Landslide Susceptibility Assessment And Landslide Prediction Impact Range In Maiji Area

The susceptibility evaluation of landslides is the prediction of the spatial distribution of landslides,which is reflected in the probability of occurrence of landslides at a macro level.This article analyzes the distribution and influencing factors of landslides based on the collection of geological data in the study area,and establishes an evaluation index system.The landslide susceptibility of the study area is divided by the information model,the random forest model and the certainty coefficient model.Based on the evaluation results of the susceptibility,the potential landslide impact area of the study area is predicted by the potential landslide in the landslide-prone area in the study area.Put forward new ideas and references for the study of landslide disaster prevention and mitigation in the study area.The main conclusions drawn from the research in this article are as follows:(1)A total of 97 historical landslide hazards were obtained through site investigation of the study area.Most of these landslides are classified small and medium with volume less than100×104m3,including 38 landslides in the loess layer,52 loess-mudstone interface landslides,6 accumulation landslides and 1 artificial filling landslide.Note that most of these landslides occurred on both banks of the Weihe River and gullies,with slope angles of around 20°-30°.Rain and human engineering activities are the main causes of landslides.factor.(2)Based on the selection principle of evaluation factors,8 geological hazard evaluation factors,including elevation,slope,aspect,curvature,rock and soil type,distance from fault,distance from river,and distance from road,are selected in this paper.Pearson's correlation coefficient is used to analyze the correlation of the selected eight,and establish a landslide hazard evaluation index system in the study area.(3)The information volume model,random forest model and certainty coefficient model were used to study the landslide susceptibility of the study area,and the susceptibility evaluation index obtained by the three models was imported into the ArcGIS software,according to the natural break point The study area is divided into five levels:low-prone area,low-prone area,medium-prone area,high-prone area and extremely high-prone area.The three models of geological hazard susceptibility zoning maps are drawn,and the three models predict The area of extremely high-high-prone areas accounted for 42.3%,24.2%,and 34.7% of the total area of the study area respectively,and the number of landslides accounted for 84.5%,90.7%,and 80.4% of the total area of the study area.They are mainly distributed along the Weihe River and G310 National Road in the study area.On both sides,the prediction results of the three models are consistent with the actual situation of the disaster.(4)Use ROC curve and AUC value to test the evaluation accuracy of the three models.Random forest algorithm model,information model and certainty coefficient model ROC curve area under the AUC value ratio:0.820,0.673,0.731,for three The accuracy comparison of the two models found that the algorithm performance of the random forest model is superior.(5)Based on the evaluation results of the susceptibility of the study area,predict the impact range of potential landslides in the areas with high and extremely high susceptibility to landslide hazards in the study area,and use Johnson and Box-Cox conversion methods to process the original landslide parameters to construct The multi-dimensional normal probability prediction model takes the upper boundary of the 90% confidence interval of the prediction result as the safety value of the influence range and fits it to obtain the unstable slope height YH and the landslide body area YA in the study area,the landslide body width Yw,and the landslide in the accumulation area.In the prediction formula between the body length YL,only the unstable slope height YH is used as the independent variable in the prediction formula,which provides great convenience for on-site rapid assessment of the potential landslide hazard and impact range of unstable loess.