn J, Craven D, Bongers F, Hall JS. 2011. Estimating carbon stock in secondary forests: decisions
and uncertainties associated with allometric biomass models. For Ecol Manage. 262:1648–1657.
Brown S. 1997. Estimating biomass and biomass change of tropical forests: a primer. (FAO Forestry Paper – 134).
Department of Natural Resources and Environmental Sciences,University of Illinois,Urbana, Illinois, USA: A Forest
Resources Assessment publication. Available from: http://www.fao.org/docrep/w4095e/w4095e00.HTM
Cao L, Coops N, Innes J, Dai J, She G. 2014. Mapping above- and below-ground biomass components in subtropical
forests using small-footprint LiDAR. Forests. 5:1356–1373.
Chave J, Andalo C, Brown S, Ma Cairns, Chambers JQ, Eamus D, Fromard F, Higuchi N, Kira T, Lescure JP, et al. 2005.
Tree allometry and improved estimation of carbon stocks and balance in tropical forests. Oecologia. 145:87–99.
Chave J, Réjou-Méchain M, Búrquez A, Chidumayo E, Colgan MS, Delitti WB, Duque A, Eid T, Fearnside PM, Goodman
RC, Henry M, et al. 2014. Improved allometric models to estimate the aboveground biomass of tropical trees. Glob
Change Biol. 20:3177–3190.
Chen Q, Qi C. 2013. LiDAR remote sensing of vegetation biomass. Remote Sens Nat Res. 20135777:399–420.
Chen Q, Baldocchi D, Gong P, Kelly M. 2006. Isolating individual trees in a Savanna woodland using small footprint
lidar data. Photogramm Eng Remote Sens. 72:923–932.
Clark ML, Roberts DA, Ewel JJ, Clark DB. 2011. Estimation of tropical rain forest aboveground biomass with small-
footprint lidar and hyperspectral sensors. Remote Sens Environ. 115:2931–2942.
Cutler MEJ, Boyd DS, Foody GM, Vetrivel A. 2012. Estimating tropical forest biomass with a combination of SAR
image texture and Landsat TM data: An assessment of predictions between regions. ISPRS J Photogramm Remote
Sens. 70:66–77.
Dalponte M, Ørka HO, Ene LT, Gobakken T, Næsset E. 2014. Tree crown delineation and tree species classification in
boreal forests using hyperspectral and ALS data. Remote Sens Environ. 140:306–317.
714 N. A. MOHD ZAKI AND Z. ABD LATIF

De Sy V, Herold M, Achard F, Asner GP, Held A, Kellndorfer J, Verbesselt J. 2012. Synergies of multiple remote sensing
data sources for REDD+ monitoring. Curr Opin Env Sust. 4:696–706.
Deng S, Katoh M, Guan Q, Yin N, Li M. 2014. Estimating forest aboveground biomass by combining ALOS PALSAR
and WorldView-2 Data: a case study at Purple Mountain National Park, Nanjing, China. Remote Sens. 7878–7910.
Drake JB, Dubayah RO, Clark DB, Knox RG, Blair JB, Hofton MA, Chazdon RL, Weishampel JF, Chazdon Robin L. 2002.
Estimation of tropical forest structural characteristics using large-footprint lidar. Remote Sens Environ. 79:305–319.
Eckert S. 2012. Improved forest biomass and carbon estimations using texture measures from worldview-2 satellite
data. Remote Sens. 4:810–829.
Ene LT, Næsset E, Gobakken T, Gregoire TG, Ståhl G, Nelson R. 2012. Assessing the accuracy of regional LiDAR-based
biomass estimation using a simulation approach. Remote Sens Environ. 123:579–592.
FAO. 2015. Global Forest Resources Assessment 2015 (ISBN 978-9.). Available from: http://www.fao.org/forestry/
fra2005/en/
Fassnacht FE, Hartig F, Latifi H, Berger C, Hernández J, Corvalán P, Koch B. 2014. Importance of sample size, data
type and prediction method for remote sensing-based estimations of aboveground forest biomass. Remote Sens
Environ. 154:102–114.
Fayolle A, Doucet J-L, Gillet J-F, Bourland N, Lejeune P. 2013. Tree allometry in Central Africa: testing the validity of
pantropical multi-species allometric equations for estimating biomass and carbon stocks. For Ecol Manage. 305:29–37.
