You are on page 1of 4

International Journal of Remote Sensing Applications Volume 3 Issue 4, December 2013

doi: 10.14355/ijrsa.2013.0304.07

www.ijrsa.org

The Retrieval of Moisture and Temperature


Index Based on ASTER Data
Jinming Sha *1 Xiaomei Li 2 Bin Xiao 3
China-Europe Center for Environment Landscape Management Fujian Normal University
*1jmsha@fjnu.edu.cn, 2Lixiaomei@fjnu.edu.cn, 3xb@163.com
Abstract
Soil moisture and land surface temperature are the
important factors in the physical process of the regional and
global surface. Accordingly, multiple thermal infrared
sensors have been sent into the outer space to monitor the
hydrothermal factor of the global or regional area
dynamically. This research used ASTER L1B data transited
on the morning of May 23rd, 2001. L 1B data is based on L
1A data, produced by using its own parameters to finish
radiometric correction as well as geometric correction. In the
case of default, UTM projection is utilized. This research was
performed in a region of Fuzhou. Because of the high spatial
and spectral resolution of ASTER data, at the same time,
with the help of MODIS data and split-window algorithm,
the authors retrieved the moisture and temperature index in
the region of complex topography.
Key words
ASTER; Land Surface Temperature; VSWI; Emissivity

Introduction
Soil moisture and land surface temperature are the
important factors in the physical process of the
regional and global surface. Thermal infrared remote
sensing is one of the important means to study the
thermal radiation characteristics as well as soil
moisture monitoring. Accordingly, multiple thermal
infrared sensors have been sent into the outer space to
monitor the hydrothermal factor of the global or
regional area dynamically. In 1999, the earth
observation satellite (Terra), equipped with ASTER
and MODIS sensors, was successful launched. This
success opened up new avenues for the dynamic
monitoring of the global and regional resources and
environment. It has more thermal infrared bands and
higher spectral resolution compared with TM or ETM.
Also, its utilization potentialities of surface emissivity,
temperature retrieval etc. are reflected more and more
fully. There are five high-resolution bands among the
fifteen thermal infrared bands which are very suitable
to study the surface hydrothermal factor of cities or
small regions.

Collection of Data Material and Treatment


This research area is Fuzhou parts, where the latitude
range is from 254934.02N to 262811.86N and the
longitude range is from 1185018.61E to
1193516.82E. This research used ASTER L1B data
transited on the morning of May 23rd, 2001. L 1B data
is based on L 1A data, produced by using its own
parameters to finish radiometric correction as well as
geometric correction. The necessary pretreatments are
calibration processing, Bowtie handling, geometric
correction and the revision of solar altitude angle.
Radiant remittances correction is to transform the
enumerated data observed from satellites into
available physical values, namely radiometric
calibration. Regarding the reflection channels, the
calibration results are the reflectance values of pixels:
R=reflectan_scale (DN-reflectance offsets)
In the above formula, R is the reflectivity after
calibration; DN refers to the observed value gained
from satellites; Reflectance scale indicates the
pantograph ratio or yield value, reflectance offsets is
zoom intercept or deviation value. These parameters
can be searched from the header files of 1B data.
As to the research, the calibration results are the
radiant brightness values of pixels.
L=radiance scales (DN-radiance offsets)
In the above expression, L is the emissive after
calibration; radiance scales is the pantograph ratio or
gain coefficient of the radiation channels; and
Radiance _offsets is the zoom intercept or offset
coefficient of the radiation channels. These parameters
can be searched from the header files of 1B data.
The Retrieval of LST
This study utilized the split-window algorithm
suggested by MaoKebiao etc. to retrieve the LST of the
study area. Among the parameters of this algorithm,
the atmospheric transmissivity and the land surface
emissivity are two key ones. Next, this study would
225

www.ijrsa.org

International Journal of Remote Sensing Applications Volume 3 Issue 4, December 2013

calculate the necessary parameters first.

