Background

High-resolution retrieval of surface soil moisture (SSM), at spatial scales from 0.1 to 1.0 km, is critical for improving land surface monitoring, resource management, and emergency response operations. However, achieving both accuracy and robustness remains a major challenge. Although physically-based scattering models offer valuable insights into the relationship between radar backscatter and surface properties, their inversion is frequently unstable due to the complexity of their parameterization. Empirical and machine learning approaches show promise but require extensive training datasets and often struggle with generalization.

A robust alternative is multidate change detection, which leverages the distinct temporal dynamics of SSM, vegetation, and surface roughness. SMOSAR (Soil MOisture retrieval from multi-temporal SAR data) is a software tool implementing the “Short-Term Change Detection” (STCD) algorithm for SSM retrieval. Initially developed for C-band data in preparation for the Sentinel-1 (S-1) mission, SMOSAR has evolved through several projects (e.g., ESA-GMES, ESA-SEOM-Exploit-S1, EU-Sensagri, ASI-SARAGRI/TETI).

Numerous publications document its performance and potential improvements (see References section). C-band SAR data have demonstrated strong potential for high-resolution SSM retrieval over bare soils and under not dense crop canopies. However, vegetation dynamics — influenced by phenology — can impair sensitivity to SSM. Conversely, C-band’s sensitivity to vegetation offers opportunities to retrieve vegetation water content (VWC) and assess crop structure.

References