Soil Moisture Estimation in Chambal Irrigation Command of Rajasthan Based on Radarsat-1 Multi-Incidence Angle Data and SAR-optical Synergy

Soil moisture is one of the key variables in agricultural water management particularly in scheduling of irrigation in the major irrigation commands. Large number of studies has proved that radar backscatter is sensitive to soil moisture due to its dependence on the complex dielectric permittivity of soil water at microwave frequencies. However, surface roughness and presence of vegetation cover introduces complexities in the precise estimation of soil moisture in large agricultural areas. Numerous approaches have been put forward to predict the backscatter as a function of sensor configuration and surface, as well as vegetation characteristics towards deriving soil moisture information. However, for operational applications in larger areas, a parameterization of these analytical and theoretical models is difficult to achieve. It has always remained a challenge as to how to estimate soil moisture using single configuration radar without field measurement of surface roughness and vegetation cover. In the present study, soil moisture was modeled in two steps. Firstly, SAR angular response was used for surface roughness characterization assuming that the surface roughness and soil moisture does not change in a short period of time in absence of rainfall or irrigation and surface perturbation. The roughness normalized backscattering (RNBS) thus generated consists of contribution from soil moisture and surface vegetation cover. In the second step normalized difference vegetation index (NDVI) generated from IRS LISS III data and RNBS was plotted. Their relationship was studied and iso-moisture lines were generated at different intervals. Specific polynomial model was derived for each level of soil moisture and using those models the SAR image was classified and usable soil moisture map was generated. The approach is valid for C-band radar and gives good result in fallow, and crop cover areas where normalized difference vegetation index (NDVI) value less than 0.4 while using IRS LISS III data.