Shallow landslides predisposing and triggering factors in developing a regional early warning system.

Landslides are the most dangerous and common natural phenomena causing higher number of victims and damages in Piemonte (north-western Italy). An integrated statistical approach to identify the main predisposing and triggering factor causing shallow landslides occurrence based on a large dataset of historical rainfall-induced shallow landslides and a long-time series of rainfall data is presented. Historical shallow landslide events were compared with territorial and rainfall characteristics. Exploratory Spatial Data Analysis (ESDA) has been performed in order to explore the spatial distributions of landslides with the environmental quantitative variables (morphometry and rainfall) to identify possible patterns of spatial autocorrelation. ESDA showed strong autocorrelation of landslide locations with rainfall variables, while resulting in low values for morphometric parameters at the investigated scale. Moreover, by intersecting landslides distribution with categorical land thematic parameters (land thematic maps) allowed to identify areas with different landslide density occurrence. Resulting susceptibility-related areas were used to improve the landslides regional Early Warning System of Piemonte based on refined empirical intensity vs duration (I-D) rainfall thresholds. [ABSTRACT FROM AUTHOR]