6533b836fe1ef96bd12a1225

RESEARCH PRODUCT

Predicting storm triggered debris flow events: application to the 2009 Ionian-Peloritan disaster (Sicily, Italy)

subject

description

Abstract. The main assumption on which landslide susceptibility assessment by means of stochastic modelling lays is that the past is the key to the future. As a consequence, a stochastic model able to classify a past known landslide scenario should be able to predict a future unknown one as well. However, storm triggered landslide events in the Mediterranean region could pose some limits on the operative validity of such expectation, as they typically result by a randomness in time recurrence and magnitude. This is the case of the 2007/09 couple of storm events, which recently hit north-eastern Sicily resulting in largely different disaster scenarios. The purpose of this study is to test whether a susceptibility model based on stepwise binary logistic regression is able to predict a storm triggered debris flow scenario. The study area is the small catchment of the Itala torrent (10 km2), which drains from the southern Peloritan Mountains eastward to the Ionian sea, in the province of the Messina territory (Sicily, Italy). The shallow landslides activated in the occasion of two close intense rainfall events have been mapped by integrating remote and field surveys, producing two event inventories which include 73 landslides, activated in 2007, and 616 landslides, triggered by the 2009 storm. The set of predictors were derived from a 2 m cell digital elevation model and a 1 : 50 000 scale geologic map. The topic of the research was explored by performing two types of validation procedures: self-validation, based on the random partition of each event inventory and chrono-validation, based on the time partition of the landslide inventory. It was therefore possible to analyse and compare the performances both of the 2007-calibrated model in predicting the 2009 landslides (forward chronovalidation) and vice versa of the 2009-calibrated model in predicting the 2007 landslides (backward chronovalidation). Both the two predictions resulted in largely acceptable performances, in terms of fitting, skill and reliability. However, a loss of performance and differences in the selected predictors between the self-validated and the chrono-validated models which are linked to the characteristics of the two triggering storms are highlighted.