Probability hazard map for future vent opening at Etna volcano (Sicily, Italy).

Abstract

Mount Etna is a composite stratovolcano located along the Ionian coast of eastern Sicily. The frequent occurrence of flank eruptions (at an interval of years), mostly concentrated along the NE, S and W rift zones leads to a high volcanic hazard that, linked with intense urbanization, poses a high volcanic risk. In the framework of the project PON SIGMA (Integrated Cloud-Sensor System for Advanced Multirisk Management), we develop a near real-time computer-assisted analysis and probabilistic evaluations that provides the identification of the areas prone to the highest vent opening hazard. A longterm volcanic hazard assessment, mainly based on the past flank activity of the Mt. Etna volcano, is the basic tool for the evaluation of this risk. Then, a reliable forecast of where an impending eruption will occur is needed. The use of a code such BET_EF (Bayesian Event Tree_Eruption Forecasting) delivers a long-term hazard map, that, if additional data are provided, switches into a short-term future vent opening map. The present application is based on incoming seismic and ground deformation data. Analytic inversion of high frequencies deformation data is performed to find the key parameters of a magmatic source in an elastic, isotropic and homogeneous half-space. Seismic data allow us to set the boundary of the investigated area. The inversion is performed by using the genetic algorithms (GAs) approach, a well-known search technique widely used to solve optimization problems and categorized as global search heuristics (Goldberg, 1989). Hence the magmatic source is located, a forward model is computed to evaluate the deformation field over Mt. Etna surface. Therefore, for each cell, the displacement vector modulus is estimated and the density probability function is calculated. A higher probability value matches with the cells with larger modulus, whereas lower estimate is found where the modulus is close to zero, being the sum of the probability values normalized to one over the investigated area. We modelled the final intrusion of the May 2008 – July 2009 flank eruption at Mt. Etna, whose onset was preceded by an intense seismic swarm and marked by ground deformation recorded at GPS stations. The future vent forecast highlights the area with higher probability, increasing the difference in relative values between that zone and the rest of the volcano edifice. It is worthy notice that a good accordance is evident if the highest probability area is compared with the real vent occurrence.