Abstract: In order to further improve the forecasting and early warning ability of small reservoirs in Anhui Province, based on the characteristics of large number, wide range and poor data conditions of small reservoirs in Anhui Province, the mountainous area of southern Anhui was divided into 3 regions, and 8 hydrological stations were selected to construct the Xin'anjiang model, and the parameter calibration analysis was carried out. On the basis of analyzing the characteristics of each small reservoir basin, the Xin'anjiang model was constructed. The parameters of representative stations are transplanted to small reservoirs by using the watershed similarity principle, so that the inflow of reservoirs was calculated. Using reservoir capacity data and discharge curve, small reservoirs dispatching model was constructed to calculate water level process. Select some small reservoirs and use the reservoir monitoring data from June 18 to 20 in 2024 to verify and analyze the forecast results. The result shows that: This method can be used to forecast the water level of small reservoirs in the whole province, the relative error of the highest water level can be controlled within 30%, the pass rate reached 70%. However, the quality of reservoir monitoring data is still not high, The basic information of reservoir is incomplete and the lack of knowledge of the actual situation of reservoir all affect the accuracy of flood prediction in small reservoirs.