S1 Data -

Soil moisture is a key parameter for the exchange of substance and energy at the land-air interface, timely and accurate acquisition of soil moisture is of great significance for drought monitoring, water resource management, and crop yield estimation. Synthetic aperture radar (SAR) is sensitive to soil moisture, but the effects of vegetation on SAR signals poses challenges for soil moisture retrieval in areas covered with vegetation. In this study, based on Sentinel-1 SAR and Sentinel-2 optical remote sensing data, a coupling approach was employed to retrieval surface soil moisture over dense vegetated areas. Different vegetation indices were extracted from Sentinel-2 data to establish the vegetation water content (VWC) estimation model, which was integrated with the Water Cloud Model (WCM) to distinguish the contribution of vegetation layer and soil layer to SAR backscattering signals. Subsequently, the Oh model and the Look-Up Table (LUT) algorithm were used for soil moisture retrieval, and the accuracy of the result was compared with the traditional direct retrieval method. The results indicate that, for densely vegetated surfaces, VWC can be better reflected by multiple vegetation indices including NDVI, NDWI2, NDGI and FVI, the R2 and RMSE of VWC estimation result is 0.709 and 0.30 kg·m-2. After vegetation correction, the correlation coefficient increased from 0.659 to 0.802 for the VV polarization, and from 0.398 to 0.509 for the VH polarization. Satisfactory accuracy of soil moisture retrieval result was obtained with the Oh model and the LUT algorithm, VV polarization is found to be more suitable for soil moisture retrieval compared to VH polarization, with an R2 of 0.672 and an RMSE of 0.048m3·m-3, the accuracy is higher than that of the direct retrieval method. The results of the study preliminarily verified the feasibility of the coupling method in soil moisture retrieval over densely veg etated surfaces.