is lower than that for trees even the area is fully covered
by vegetation. The variation for SWI is considered to be
caused by the difference of surface roughness, which is
one of the dominant factors affecting SAR backscatter.
3.SEASONAL CHANGES OF SAR BACKSCATTER
IN PADDY FIELD
The authors studied the actual changing patterns of SAR
backscatter concerning to the change of land cover
conditions in the paddy field at the same test site as Fig. 1.
The flow of data analysis is shown in Fig.5. The three
temporal SAR data in Fig.1 were co-registered each other
and a MAP filter was applied in order to remove speckle
noise. After principal component analysis was performed
to the three SAR images, clustering was adopted as
non-supervised classification method to obtain land cover
classification image at the test site. As the result of
clustering, total eight land cover classes were obtained.
The land cover classification image is shown in Fig.6
The graph in Fig.7 shows the changing patterns of NRCS
derived from the SAR standard product by NASDA for
eight land cover classes obtained by above analysis.
The solid lines (C2, C3, C4, and C6) in the graph
correspond to the patterns of paddy fields. The change
of C4 and C6 between March and September was proved
to be caused by the difference between bare soil condition
and the rice growth from the ground survey. The change
of C2 was considered due to the planting of second rice in
the dry season. The change of C3 between March and
April was estimated due to the change of soil moisture
according to the result of regression analysis described in
the previous section. Natural levee which distributes
widely in the test site was easily identified as the class C8,
which always resulted in high backscatter.
As described above, in even L-band SAR images, much
information about the seasonal changes of paddy and soil
condition can be extracted from multitemporal data.
These results seem to suggest the capability of JERS-1
SAR data for actual monitoring of land cover conditions in
tropical regions.
4.CHANGES OF SAR BACKSCATTER
BY FLOODING
The authors studied the change of SAR backscatter in
paddy fields when they were flooded. Fig.8 shows two
SAR images observed in the rainy season in 1993 and
684
1995. The right image shows the SAR image taken on
September 26th, 1995 when a big flood occurred in the
Central Plain of Thailand. This image can be compared
with the left image, which was taken almost in the same
rainy season but in non-flood condition.
According to a ground survey at the same time of SAR
observation in 1995, the SAR data were proved to result
in very low backscatter in the flooded areas as shown in
Fig.9. Therefore, the flooded areas in the paddy fields
were extracted by subtracting NRCS values from those of
the data in non-flooded conditions, namely the data in
September, 1993. According to all weather character-
istics of SAR, the SAR data are considered to be effective
data source for the detection of flood in the rainy season.
5.CONCLUSION
Regression analysis using JERS-1 SAR and Landsat TM
data acquired on the same day suggested that the change
of L-band SAR backscatter is related to the change of soil
moisture and vegetation coverage conditions with
positive correlation. Case studies concerning to the
change of land cover in paddy fields by seasonal variation
and flooding proved that these changes can be detected
as the change of L-band SAR backscatter. Therefore
the result of these case studies indicates the effectiveness
of L-band SAR data by JERS-1 for the monitoring of
several types of changes of land cover conditions in
tropical regions, where data acquisition is necessary even
in bad weather conditions of the rainy season.
This study was conducted as the joint research between
RESTEC and NRCT and also supported by the Special
Coordination Fund of the Science and Technology
Agency of Japan.
REFERENCES
Takeuchi,S. and R. Suwanwerakamtorn, 1995a. Analysis
of the Influence of Land Cover Conditions on SAR
Backscatter Using Simultaneous SAR and TM Data.
J. of JSPRS, vol. 34, no. 5, pp.45-48.
Takeuchi,S. and R. Suwanwerakamtorn, 1995b. Land
Cover Change Analysis Using Multitemporal SAR Data
- A Case Study in the Central Plain of Thailand -.
Proceedings of the 19th Japanese Conference on
Remote Sensing, pp.113-114.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B7. Vienna 1996