A SAR-Only Inversion Framework for Soil Moisture Using Multi-Index Comparative Analysis

Abstract

Soil moisture is critical to the ecological stability and hydrological cycle. Optical remote sensing is severely constrained by cloud cover, snow, and frozen soil in alpine regions, hindering long-term soil moisture monitoring. Taking Nagqu region in the Tibetan Plateau as the study area, this paper constructs a pure microwave soil moisture inversion framework based on multi-temporal Sentinel-1 SAR data to avoid optical data dependence. Four SAR vegetation indices (DpRVI, RVI, DpSVI and PRVIc) were integrated into the coupled WCM–Oh2004 model to dynamically correct vegetation attenuation and surface scattering. The results show that the DpRVI-based model performs best, with R = 0.85 and RMSE = 0.0709 cm3/cm3, outperforming other indices. The framework maintains stable accuracy in the growing season and effectively captures spatiotemporal soil moisture variations. The proposed SAR-only method agrees well with official downscaled soil moisture products, proving its applicability for continuous soil moisture monitoring in optically inaccessible alpine regions.