Fazakas Z, Nilsson M, Olsson H. 1999. Regional forest biomass and wood volume estimation using satellite data and
ancillary data. 99:417–425.
Feldpausch TR, Lloyd J, Lewis SL, Brienen RJW, Gloor M, Mendoza Monteagudo A, Lopez-Gonzalez G, Banin L, Abu
Salim K, Affum-Baffoe K, et al. 2012. Tree height integrated into pantropical forest biomass estimates. Biogeosciences.
9:3381–3403.
Fernández-Manso O, Fernández-Manso A, Quintano C. 2014. Estimation of aboveground biomass in Mediterranean
forests by statistical modelling of ASTER fraction images. Int J Appl Earth Obs Geoinf. 31:45–56.
Foody GM, Cutler ME, McMorrow J, Pelz D, Tangki H, Boyd DS, Douglas IAN. 2001. Mapping the biomass of Bornean
tropical rain forest from remotely sensed data. Glob Ecol Biogeogr. 10:379–387.
Frazier RJ, Coops NC, Wulder MA, & Kennedy R.. 2014. Characterization of aboveground biomass in an unmanaged
boreal forest using Landsat temporal segmentation metrics. ISPRS J Photogramm Remote Sens. 92:137–146.
Gibbs HK, Brown S, Niles JO, Foley JA. 2007. Monitoring and estimating tropical forest carbon stocks: making REDD
a reality. Environ Res Lett. 2:045023.
Gleason CJ, Im J. 2012. Forest biomass estimation from airborne LiDAR data using machine learning approaches.
Remote Sens Environ. 125:80–91.
Gobakken T, Næsset E, Nelson R, Martin O, Gregoire TG, Ståhl G, Holm S, Ole H, Astrup R. 2012. Remote Sensing
of Environment Estimating biomass in Hedmark County, Norway using national forest inventory fi eld plots and
airborne laser scanning. Remote Sens Environ. 123:443–456.
Gómez C, White JC, Wulder MA, & Alejandro P. 2014. Historical forest biomass dynamics modelled with Landsat
spectral trajectories. ISPRS J Photogramm Remote Sens. 93:14–28.
Gonzalez P, Asner GP, Battles JJ, Lefsky MA, Waring KM, & Palace M. 2010. Forest carbon densities and uncertainties
from Lidar, QuickBird, and field measurements in California. Remote Sens Environ. 114:1561–1575.
Gwenzi D, & Lefsky MA. 2014. Modeling canopy height in a savanna ecosystem using spaceborne lidar waveforms.
Remote Sens Environ. 154:338–344.
Hansen MC, Potapov PV, Moore R, Hancher M, Turubanova SA, Tyukavina A, … Townshend JRG. 2013. High-resolution
global maps of 21st-century forest cover change. Science. 342:850–853.
Hansen MC, Potapov PV, Moore R, Hancher M, Turubanova SA, Tyukavina A, Thau D, Stehman SV, Goetz SJ, Loveland
TR, et al. 2013. High-resolution global maps of 21st-century forest cover change. Science. 342:850–853.
He Q, Chen E, An R, Li Y. 2013. Above-ground biomass and biomass components estimation using LiDAR data in a
coniferous Forest. Forests. 4:984–1002.
Huang W, Sun G, Dubayah R, Cook B, Montesano P, Ni W, Zhang Z. 2013. Mapping biomass change after forest
disturbance: applying LiDAR footprint-derived models at key map scales. Remote Sens Environ. 134:319–332.
Hudak AT, Strand EK, Vierling LA, Byrne JC, Eitel JUH, Martinuzzi S, & Falkowski MJ. 2012. Quantifying aboveground
forest carbon pools and fluxes from repeat LiDAR surveys. Remote Sens Environ. 123:25–40.
Hunter MO, Keller M, Victoria D, Morton DC. 2013. Tree height and tropical forest biomass estimation. Biogeosciences.
10:8385–8399.
Hussin YA, Gilani H, van Leeuwen L, Murthy MSR, Shah R, Baral S, Tsendbazar NE, Shrestha S, Shah SK, Qamer
FM. 2014. Evaluation of object-based image analysis techniques on very high-resolution satellite image for biomass
estimation in a watershed of hilly forest of Nepal. Appl Geomatics. 6:59–68.
