Keywords
lava flow, mud flows, modelling and simulation, susceptibility, hazard
Start Date
1-7-2008 12:00 AM
Abstract
A method for mapping mud flows and lava flows, and for evaluating related susceptibility and hazard has recently been tested at different study areas of Southern Italy (located in Campania, Calabria, and Sicily), which have repeatedly been affected by damaging events in historical time. The approach is based on numerical models, GIStechniques, geo-environmental and historical evaluations. Results have been obtained through a statistical approach by simulating a large number of events on a cluster. Employed cellular automata models have first been calibrated, either manually (by trial and error) or by means of Genetic Algorithms, and then validated against past flow events that occurred in the same study areas, or in similar geo-environmental settings. Simulations have been quantitatively evaluated with respect to real cases by means of two distinct functions of fitness based on 1) the affected areas for mudflows, and on 2) affected areas and duration for lava flows. Aiming at susceptibility/hazard assessment, a grid of possible sources has been hypothesised on the basis of historical/geological knowledge and statistics of past events. For each source, a high number of simulations has been planned by adopting combinations of sources’ and materials’ characteristics. Probabilities of activation, empirically based on past events, have been assigned to each source of the grid by considering its location and geological information. Different probabilities have also been assigned to each “type of event” by taking into account their observed historical frequencies. Two different types of maps were realized in a GIS: 1) a susceptibility map, realised by simply counting the frequencies of flows affecting each site; 2) a hazard map, in which probabilities have been “empirically” attributed to each simulation based on location of sources and types of event. Preliminary results (based on a subset of the overall planned simulations) clearly depict the most susceptible and hazardous sectors.
Mud-flow and lava-flow susceptibility and hazard mapping through numerical modelling, GIS techniques, historical and geoenvironmental analyses
A method for mapping mud flows and lava flows, and for evaluating related susceptibility and hazard has recently been tested at different study areas of Southern Italy (located in Campania, Calabria, and Sicily), which have repeatedly been affected by damaging events in historical time. The approach is based on numerical models, GIStechniques, geo-environmental and historical evaluations. Results have been obtained through a statistical approach by simulating a large number of events on a cluster. Employed cellular automata models have first been calibrated, either manually (by trial and error) or by means of Genetic Algorithms, and then validated against past flow events that occurred in the same study areas, or in similar geo-environmental settings. Simulations have been quantitatively evaluated with respect to real cases by means of two distinct functions of fitness based on 1) the affected areas for mudflows, and on 2) affected areas and duration for lava flows. Aiming at susceptibility/hazard assessment, a grid of possible sources has been hypothesised on the basis of historical/geological knowledge and statistics of past events. For each source, a high number of simulations has been planned by adopting combinations of sources’ and materials’ characteristics. Probabilities of activation, empirically based on past events, have been assigned to each source of the grid by considering its location and geological information. Different probabilities have also been assigned to each “type of event” by taking into account their observed historical frequencies. Two different types of maps were realized in a GIS: 1) a susceptibility map, realised by simply counting the frequencies of flows affecting each site; 2) a hazard map, in which probabilities have been “empirically” attributed to each simulation based on location of sources and types of event. Preliminary results (based on a subset of the overall planned simulations) clearly depict the most susceptible and hazardous sectors.
