Soil Moisture Retrieval from Dual-Polarized Sentinel-1 SAR Data over Agricultural Regions Using a Water Cloud Model

Abstract

The accurate retrieval of soil moisture plays a pivotal role in agriculture, especially in effective irrigation water management, as it significantly affects crop growth and crop yield. For accurate estimation of surface soil moisture (SSM) over agricultural landscapes, the Water Cloud Model (WCM) using synthetic aperture radar (SAR) data is one of the promising and widely used semi-empirical models. However, estimating SSM across vegetated regions is still challenging due to the considerably backscattered radar signal affected by vegetation. The present study mainly focuses on the robustly investigated capability of dual-polarized Sentinel-1 SAR-derived vegetation descriptors in the WCM in SSM retrieval over wheat crops. The vegetation descriptors used in the study are radar vegetation index (RVI), backscattering ratio, Polarimetric radar vegetation index (PRVI), dual Polarization SAR vegetation Index (DPSVI), and Dual Polarimetric radar vegetation index (DpRVI). The performance of different vegetative descriptors in WCM was evaluated using statistical indicators, i.e., coefficient of determination (R²), Nash Sutcliffe efficiency (NSE), percent bias (PBIAS), and root mean square error (RMSE). The results of the WCM model illustrate that all the models show acceptable results, which confirms that this vegetative descriptor can be useful to estimate the soil moisture over the wheat crop in the study area, except for DPSVI. Furthermore, the results revealed that model performances gradually decrease as the crop enters the complex stages. In addition, WCM model results suggest that models are performing better in predicting the higher moisture content (> 30%), followed by medium moisture levels (15–30%) and lower moisture levels (< 15%). In summary, the overall finding demonstrates that PRVI outperformed other models in terms of statistical indicators value for calibration (R² = 0.728, NSE = 0.727, PBIAS = -2.67%, and RMSE = 2.985%) and validation (R² = 0.728, NSE = 0.684, PBIAS = -13.666%, and RMSE = 4.106%). Thus, overall results proved that the WCM model has considerable potential to retrieve SSM over wheat crops from Sentinel-1 satellite data.