Application of frequency ratio and weights of evidence models in landslide susceptibility mapping for the Shangzhou District of Shangluo City, China

Abstract

The current research presents landslide susceptibility mapping by frequency ratio (FR) and weights of evidence (WoE) models based on geographic information system (GIS) and an assessment of their performances for the Shangzhou District of Shangluo City, China. Firstly, landslide locations of the study area were detected using aerial photographs as well as by carrying outfield survey. Then, a total of 145 landslides were mapped out of which 101 (70 % landslide locations) were randomly selected for training the models, and the remaining 44 (30 % landslide locations) were used for validating the models. The following ten landslide conditioning factors, such as slope aspect, curvature, slope angle, elevation, distance to rivers, distance to faults, lithology, peak ground acceleration, distance to roads and precipitation, were considered in this study. Subsequently, landslide susceptibility maps were produced using FR and WoE models in ArcGIS 10.0 software. The validation of landslide susceptibility maps were carried out using areas under the curve. The validation results showed that the training accuracy were 0.7635 (76.35 %) and 0.7450 (74.50 %) for the FR and WoE models, with predictive accuracy 0.7395 (73.95 %) and 0.7102 (71.02 %), respectively, indicating that landslide susceptibility mapping using FR model is more accurate than WoE model for the study area.