Numerical Simulation And Risk Assessment Of Mudflow In The Sijian River Basin,Yongji County,Jilin Province

Jilin Province’s Yongji County is one of the areas with frequent debris flow disasters,and the Sijian River Basin is the most densely distributed area of debris flow in Yongji County,with losses caused by debris flow disasters each year that are difficult to estimate.In addition to the impact of climate change and frequent extreme weather events,there is significant uncertainty in disaster prevention and mitigation work in the area,threatening the lives and property of the people in the Sijian River Basin.Therefore,this paper takes the Sijian River Basin as the study area,carries out a systematic risk assessment,and simulates the movement process of typical debris flow disasters in the area,hoping to provide a certain reference basis for disaster prevention and mitigation work in the study area.Based on the "Geological Hazard Risk Survey Project in Yongji County,Jilin Province," this study combined historical data with field investigations to first carry out a risk assessment of debris flow disasters in the study area.Based on 12 selected factors,the geographic information similarity method was used to select negative samples to improve the data quality.A deep learning model based on the VGG network was adopted to improve the model accuracy,ultimately obtaining the hazard assessment results of the study area.Based on the three major influencing factors of population vulnerability,asset vulnerability,and resource vulnerability,the analytic hierarchy process was used to obtain the vulnerability assessment results of the study area,coupled with the pre-order hazard assessment results to obtain the final risk assessment results of the study area.Subsequently,numerical simulations of the typical debris flow disaster of the Dashigou debris flow in the study area were conducted.Using the Massflow software,a debris flow basic model based on the Voellmy base friction model was established.By inverting the motion parameters of the 2017 debris flow disaster data,the debris flow movement process under different recurrence intervals(50 years,100 years,and 200 years)was simulated.The danger assessment of the typical debris flow was completed through the maximum depth and maximum momentum of the debris flow and the outbreak frequency index.Combined with the vulnerability assessment results mentioned above,the risk assessment results of the typical debris flow were ultimately obtained.The main contents and achievements of this study are as follows:1.The basic environmental information of the study area,such as geographic location,meteorological hydrology,topography and geomorphology,lithology,geological structure,earthquake conditions,hydrogeological conditions,engineering geological conditions,and vegetation coverage conditions were obtained.The main reasons for the frequent debris flow disasters in the study area were analyzed,namely the topographical conditions of being surrounded by mountains on the southeast,west,and middle sides,and the presence of abundant material sources for slope and gully erosion and gully bed accumulation,as well as the water source conditions of high rainfall,short runoff,fast drainage,and strong water dynamic intensity.2.The present study aimed to evaluate the risk of debris flow in the study area using various influencing factors,such as elevation,slope,aspect,curvature,TWI(Topographic Wetness Index),landform type,lithology,vegetation cover,distance to roads,distance to watercourses,distance to faults,and annual precipitation.A multiple collinearity analysis was conducted to select the factors for the study,and the results showed that none of the 12 factors had collinearity.Furthermore,a geographic information similarity-based approach was employed to select negative samples,and it was found that selecting negative samples with a similarity value between 0 and 0.5could improve both the modeling accuracy and land use value.A deep learning model was developed using Python,and the VGG model was adopted as the main body of the convolutional neural network to evaluate the risk of debris flow in the study area.The model used two-dimensional images of the sample point data as inputs,and the results were evaluated using the receiver operating characteristic curve and AUC value.Four maps of the risk of debris flow in the Sijian River Basin were produced.3.The vulnerability of the study area was assessed using the Analytic Hierarchy Process method by selecting three categories of factors that could influence vulnerability,including population,assets,and resources.The weights of these factors were obtained through the Analytic Hierarchy Process and were calculated in Arc GIS,and four maps of the vulnerability of the Sijian River Basin were generated.Based on the maps of the risk and vulnerability of debris flow,a map of the risk assessment of debris flow was produced.4.Massflow software was used to simulate the typical Dashigou area in the study area.The Voellmy friction model was selected as the base model,and the parameters of the debris flow in Dashigou were obtained by inverting the data of the debris flow disaster in 2017.Three recurrence periods(50-year,100-year,and 200-year)were selected to calculate the debris flow discharge curves,and the motion elements,including the impact range,mud depth,and flow velocity,were simulated and plotted at different times.5.The risk assessment of the typical debris flow was conducted by combining the maximum depth and momentum of the debris flow as the strength index,and integrating the frequency of debris flow eruption.The risk assessment of the typical debris flow disaster was produced based on the vulnerability assessment results of the previous section.