GNGTS 2015 - Atti del 34° Convegno Nazionale

GNGTS 2015 S essione 2.1 see Azzaro et al. , 2012a). They are defined by the distribution of the earthquake epicenters of the instrumental data set and also include the sources of the strongest earthquakes occurring at Etna during the last 3 centuries. The characterization of SZs includes the estimation of the effective depth, i.e. the seismogenic layer where most of the seismic energy is released. For this purpose we calculated, for each SZ, the distribution of the number of earthquakes and the related strain release vs. depth, with steps of 1 km, using the 3-D re-located instrumental data. Results indicate that the seismogenic thickness is mainly confined in the first 7 km below sea level (b.s.l.), which is in agreement with the overall 1-D depth distribution of seismicity in the Etna area (Alparone et al. , 2015). In particular the 3-D data set shows a clustering of hypocenters, allowing to recognize a main seismogenic layer at about 1 km b.s.l. and, in some cases, also a second layer at about 5 km b.s.l. defining the bottom of the SZs. Seismic rates from the instrumental catalogue have been obtained for each SZ by using the ZM ap tools (Wiemer, 2001). Detailed analyses show that the structures belonging to the Timpe fault system have similar b -value coefficients of the GR relationship, while the Pernicana fault, though it retains about the same annual rate ( a -value) of earthquakes, it has a much lower b - value. The above frequency-magnitude distributions (FMD) were then compared with those obtained from the historical macroseismic catalogue (CMTE Working Group, 2014), covering a time-span of about 150 years for all SZs except for the Pernicana fault, whose seismic history is limited at the last 35 years. Since the extension of the two catalogues is different, FMD were normalized to one year. b -values calculated from the instrumental and macroseismic data sets are consistent each other, so we can affirm that instrumental seismicity occurring in a time- window of 9 years, in which no seismic swarm due to flank eruptions altered the “normal regime” of our SZs, is representative for a long-term seismogenic behavior. In order to calculate the seismic hazard by the innovative approach of Branch 3, which uses the distributed seismicity as background model, the a - and b -value coefficients of the GR were calculated using a three-dimensional grid with inter-nodal distance of 2 km and 3 km search radius; grid nodes with less than 20 earthquakes have been discarded, and other a -values have been normalized accordingly to the volume represented. For the individual faults of Etna’s eastern flank which generated major earthquakes – we consider characteristic those having an epicentral intensity I 0 ≥ VIII EMS (European Macroseismic Scale, see Grünthal, 1998), corresponding to magnitude M w ≥ 4.6 (Azzaro et al. , 2011) - the expected M max and the mean recurrence time (T mean ) are estimated using two different approaches: i) historical earthquake catalogue data and ii) fault data, which are representative of tectonic activity. Branches 2 and 3 fork further to take into account stationarity (Poissonian approach) or time-dependency on faults. In the first “historical” approach, already described in Azzaro et al. (2012b; 2013), T mean is computed by the inter-event times occurred on the same structure as defined by the fault seismic histories. Assuming that there are no significant differences between faults, we grouped all inter-event times in order to obtain a more statistically significant sample. In this approach the aperiodicity value α is obtained from the instrumental b -value of GR (Zoller et al. , 2008). As an alternative method, T mean referred to major earthquakes generated by the Pernicana and Timpe faults is estimated by using the geometric-kinematic fault parameters such as 3-D dimension, kinematics and slip-rate (Azzaro et al. , 2014). This analysis has been carried out through the software FISH, a Matlab ® tool developed in the framework of the DPC-INGV S2 Project to turn fault data into seismic hazard models (Pace et al. , 2015). The FISH code “quantifies” the seismic activity fromgeometry and slip-rate of a fault through different empirical and analytical scaling relationships between dimension of the source and characteristics of the expected earthquake, providing several values of M max and associated T mean . FISH, therefore, formally propagates the errors of magnitude and slip-rate obtaining, for the characteristic magnitude expected in each fault, the most probable value of T mean with the associated standard