Soil Moisture Retrieval Using SAR Derived Vegetation Descriptors in Water Cloud Model

Abstract

In radar remote sensing applications, soil moisture retrieval over the vegetated surface is a challenging issue due to complex interaction of radar waves with vegetation layer and the underlying soil. Several studies utilized the Water Cloud Model (WCM) directly or by coupling it with surface inversion models, to compensate vegetation effects while estimating soil moisture. The realization of vegetation component in the original form of WCM utilizes various plant descriptors (e.g., vegetation water content (VWC) and plant area index (PAI)). These descriptors were eventually replaced with vegetation metric obtained from ancillary sources (e.g., the Normalized Difference Vegetation Index -NDVI derived from the optical sensor). To overcome this dependency on ancillary data, we utilize radar derived vegetation descriptors to estimate soil moisture over canola fields. We investigated the performance of WCM for soil moisture retrieval utilizing the PAI and radar derived vegetation descriptors, i.e., the ratio of backscatter intensities (HH/VV and VH/VV) and indices (viz., Radar Vegetation Index (RVI), and Generalized Radar Vegetation Index (GRVI)) in WCM. The radar data derived vegetation descriptors provides encouraging retrieval accuracy with RMSE ranging from 0.04 (for GRVI) to 0.08 m3 m−3 (for HH/VV). This comparative analysis using different polarizations indicates that the HH polarization outperforms VV, while VH has marginal deviations for all descriptors.