Research On Simulation And Analysis Technology Of Mountain Torrent Disasters In Small Watersheds In Mountainous Area

This study makes comprehensive use of GIS technology and HEC-HMS hydrological model,fully considers the relationship between underlying surface conditions and rainstorm flood in the study area,and highlights the main solutions and methods for design rainstorm flood calculation,watershed runoff generation,watershed confluence,river flood routing and risk analysis of small watershed in mountainous area.Taking Qishan Lake Basin as the research object,the spatial distribution and characteristics of rainfall in the study area are comprehensively analyzed by using the methods of climate change trend analysis and catastrophe detection;based on the calculation method of design rainstorm and flood,the simulation function of rainfall process in multi scenario recurrence period is established;the simulation model of rainstorm and mountain flood yield and concentration is constructed to simulate the flood runoff process in different return periods;Through the research on the dynamic collaborative evolution algorithm of mountain flood,quantitatively analyze the process of flood production and concentration in the basin,determine the flood water level elevation in different return periods,realize the dynamic evolution of flood in time and space,realize the risk analysis of mountain flood disaster,and provide technical support for mountain flood disaster prevention and reduction in mountain areas.The main results of this paper are as follows:(1)Using historical rainfall data,the rainfall characteristics are analyzed by using climate change trend analysis and catastrophe detection methods such as moving average,cumulative anomaly and Mann-Kendall method.The average annual rainfall in Qishan lake basin generally decreases from south to north,showing the distribution characteristics of less in the East,more in the west,more in the South and less in the north.The rainfall has obvious seasonal characteristics.The annual average monthly rainfall is mainly concentrated in July and August,accounting for 57.13%of the annual rainfall;The average annual rainfall from1970 to 2020 is about 472.6mm.From the 1970s to the middle and late 1980s,the annual rainfall showed a downward trend,began to increase in the late 1980s,and the upward trend continues to this day;The linear trend line of annual rainfall is"y=3.0033x+394.48",and the annual rainfall shows a fluctuating state,but generally shows an upward trend;The analysis of UF and UB curves shows that there are many abrupt changes in rainfall from 2013to 2020,but the intersection is within the critical line ofα=0.05 significance level,the sudden change trend of rainfall is not obvious.(2)Through comparative analysis,the Gumbel distribution curve is selected for probability distribution model fitting,and the heavy rain intensity formula and single return period rainstorm intensity formula are solved by the least square method.The average absolute variance is 0.049(mm/min),and the average relative variance is 6.64%,which meets the requirements of fitting accuracy.Based on the rainstorm intensity formula of single return period and the division of rain pattern scenarios,the Chicago rain pattern method and smote sample expansion algorithm are used to establish the multi scenario rainfall process database,the K-means clustering algorithm is used to cluster the multi scenario rainfall pattern curve,the multi scenario recurrence rainfall process simulation function is established,and the rain patterns with different return periods and different durations are obtained.By establishing the simulation function of rainfall process in multi scenario recurrence period,the influence of factors such as insufficient rainfall data,determined rain peak position,too sharp and thin rain peak and uniform rain pattern in the traditional rain pattern model is overcome,and some uncertainty is reduced in the simulation and analysis of mountain torrent disaster.(3)Based on geographic basic data and GIS analysis technology,complete the preprocessing of basic data,the establishment of meteorological and watershed generalization model,calculate the model parameters,and complete the construction of rainstorm and mountain flood yield and concentration simulation model in Qishan Lake Basin.In view of the lack of hydrological data in the basin and the difficulty of verifying the model simulation results from the perspective of measured hydrological data,the runoff coefficient method is innovatively used to evaluate the applicability of the model.The results show that the simulation results of 20150629,20150910,20160714 and 20170522 rainfall basically fluctuate between the minimum runoff and the maximum runoff.The average certainty coefficient R~2 of runoff and average runoff is 0.96 and the average Nash efficiency coefficient is 0.60,It shows that HEC-HMS runoff simulation model has a good effect in Qishan Lake Basin.(4)The analysis of parameter sensitivity results shows that CN value is a highly sensitive parameter for peak discharge and total runoff.When CN value changes between-20%～20%,the fluctuation range of peak discharge is-24.4%～46.4%,and the fluctuation range of total runoff is-19.9%～43.4%;For the peak arrival time,the peak delay time and regulation and storage parameters have an obvious impact on the peak arrival time.With the increase of the two parameters,the peak arrival time is delayed.Land use type is an important factor affecting CN value.The influence of land use change on flood process is quantitatively expressed from two aspects of Interdecadal and different resolution.The results show that the interdecadal land use features are relatively fixed,will not change greatly with time,and have little impact on the flood runoff process;Different resolutions of land use have different degrees of expression of feature details.The higher the resolution,the more accurate and detailed the expression of feature details,which has a great impact on the flood runoff process.(5)Based on the 24h rainfall data of different return periods in the pre peak scenario,combined with the rainstorm mountain flood yield and concentration simulation model,the24h flood runoff process of different return periods in the basin is simulated;Through the optimization research of mountain flood dynamic co evolution algorithm,calculate the ponding volume of unit rainfall process in different return periods,determine the flood water level elevation in different return periods,and quantitatively analyze the process of flood production and confluence in the basin in combination with water dynamic balance and seed spread method,so as to realize the dynamic evolution of flood in time and space;According to the risk classification standard,the risk analysis of mountain flood disaster in Qishan lake basin is realized.