Ioki K, Tsuyuki S, Hirata Y, Phua M-H, Wong WVC, Ling Z-Y, Saito H, Takao G. 2014. Estimating above-ground
biomass of tropical rainforest of different degradation levels in Northern Borneo using airborne LiDAR. For Ecol
Manage. 328:335–341.
Jing L, Hu B, Noland T, Li J. 2012. An individual tree crown delineation method based on multi-scale segmentation of
imagery. ISPRS J Photogramm Remote Sens. 70:88–98.
Geocarto International  715

Karna YK, Hussin YA, Gilani H, Bronsveld MC, Murthy MSR, Qamer FM, Karky BF, Bhattarai T, Aigong X, Baniya
CB. 2015. Integration of Worldview-2 and airborne LiDAR data for tree species level carbon stock mapping in Kayar
Khola watershed. Nepal. International Journal of Applied Earth Observation and Geoinformation. 38:280–291.
Ketterings QM, Coe R, van Noordwijk M, Ambagau Y., & Palm CA. 2001. Reducing uncertainty in the use of allometric
biomass equations for predicting above-ground tree biomass in mixed secondary forests. For Ecol Manage. 146:199–
209.
Kronseder K, Ballhorn U, Böhm V, Siegert F. 2012. Above ground biomass estimation across forest types at different
degradation levels in Central Kalimantan using LiDAR data. Int J Appl Earth Obs Geoinf. 18:37–48.
Kumar P, Sharma LK, Pandey PC, Sinha S, Nathawat MS. 2013. Geospatial strategy for tropical forest-wildlife reserve
biomass estimation. IEEE J Selected Topics Appl Earth Obs Remote Sens. 6:917–923.
Latif ZA, & Blackburn GA. 2010. The effects of gap size on some microclimate variables during late summer and autumn
in a temperate broadleaved deciduous forest. International Journal of Biometeorology. 54:119–129.
Latif ZA, Blackburn GA. 2012. Forest microclimate modelling using remotely sensed data. Int Survey Res J. 2:19–25.
Latifi H, Nothdurft A, Koch B. 2010. Non-parametric prediction and mapping of standing timber volume and biomass
in a temperate forest: application of multiple optical/LiDAR-derived predictors. Forestry. 83:395–407.
Latifi H, Fassnacht F, Koch B. 2012. Forest structure modeling with combined airborne hyperspectral and LiDAR data.
Remote Sens Environ. 121:10–25.
Lefsky MA, Cohen WB, Parker GG, Harding DJ. 2002. Lidar remote sensing for ecosystem studies. Bioscience. 52:19–30.
Li M, Im J, Quackenbush LJ, Liu T. 2014. Forest Biomass and Carbon Stock Quanti fi cation Using Airborne LiDAR
Data: a case study over Huntington Wildlife Forest in the Adirondack Park. 7:3143–3156.
Lu D, Mausel P, Brondizio E, Moran E. 2002. Above-ground biomass estimation of successional and mature forests using
TM images in the Amazon Basin. Advances in Spatial Data Handling. 183–196.
Lu D, Batistella M, Moran E. 2005. Satellite Estimation Above Ground Biomass and Impact of Forest Stand. Photogramm
Eng Remote Sens. 71:967–974.
Lu D, Chen Q, Wang G, Liu L, Moran E. 2014. A survey of remote sensing-based aboveground biomass estimation
methods in forest ecosystems. Int J Digital Earth. 37–41.
Lu X, Guo Q, Li W, Flanagan J. 2014. A bottom-up approach to segment individual deciduous trees using leaf-off lidar
point cloud data. ISPRS J Photogramm Remote Sens. 94:1–12.
Lumbres RIC, Lee YJ. 2014. Aboveground biomass mapping of La Trinidad forests in Benguet, Philippines, using Landsat
Thematic Mapper data and k -nearest neighbor method. Forest Sci Technol. 10:104–111.
Mascaro J, Asner GP, Knapp DE, Kennedy-Bowdoin T, Martin RE, Anderson C, Chadwick KD. 2014. A tale of two
“Forests”: Random Forest machine learning aids tropical Forest carbon mapping. PLoS One. 9:12–16.
