Research On Flood Classification Forecast And Design Flood In Shiquan-Ankang Interval Tributary

Affected by climate change,extreme rainfall events occur frequently,and flood disasters have become the most frequent and destructive natural disasters in china,seriously threatened the safety of people’s lives and the stability of the society.As the most effective non-engineering measure for flood prevention,flood forecasting has attracted more and more attention.In order to improve the accuracy of flood prediction,the flood is classified and forecasted and the forecast error is corrected in real time.This study took China’s strategic water resources allocation strategy water source area of the upper Han River Shiquan-Ankang Reservoir interval tributary basin as the research object.Based on the future rainfall process in the climate model simulated by the stochastic rainfall simulation model,the hydrological model is used to simulate the future flood process.The future design flood of the basin is deduced and the flood evolution process is analyzed.The main research contents and results are as follows:(1)The total rainfall amount,maximum rainfall intensity and average rainfall intensity were taken as clustering factors,and the SOM neural network model was used to cluster the historical floods in the basin.The BP neural network model was trained based on the clustering results,and the flood was classified by the trained model.The results show that the average peak discharge of different types of flood varies significantly after flood clustering,and the classification accuracy of each type of flood in each basin is more than 80%in the process of flood classification,which indicates that the flood clustering and classification model in this study have good applicability in the study area.(2)The TOPMODEL model was constructed,and the model parameters of different types of floods were determined by PSO optimization algorithm,and the parameters were verified.In order to reduce the prediction error,ensemble Kalman filter is used to correct the flood prediction results in real time.The results show that TOPMODEL model has a good simulation effect in the flood prediction of main tributaries in the upper reaches of Han River,and the prediction accuracy is significantly improved after real-time correction,which indicates that TOPMODEL model can be used to simulate the flood process in the upper reaches of Han River,and real-time correction technology can significantly reduce the error of flood prediction.(3)The NSRP rainfall stochastic simulation model was constructed,according to the historical hourly rainfall estimated model parameters and verify the applicability of the model,combined with the climate model under different discharge situation of future rainfall data,with rainfall as constraint,the stochastic simulation of future years 3rd largest rainfall.The TOPMODEL model was used to simulate the flood process,extract the annual maximum flood peak discharge sequence,calculate the design flood,and analyze the change of the future design flood under different discharge scenarios.It is found that the design flood peaks in different recurrence periods of future design floods under the three emission scenarios of RCP2.6,RCP4.5 and RCP8.5 are increased compared with the historical design flood peaks,indicating that both the frequency and magnitude of future floods are increasing.