An Extended Approach for Biomass Estimation in a
Mixed Vegetation Area Using ASAR and TM Data
Minfeng Xing, Binbin He, Xingwen Quan, and Xiaowen Li
Abstract
The use of microwave remote sensing for estimating vegeta-
tion biomass is limited in arid regions because of the het-
erogeneous distribution of vegetation, variable scattering
mechanisms from different vegetation components, and the
strong influence from underlying ground surface. In order to
minimize this problem, a synergistic method of optical and
microwave remote sensing data for the retrieval of aboveg-
round biomass (
AGB
) based on the modified water cloud
model (
WCM
) was developed in this paper. Vegetation cover-
age which can be easily estimated from optical data as ad-
ditional information was combined in this method. Dimidiate
pixel model (
DPM
) and phenological subtraction methodology
(
PSM
) were used to estimate vegetation coverage and differen-
tiate vegetation types in the sub-pixel domain, respectively.
The percentage cover of unmixed vegetation was incorporated
to minimize problems associated with heterogeneous vegeta-
tion and sparse vegetation cover. Finally, the accuracy and
sources of error in this novel
AGB
retrieval method were evalu-
ated. The results showed that the predicted
AGB
correlated
with the measured
AGB
(R
2
= 0.8007;
RMSE
= 0.2808 kg/m
2
).
Introduction
Grasslands are important ecosystems in arid regions. As a
significant portion of a dry region is covered with grassland, a
monitoring program focused on grassland dynamics is highly
desirable. Vegetation biomass is an important indicator of
ecosystem health because it reflects the ability of that eco-
system to obtain energy. Thus, using biomass as an indicator
for monitoring a given grassland is of great importance for
understanding the current status of that environment. Satel-
lite remote sensing provides a uniquely effective and efficient
means of monitoring and assessing vegetation biomass.
Although biomass cannot be directly measured from space,
remote sensing information extracted from both optical and
synthetic aperture radar (
SAR
) (e.g., the reflectance of optical
remote sensing or the backscattering coefficient of
SAR
) can
be related to biomass. In previous studies, vegetation biomass
was effectively estimated through both optical (Clevers
et
al
., 2007; Coppin
et al
., 2001; Liu
et al
., 2010; Lu
et al
., 2005;
Muukkonen and Heiskanen, 2005; Phua and Saito, 2003;
Zheng
et al
., 2004) and
SAR
remote sensing data (Englhart
et
al
., 2011; Li and Potter, 2012; Ni
et al
., 2013; Tsolmon
et al
.,
2002; Wang and Ouchi, 2008 and 2010). However, there are
limitations to the estimation of biomass with optical or mi-
crowave remote sensing data alone. Visible and near infrared
wavelengths are easily scattered or absorbed within the plant
canopy. Optical remote sensing is also limited in detecting
woody vegetation structure (Englhart
et al
., 2011; Li and
Potter, 2012). Therefore, aboveground biomass (
AGB
) estima-
tion using optical remote sensing is hindered by a lack of veg-
etation structure. In addition, monitoring vegetation biomass
using optical data is affected by signal saturation for leaf area
index (
LAI
) values higher than 2 (Shoshany, 2000). A radar
image is capable of penetrating a vegetative canopy (Wong
and Fung, 2013), and consequently, this image captures
SAR
backscattering from the surface of the canopy, the sub-surface,
and the ground surface. Because ground surface scattering
can affect
AGB
estimation from
SAR
, the synergistic use of both
optical and
SAR
data could potentially overcome problems
inherent to using optical or
SAR
data alone. For example,
previous research (Svoray and Shoshany, 2002 and 2003) has
been conducted in a humid to semi-arid transition region that
combined optical and
SAR
data. In addition, other studies
(e.g., Amini and Sumantyo, 2009; Chen
et al
., 2009; Häme
et
al
., 2013; Moghaddam
et al
., 2002; Wang and Qi, 2008) have
suggested that the accuracy of biomass estimates was signifi-
cantly improved when optical and radar data were combined
compared to estimates from a single data type alone. It was
also reported that using a synergistic approach with opti-
cal and microwave data could reduce canopy effects on
SAR
signals (Wang
et al
., 2004) and create the possibility to extend
the range of validity of the biomass estimates (Moghaddam
et al
., 2002).The effects of vegetation on
SAR
backscattering
are controlled by vegetation’s biophysical parameters (e.g.,
vegetation coverage;
LAI
), which can be determined through
optical remote sensing data. The vegetation’s biophysical
parameters could be used to quantify canopy attenuation to
SAR
signals in the radiative transfer functions model and thus
extend the threshold of signal saturation (Wang and Qi, 2008).
The objective of this study was to develop a synergistic
method for applying optical and microwave remote sensing
data to estimate
AGB
in a region with mixed vegetation. The
coverage of different vegetation types was delineated using a
dimidiate pixel model (
DPM
) (Gutman and Ignatov, 1998) and
the phenological subtraction methodology (
PSM
) (Shoshany
and Svoray, 2002) with optical datasets. The contribution of
radar backscattering from different vegetation types was then
quantified and incorporated into the coverage data. This meth-
odology was ultimately applied to a case study in a mixed veg-
etation area of Wutumeiren Prairie, Qinghai Province, China.
School of Resources and Environment, University of Electronic
Science and Technology of China, No. 2006, Xiyuan Ave, West
Hi-Tech Zone, Chengdu, China, 611731 (
).
Photogrammetric Engineering & Remote Sensing
Vol. 80, No. 5, May 2014, pp. 429–438.
0099-1112/14/8005–429
© 2014 American Society for Photogrammetry
and Remote Sensing
doi: 10.14358/PERS.80.5.429
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
May 2014
429
