The water conservation function plays a pivotal role in regulating ecosystem supply-demand dynamics. Under intensifying global climate change, understanding the spatiotemporal variations in soil moisture (SM) is crucial for assessing the water retention potential of terrestrial ecosystems. However, due to issues including data homogeneity and model uncertainty, the long-term response mechanisms of SM to topographic variations in the Gansu section of the upper Yellow River remain unclear, introducing uncertainty in estimating the soil moisture capacity of the Loess Plateau ecosystem. Employing multi-source data from the Wanchuan River Basin, a major tributary of the Yellow River, this study constructed random forest model and multiple linear regression models to simulate the spatiotemporal distribution of SM at depths of 0-30 cm and investigate its response to topographic factors. Key indicators for SM inversion in the Wanchuan Basin included bulk density (BD), capillary porosity (Pcap), non-capillary porosity (Pnon), cross-polarized radar reflectivity (VH), and the normalized difference vegetation index (NDVI). Among the evaluated models, the linear regression approaches outperformed their machine learning counterparts. Specifically, ridge regression achieved higher R2 and correlation coefficients, along with lower RMSE, MAE, MAPE, and p-values across all three soil depths, demonstrating its optimal performance for soil moisture inversion in the study area.Spatially, high SM values were concentrated in the Xinglong Mountain Forest Park along the southern bank, while the northern bank exhibited lower values. Topographic analysis revealed that SM increased with elevation but decreased with slope gradient, showing higher levels in north-facing aspects. The importance scores indicated that elevation was the topographic factor with the strongest influence on SM.These findings provide critical data and a theoretical foundation for understanding the response of soil moisture to topographic changes in the Yellow River's Gansu reach.