Soil temperature is closely related to energy exchange and water cycle between land surface and atmosphere, and further affects the vegetation growth and regional ecological security. This study first analyzed the change characteristics of measured soil temperature at different soil depths from October 2018 to July 2020 in Qinghai Lake watershed, then calibrated the SHAW model based on the soil temperature from October 2018 to August 2019 and further verified it using the measured data from September 2019 to July 2020. At last, the effects of vegetation and soil parameters on soil temperature change were explored. The results showed that: 1) The change range of soil temperature decreased with increasing soil depth, i.e., the range of mean daily soil temperature was 25.50 ℃ at 5 cm depth, and 20.19 ℃ at 35 cm depth, both of which were much lower than the corresponding range of air temperature (36.32 ℃). 2) During the calibration period of the SHAW model, the Nash-Sutcliffe efficiency coefficient (NSE) of soil temperature was higher than 0.94 at all layers, and the root mean square error (RMSE) gradually decreased from the surface (1.91 ℃) to the deep layer (0.86 ℃). In general, the simulation accuracy of the model increased with the increase of soil depth. The evaluation indexes during the validation period were all slightly lower than those in the calibration period. However, the NSE of each layer exceeded 0.93, and the RMSE reduced from 1.98 ℃ in the shallowest layer to 0.98 ℃ in the deepest layer, indicating that the SHAW model could be used to simulate the change of soil temperature in the Qinghai Lake watershed. 3) The change of soil temperature was negatively correlated with the leaf area index and soil bulk density. In addition, soil temperature responded significantly to saturated conductivity, pore-size distribution index and air-entry potential only when they decreased by 60%, 40% and 30%, respectively.