Landslide susceptibility zonation at the regional scale: the Molise case study (Italy)

Molise is an Italian region with a high concentration of landslides. However, an estimation of landslide susceptibility for the entire region has never been officially performed. In this paper, a landslide susceptibility analysis is illustrated based on the implementation of two different statistical methods (bivariate and multivariate). The aim is to obtain a susceptibility map that provides a good representation of the criticality of the study area together with a reliable statistical performance. Various combinations of conditioning factors were tested with the two methods and, based on expert judgment and ROC (Receiver Operating Curve) values, the one obtained through the multivariate approach was selected as the best map.