Abstract: This paper uses digital elevation model data with a spatial resolution of 90 m to construct a Liuxihe model, and discusses its applicability to the Shangyoujiang Reservoir basin with different river channel grades and rainfall station densities. Its optimal parameters are determined using the particle swarm optimization algorithm, and then it is used to simulate 18 flood events measured in 2009-2020. The results show that for this model, its simulation index is slightly better when the river channels are divided into three grades for model construction than the case of four grades, and the flood forecasts with the former are more accurate. Comparison of the model schemes based on 9 and 15 rainfall stations and their simulations of 31 flood events of 2000-2020 shows that the scheme of 15 stations is better, and high density of gauge stations is conducive to improving the Liuxihe model and its forecasting accuracy. Therefore, the Liuxihe model can be used in the flood forecasts of the Shangyoujiang Reservoir.