Landslide Susceptibility Assessment Based On Random Forest Model In Shangnan County

Landslide disasters occur frequently in China,which are widely distributed and destructive,resulting in serious damage to the country's social and economic construction and people's lives and property.With the continuous development of economic construction,the environment on which human beings depend is increasingly disturbed by engineering activities,which makes the frequency and intensity of landslides show a growing trend.Landslide susceptibility assessment predicts the spatial distribution and occurrence probability of landslides,and provides important scientific basis for risk management and urban planning of landslides.Shangnan County is located in the Qin-Ba mountain in the southeastern part of Shaanxi Province.Its geological structure is complex,geological environmental conditions is poor,human engineering activities are strong,and landslide hazards become more and more serious.Therefore,it is of great significance to assess the susceptibility of landslides for the prevention and control of landslides.On the basis of collecting geological data of Shangnan County,this paper analyses the characteristics and influencing factors of landslides in the study area,establishes an evaluation index system,and obtains landslide susceptibility zoning of Shangnan County through random forest model based on GIS technology and R language.The research work and achievements in this paper are as follows:(1)Based on the collected geological data and satellite images of Shangnan County,the development and distribution characteristics of landslides in the study area are analyzed,and the influencing factors such as topography and geomorphology,geological structure,formation lithology,rock soil mass types,neotectonics,earthquake,climate hydrology and human activities are studied.(2)Based on the experience of predecessors and the selection principle of evaluation index,topographic and geomorphological factors such as elevation,slope,aspect,ground roughness,plane curvature,section curvature,curvature and relief amplitude are selected.Rock soil mass types,fault of geotechnical structure factors are selected.Rainfall,river of climate hydrology factors are selected,Roads of human engineering activity factors and normalized vegetation index of vegetation factors are selected.Landslide geological hazard evaluation index system in Shangnan County has been established initially.(3)Using ArcGIS to extract grid data of the study area,building random forest model based on R language and drawing ROC curve evaluation model,the landslide susceptibility distribution map,the factor importance distribution maps based on average Gini index reduction and average accuracy reduction in Shangnan County are finally obtained.Fault bufrer zone elevation,aspect,rock soil mass types,profile curvature,slope factors are the main causes of landslide occurrence in Shangnan County(4)The AUC value of area under ROC curve is 0.835,which indicates that the accuracy of random forest model zoning is high.The study area is divided into four parts:low,medium,high and extremely high vulnerable areas by natural breakpoint method.The area of low vulnerable areas is 480.19 km2,accounting for 20.81%of the whole county area,the area of medium vulnerable areas is 577.89 km2,accounting for 25.05%of the whole county area,the area of high vulnerable areas is 735.26 km2,accounting for 31.87%of the whole county area,the area of extremely high vulnerable areas is 513.66 km2,accounting for 22.27%of the whole county area.