Study On Evaluation Method Of Highway Landslide Susceptibility

Landslides are one of the most common natural disasters in China,usually causing huge casualties and property damage,among which highway landslides have a great impact on traffic roads.Highway is an important carrier to serve the economy,society and the public.With the rapid development of China’s transportation infrastructure construction and the increasing frequency and intensity of natural disasters,the losses and social costs caused by traffic disruptions caused by natural disasters far exceed the losses caused by the destruction of transportation infrastructure.Highway landslide susceptibility evaluation is extremely important for mapping out the status of highway disasters and their distribution,and doing a good job of statistical analysis of highway disaster data.The terrain of Laibin city of Guangxi Zhuang Autonomous Region is mainly mountainous and hilly,with a dense road network and dense distribution of highway landslide points.The whole region of the Autonomous Region has a complex and basin-like geomorphology,and landslide disasters occur frequently.In this paper,firstly,using Laibin city as the research region,nine evaluation factors,including elevation,slope,aspect,height difference,plan curve,profile curve,precipitation,TWI(Topographic Wetness Index)and FVC(Fractional Vegetation Cover),and five models,including FR(Frequency Ratio),FR-AHP(Analytic Hierarchy Process),FR-LR(Logistic Regression),FR-BPNN(Back Propagation Neural Network)and FR-SVM(Support Vector Machine),are selected to evaluate the susceptibility of highway landslides,and initially verified the reliability of the relevant models.Secondly,the scope of the research region is expanded,and Guangxi Zhuang Autonomous Region is taken as the research region,and twelve evaluation factors,including elevation,slope,aspect,plan curve,profile curve,precipitation,FVC,TWI,soil moisture,TPI(Topographic Position Index),population and distance from road,and fifteen models,including FR,FR-AHP,FR-LR,FR-BPNN,FRSVM,IV(Information Value),IV-AHP,IV-LR,IV-BPNN,IV-SVM,CF(Certain Factor),CF-AHP,CF-LR,CF-BPNN and CF-SVM,are selected to evaluate the susceptibility of highway landslides,and compare and analyze the advantages and disadvantages of each model,in order to provide experience and reference for landslide susceptibility evaluation in other regions.The key findings of this paper are as follows:(1)In the susceptibility evaluation of highway landslides in Laibin city,the very highand high-susceptibility regions are mainly located at low elevation,low vegetation cover and frequent human activities,while the very low-susceptibility regions are mainly located in the western and eastern parts of the research region,where the elevation is high and human activities are less.The validation accuracy of the ROC curves of the five evaluation models is greater than 0.700,which has certain reliability.Among them,the FR-LR model has the highest accuracy of 0.804,and this model is more suitable for the evaluation of landslide susceptibility of highway in Laibin city.(2)The twelve evaluation factors selected for the susceptibility evaluation of highway landslides in Guangxi Zhuang Autonomous Region all passed the independence test,avoiding the errors caused by the repeated use of the same factors,especially the inclusion of the highway buffer zone factor,which is in line with the characteristics of highway linear features.The very high-and high-susceptibility regions in the research region are mainly distributed around the highway,where human activities are frequent,the elevation is low and the slope is gentle,which is also in line with the characteristics of highway distribution.Very low-susceptibility regions are mainly distributed in the northwest and northeast of the research region.These areas have steep slopes and relatively few highways.The ROC curves verification accuracy of the fifteen selected models is greater than 0.750 with high reliability,among which,the CF-LR model has the highest accuracy and is more suitable highway landslide susceptibility evaluation model for Guangxi Zhuang Autonomous Region.(3)Some previous studies have analyzed or compared one or more of the fifteen models,however,few studies have analyzed and compared these fifteen models and applied them to highway landslides.In this paper,the evaluation results of these fifteen models are comprehensively analyzed and compared in order to provide experience and reference for landslide susceptibility evaluation in other regions.The research results of this paper are helpful to the targeted development of highway infrastructure maintenance business,and will also provide strong support for rescue and maintenance after highway landslides and post-disaster reconstruction.