Mbaabu PR, Hussin YA, Weir M, Gilani H. 2014. Quantification of carbon stock to understand two different forest
management regimes in Kayar Khola watershed, Chitwan, Nepal. J Indian Soc Remote Sens. 42:745–754.
McRoberts RE, Tomppo EO, Finley AO, Heikkinen J. 2007. Estimating areal means and variances of forest attributes
using the k-Nearest Neighbors technique and satellite imagery. Remote Sens Environ. 111:466–480.
Meyer V, Saatchi SS, Chave J, Dalling JW, Bohlman S, Fricker GA, Robinson C, Neumann M, Hubbell S. 2013. Detecting
tropical forest biomass dynamics from repeated airborne lidar measurements. Biogeosciences. 10:5421–5438.
Mitchard ETA, Feldpausch TR, Brienen RJW, Lopez-Gonzalez G, Monteagudo A, Baker TR, Lewis SL, Lloyd J, Quesada
CA, Gloor E, et al. 2014. Markedly divergent estimates of Amazon forest carbon density from ground plots and
satellites. Glob Ecol Biogeogr. 23:935–946.
Mongus D, Žalik B. 2015. An efficient approach to 3D single tree-crown delineation in LiDAR data. ISPRS J Photogramm
Remote Sens. 108:219–233.
Morel AC, Saatchi SS, Malhi Y, Berry NJ, Banin L, Burslem D, Nilus R, Ong RC. 2011. Estimating aboveground biomass
in forest and oil palm plantation in Sabah, Malaysian Borneo using ALOS PALSAR data. For Ecol Manage. 262:1786–
1798.
Neigh CSR, Nelson RF, Ranson KJ, Margolis HA, Montesano PM, Sun G, Kharuke V, Næssetf E, Wulderg MA, Andersen
H-E. 2013. Taking stock of circumboreal forest carbon with ground measurements, airborne and spaceborne LiDAR.
Remote Sens Environ. 137:274–287.
Ngo KM, Turner BL, Muller-Landau HC, Davies SJ, Larjavaara M, Nik Hassan NFB, Lum S. 2013. Carbon stocks in
primary and secondary tropical forests in Singapore. For Ecol Manage. 296:81–89.
Patenaude G, Hill R, Milne R, Gaveau DLA, Briggs BBJ, Dawson TP. 2004. Quantifying forest above ground carbon
content using LiDAR remote sensing. Remote Sens Environ. 93:368–380.
Pirotti F. 2011. Analysis of full-waveform LiDAR data for forestry applications: a review of investigations and methods.
IForest. 4:100–106.
Popescu SC, Wynne RH, Nelson RF. 2003. Measuring individual tree crown diameter with lidar and assessing its influence
on estimating forest volume and biomass. Can J Remote Sens. 29:564–577.
Popescu SC, Zhao K, Neuenschwander A, Lin C. 2011. Satellite lidar vs. small footprint airborne lidar: comparing the
accuracy of aboveground biomass estimates and forest structure metrics at footprint level. Remote Sens Environ.
115:2786–2797.
716 N. A. MOHD ZAKI AND Z. ABD LATIF

Propastin P. 2012. Modifying geographically weighted regression for estimating aboveground biomass in tropical
rainforests by multispectral remote sensing data. Int J Appl Earth Obs Geoinf. 18:82–90.
Rana P, Korhonen L, Gautam B, Tokola T. 2014. Effect of field plot location on estimating tropical forest above-ground
biomass in Nepal using airborne laser scanning data. ISPRS J Photogramm Remote Sens. 94:55–62.
Reitberger J, Schnörr C, Krzystek P, Stilla U. 2009. 3D segmentation of single trees exploiting full waveform LIDAR
data. ISPRS J Photogramm Remote Sens. 64:561–574.
Roy PS, Ravan SA. 1996. Biomass estimation using satellite remote sensing data – an investigation on possible approaches
for natural forest. J Biosci. 21:535–561.
Rutishauser E, Noor’an F, Laumonier Y, Halperin J, Hergoualc’h K, Verchot L. 2013. Generic allometric models including
height best estimate forest biomass and carbon stocks in Indonesia. For Ecol Manage. 307:219–225.
