Comparing landslide susceptibility models in the Río El Estado watershed on the SW flank of Pico de Orizaba volcano, Mexico

Abstract

A comprehensive study of landslide susceptibility models is carried out in the Rio El Estado watershed on the SW flank of Pico de Orizaba volcano. A detailed multitemporal landslide inventory map in the watershed is used as a framework for the quantitative comparison of three landslide susceptibility models. The first landslide susceptibility map is created by using the Stability Index MAPping model. The second and the third landslide susceptibility maps are created using multiple logistic regression (MLR) and multicriteria evaluation models. The validation of the resulting susceptibility maps is performed by comparing them with an inventory map in a contingency table and through the area under the receiver operating characteristic curve. The results point out that the models tend to over-predict and have a moderate to high match with the landslide areas. In this research, MLR is preferred over the other two models because MLR obtains similar or better results with fewer significant variables.