A multi-source data fusion method to retrieve soil moisture dynamics and its influencing factors analysis in the ecological zone of the eastern margin of the Tibetan Plateau

• Soil moisture dynamics retrieval based on multi-source data fusion. • Exploring the evolution of soil moisture during the growing season. • The fusion results of V-polarized data were better than those of H-polarized data. • TB and RMS-h had the greatest impact on the SMAP and Sentinel-1 data retrieval models, respectively. Accurate soil moisture monitoring is crucial for understanding the role of hydrological processes in the climate system within the ecological region in the Eastern Margin Ecotone of the Qinghai-Tibetan Plateau (EMETP). This study proposed a data fusion method to acquire soil moisture with high resolution (1 km) and more accuracy. This study combines the fine-scale (1 km) spatial details of soil moisture obtained from active microwave data with the broader-scale (5 km) soil moisture information derived from passive microwave data in both V-polarization and H-polarization modes. The results showed that the fusion results based on V-polarization data were better than that of H-polarization data. The root mean square error (RMSE) is 0.0410, and the correlation coefficient is 0.8809 With the Geodetector, we found that the spatial pattern of soil moisture in the EMETP was influenced by multiple interacting factors. Among these, the interactions between the Normalized Difference Vegetation Index (NDVI) and elevation, as well as NDVI and soil type, showed the strongest influence on the spatial distribution of soil moisture. In addition, the brightness temperature and root mean square height of the surface were the most influential parameters for the retrieval model based on SMAP data and sentinel-1 data, respectively. The conclusions provide a valuable reference for the water resources estimation and management in the region of the Tibetan Plateau.