Spatial series approach to estimate soil moisture over wheat fields from a single SAR image

The coupling of soil moisture (SM) with other factors, such as surface roughness and vegetation coverage, impedes the generation of large-scale high-precision SM products. Eliminating the effects of vegetation and roughness based on backscattering intensity ratios (BIRs) from time-series data is the primary approach for addressing this problem. However, this method is susceptible to the low temporal resolution of SAR images, leading to unstable retrieval results. In this study, we propose to decouple SM from other factors using a spatial series approach (SSA), which requires only a single synthetic aperture radar (SAR) image and utilizes the BIRs of spatial series points to retrieve SM. The core idea is to employ BIRs to implicitly compensate for vegetation and roughness effects, thereby avoiding explicit parameter estimation with inherent biases. For this purpose, spatial series points are selected based on the incidence angle, volume scattering power, canopy dominant orientation, and roughness. Then, the BIRs of these points are used for SM retrieval. Experiments are conducted using L-band UAVSAR data acquired at different times. The experimental results show that the root mean square error (RMSE) of SM retrieved by the SSA is around 10 %, and the correlation coefficient exceeds 0.7. This approach holds value by expanding the observational dimensions and enhancing SM monitoring capabilities, particularly in scenarios where time-series SAR data acquisition is unfeasible. Future integration with temporal-series methods could establish a spatiotemporal-series inversion framework, which would markedly advance large-scale SM retrieval research.