Dynamic Early Warning Model And Application Of Mountain Flood Disasters In Typical Small Watershed In Henan Province

The flash flood disaster occurs frequently in China,which will result in a major disaster casualties and huge property losses.Flash flood warning has played an important role in the prevention of the flash flood.The complex causes of flash flood make it very difficult to warn it.This paper focuses on the mountainous watershed of semiarid and semihumid areas,and then a forecasting approach is proposed on the basis of according to dynamic critical rainfall index,which is used in the small watershed of the Peihe River in Henan province.The result can provide technical reference for other small watershed.The main contents and achievements are as follows:(1)Watershed runoff calculation has played an important role in calculation of critical rainfall.The accuracy of the results directly affects the accuracy of flash flood warning.So far as some watershed of semiarid and semihumid areas,the models of runoff yield should include excess infiltration and excess storage at the same time.On the basis of this fact,through analyzing mechanism of runoff yield in the watershed,a model is built up which reflects excess infiltration and excess storage simultaneously.It is used in the small watershed of the Peihe River in Henan province.The model consists of excess storage-excess infiltration and instantaneous unit hydrograph.The results show that the simulation result was ideal and it was the second-grade forecast accuracy,which suggests that the model can be used to determine the critical rainfall for early-warning of mountain flood in Peihe River Basin.(2)The critical rainfall is used as the sample basis of dynamic warning.According to the rainfall and flood data in the Peihe River Basin,20 rainfalls were selected and the critical rainfall was determined by trial algorithm.The factors affecting the critical rainfall of mountain flood disaster are analyzed comprehensively,and the main factors of antecedent precipitation,rainfall intensity and accumulated precipitation were determined.Three methods were used to establish the relationshipbetween influencing factors of critical rainfall and critical rainfall,and the dynamic early warning models were constructed.(3)The dynamic early warning model based on multiple linear regression,the dynamic early warning model based on BP neural network and the dynamic early warning model based on support vector machine were constructed.The prediction accuracy of each model compared by mean relative error,root mean square error and determination coefficient.The results show that the accuracy of the dynamic early warning model based on multiple linear regression is better than the other two models,and the results of the dynamic critical rainfall is closer to critical rainfall.So the dynamic early warning model based on multiple linear regression is more suitable for Peihe River Basin.(4)The three dynamic early warning models were applied to 20 floods to test the accuracy of dynamic critical rainfall.The correct forecast rate is over 80%,demonstrating that the method is reliable and can be used to flash flood forecasting and early warning.