Research On Flood Forecasting Of Xiaoqing River Basin Based On HEC-HMS Hydrological Model And GRU Intelligent Model

Affected by climate change and human activities,extreme rainfall events increased,leading to frequent hydrological incidents such as mountain floods and urban floods.The Xiaoqing River Basin has aggravated the flood disaster in the basin due to special geography.In response to the hoods of the Xiaoqing River Basin in recent years,the flood prediction accuracy and serious disaster loss are serious.Based on the analyzing the variation of drought and flood in Xiaoqing River Basin and data transfer research is carried out,establishing the HEC-HMS hydrological model and GRU intelligent flood forecast model of Huangtai Bridge Station,from multi-angle comparison analysis,and verify the reliability of the GRU intelligent model.The main research contents and results are as follows:(1)Two rainfall indexes using SPI and Pa analyze drought and flood changes from multi-scale analysis,and the lack of rainfall data is used to use CEEMD-SPA method for shift.The results show that the increase in rainfall increases with the calculation scale of Pa;the SPI increases with the increase in the calculation scale,the frequency of floods above 5% increases,and the frequency of normal events is reduced,and the frequency of drought and floods is increased.In response to the lack of rainfall data in Wujiapu rainstorm,the CEEMD-SPA method is proposed to transfer the neighboring rainstorm and improve the accuracy of rainfall data transfer.Increase rainfall information to provide full rainfall for flood forecasting data.(2)Establishing a distributed hydrological model of the Legend of the Huangtai Bridge Station by HEC-HMS,based on the Arc GIS extraction water system,the basin is divided into 6 sub-basins,and the CN value of the calcined sub-basin is used to use the soil type and land using,the basis of 8 parameter ratios for each sub-partition in HEC-HMS,flood prediction research is carried out.The results show that the average absolute error of the flood forecast value of the field is bigger,the flood time error is ±1h,the flood flow error is about 10%,and the diameter depth error is less than 20%,which meets the specifications belonging to qualified forecast,pass rate reaches 100%,the deterministic coefficient is 0.70-0.81,and the model forecast accuracy reaches the grade B.(3)Establishing the GRU intelligent flood forecasting model of the Huangtai Bridge Station by the Rainfall Classification and flood process between least rainfall flood process relationship,and flood prediction verification is conducted.The results show that the average absolute error of the flood forecast value and the root mean square error are small.The peak current error is within 1 hours,the flood flow error is about 10%,the flood vectors of the field is about 2%,and the runoff is obviously reduced by the HEC-HMS model,all errors meet the specification,the qualified rate reaches 100%,and the deterministic coefficient is more than 0.95,and the deterministic coefficient is increased by about 0.2 by the HEC-HMS model,the forecast accuracy reaches the grade A.(4)The applicability verification of the K-Means rainfall classification GRU intelligent model is adopted during “Liqima” Typhoon.The results show that the flood flow error of the GRU model forecast is 1.92%,the diameter depth error is0.28%,the peak current error is 0 h,and the deterministic coefficient is 0.90 or more,the forecast accuracy reaches the grade A,the GRU model is significantly reduced by the HEC-HMS model flood prediction,and the accuracy of the deterministic coefficient is improved.Which Verifies the applicability of the GRU model in the flood forecast of Xiaoqing River Basin,it shows that the GRU intelligent model has achieved good results in flood forecasting.