Abstract: The aim of this study is to explore the applicability of Liuxihe model in forecasting flood of small and medium sized watershed and the impact of DEM-based digital river channel delineation and classification on flood forecasting result. A Liuxihe model for flood forecasting in the Longhua river watershed is established, with model parameters optimized by using Particle Swarm Optimization (PSO) algorithm. Moreover, the effect of digital river channel classification on flood forecasting is discussed. Results show that higher order of digital river channel would result in higher peak flow of flood, earlier peak flow, larger runoff coefficient, and simulation result closer to factual values. Three-order of river system, rather than one order, is more appropriate when adopting the Liuxihe model for the flood forecast of small and medium sized watershed. As parameters in the Liuxihe model are optimized by PSO algorithm, only one observed flood event is required for parameter optimization in practical application, hence improving the performance of the model effectively. Validated by 50 flood events, the Liuxihe model with three orders of river system for flood forecasting of Longhua river watershed is proved feasible in real-time flood forecasting for the Longhua river watershed

Key words: flood forecasting, small and medium sized watershed, distributed hydrological model, Liuxihe model, digital river channel