Simulation of debris flow events in Sicily by cellular automata model SCIDDICA_SS3

Debris flow models are widely used for hazard mapping or for evaluating the effectiveness of risk mitigation measures. Several models analyze the dynamics of debris flow runout solving Partial Differential Equations. In use of such models, difficulties arise in estimating kinematic geotechnical soil parameters for real phenomena. In order to overcome such difficulties, alternative semi-empirical approaches can be employed, such as macroscopic Cellular Automata (CA). In particular, for CA simulation purposes, the runout of debris flows emerges from local interactions in a dynamical system, subdivided into elementary parts, whose state evolves within a spatial and temporal discretum. The attributes of each cell (substates) describe physical characteristics. For computational reasons, the natural phenomenon is splitted into a number of elementary processes, whose proper composition makes up the CA transition function. By simultaneously applying this function to all the cells, the evolution of the phenomenon can be simulated in terms of modifications of the substates. In this study, we present an application of the macroscopic CA semi-empirical model SCIDDICA_SS3 to the Peloritani Mountains area in Sicily island, Italy. The model was applied using detailed data from the 1 October 2009 debris flow event, which was triggered by a rainfall event of about 250 mm falling in 9 hours, that caused the death of 37 persons. This region is characterized by river valleys with large hillslope angles (30°-60°), catchment basins of small extensions (0.5-12 km2) and soil composed by metamorphic material, which is easy to be eroded. CA usage implies a calibration phase, that identifies an optimal set of parameters capable of adequately play back the considered case, and a validation phase, that tests the model on a sufficient (and different) number of cases similar in terms of physical and geomorphological properties. The performance of the model can be measured in terms of a fitness function that compares the observed landslide with the simulated one. This function returns values from 0 (completely wrong simulation) to 1 (perfect match); values greater than 0.7 are considered acceptable. The adopted version SCIDDICA_SS3 was calibrated on debris-flows occurred in Torrente Sopra Urno, that have caused most of the damage in Giampilieri town. Other 5 events, occurred in the same day and on the same area, were used for validation with fitness function ranging from 0.72 to 0.78. Simulations show a good capability of the model to describe the runout of debris in such highly-urbanized area, according to several performance indices. The calibrated parameters may be reasonably used to simulate debris flow runout in the nearby catchments for predictive purposes, aimed at risk assessment. Acknowledgements: This research was funded by the Italian Education, University and Research Ministry (MIUR), PON Project No. 01_01503 'Integrated Systems for Hydrogeological Risk Monitoring, Early Warning and Mitigation Along the Main Lifelines', CUP B31H11000370005