GNGTS 2022 - Atti del 40° Convegno Nazionale

326 GNGTS 2022 Sessione 2.2 EMPIRICAL NONERGODIC GROUND SHAKING SCENARIOS FOR CENTRAL ITALY: METHODOLOGY AND APPLICATIONS S. Sgobba, F. Pacor, G. Lanzano Istituto Nazionale di Geofisica e Vulcanologia (INGV), Sezione di Milano, Italy Earthquake-induced shaking scenarios represent increasingly crucial elements for the description of seismicmotion and its spatial variability, which are key-elements in understanding the processes of past events as well as for seismic risk purposes in engineering applications and civil protection strategies. Today, great resources are devoted to the generation of such scenarios mainly based on the adoption of physically-based 3D numerical simulations, performed with different techniques, which provide a physical representation of the finite fault rupture process, seismic wave propagation and attenuation. However, these methods, while providing more realistic fields of motion, often require complex modeling of the fault rupture and the crustal medium, apart from being computationally expensive. Empirical approaches to shaking scenarios generation have also emerged in recent years (e.g. USGS ShakeMap; Worden et al. , 2018), which are essentially based on the use of existing ground motion models (GMMs), providing the probability distribution of a given ground motion parameter as a function of basic explanatory variables, such as event magnitude, source-site distance, and soil conditions. Although GMMs provide robust estimates of the median ground motion, they are affected by inherent limitations such as the insufficiency in dealing with the spatial correlation that causes isotropic and stationary description of ground motion, a poor link with the physical parameters and, mostly, the assumption of ergodicity. The latter means that the models are calibrated on a global scale, assuming that the variability at a single site from a specific source is identical to that derived from multiple sites over large regions (Anderson and Brune, 1999). This hypothesis causes a great uncertainty associated with the model’s predictions, thus essentially neglecting those characteristics that imprint to seismic motion peculiar and systematic effects on a regional scale, such as source scaling and attenuation features. More accurate predictions can be obtained by relaxing the ergodic assumption in favor of non- ergodic approaches, in which the repeatable terms of variability due to source-, path- and – site effects are used to provide region-specific corrections of the median predictions, as well as to move part of the aleatory variability into epistemic uncertainty (e.g. Rodriguez-Marek et al., 2013; Landwehr et al., 2016, Kuehn et al., 2019; Abrahamson et al., 2019, Sgobba et al., 2019, Lanzano et al., 2021, Parker et al., 2022, among others). An example of a fully nonergodic methodology for Central Italy is provided by Sgobba et al. (2021), which is the base of the approach presented here. Methodology. Following the study by Sgobba et al. (2021), the estimates of an ad-hoc regional GMM calibrated at reference level (i.e. without site amplification) are adjusted for between-event and source region correction and combined with the maps of the source-to- site paths and site-response, being capable of generating nonergodic predictions of ground shaking on a regular grid (1.6x1.6 km) . The systematic and repeatable effects of ground motion, estimated via residual analysis, are identified in the following components (more details can be found in Sgobba et al. , 2021): • A correction of the location-to-location terms (δL2L) defined as the average between- event term within the source region. The location terms are computed by means of a clustering approach in which the events are aggregated and averaged within polygonal areas identified on the basis of spatial-temporal criteria of clustering. δL2L reflect the average stress-drop within the source region; • A map of the propagation effects based on the (non-stationary) spatial correlation of the path-to-path terms (δP2P), i.e. the systematic deviations of ground motion along each