Mapping the landslide susceptibility based on an IV-RF Model with revised slope units：A case study of Junlian County，Sichuan Province

Abstract

Abstract Evaluating landslide susceptibility is a key objective of preliminary research in landslide management and land use planning. Using the revised slope units as an evaluation unit, an information volume method-random forest (IV-RF) model is used to assess the landslide susceptibility in the transitional zone from the basin to the mountainous area. First, based on historical landslide data, an inventory of landslide was compiled using high-precision satellite identification technique supplemented by field surveys. Nine landslide influence factors, including elevation, slope, aspect, land surface relief, lithology, distance from fault, distance from river, distance from road, and land use, were selected via Pearson correlation analyses, excluding the rainfall factor and land surface relief factor, to construct the model and calculate the information values. The RF model was used to calculate the weights of the 8 influential factors. The accuracy of the landslide susceptibility map generated by the IV-RF model was examined by the ROC curve, and the AUC value was 0.945. The results showed that the IV-RF model based on the revised slope units for landslide susceptibility assessment is a reliable and effective method.