National and regional empirical rainfall thresholds for possible shallow landslide occurrence in Italy

Silvia Peruccacci, Maria Teresa Brunetti, Stefano Luigi Gariano, Massimo Melillo, Mauro Rossi, Fausto Guzzetti, 2019, National and regional empirical rainfall thresholds for possible shallow landslide occurrence in Italy, Congresso SIMP-SGI-SOGEI 2019 - Il tempo del pianeta Terra e il tempo dell'uomo: Le geoscienze fra passato e futuro, pp. 570–570, Parma, 16-19/09/2019,

The Italian territory is highly prone to shallow rainfall-induced landslides. Moreover, a large physiographic, geological and climatic variability characterizes the national landscape. For these reasons - and for the abundance of available rainfall measurements - Italy is an ideal test site to investigate how rainfall conditions that induced shallow landslides may vary in different environmental settings. Since 2007, we have been building a catalogue of rainfall events responsible for the triggering of landslides in Italy. Using accurate landslide information and hourly rainfall data recorded by a network of 2228 rain gauges, we reconstructed 2309 rainfall events that induced 2819 (mostly) shallow landslides in Italy in the period January 1996 - February 2014. We calculated the rainfall duration D (in hours) and the cumulated event rainfall E (in mm) presumably responsible for each failure. Using a well-established frequentist method, we calculated empirical cumulated event rainfall-rainfall duration (ED) thresholds at different non-exceedance probabilities and the uncertainties associated with their parameters. Using the entire dataset, we determined a national threshold representing the rainfall conditions that can likely result in landslides in Italy. By considering six environmental subdivisions based on topography, lithology, soil regions, land cover, climate, and precipitation regime, we defined 26 regional thresholds identifying the rainfall conditions responsible for landslide triggering in different environmental settings in Italy. Overall, the resulting national and environmental thresholds are similar, and cover a small part of the possible DE domain. Nevertheless, thresholds for meteorological domains, which are classified according to the mean annual precipitation (MAP) become higher at increasing values of the MAP. This confirms the idea that the landscape adjusts to the regional meteorological conditions. We also observed that the national threshold at 20% non-exceedance probability was capable of predicting all the rainfall-induced landslides that caused casualties between 1996 and 2014. We suggest that this threshold can be used as lower limit to forecast fatal rainfall-induced landslides in Italy. The findings encourage the use of empirical rainfall thresholds for landslide forecasting in Italy, but poses an empirical limitation to the possibility of defining accurate thresholds for small geographical areas, with insufficient datasets.

Data from