Saatchi SS, Houghton RA, Dos Santos Alvalá RC, Soares JV, Yu Y. 2007. Distribution of aboveground live biomass in
the Amazon basin. Glob Change Biol. 13:816–837.
Shimamoto CY, Botosso PC, Marques MCM. 2014. How much carbon is sequestered during the restoration of tropical
forests? Estimates from tree species in the Brazilian Atlantic forest. For Ecol Manage. 329:1–9.
Sousa AMO, Cristina A, Mesquita P, Marques JR. 2015. Biomass estimation with high resolution satellite images: A case
study of Quercus rotundifolia. ISPRS J Photogramm Remote Sens. 101:69–79.
Stibig H-J, Achard F, Carboni S, Raši R, Miettinen J. 2014. Change in tropical forest cover of Southeast Asia from 1990
to 2010. Biogeosciences. 11:247–258.
Sun G, Ranson KJ, Guo Z, Zhang Z, Montesano P, Kimes D. 2011. Forest biomass mapping from lidar and radar synergies.
Remote Sens Environ. 115:2906–2916.
Tonolli S, Dalponte M, Neteler M, Rodeghiero M, Vescovo L, Gianelle D. 2011. Fusion of airborne LiDAR and satellite
multispectral data for the estimation of timber volume in the Southern Alps. Remote Sens Environ. 115:2486–2498.
UNFCC. 2015. UNFCC. [cited 2015 May 20]. Available from: http://unfccc.int/kyoto_protocol/items/2830.php
Van der Laan C, Verweij PA, Quiñones MJ, Faaij AP. 2014. Analysis of biophysical and anthropogenic variables and
their relation to the regional spatial variation of aboveground biomass illustrated for North and East Kalimantan.
Borneo Carbon Balance Manage. 9:1–12.
Vega C, Hamrouni A, El Mokhtari S, Morel J, Bock J, Renaud J-P, Bouvier M, Durrieu S. 2014. PTrees: a point-based
approach to forest tree extraction from lidar data. Int J Appl Earth Obs Geoinf. 33:98–108.
Viana H, Aranha J, Lopes D, Cohen WB. 2012. Estimation of crown biomass of Pinus pinaster stands and shrubland
above-ground biomass using forest inventory data, remotely sensed imagery and spatial prediction models. Ecol
Model. 226:22–35.
Vincent G, Sabatier D, Blanc L, Chave J, Weissenbacher E, Pélissier R, Fonty E, Molino JF, Couteron P. 2012. Accuracy
of small footprint airborne LiDAR in its predictions of tropical moist forest stand structure. Remote Sens Environ.
125:23–33.
Vincent G, Sabatier D, Rutishauser E. 2014. Revisiting a universal airborne light detection and ranging approach for
tropical forest carbon mapping: scaling-up from tree to stand to landscape. Oecologia. 175:439–443.
Vosselman G, Maas H-G. 2010. Airborne and terrestrial laser scanning. Scotland: Whittles Publishing.
Wang C, Qi J. 2008. Biophysical estimation in tropical forests using JERS-1 SAR and VNIR imagery. II. Aboveground
woody biomass. Int J Remote Sens. 29:6827–6849. doi:http://dx.doi.org/10.1080/01431160802270123
West PW. 2009. Tree and Forest Measurement. Berlin, Heidelberg: Springer-Verlag.
Wulder MA, White JC, Nelson RF, Næsset E, Ørka HO, Coops NC, Hilker T, Bater CW, Gobakken T. 2012. Lidar sampling
for large-area forest characterization: A review. Remote Sens Environ. 121:196–209.
Yan F, Wu B, Wang Y. 2015. Estimating spatiotemporal patterns of aboveground biomass using Landsat TM and MODIS
images in the Mu Us Sandy Land, China. Agric For Meteorol. 200:119–128.
Zheng G, Chen JM, Tian QJ, Ju WM, Xia XQ. 2007. Combining remote sensing imagery and forest age inventory for
biomass mapping. J Environ Manage. 85:616–623.
Zhu X, Liu D. 2015. Improving forest aboveground biomass estimation using seasonal Landsat NDVI time-series. ISPRS
J Photogramm Remote Sens. 102:222–231.

View publication stats

You might also like