Land Surface Emissivity

Atmospheric Transmissivity

Land surface emissivity, the key parameter for the


retrieval of LST, is mainly determined by the structure of
land surface as well as the wavelength, and it is usually
assumed known in the split-window algorithm.
Through analyzing the emissivity in the ASTER spectral
library, it was found that the emissivity of most surface
features in the 8-14 m wavelength range changed
slightly and was all above 0.9. This research used NDVI
(Normalized Difference Vegetation Index) which was
calculated by ASTER Visible Light/Near-Infrared. Then
the study area was classified briefly according to the
emission characteristics of the various surface features in
the visible light/near-infrared band. Finally, the
corresponding emissivity was given based on the
ASTER spectral library, which was listed in Table 1.

Atmospheric transmissivity is the key parameter for


the retrieval of LST. Numerous researchers have
raised lots of effective estimation methods for different
conditions. Atmospheric modeling software (6S,
MODTRAN, LOWTRAN etc.) is usually applied to
analog the relationship between transmissivity and the
water content of atmosphere. Then the atmospheric
transmissivity is estimated by reckoning the water
content of atmosphere. According to the fact that the
near-infrared band of MODIS shows diverse features
for different water content of atmosphere, this study
adopted 2-band ratio method to estimate the water
content of atmosphere, thus the atmospheric
transmissivity of the thermal infrared band was
obtained. The 2-band ratio method is expressed as
follows:
Tabs (0.940m) = (0.940m)/ (0.865m)
Kaufman and Gao provided the following expression
by contrasting the relationship between the values and
the water content of atmosphere:
Tw(940/865) = exp(-w1/2)

R2 = 0.999

For complex earth surface, 0.02, 0.651,then :


W = [(- ln Tw)/ ]2
This study used the ratio of band2 and band19 of
MODIS to estimate the water content of atmosphere.
As the resolutions didnt match, it was needed to
resample the image of the water content of
atmosphere to 90 meter so that MODIS data matched
ASTER thermal infrared band data.
MaoKebiao etc. (2006) exploited MODTRAN to
simulate the relationship between the atmospheric
transmissivity of ASTER thermal infrared band and
the water content of atmosphere, as well built their
relationship. The linear relationship between the
atmospheric transmissivity of ASTER band13 or
band14 and the water content of atmosphere is shown
as below:
For band 13: =1.02-0.104 W

R2=0.985

For band 14: =1.04-0.133 W

R2=0.9915

In the above formula, is the atmospheric


transmissivity of a band, w is the corresponding water
content of atmosphere.
With this expression and the above estimated water
content of atmosphere, the atmospheric transmissivity
of this study area was gained.
226

TABLE 1 LAND SURFACE CLASSIFICATION AND EMISSIVE SETTING


Open
Woodforest
lands
/Grassland
NDVI
<0.0 0.0-0.1 0.1 0.35 0.350.70 0.70.85
0.91.0
Emissive 0.9927 0.9236
0.9426
0.9537
0.9752
0.9853
Setting

Water

Bare ConstrucArea tion Zone

Mixing
Zone

According to the emissive and land types set in Table1,


the emissive map was gained through ENVI software.
Because the resolution of this emissive map is 15 meter,
it was required to resample it to 90 meter in order to
match the thermal infrared image. The emissive image
is showed in Figure.
Retrieval of LST
So far, given the influences of factors such as region
features and data and so on, the proposed algorithm
models are listed as follow: the single-window
algorithm which is applied for LANDSAT TM/ETM
MODIS and ASTER data, the split-window algorithm
which is employed for NOAA/AVHRR,MODIS and
ASTER data, the split-window algorithm and
day/night algorithm for MODIS and the algorithms
based on them, for example, TanZhihao etc.(2004)
suggested split-window algorithm and single-window
algorithm for MODIS. MaoKebiao etc. (2005) proposed
split-window algorithm, single-window algorithm and
four-channel algorithm for ASTER.
This issue will adopt the split-window algorithm for
ASTER, provided by Tan Zhihao, MaoKebiao etc., to
quantify the hydrothermal conditions of the remote
sensing monitoring areas. The LST of study area will
be retrieved using the gained emissive and the
simulated transmission rate.
Ts=(C14*(D13+B13)C13*(D14+B14))/(C14*A13C13*A
14)

International Journal of Remote Sensing Applications Volume 3 Issue 4, December 2013

The LST of this study area was got through by


calculating
the
emissivity
and
atmospheric
transmissivity, which was shown in Figure.

