Lava-flow hazard on the SE flank of Mt. Etna (Southern Italy)

A method for mapping lava-flow hazard on the SE flank of Mt. Etna (Sicily, Southern Italy) by applying theCellular Automata model SCIARA-fv is described, together with employed techniques of calibration andvalidation through a parallel Genetic Algorithm. The study area is partly urbanised; it has repeatedly beenaffected by lava flows from flank eruptions in historical time, and shows evidence of a dominant SSE-trendingfracture system. Moreover, a dormant deep-seated gravitational deformation, associated with a largervolcano-tectonic phenomenon, affects the whole south-eastern flank of the volcano.The Etnean 2001 Mt. Calcarazzi lava-flow event has been selected for model calibration, while validation hasbeen performed by considering the 2002 Linguaglossa and the 1991–93 Valle del Bove events — suitable datafor back analysis being available for these recent eruptions. Quantitative evaluation of the simulations, withrespect to the real events, has been performed by means of a couple of fitness functions, which considereither the areas affected by the lava flows, or areas and eruption duration. Sensitivity analyses are in progressfor thoroughly evaluating the role of parameters, topographic input data, and mesh geometry on modelperformance; though, preliminary results have already given encouraging responses on model robustness.In order to evaluate lava-flow hazard in the study area, a regular grid of n.340 possible vents, uniformlycovering the study area and located at 500 m intervals, has been hypothesised. For each vent, a statisticallysignificantnumber of simulations has been planned, by adopting combinations of durations, lava volumes,and effusion-rate functions, selected by considering available volcanological data. Performed simulationshave been stored in a GIS environment for successive analyses and map elaboration.Probabilities of activation, empirically based on past behaviour of the volcano, can be assigned to each vent ofthe grid, by considering its elevation, location with respect to the volcanic edifice, and proximity to its mainweakness zones. Similarly, different probabilities can be assigned to the simulated event types (combinationsof durations and lava volumes, and to the effusion-rate functions considered). In such a way, an implicitassumption is made that the volcanic style will not dramatically change in the near future. Depending onadopted criteria for probability evaluation, different maps of lava-flow hazard can be compiled, by taking intoaccount both the overlapping of the simulated lava flows and their assumed probabilities, and by finallyranking computed values into few relative classes.The adopted methodology allows to rapidly exploring changes in lava-flow hazard as a function of varyingprobabilities of occurrence, by simply re-processing the database of the simulations stored in the GIS. ForCivil Protection purposes, in case of expected imminent opening of a vent in a given sector of the volcano, reprocessingmay help in real-time forecasting the presumable affected areas, and thus in better managing theeruptive crisis. Moreover, further simulations can be added to the GIS data base at any time new differentevent types were recognised to be of interest.In this paper, three examples ofmaps of lava-flowhazard for the SEflank ofMt. Etna are presented: the first has beenrealisedwithout assigning any probability to the performed simulations, by simply counting the frequencies of lavaflows affecting each site; in the second map, information on past eruptions is taken into account, and probabilitiesare empirically attributed to each simulation based on location of vents and types of eruption; in the third one, astronger role is ascribed to the main SSE-trending weakness zone, which crosses the study area between Nicolosiand Trecastagni, associated with the right flank of the above-cited deep-seated deformation. Despite being onlypreliminary (as based on a sub-set of the overall planned simulations), themaps clearly depict themost hazardoussectors of the volcano, which have been identified by applying the coupledmodelling-GISmethod here described.