Prioritization of landslide conditioning factors and its spatial modeling in Shangnan County, China using GIS-based data mining algorithms

摘要: The main objective of the current study is to apply a random forest (RF) data-driven model and prioritization landslide conditioning factors according this method its comparison multivariate adaptive regression spline (MARS) for susceptibility mapping in China. For purpose, at first, locations were identified by earlier reports, aerial photographs, field surveys total 348 landslides mapped from various sources GIS. Then, inventory was randomly split into training dataset (70% = 244 landslides) remaining (30% = 104 used validation. In study, 12 applied detect most susceptible areas. These slope aspect, altitude, distance faults, lithology, normalized difference vegetation index, plan curvature, profile rivers, roads, angle, stream power topographic wetness index. relationship between each factor finalized using frequency ration (FR) model. Subsequently, landslide-susceptible areas MARS RF models. results revealed that important accuracy measure (mean decrease) are lithology (23.47%), faults (22.21%), altitude (19.58%). We also notice (19.04%), (18.83%), roads (15.29%) have highest importance Gini measure. Finally, maps produced two models verified receiver operating characteristics curve. showed map has higher prediction rate than area under curve values 87.51 77.32%, respectively. According validation results, exhibits better could be proposed land-use planning area.