GNGTS 2017 - 36° Convegno Nazionale

GNGTS 2017 S essione 2.1 277 - ���� ���� ��������� � � � ���� ��������� �� ����� ���� ��������������� MDPs with intensity I < 3 were discarded to avoid data incompleteness; - ������ ���� ��������� ������� ��� �� �������� ����������� ���� �������� events with cumulated effects due to damaging aftershocks were removed; - ������ ���� ����� ����� � ���� ���� ���������� events with focal depth > 33km were discarded; - ��������� ��� �������������� ����������� ���� ���� ��������� ��� �� ���������� ������������ off-shore and cross-national earthquakes were also discarded due to incomplete distribution of MDPs and ill-constrained epicenter location; - ������ ������������� �� ��� � events characterized by low I 0 and/or small Mw because the study will be focused on strong earthquakes; - ������ ���� ��� ���� ���� �������� ������� ����� ����� ���� ��� ���������� ��������� events with few MDPs were removed, because these could bias the regression analysis; - ��������� ����������� ����� ���� ���������� �� ������ ���������� �� ������� ����� �� carefully considering those MDPs associated to places identified as special cases by DBMI, such as TE (large territory), IB (isolated building), SS (small settlement), MS (multiple settlement), DL (deserted locality), AL (absorbed locality), CQ (city quarter), because the statistical nature of intensity might not be full filled; - ��������� ����������� ����� ���������������� ������������ ����������� �� ���� ����� carefully considering those non-conventional macroseismic intensities by DBMI (e.g. Felt, Damage, etc). After the application of these selection criteria, the resulting set of macroseismic intensities is made of 16.261 MDPs related to 118 events (i.e. input calibration dataset, Fig. 1) that occurred from 1908-2013. Modelling and conclusions. A typical functional formula is chosen in the present study to model the attenuation of Italian macroseismic intensity data as a function of epicentral distance and Mw using five free parameters: I = a – bR – cLogR + dMw with R=(x 2 +(e) 2 ) 0.5 i.e. x is the epicentral distance. The model parameters (a, b, c, d, e) are determined through a non-linear least squares method using KaleidaGraph software (Synergy Software, 2014) and the calibration dataset. The calibrated model assumes a point source and an isotropic macroseismic field, centered at the instrumental epicenter. The between-events residual (also called inter-event residual or inter-event term, Al Atik et al. , 2010) was computed to obtain the best possible performance earthquake-to-earthquake variability of the calibrated macroseismic intensity attenuation model that is: I = (1.8125 ± 0.10329) - (0.0038551 ± 0.000266)*R - (2.6096 ± 0.066535)*LogR + (1.4206 ± 0.0066)Mw with R=(x 2 +(9.87) 2 ) 0.5 . Tab. 1 summarizes the range of the calibrated model. Time No No Mw x (km) h (km) I Ixx /Io range Events MDPs Range Range focal depth Range Range [1908-2013] 118 16.261 [3.82-7.10] [0.11-634] h<33 [3-11] [5-11] The calibrated macroseismic intensity attenuation model (Figs. 2 and 3) is characterized by ΔI-residuals (observed –predicted) as Gaussian curves (normal distributions, Gomez Capera, 2006). The calibrated macroseismic intensity attenuation is characterized by a lower standard deviation (σ=0.75) than the one reported in Gomez-Capera et al. (2010), and resulting in better quality fit factors for each calibrated attenuation coefficient. A validation process will be the next phase of this study. Tab. 1 - Range of the calibrated model.