FIGURE 1 DENSITY SPLIT PLOT OF TS

FIGURE 2 DENSITY SPLIT PLOT OF VSWI

Retrieval of Soil Moisture


So far, the remote sensing quantitative monitoring
industry for soil moisture has made great progress.
Domestic and foreign experts and scholars have
proposed many algorithms and models for remote
sensing retrieving soil moisture. To sum up, the
algorithms can be sorted into the following types: visible
light-near infrared method, thermal inertia method,
temperature-vegetation index method, microwave
method. Each method or model is established and
developed under specific conditions and certain district.
Consequently, on one hand, it has some dominants and
properties, on the other hand, it leads to some
limitations and regional applicability inevitably.
According to the natural geographical conditions of
the study area and taking the traits of each remote
sensing algorithm and model for soil moisture into
account, this research will use temperature-vegetation
index method to realize the remote sensing monitoring
of the water environment in this study area. Under the
condition that the vegetation cover is not complete, if

www.ijrsa.org

the LST, obtained by remote sensing, is taken as an


indicator alone to monitor the retrieval of soil
moisture, and then the high soil background
temperature will affect the retrieval process and result
comparatively great, even leading to distortion.
Vegetation index can provide the growth status of
vegetation and coverage information. Accordingly,
combining the spectral vegetation index with LST
information will not only realize the retrieval of soil
moisture, but also eliminate the effects of soil
background. Domestic and foreign scholars have
researched the relationship between the LST and
vegetation index in various spatial and temporal scales
and have found that there were significant negative
correlation between them. They obtained the feature
space which consists of Ts and NDVI through remote
sensing technique. Besides, they tried various methods to
retrieve soil moisture and built the integrated application
models which were applied for drought monitoring.
Vegetation water index is sensitive to the water
content of regional soils as well as the water
conditions of crops. Jiangdong etc. (2001) utilized the
ratio of vegetation index and land surface temperature,
namely NDVI/Ts, to be the indicator of the surface
water conditions, especially the water stress of crops.
NDVI reflects the vegetation growth conditions in
different periods, and Ts implies canopy temperature.
NDVI/Ts index combines soil water with crops growth
condition to highlight the thoughts of Crop Available
Water (or Effective Water). As a result, the vegetation
water index is defined as:
VSWI = NDVI/Ts
NDVI = (CH2 CH1)/(CH2 + CH1)
In the above formula, Ts indicates canopy temperature
of vegetation, valued approximately by LST. NDVI is
short for Normalized Difference Vegetation Index.
CH2 and CH1 are the reflectance ratios of
Near-infrared band and visible light-red band of
ASTER respectively. If there is sufficient water content
in soil, then crops will grow fast and well and the
NDVI value will be high, in this case, it will lead to
strong crop transpiration, reduction of evaporationimpedance of the whole pixel, augment in the
proportion of latent heat and the fall of pixel surface
temperature, so the VSWI value will increase.
Conversely, the VSWI value will decrease.
Liuli etc. (1998, 1999) made use of the meteorological
satellite materials and vegetation index method to
build the drought monitoring models for Guizhou
Province, then determined VSWI remote sensing
drought index and arid area, and contrasted them
227

www.ijrsa.org

International Journal of Remote Sensing Applications Volume 3 Issue 4, December 2013

