Landslide susceptibility mapping using logistic regression analysis in Latyan catchment

Abstract

Every year, hundreds of people all over the world lose their lives due to landslides. Landslide susceptibility map describes the likelihood or possibility of new landslides occurring in an area, and therefore helping to reduce future potential damages. The main purpose of this study is to provide landslide susceptibility map using logistic regression model at Latyan watershed, north Iran. In the first stage, 208 Landslide locations were identified and mapped using extensive field surveys. 75 % of these landslides were used for training and 25 % of them for validation of the model. The mapped landslides were then georeferenced using ArcGIS 10 to provide the landslide inventory map. In the second stage, maps of factors affecting the occurrence of landslides were prepared in ArcGIS 10. Finally in the last stage, the relationships between these affecting factors and the landslide inventory map were calculated using Logistic regression algorithm. The amount of pseudo R2 (0.32) and AUC (0.85) shown the high efficiency of Logistic regression model. According to the coefficients obtained by the model, lithology is the most important factor affecting landslide occurrence (coefficient= +12.032). Most landslides (69%) are located within Ek Formation. The results indicated that 7.56% of the basin is located in high susceptibility class and 2.88% in very high susceptibility class.