GNGTS 2022 - Atti del 40° Convegno Nazionale

GNGTS 2022 Sessione 2.2 317 Tab. 1 - Best Matching scenarios earthquakes calculated for the three reference sites, compared with the disaggregation results from Barani et al.,-- 2009, for the same sites and return period.       BMSE     Mean values from Barani et al. (2009) Site Mw Repi [km] ε T (s) M^ w Re^ pi [km] ε ^ Milano 5.8 31 0 0.2 5.1 65 1.88 1 5.5 109 1.58 Bologna 6 14 0 0.2 5 9.6 1.11 1 5.4 18.6 1.4 L’Aquila 6.6 20 0 0.2 5.8 9.3 1.13 1 6.5 18.2 0.91 shown on the left together with the nearest seismogenic sources of the ZS9 model used in the MPS04 SH analyses (Meletti et al. , 2008). The circles show the scenario epicentral distances R epi . The best fitting spectra are shown on the right (in red dashed line) together with the corresponding target UHS spectra (grey solid line). As remarked, ε was considered equal to zero. Results are in close agreement with the seismogenic framework of the analysis: the best matching scenario is an earthquake that is compatible with the sources that have been used to calculate the hazard. For Milano, for instance, the best fitting scenario is an earthquake that may be generated on source 907 or 911, at a distance of about 30 km, with a magnitude that is quite near the maximum magnitude that characterizes the seismogenic sources (shown on each plot). Table 1 shows the calculated BMSE for the three reference sites compared with the probabilistic disaggregation results from Barani et al. 2009 study, for the same return period but different structural periods. The disaggregation performed by Barani et al. shows different scenarios for each site: a smaller magnitude local scenario, mostly contributing to the higher frequencies, and a higher magnitude distant scenario, dominating at longer periods. Note that, while the results from the BMSE are related to the median predictions of the GM model ( ε =0), the results from Barani et al. have ε values considerably higher than 1, and this is the reason for the difference in the (M w – R) scenario values. The higher ε values are related to the coda of the distributions of the GM models having higher uncertainties. The discrimination among different best matching scenarios for the same site is currently not possible, but further research is being under development. Conclusions. As a first conclusion, this work shows how the results obtained with the Best Matching Scenario Earthquake fitting procedure were found to be generally consistent with the seismotectonic framework of Italy: the calculated earthquake scenarios showed values of magnitude and distance in general agreement with the seismogenic sources used to calculate the hazard at most of the considered sites. Although these results are just preliminary, they are indeed quite appealing in the sense of finding an alternative approach that may overcome the drawbacks of probabilistic disaggregation. As main advantages, the resulting scenarios: (i) are not related to a specific structural period, (ii) do not depend on the choice of the moment value of the statistical distribution of magnitudes and distances, (iii) can be obtained using the same GMMs of a PSHA, with the same weighting, (iv) can be of immediate and simple use even to non-expert users. As future research, the authors are working on further site-specific validations, as well as on the possible identification of more than one ‘best’ scenario, which is of particular interest in areas where the hazard is dominated by different sources.