IAPRS & SIS, Vol.34, Part 7, “Resource and Environmental Monitoring”, Hyderabad, India, 2002 
  
  
4.5 Image registration 
IRS LISS III image was geo-referenced using the GCPS from 
1:50,000 scale topographic map. The SAR images were co- 
registered with this base image using affined transformation 
and resampling. The road/rail, canal/drainage networks and 
ground truth locations were digitized and transformed to the 
image. 
4.6 Signature Extraction 
Backscattering coefficient values for sampling locations were 
extracted from SAR images. IRS LISS-III data was classified 
using ground truth information to delineate the rice area. The 
area was categorized based on crop growth as early and late 
rice. Normalized Difference Vegetation Index (NDVI) image 
of the rice pixels was generated using the red and NIR bands of 
LISS III data. A density slicing was applied to stratify the rice 
area NDVI as low, medium and high. 
4.7 Soil moisture retrieval and validation 
An existing validated soil moisture model derived using 
extended low data for rice/fallow fields was used to retrieve soil 
moisture of rice fields. A soil moisture map of rice area was 
derived by density slicing method. The results were validated 
using the moisture value of sampling sites. 
S. RESULTS 
Rice was transplanted during late January to early February. 
The crop was in grain filling to soft dough stage during end of 
March in late planting and early planting crop respectively. The 
farmers in the area followed a uniform management and 
planting practice. A single crop variety- Lal Swarn dominated 
the area which attained 4-5 Leaf Area Index (LAI). Figure 1 
shows the False Color Composite (FCC) of IRS LISS III data. 
One can observe the large contiguous rice cultivation practice 
of the area. The early sown crop appearing dark red was 
confined to the southeast part of the area. Major rice area was 
under late sown category, where the crop was in grain filling 
stage during March. This was reflected in the derived NDVI 
range of the rice fields, which was from 0.38 to 0.49. The 
average NDVI of early sown crop was around 0.40 and that of 
late sown crop was 0.45. 
The higher incidence angle data (S6) of March 23 did not show 
significant variation in backscatter among 
Figure 1. IRS-LISS-II FCC (31-Mar-2000) showing 
distribution of late and early sown rice in the study area, which 
re in grain filling and soft dough stage respectively. 
the fields. This was attributed to signal saturation due to 
uniform and dense vegetative cover of the rice fields (Figure 2). 
The total range was between —7 to —9 dB. In contrast, 
significant variation of 1.5 to —6 dB in backscatter was 
observed in the extended low data (Figure 3). The c? for the 
areas with late sown crop having standing water was -5 to -6 
dB. Fields under wet condition had high backscatter of —2 
toldB. The flooded fields could be easily delineated using this 
threshold value. 
776 
Figure 2. Radarsat SAR, Standard-6 beam mode (28-Mar- 
2000), image of the study area. Note that there is no significant 
difference in backscatter irrespective of field condition beneath 
the rice crop, A soil moisture model was developed to retrieve 
soil moisture from backscatter (Figure 4). For the rice fields, 
the derived soil moisture from o° values was segmented in to 
two levels as saturated and high soil moisture as shown in 
Figure 5. The non-rice areas have been masked out. 
Figure 6 shows the LISS III derived NDVI map of the area. 
NDVI values were segmented in to three colors as red (high 
NDVI average NDVI value as .49), cyan (medium NDVI with 
average value as 0.44) and low NDVI with average NDVI as 
0.38 is shown in yellow color. It was interesting to note that 
saturated soil moisture status (shown in red in Figure 5) was 
found to be associated with the early sown crop which were in 
grain filling stage and had low NDVI (shown in yellow color in 
Figure 5). Ancillary information showed that the fields 
belonging to this category were of clay-loam soil and the 
physiographic was comparatively low. These conditions had led 
to saturated soil moisture conditions even long after the 
irrigation was stopped. The other fields belonging to late sown 
category with no standing water belonged to sandy loam soil 
and were relatively of higher gradient. Hence the soil though 
wet was not under saturated conditions. 
  
  
  
Figure 3. Radarsat SAR, Extended Low beam (23-Mat-2000), 
image showing rice fields with (1) saturated soil condition (2) ' 
high soil moisture (3) standing water.