-B3, 2012
try and Remote
N., 2009. A
lding extraction
"Selected Topics
sing 2(1):11-20.
_IDAR data and
description. In:
0. DGPF project:
1era systems -
nerkundung —
P., Koehl, M.,
es using LIDAR
al Archives of
ial Information
2-1750, pp. 87-
3 reconstruction
Congress of the
Remote Sensing
med, M., 2011.
gration of Lidar
Archives of
jal Information
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXIX-B3, 2012
XXII ISPRS Congress, 25 August — 01 September 2012, Melbourne, Australia
RETRIEVING SURFACE SOIL MOISTURE FROM MODIS AND AMSR-E DATA: À
CASE STUDY IN TAIWAN
C.F. Chen ***, Y J. Lin", LY. Chang, N T. Son?
* Center for Space and Remote Sensing Research — (cfchen, lychang, ntson)@csrsr.ncu.edu.tw
® Department of Civil Engineering - eugenelin151121@yahoo.com.tw
National Central University, Jhongli, Taiwan
Commission III, WG III/5
KEY WORDS: MODIS data, Surface soil moisture, Taiwan.
ABSTRACT:
Soil moisture is a key factor that controls the exchange of water between land surface evaporation and plant transpiration. Information
on surface soil moisture variations in both time and spatial domains is important for numerous applications, especially agricultural and
environmental monitoring. This study aimed at retrieving surface soil moisture from daily MODIS and AMSR-E (Advanced
Microwave Scanning Radiometer - Earth Observing System) data. A case study was conducted in Taiwan for 2009. Data were
processed using the Temperature Vegetation Dryness Index (TVDI). This index is developed based on an empirical analysis of the
relationship between land surface temperature (LST) and normalized difference vegetation index (NDVI) data. The comparison
between the TVDI results and the daily precipitation data collected from meteorological stations throughout the study area indicated
that there were close relationships between the two datasets. The TDVI results (values range from 0 to 1) were converted to the same
unit with the AMSR-E soil moisture data (i.e., g cm™) by linear regression analysis between these two datasets. The results achieved by
this analysis were soil moisture maps that had a better spatial resolution (1 km x 1 km) than the AMSE-E soil moisture data (25 km x 25
km). The soil moisture achieved by TVDI — AMSR-E regression analysis showed the comparable spatial patterns with those from the
AMSR-E soil moisture data. A quantitative analysis between the soil moisture (deduced from TVDI-AMSR-E analysis) and the
AMSR.E soil moisture data also reaffirmed significant correlations between the two datasets. This study has demonstrated a method of
surface soil moisture retrieval from MODIS and AMSR-E data.
1. INTRODUCTION
Information on surface soil moisture conditions is important for
water management for specific crops. This parameter influences
the transpiration rate of crops (Kramer, 1969). Thus, monitoring
soil moisture is an important activity because agricultural
planners need such information to devise better water and crop
management strategies for agricultural development. Surface soil
moisture can be monitored using climatic indices such as Palmer
drought severity index (PDSI) (Palmer, 1965) and standardized
precipitation index (SPI) (McKee et al., 1993). The use of these
indices for regional soil moisture monitoring is infeasible
because the data are usually insufficient over a large area and the
cost of data acquisition is expensive.
Satellite-based indices have been developed and used for
large-scale monitoring of soil moisture in the top few centimeter
of soil. For instance, AMSR-E (Advanced Microwave Scanning
Radiometer-Earth Observing System onboard Aqua satellite)
index (g cm?, 25-km resolution) can be used for global soil
moisture monitoring. However, this index is coarse and
unsuitable to be applied for the monitoring purpose in Taiwan.
The temperature vegetation dryness index (TVDI) developed
based on the empirical interpretation of the relationship between
the normalized difference vegetative index (NDVI) and land
surface temperature (LST) has widely been applied for soil
moisture monitoring (Patel et al., 2008; Sandholt et al., 2002;
Sun et al., 2008). The NDVI and LST were used to derive TVDI
because NDVI characterizes the greenness of vegetation
indicating water stress, whilst LST reflects soil moisture. Thus,
the combination of these two NDVI and LST can provide more
complete information on soil moisture (Carlson, 2007).
In this study, we aimed to develop an approach to retrieve
surface soil moisture (g cm?) from MODIS data for Taiwan. The
TVDI results were verified with precipitation data.
* Corresponding author.
2. STUDY AREA
The study area (Taiwan) covers approximately 36,000 km? and
lies between 23?46'N 121?00'E (Figure 1). The elevation ranges
from 0 to 3,952 m above the sea level. The study area has
subtropical monsoonal climate with the mean annual
precipitation of approximately 2,500 mm. The hottest month is
July with the mean temperature of 27 - 28 °C and the coolest
month is February (15 °C).
High: 3952
Figure 1. The study area showing the elevation levels and
locations of two rain gauge stations used for verification of the
TDVI results.
