Landslide susceptibility mapping in the Sichuan-Tibet traffic corridor using logistic regression- information value method

Located in east-central Qinghai-Tibet Plateau, the Sichuan-Tibet traffic corridor is one of fastest uplifting and geomorphic evolution regions on the earth. Under the coupling of internal and external dynamics, the landslide in this region is extremely developed, which seriously restricts the planning and construction of highways, railways and hydropower projects. Based on the data collection and analysis of regional geological data, this paper selects lithology, slope gradient, aspect, slope shape, topographic relief, terrain roughness, fault density and distance to rivers as contributing factors. Combined the advantages of traditional information value method and logistic regression, this paper uses the logistic regression-information value method to evaluate the landslide susceptibility of the study area. Through the multi-collinearity test and significance test of the contributing factors, it is found that the selected contributing factors have no multi-collinearity and have a significant impact on the occurrence of landslides. ROC curve is used to test the results of landslide susceptibility, and the AUC value is 0.81, which shows that the model can well predict the occurrence of landslides. The results show that the high risk areas in the study area mainly occur in the regions of the Longmenshan fault zone, Jinshajiang fault zone, Lancangjiang fault zone, Nujiang fault zone and Bianba-Luolong fault zone, as well as on the sides of deep valleys of large rivers with steep slope and large topographic relief. The middle risk areas widely exist on both sides of the tributaries of large rivers. The results are helpful in understanding the development and distribution of landslides in the Sichuan-Tibet traffic corridor, and also provide a scientific basis for the project planning and construction, disaster prevention and mitigation in the study area.