Study On Simulation And Early Warning Indicators Of Flash Floods In Xichuan River Based On Hydrological Model

Flash floods are a type of natural disasters caused by short-lived,heavy rainfall,which are characterized by rapid and destructive effects and seriously endanger and affect the lives and properties of people in the region.Shaanxi Province is located in the hinterland of China and is also a flash flood-prone area.The study of model-based flash flood hazard simulation and early warning indicators for small watersheds can supplement the way of calculating flash flood early warning indicators in the region and provide theoretical basis and technical reference for flash flood disaster prevention and management.This study takes the Xichuan River basin in Yan’an City,northern Shaanxi Province as the research object,investigates the target basin profile,historical flash flood occurrence and prevention status,and analyzes the spatial distribution and center of gravity shift of historical flash floods.Based on the spatial and temporal trends of rainfall in the watershed,and in response to the monotonic status of rainfall models used in the hydrological handbook calculations,research calculations of representative rainfall patterns were carried out.While comparing and analyzing the simulation results and applicability of two semi-distributed models for flash floods in the watershed,the better performing model is selected for further calculation of early warning indicators for disaster prevention objects in the watershed.The following conclusions are obtained:(1)Flash floods in the Xichuan River basin are caused by heavy rainfall,which is mostly concentrated in July and August.The distribution of historical flash floods in the basin tends to expand,the number of flash floods in the morth increases,the distribution of historical flash floods in the direction gradually decreases,and the center of gravity of historical flash floods migrates from the east to the west.(2)The typical rainfall distribution sets in the Xichuan River basin can be divided into four types,summarized as the rainfall peak in front and concentrated(X1),the rainfall peak in the back and dispersed(X2),the rainfall peak in the middle and concentrated(X3),and the rainfall peak in the back and concentrated(X4).The highest probability of occurrence is the type with concentrated rainfall in front of the peak,and the lowest is the type with concentrated rainfall behind the peak.(3)Both TOPMODEL and HEC-HMS semi-distributed hydrological models achieve the flood forecasting class B accuracy of the《Hydrological Information Forecasting Specification》in the flood simulation of Xichuan River Basin,but TOPMODEL model is slightly better than HEC-HMS model.The Nash coefficients of TOPMODEL model in the calibration period and validation period are 0.83 and 0.82,and the pass rates are 90%and 80%respectively;The Nash coefficients of the HEC-HMS model are 0.79 and 0.80 in the calibration period and validation period,and the pass rates are 80%.(4)The analysis of flash flood simulation results in Xichuan River basin was carried out.The results show that the number of simulated peaks smaller than the measured peaks is higher in the TOPMODEL model than in the HEC-HMS model;the two models are better than the multi-peak process in simulating the single-peak process,and the simulation effect of the main peak is better than that of the secondary peak,and the semi-distributed model does not perform best when the simulated flood peaks are close to each other;the simulated runoff depths compared with the measured runoff depth model are more.The simulated flood peak process is better than the simulated rising and receding flood process;the peak present time errors exist in advance or delay,but they all meet the accuracy evaluation criteria in the text,which is related to the rainfall distribution area and confluence time.(5)Based on the TOPMODEL model,the influence on the critical rainfall under different precipitation amounts in advance,different warning periods and rain type backgrounds is explored.The results show that:①The 1h critical rainfall values of the disaster prevention objects in the watershed are distributed in the range of 20-50mm.② The 6h critical rainfall values of X1-type rain are higher than those of X2 and X3-type rain,and most of the critical rainfall values of X3-type rain in the context of each warning period show less than those of X1 and X2-type rain.