GNGTS 2022 - Atti del 40° Convegno Nazionale

GNGTS 2022 Sessione 2.1 197 PROPOSAL OF ZONATIONS FOR THE REGIONALIZATION OF THE GROUND MOTION MODELS IN ITALY G. Brunelli, G. Lanzano, L. Luzi, S. Sgobba Istituto Nazionale di Geofisica e Vulcanologia (INGV), Milano, Italy Introduction . The aim of this work is to identify homogeneous areas (e.g. zonation) where ground motion observations are spatially clustered and biased with respect to the median predictions of a GMM calibrated at a national scale. This goal is reached by the decomposition of the residuals, i.e. the logarithmic difference between observations and the predictions of a GMM, and their spatial analysis. The reference GMM for Italy, adopted in this study, is the most recent model calibrated for active shallow crustal events by Lanzano et al. (2019a), hereinafter ITA18, and for the ordinates of the acceleration response spectra (5% damping) of the horizontal components. The zonation may be adopted to apply regional adjustments to the reference GMM, thus accounting for local source and attenuation effects. This may serve to improve the model accuracy for PSHA purposes or site-specific hazard studies, as well as for empirical earthquake scenarios and ShakeMap (Worden et al. , 2018; Michelini et al. , 2020). Data and methods . For the aim of the proposed study, we exploit the ITACAext dataset (Brunelli et al. , 2022), that includes recordings of earthquakes of magnitude equal to or greater than 3.0, that occurred during the period 1972-2020, in Italy and in the neighboring countries. Further details are reported in Brunelli et al. (2021). The final subset for the analyses consists of 37,098 records of 1,863 earthquakes, recorded by 1,922 recording stations. In order to interpret the results of the residual analysis, we homogenize the estimates of moment magnitude (Mw) of the events by converting Mw calculated with the TDMT method (Time Domain Moment Tensor; Dreger et al. , 2003) into Mw estimated with the RCMT method (Regional Centroid Moment Tensor, RCMT; Pondrelli et al. , 2002), because in ITA18, the explanatory variable for magnitude scaling is the RCMT-Mw. For this purpose, we adopt the relations by Brunelli et al. (2021), modifying the empirical equations, initially proposed by Gasperini et al. (2012-2013). The core of the analysis is the residual computation, i.e. difference between the (natural) logarithm of the observed ground motion log(Y) obs and the corresponding prediction log(Y) pred , from a reference GMM. R es is computed using ITA18, assuming R rup as distance metrics and the ground motion parameters are the ordinates of the acceleration response spectra (5% damping) in the interval of vibration periods 0.01-10s. According to the well-established Al Altik et al. (2010) approach, the residuals R es are decomposed into repeatable terms referring to event (δB e ), station (δS2S s ) and event- and site- corrected residuals (δW es ), by using a mixed- effect regression (Stafford, 2014; Bates, 2015). Several authors (Bindi et al. , 2018a; Bindi and Kotha, 2020) show that δB e is correlated with source characteristics not captured by the standard magnitude scaling in GMMs, such as stress drop. The δW es , instead, represent the aleatory leftover part of the residuals and should include regional propagation features or effects that are not captured by the reference model, such as directivity effects (Colavitti et al., 2022). In this study, we investigate the spatial trend of between-event residuals, δB e , to study the characteristics of the seismic sources and the δWes to find the anisotropies of the anelastic attenuation. Proposed zonations . Source effects . For the purpose of regionalizing the source contribution to the ground motion, we consider, as a starting point, the seismogenic zonation ZS16 (Meletti et al. , 2019), developed ad-hoc in the framework of the latest probabilistic seismic hazard (PSH) model for Italy (MPS19, Meletti et al., 2021). The proposed zonation modifies the initial ZS16, on the basis of the spatial distribution of the δB e at short period (the vibration period interval