GNGTS 2013 - Atti del 32° Convegno Nazionale

The purpose of this study is to determine an empirical relationship M D - M L for Mt. Etna earthquakes by applying the GOR relation (Carrol and Ruppert, 1996), for the dataset of M L and M D available for the earthquakes recoded from 2005 to 2012. The dataset consists of 3361 earthquakes selected from the INGV-OE catalogue (Catalogo dei terremoti della Sicilia Orientale – Calabria Meridionale, 1999-2013, Gruppo Analisi Dati Sismici). Earthquakes are spread all over the area and focii are mainly from the surface to 20 km. For the whole dataset both M D and M L are given. As shown in Fig. 1, the dispersion of M D is very large and it is caused by the uncertainty in the duration estimates, which is mainly related to the presence of noise which mostly affects short-duration (less-energetic) earthquakes (Murru et al. , 2007). Therefore, the M D scale, overestimating small size events and underestimating greater events, appears to be compressed respect to the M L scale. In particular, M D are overestimated for M L <=2.0 and underestimated for M L > 2.5. This behaviourmay be clearly observed also computing the frequencymagnitude distribution of the catalogue and the energy seismic strain release for the whole period. In particular, the slope of the Gutenberg-Richter (G-R) relation (Fig. 2a) calculated by considering M D is higher than that of M L . In Fig. 2a, it is clear that small magnitude values are well represented by M L but not by M D which includes in the classes 1.0-1.4 all low energy events. The distributions seems very similar from 2.0 to 2.5 and, again, the events more energetic are better represented in M L distribution. This has to be taken into account in carrying out studies that use the b -value or seismicity rate because it may have implications for both source processes and hazard estimation. Similarly, the plot of the seismic energy released in time (Fig. 2b) enhances that also this estimate is affected by the M D - M L behaviour. In our regression analyses, we put a great deal of effort into assessing the magnitude uncertainties that are required for the application of GOR method. The standard deviation (σ) for M L and M D are 0.61 and 0.49 respectively, giving an error variance ratio η of 1.24. Being this value in the range 0.7 - 1.8 (Castellaro and Bormann, 2007) we may feel confident that the application of GOR gives better results than SLR. The basic procedure for GOR is described in detail in the literature (Carroll and Ruppert 1996; Das et al. 2011) and thus we will briefly describe it. Considering events data of magnitude pairs ( M Xobs , M Yobs ), a GOR relation is expressed in the following form: M Y 1 = a M X * + b (3) Fig. 2 – a) Frequency magnitude distributions, in normal and cumulative scale, calculated by considering M D (blue) and M L (red); b) seismic energy strain release calculated from both M D (blue) and M L (red). 36 GNGTS 2013 S essione 2.1