Defining physically-based rainfall thresholds for early warning systems

Empirical rainfall thresholds compiled on correlations between recorded data show that precipitation intensities and durations required to trigger shallow landslides vary with climatic, geotechnical and topographic conditions; consequently, thresholds exhibit a high degree of spatial variability, even across relatively small geographic areas (see, e.g., Baum and Godt (Landslides 7:259-272, 2010; Guzzetti et al. Landslides 5:3-17, 2008). In order to define intensity/duration rainfall thresholds capable of considering the site-specific hillslope characteristics, GIS-based modelling techniques have been developed and successfully applied starting from Iverson's theory (Baum et al. TRIGRS - a Fortran program for transient rainfall infiltration and grid-based regional slope-stability analysis, version 2.0, 2008; Godt et al. Rev Eng Geol 20:137-152, 2008; Salciarini et al. Eng Geol 102:227-237, 2008). In this work A GIS-based code is presented which permits the assessment of the spatial distribution of the minimum rainfall intensity that triggers shallow landslides and debris flows over a given study area, based on the rainfall duration and the local geometric, hydrologic and mechanical characteristics of the slopes. Such an approach is used for predicting landslide scenarios produced by short-duration rainfalls. An example of application to a study area of the Umbria Region in central Italy is presented, describing the capability of the model of providing site-specific thresholds for different rainfall scenarios and issuing different levels of hazard warning. The application illustrates some challenges on the technically feasibility of shallow-landslide early warning systems, capable of including specific information on the affected areas, probability of landslide occurrence and expected timing.