Soil Moisture Estimation for Wheat Crop Using Dual-Pol L-Band SAR Data

Soil moisture retrieval over the vegetated soil surfaces using Synthetic Aperture Radar (SAR) data is a challenging issue. Presence of vegetation over soil surface makes the interaction of the radar signal with the soil more complex. Several studies used the Water Cloud Model (WCM) to separate vegetation effect on the soil backscatter while estimating the soil moisture. The general form of WCM utilizes one or two vegetation descriptors (e.g., Vegetation Water Content (VWC) and Leaf Area Index (LAI)) in determining the vegetation contribution. Eventually, these descriptors replaced by vegetation metric derived from ancillary sources (e.g., Normalized Difference Vegetation Index-NDVI). This ancillary data may not be available close to the date of SAR data acquisition due to several reasons. To circumvent these challenges, we use SAR derived vegetation descriptors in estimating soil moisture over wheat fields. We studied the performance of four different descriptors (viz., VWC, NDVI, cross-pol ratio (CPR), Dual-pol Radar Vegetation Index (DpRVI)) for estimating soil moisture using WCM. SAR derived vegetation descriptors for dual-pol data provided a reliable accuracy with a r value of 0.86 and RMSE of 5.9% (DpRVI-VV) as compared to NDVI. HH polarisation outperformed VV polarisation agreeing with the fact that vertically oriented crops less affects horizontally polarized signal.