Abstract: Under the impact of extreme climate events, frequent occurrences of floods have resulted in severe human casualties and economic losses. However, the accuracy of flood early warning and forecasting for small and medium-sized reservoirs is relatively low. Therefore, this study focuses on the Doushanqiao Reservoir in Hunan Province to investigate the rainfall-runoff process and characteristics of storm-induced floods. It introduces a spatiotemporal variable source distributed hydrological model that considers changes in soil moisture and geomorphic hydrological response units to simulate the rainfall-runoff process. The model demonstrates the simulated results align with the runoff coefficient, runoff depth, and runoff volume corresponding to the multi-year average monthly runoff at Doushanqiao. The model exhibits good performance in predicting flood events, thus providing a reference for reservoir flood management.