A Semiempirical Modeling for Soil Moisture Retrieval Using High-Resolution SAR Data

The NASA-ISRO Synthetic Aperture Radar (NISAR) mission will advance global soil moisture monitoring through high-resolution L-band observations. However, accurate fine-scale retrieval remains challenging due to the 12-day revisit interval and complex, scale-dependent influences of vegetation and surface roughness on backscatter. This study introduces the SMAP-AVS (Attenuation-Volume scattering-Surface scattering) model, a semi-empirical framework evaluated using ALOS-2 PALSAR observations. Formulated within the Water Cloud Model framework, the methodology moves beyond conventional land-cover-based parameters by adopting pixel-wise parameterization. Leveraging soil moisture temporal stability, the framework assumes strong correlation between coarse-scale (9 km) SMAP L3 products and fine-scale (1 km) variations, enabling separation of vegetation and surface roughness contributions to the SAR signal. The SMAP-AVS framework is resolution-agnostic; while demonstrated at 1 km, it scales to meter-resolution expected from NISAR. Once parameterized, the model operates in snapshot retrieval mode, deriving soil moisture from single SAR acquisitions and NDVI data without further temporal information. Validation across four hydro-climatically diverse U.S. regions (2021–2024) included temporal comparison against International Soil Moisture Network observations and spatial validation using Multi-Radar Multi-Sensor precipitation fields to distinguish physical moisture heterogeneity from artifacts. SMAP-AVS resolves rainfall-driven patterns often missed by coarser products, with spatial correlation coefficients exceeding 0.5 in semi-arid and agricultural regions, demonstrating that the framework captures physically significant hydrological variability and offers a scalable methodology for operational high-resolution soil moisture products.