Forewarning Model For Landslides Triggered by Rainfall Based on the Analysis of Historical Data

Landslides triggered by rainfall could be forecast by modelling the relationship existing between landslide occurrences and antecedent precipitation events. The FLaIR hydrological model allows the forecast through the joint employment of two modules: RL (Rainfall-Landslide) and RF (Rainfall Forecasting). The first module identifies the relationship between rainfalls and landslides by means of a "mobility function" Y(t), obtained through the convolution of the infiltrated rainfalls and a "transfer function" ψ(t). The second module uses stochastic models of rainfall for a probabilistic assessment of the mobility function evolution. In the present paper a selected group of case-studies, referred to Italian territory, has been analysed in order to test the performances of the FLaIR model for landslides with different characteristics. The applications underline a similarity of the estimated transfer functions for landslides with similar lithological characteristics. This result could be very useful in landslides forecasting when there is a lack of data.