Soil moisture estimation using Simulated NISAR Dual Polarimetric GRD Product over croplands

Synthetic Aperture Radar (SAR) has immense potential in estimating soil moisture with high-resolution imaging capability and cloud independent acquisition ability. Nevertheless, estimation of soil moisture under vegetation cover is a challenging task. Notably, existing literature used ancillary data sources, such as optical data, to segregate the vegetation contribution in the backscatter. In this study, we propose a new Ground Range Detected (GRD) radar vegetation index for dual-pol data, DpRVI c that overcomes the typical shortcomings (such as cloud cover, asynchronous observations and saturation effect for denser canopies) associated with different optical data derived indices. This proposed descriptor jointly utilizes the copol purity of the wave and normalized co-pol intensity parameter. We then use this index in the Water Cloud Model to estimate soil moisture over croplands. Furthermore, the performance of DpRVI c is compared with Normalized Difference Vegetation Index (NDVI) by utilizing the simulated NISAR L-band dual-pol data (VV-VH, HH-HV) over a Canadian test site. The proposed method has proven to be a potential alternative to synergetic approaches with Root Mean Square Error (RMSE) ranging from 5.6% to 6.0% with DpRVI c as a vegetation descriptor. Thus, the proposed vegetation descriptor provides new insights to quantify the vegetation using dual-pol GRD SAR data. Further, the adapted soil moisture technique has opened up a new avenue for soil moisture estimation using dual-pol GRD SAR data in the presence of vegetation.