Investigating landslide susceptibility procedures in Greece

The study aims to present quantitative-based landslide susceptibility mapping through different procedures. Medium- to small-scale analysis was performed applying the most common statistical methods used for this purpose. For that reason, a complete landslide inventory is also presented consisting of about 209 events from 1913 to 2015 in a selected area in Western Greece. In addition, five predisposing factors (variables) (lithology, slope, elevation, rainfall, land use) were also selected to join the procedure. The methodology avoids susceptibility overestimation by the use of different validation methods. As a consequence, the most suitable and reliable model for the study area is discussed. The susceptibility assessment based on frequency ratio, landslide relative frequency, statistical index, likelihood ratio, and weights-of-evidence statistical models as well as success and prediction curves has also been used for validation. Landslide susceptibility index (LSI) was expressed as an algebraic summary according to bivariate analysis. Taking into account the impact of each variable on landslide susceptibility individually, the LSI is being converted into LSIweight. For that purpose, the results arising from validation procedure are used to attribute statistical weights on each variable included to landslide susceptibility estimation.