European Seismological Commission, 32nd General Assembly, Sept. 6-10, 2010, Montpellier, France
Caccavale M., Convertito V., Zollo A., Emolo A., Gallovic F.
Classic probabilistic seismic hazard analysis (P SHA), proposed by Cornell (1968), is performed by assuming: a uniform space distribution for the seismicity in a given magnitude range and a poissonian time-independent earthquake occurrence model. The effect of each earthquake is estimated by using a proper ground-motion predictive equation (GMP E). The application of classical P SHA is not obvious when a single causative fault and an associated maximum earthquake are considered as the threat for the site of interest. P articularly for volcanic areas and largest earthquakes, two main limitations have to be faced in order to apply classical P SHA. First consists in the estimation of the expected return periods, the second one in the characterization of a strong seismic wave attenuation media.
In this work an extension to time-dependent model of the integrated probabilistic–deterministic method proposed by Convertito et al. (2006) is presented with an application to Campi Flegrei and Vesuvius volcanic areas. This method can be used to overcome some of the limitations of both P SHA and deterministic approaches when applied to a single fault for a scenario-like description of the hazard. The distance and azimuthal distribution of ground motion parameters and their expected ranges of variability, relevant for estimating the earthquake effects, are obtained through a statistical analysis of a synthetic waveforms database. The waveforms have been produced by a hybrid simulation method by Gallovic and Brekesova (2007) for a large number of possible scenarios for M4.0 and M5.5 earthquake. Different nucleation points, rupture velocity, and fault mechanism have been considered. For Campi Flegrei surface fault projection are located on the caldera border, for Vesuvius volcano the fault are located in the center of crater. P eak-ground motion estimates are validated through a comparison with an ad-hoc GMP E retrieved for the area of interest by using the stochastic approach of Boore (1983). Concerning the time-dependent approach, different renewal models (i.e. log-normal and Brownian passage time) have been considered and applied. The measure of the gain between the time-independent and time-dependent model has been evaluated in order to estimate the better time-dependent approach.