Soil moisture inversion in the Shandian River Basin based on multi-temporal sentinel data

Soil moisture is a key variable in terrestrial surface systems, and obtaining high-resolution long-term soil moisture information is of great significance. This article conducts inversion experiments based on multi temporal Sentinel-1 radar data and Sentinel-2 optical data from 2020 in the Shandian River Basin. Firstly, within the framework of the water cloud model, three different optical remote sensing indices were used to estimate the vegetation moisture content in the model, in order to correct the scattering contribution of the vegetation layer. On this basis, a soil moisture observation equation system is constructed based on the Dobson soil dielectric model and the Alpha approximation model. The least squares method is used and a regularization term is added to solve the soil dielectric constant, thereby achieving the inversion of soil moisture. The study used corresponding ground observation data for verification, and the results showed that the inversion of soil moisture values had good consistency with ground observation values. Especially based on normalized difference moisture index (NDMI), the inversion effect is the best after removing the backscatter effect of vegetation. The correlation coefficient (R) between the inversion results and ground measurement data reaches 0.864, and the root mean square error (RMSE) is 0.038cm³/cm³, indicating that this method can accurately estimate soil moisture. At the same time, analyzing soil moisture changes in time series, reflecting their seasonal fluctuations and correlation with regional precipitation events, verified the applicability of the inversion method considering vegetation phenology in the Shandian River Basin.