with the drought shown by the land surface drought


index. Furthermore, the regression equation of VSWI
and the land surface drought index are established.
Summary
According to the above retrieval results, the general
scope of the temperature and humidity of each surface
features in this phase were obtained.
(1)It is a feasible experiment to receive the atmospheric
transmissivity, surface emissivity and radiation
measurements based on ASTER and MODIS data and
retrieve the hydrothermal parameters of this study
area with complex terrain by using split-window
algorithm and based on ASTER and MODIS data
(2)It can be concluded from the retrieval result that
both the LST and soil relative humidity reflect the
general trend of its changes with elevation. LST
decreased with elevation, and this phenomenon was
caused by the coupling between temperature and land
surface. Temperature was reduced with height. This
study area belongs to the monsoon climate distinct, so
the precipitation of northwest and southeast districts
are more than that of east and northeast regions, hence,
the general soil relative humidity distributes similarly.
(3)Soil humidity distribution mainly depends on LST
and vegetation coverage, whose change varies in
different districts. If the vegetation coverage rate of a
region is relatively higher, then the soil humidity will
be affected by NDVI mainly. If the vegetation
coverage rate of a region is zero or relatively lower,
then the soil humidity will be affected by LST mainly,
namely the influence of solar radiation. When the LST
is high, strong evaporation will take place in soil so
that the soil humidity will decline relatively in this
area.
(4)In this study, because of the impact of various
conditions, there was a certain error needed to be
verified by simultaneous observation data of land
surface. Moreover, the algorithms of radiometric
calibration and atmospheric transmissivity needed to
be improved. The emissivity was also difficult to be
gained precisely. The ground resolution of ASTER
data is relatively low so that it affected the precision of
retrieval of hydrothermal index directly. The
circumstances of the changes with elevation of surface
temperature and soil moisture, along with the
relationship between soil moisture and actual relative
soil moisture content should be further studied and
improved in the future researches.

228

ACKNOWLEDGEMENTS

This paper has been supported by the key subject of


MOST and Marie Curie Actions "Integrated
geo-spatial information technology and its application
to resource and environmental management towards
the GEOSS ";Fujian Haixia resources and environment
key project Research on building eco-civilization
Fujian Surveying and GIS bureau Project Monitoring
Soil Erosion
REFERENCES

Denghui. Research on remote sensing monitoring of soil


moisture in large area based on MODIS data (D).
Resources

Division

of

the

Chinese Academy

of

Agricultural Sciences, 2004


Jiang Dong etc. Ecological Connotation and Application of
the Vegetation Index-Surface Temperature Feature Space
[J]. Progress in Geography, 2001, 20 (2)
Liu Li, Liu Qing. The application of satellite remote sensing
information in the drought monitoring in Guizhou [J].
Chinese Agricultural Meteorology, 1999, 20 (3)
Liu Yu-jie, YANG Zhong-dong etc.. The Principles and
Algorithms of remote sensing information processing
based on MODIS [M]. Science Press, 2001
Mao Ke-biao, Tang Hua-jun etc. A Split-window Algorithm
for Retrieving Land-Surface Temperature from ASTER
Data [J].Remote Sensing Information, 2006
Ran Qiong etc.. DEM correction using TVDI to evaluate soil
moisture in China [J].SCIENCE OF SOIL AND WATER
COSERVATION, 2005, 3 (2):32-36
Wang Feng-min, Tian Qing-jiu etc. Retrieving land surface
temperature of mountain areas in southern China based
on ASTER data [J].Remote sensing for land & resources,
2005, 1 (63)
Wang Jun-hua, Huang Yong-lin. The progress of Study on
Retrieving Land Surface Temperature (LST) By Using
MODIS Data in Our Country [J]. Journal of Guangxi
Meteorology, 2005, 26 (4)
Yang Qing-sheng, Liu Chuang. Retrieving Land Surface
Temperature from MODIS Data [J].Remote Sensing
Technology and Application, 2004, 19 (2)
Li Xiaomei, Dr., Associated Professor, correspondence
Lixiaomei@fjnu.edu.cn.

You might also like