Research And Application Of Rainfall-runoff Simulation In The Watershed Of Tokachi Based On IFAS Model

Utilizing hydrological model to simulate watershed hydrological process is helpful for researchers to understand and recognize watershed hydrological cycle process more clearly,and is also conducive to the promotion and application of the model.The watershed of Tokachi river in Hokkaido has complex topography and landform,high vegetation coverage and frequent flood disasters.There are more than one hundred rainfall and flow observation stations of large and small scale in the watershed.The layout of the stations is reasonable and the observation data are sufficient.These characteristics mean that the watershed was endows a high research value.In this paper,the watershed of Tokachi river in Hokkaido was taken as the research area.The research results were evaluated by employing Nash-Sutcliffe deterministic coefficient,waveform error,volume error and peak error.The applicability of IFAS?Integrated Flood Analysis System?model was verified in the watershed above Maoyan Station which the downstream flow observation station of the Tokachi river watershed,The problem of IFAS model used in extreme flow case was discussed.The precision of IFAS model using different rainfall products was evaluated and the application effect of combining reviseed radar rainfall data with IFAS model used in New Tokugawa watershed was apraised.The possibility of using RNN?Recurrent Neural Network?model and LSTM?Long Short-Term Memory?model to predict hourly rainfall was explored.And the simulation accuracy of the combination application of hourly rainfall and IFAS model predicted by LSTM model was evaluated.The main conclusions of the study are as follows:?1?The applicability of IFAS model in Tokachi river watershed is pretty fine.Whatever the rainfall datausing come from GSMaP_NRT satellite,TRMM₃B42RT?V7?satellite or ground observation station,it can simulate the surface runoff process of Tokachi river watershed with high precision.?2?For the case of extreme flow,the calculation error of IFAS model is very obvious.When the watershed river flow is very small,no matter which rainfall data are used,IFAS model seriously overestimates the ground runoff.When the watershed river flow is very large,the rainfall data of GSMaP_NRT satellite and TRMM₃B42RT?V7?satellite are almost invalid,while the rainfall data of ground observation station are used,the simulate error is relatively small.?3?The IFAS model combined with the revised radar rainfall data can simulate the surface runoff process of the New Tokugawa watershed watershed with high precision.By revising the radar rainfall data,not noly the application range of radar rainfall data is expanded,but also new rainfall data are provided for the IFAS model.?4?Through a large number of experiments and the comparison of the prediction results of RNN model and LSTM model,it is found that when LSTM model operates 200 times,it can predict hour rainfall precisely.Hourly rainfall prediction based on the combination of IFAS model and LSTM model can simulate the runoff process with high precision and the simulation error mainly occurs in the peak flow.