Spatial patterns of soil water content are fundamental to regional ecosystem stability and the sustainable use of water resources. Focusing on the Qinghai Lake Basin， this study integrated field measurements and remote-sensing imagery with statistical and geospatial analyses to investigate the spatial distributions of soil water content and soil water storage at multiple depths. Machine-learning models—Random forest （RF） and Extreme gradient boosting （XGBoost）—were applied to simulate soil water content and storage at varying depths across the basin. The results indicated that soil moisture content in the Qinghai Lake Basin gradually decreased with increasing soil depth （32.06%， 27.17%， and 24.99%， respectively）， while soil water storage increased with depth （29.72 mm， 32.07 mm， and 32.34 mm， respectively）. Spatially， soil moisture in the basin exhibited a general pattern of high values in the north and east and low values in the south and west， with higher values in the upper reaches of rivers than in the lower reaches， and higher values in the mountainous areas surrounding the lake than in the lakeshore zones. Across different vegetation types， alpine meadows had the highest soil moisture content and water storage. Model simulation results revealed that air temperature， precipitation， and the Normalized Difference Vegetation Index were the primary influencing factors of the soil moisture spatial model. This study elucidates the spatial distribution patterns and environmental response mechanisms of soil moisture in the Qinghai Lake Basin， providing a scientific basis for regional ecological restoration and water resource management.

Soil water content  /  Soil water storage  /  Spatial distribution  /  Machine learning