Comparative Study of Conceptual Hydrological Model and Intelligent Model in Flood Simulation of Small and Medium Rivers
Compared with the hydrological forecasting research of large river basins,the research of small and medium-sized river basins is relatively scarce.The simulation result of LSTM(Long Short-Term Memory)model and Xin'anjiang model in flood fore-casting was studied by taking the catchment area above the hydrological station of Hexi in Yuanjiang River as a case study.By comparison,the overall simulation accuracy of the Xin'anjiang model is higher,whose average relative errors of flood volume,flood peak and peak present time are 9.39％,9.55％and 1.6h,respectively.The deterministic coefficient is 0.73.The comprehen-sive qualified rate is 100％,which reaches the Class A accuracy standard.The simulation accuracy of LSTM model is lower,and the average relative errors of flood volume,flood peak and peak present time are 11.76％,12.33％and 2.3 h,respectively.The deterministic coefficient is 0.60,and the comprehensive qualified rate is 75％,reaching the Class B accuracy standard.The results show that the Xin'anjiang model and the LSTM model are effective methods to forecast the flood of small and medium rivers.Both can be used for the forecast,and the simulation accuracy of the Xin'anjiang model is higher than that of the LSTM model for the Hexi basin.
flood forecastsmall and medium-sized riversLSTM neural networkXin'anjiang Model
万方数据
维普
