Regional landslide susceptibility model using the Monte Carlo approach– the case of Slovenia

Abstract

Based on the analyses of landslide spatial occurrence, a regional landslide susceptibility model for the area of Slovenia with medium spatial resolution was calculated. Of 3241 landslides with known locations, 67% were selected randomly but representatively as the learning sub-set and used for univariate statistical analyses (c2) to analyse the landslide occurrence in relation to the precondition factors (lithology, slope inclination, slope curvature, slope aspect, distance to geological boundaries, distance to structural elements, distance to surface waters, flow length and land-cover type). In addition, a relation to the triggering factors (maximum 24-hour rainfall intensity with a return period of 100 years, average annual rainfall, and peak ground acceleration with a return period of 475 years) was assessed. The analyses were performed using a geographic information system – GIS in raster format with 25 × 25 m pixel size. The results of the analyses were later used for the development of a weighted linear susceptibility model where more than 156 000 automatically calculated models with random weight combinations were derived. The landslide testing sub-set (33% of landslides) and representative areas with no landslides were used for the validation of all models developed. The results showed that relevant precondition factors for landslide occurrence are (with their weight in a linear model): lithology (0.33), slope inclination (0.23), land-cover type (0.27), slope curvature (0.08), distance to structural elements (0.05), and slope aspect (0.05).