GNGTS 2017 - 36° Convegno Nazionale

300 GNGTS 2017 S essione 2.1 Squares approach (GR-LS) and one adopted the Maximum Likelihood (GR-ML) method of Weicher (1980). A weight of 0.4 was applied to the ML branch and only 0.2 to the LS one. The maximum magnitude was evaluated for macroareas representing portions of the Italian territory and surroundings for which it is expected a homogeneous tectonic behaviour. It represents the magnitude value for which the rate of occurrence is considered negligible for a mean return period of 30-5000 years. The data used for the M max estimates were the historical earthquake catalogue (CPTI15) and the composite seismogenic database (DISS 3.2.0). M max1 is equal to the highest magnitude observed or computed (from the dimensions of the faults) for the zone, increased of its standard deviation. M max2 is assumed conservatively equal to M max1 +0.3 (Fig. 2). A weight of 0.5 is applied to each branch. To compute the preliminary hazard results only one ground motion prediction equation (GMPE: Akkar et al. , 2014) was applied, both using the Joyner-Boore and hypocentral distance versions. This first preliminary proposal was subsequently examined by a group of elicitorswho showed some perplexity and given some suggestions. Following the indications of the elicitors: • ��� ����� ������ ��� ���� �������� �� ����� �������� ��� �� ������ �� ����� ���� �� �� the GR-LS branch has been deleted: in fact, although the LS method is often used it is not formally suitable since magnitude is not error free, cumulative event counts are not independent, and the error distribution of the number of earthquake occurrences does not follow a Gaussian distribution; • ����� ���� ����� ����� ���� ���� �� ������� ��� ������������� �� since some small areas were poor in events, the corresponding GR b -values were in some cases abnormal values, far from the theoretical “world” value of 1. For this reason, it has been chosen to merge the poorly documented zones into wide regions when showing homogeneity, according to the information available, in themain type of failuremechanism, and in the completeness and M max macroareas, as defined by the project managers. The b -value has been calculated for these homogeneous regions and associated to each of the SZs contained inside. Conversely, the SZs have been kept independent for the a -value computation; • ����� �� ��� � rates in the Share areas covered by the CPTI15 catalogue have been calculated. In the meantime, a new version of the CPTI15 was released, again with maps annexed for statistical and also historical completeness. For these reasons, a new logic tree of eight branches has been considered (Fig. 1b): two branches to take into account historical and statistical estimates of completeness, with a weight of 0.5 applied to each branch; two branches for the seismicity model: one branch accounted for individual rates (I-R, with a weight of 0.4) and one for the GR-ML estimates of the a and b -values (with a weight of 0.6). To complete the logic tree, two branches were related to alternative M max estimates as in the previous computation: a weight of 0.5 has been again applied to each branch. To compute the new hazard results, the Akkar et al. (2014) GMPE was again applied, using the Joyner-Boore as distance metrics. The comparison between the 2016 (Fig. 3a) and 2017 (Fig. 3b) calculations, using the Joyner- Boore distance, shows a substantial consistency of the results between the two maps, except in Calabria, where the new b -value computation has led to a general increase of PGA expected values. A comparison with the results of the MPS04 (Fig. 3c) and SHARE (Fig. 3d) maps shows that the PGA values calculated in the present study are fairly consistent with those of SHARE, while they are higher than those of MPS04 throughout the national territory. This fact is probably explained by the different GMPEs used in the three analyses. In the present study, the largest values of PGA are found along the northern Apennines, and this result is probably related to the small size of some SZs in this area, compared to the MPS04 and consequently SHARE ones. References Akkar S., Sandikkaya M. A., Bommer J. J.; 2014: Empirical Ground-Motion Models for Point- and Extended- Source Crustal Earthquake Scenarios in Europe and the Middle East, Bulletin of Earthquake Engineering, 12(1), 359 -387.