Modelling the rainfall-induced mobilization of a large slope movement in northern Calabria

Abstract

Two recent phases of mobilization of a large, rainfall-induced debris slide are analysed in terms of relationships between rains and phases of displacement. The first activation at San Rocco (San Benedetto Ullano, Calabria) occurred on 28 January 2009, after extraordinary rains had stricken the region for a couple of months. Detailed geomorphologic field surveys, combined with measurements of superficial displacements at datum points, were performed to properly recognize the evolution of the phenomenon. In addition, a real-time control system of rains and superficial displacements measured at extensometers was implemented, to better analyse the evolution of the phenomenon. In early May 2009, the activity reduced to very slow displacements, persisting in the same condition for the following 8 months. On 1 February 2010, premonitory signs of a new phase of activation were noticed, again following 2 months of extraordinary rainfalls. After few days of further precipitations, the middle sector of the landslide activated since 11 February, disrupting the road network and threatening the major lifelines and some buildings. A hydrological analysis aimed at simulating the dates of activation and the main phases of acceleration of the phenomenon was carried out, by calibrating the empirical model FLaIR against the daily rainfalls and the history of known phases of mobilization since 1970. Calibration allowed to successfully simulate both the cited phases of activity of the San Rocco landslide, by predicting the beginning of the movements as well as the following paroxysmal stages, as testified by the measurements at datum points and extensometers. The set of parameters obtained through calibration reflects the influence of both prolonged antecedent rains, and of high-intensity rainfalls of shorter duration, which slightly preceded the major displacements. Once calibrated the model, a suitable threshold could be defined, by analysing the trend of the mobility function against the history of activations of the considered slope movement, and by excluding false alarms. Accordingly, a reliable tool for predicting the phases of activity of a large slope movement could therefore be obtained.