A new GIS-based multivariate statistical analysis for landslide susceptibility zoning

A common multivariate statistical analysis was applied for landslide susceptibility zoning. The novelty consists in a particular attention on the choice of the factors which influence the landslide occurrence and triggering, and in a critical review of the choice of the calibration area in terms of its extension and of distribution of the influencing factors. Geomorphologic, geological, and both climatic and anthropogenic factors are accounted for. Infrastructures, such as roadways and railways, have been analysed, both for their role as triggering factor and because their preservation plays an important role in the economic choices of the territorial protection agencies. The proposed statistical model has been calibrated using the entire Liguria and Piedmont regions (North-Western Italy), in order to statistically analyse quite different morphological characteristics. Finally the research has developed a guideline for the landslide susceptibility zoning, by proposing an automatic procedure able to map the landslide susceptibility and to provide useful design suggestions for the correct choice and/or the conscious exclusion of types of interventions.