Landslides Development Characteristics And Susceptibility Assessment Along China-Nepal Cross-Border Railway

The proposed China-Nepal cross-border railway is an important transportation system connecting China and Nepal.The railway construction will further improve China-Nepal transportation and strengthen economic,trade and cultural exchanges between China and Nepal.The railway planning area is located on the southern edge of the Qinghai-Tibet Plateau,passing through the high mountains and valleys of the Himalayan orogenic belt,with complex geological conditions,many types of engineering geological problems,and variable climate and environment.This area has nurtured a large number of landslide geological disasters,posing a major threat to the planning,construction and safe operation of the railway.Therefore,it is of great significance for railway line selection and disaster prevention and control to establish a basic database of landslide disasters in railway construction areas,study the development law and distribution characteristics of landslide disasters,and carry out quantitative evaluation of landslide disaster susceptibility.The research area of this paper is the Sino-Nepalese crossborder railway area.Methods such as remote sensing image interpretation and field investigation are used to obtain landslide disaster points in the area and establish a landslide disaster database.Using geographic information system(GIS),12 factors affecting landslide hazards were extracted,and the percentage of landslide area was used as an index to analyze the degree of influence of different factors on landslide hazards.Through the GIS platform,a variety of evaluation models are used to evaluate the susceptibility of landslide hazards in the study area,and then the susceptibility of landslide hazards in the study area is graded,and the susceptibility zoning map is drawn.The ROC curve was used to test the rationality and accuracy of the susceptibility evaluation results of the six models.The main research results obtained in this paper are as follows:(1)Based on medium and high-resolution multi-source remote sensing images,through visual interpretation combined with field investigation and verification,707 historical landslide disaster points were obtained through interpretation,with a total disaster area of 223.8km2.Landslides are divided into small,medium,large and super large landslides,accounting for30.27%,41.30%,19.66% and 8.77% of the total landslides,respectively.Small and mediumsized landslides mainly occur on both banks of rivers,and large-scale landslides mainly develop in the high Himalayas.The development of landslides in the study area has obvious spatial differentiation characteristics,showing a certain trend and inhomogeneity.(2)Based on ArcGIS software,extract the graded area of each factor,the landslide area and the number of landslides in the graded area,and analyze the impact of 12 landslide impact factors on landslide disasters;among them,large-scale landslide disasters are mainly developed in the high Himalayas on the Sino-Nepalese border.Mountainous areas are mainly affected by regional structure and stratigraphic lithology.Small and medium-sized landslide disasters are mainly distributed in Nepal,and are greatly affected by water system and vegetation coverage.Person correlation analysis and variance expansion factor are used to select the influencing factors.Finally,slope,slope aspect,ground curvature,surface cutting degree,elevation variation coefficient,geological structure,stratigraphic lithology,river system,terrain humidity index,vegetation A total of 10 factors of the coverage index were applied to the evaluation model,and the basic environmental data for landslide susceptibility evaluation in the study area were finally obtained.(3)A total of 6 evaluation models including information quantity model,deterministic coefficient model,support vector machine model,logistic regression model,logistic regressioninformation quantity model and logistic regression-certainty coefficient model were used to evaluate the landslide susceptibility in the study area.Evaluation.Cross-checking and ROC curve were used to test the results of the evaluation models.The test results showed that except the information model,the other five models had reference value.Compared with the mathematical statistical model,the evaluation accuracy of the machine learning model is about1% higher.(4)According to the evaluation results of landslide susceptibility,the study area is divided into five susceptibility levels.Compared with the two mathematical statistical models,the susceptibility evaluation results of the machine learning model and the coupled model are extremely low.The proportion of the total area of the area and the low area is generally larger than that of the extremely high and high areas,and the area of the extremely high susceptibility area has the smallest area.Eliminate the influence of artificial grading of evaluation factors.The ROC curve accuracy test of the logistic regression-deterministic coefficient model is the highest,and the evaluation result of the model is the best,which is the most suitable evaluation model for the landslide susceptibility analysis of the proposed China-Nepal cross-border railway,and can be further used to guide the landslide susceptibility analysis and disaster prevention and reduction along the China-Nepal railway.