Soil moisture estimation from ERS-2 SAR data in Solani River catchment

Backscatter coefficient estimated from ERS-2 SAR sensor can effectively be used to derive soil moisture state of a river catchment which is of great importance from hydrological point of view. However, the backscatter coefficient is highly affected by a number of factors such as topography, vegetation density, and variations in small-scale surface roughness. Analysing the effect of these factors to eliminate their effect on backscatter coefficient for accurately estimating the soil moisture is the main focus of the present study. ERS-2 SAR image of date 28(th) July 2003 (i.e., autumn season) was utilised for carrying out the study in a typical river catchment in India. Incidence angle based model was used to account the effects due to topography. The effects of vegetation on backscatter coefficient were minimised by using the semi-empirical water cloud model. Four agricultural crops and grassland compose the set of vegetation classes in the study area. A comparative study between three important parameters that describe vegetation in terms of their bulk characteristics (e.g., leaf area index; LAI, plant water content; PWC and crop height W) was carried out to identify a vegetation descriptor that had the maximum influence on backscatter coefficient. The effect of three canopy descriptors namely LAI, PWC and h were assessed on individual basis by proposing three separate models used in the water cloud model so as to simplify the model, that could use a single canopy descriptor instead of two or more as used in many other studies. Results indicated that the backscatter coefficient obtained from the model using LAI showed stronger relationship with the observed volumetric soil moisture with high R-2 values. A nonlinear least square method (LSM was implemented to estimate volumetric soil moisture. A significantly high correlation was obtained between the retrieved soil moisture and the observed soil moisture with high R-2 values of the order of 0.95 to 0.97 and low rmse values for almost all the vegetation classes and barren land. Subsequently, soil moisture map of the study area was generated from the SAR image.