Brightness-Shape-Moisture Radiative Transfer Model Improves Soil Moisture Estimation from Sentinel-2 Imagery

Soil moisture is the main water source for plants, affecting everything from germination to nutrient absorption. The physics-informed brightness-shape-moisture (BSM) radiative transfer model can be used in the inverse direction for accurate soil moisture estimation from high-resolution spectra. However, this accuracy is not maintained once spectra are resampled to fewer-band multispectral resolutions. By coupling a simulated BSM look-up table, resampled to Sentinel-2 response spectra, with gradient boosting regression, the authors developed a transferable method capable of estimating soil moisture from Sentinel-2 imagery, achieving high accuracy on simulated data (R2 = 0.96) and moderate accuracy on real-world data (R2 = 0.51), while avoiding the poorer generalization typically associated with empirical, field-calibrated models. This approach shows strong potential for operational, long-term soil moisture monitoring across Southern Africa.