GNGTS 2023 - Atti del 41° Convegno Nazionale

Session 2.1 GNGTS 2023 Development and application of probabilistic seismic hazard models at a national scale in Italy: critical issues in a historical perspective M. Stucchi 1 , C. Meletti 2 1 INGV Bologna 2 INGV Pisa While at a European scale probabilistic seismic hazard models were developed mainly looking at both building code and insurance problems (GSHAP, Giardini et al., 1999; SESAME, Jimenez et al., 2001; SHARE, Woessnet et al., 2015; GEM, Pagani et al., 2018; SERA, Danciu et al., 2021), at a national scale in the last 50 years such models have been built in the perspective of the building code, only. We are referring here first to the models - and their application - developed in the frame of initiatives between 1980 and 2000: Progetto Finalizzato Geodinamica (PFG, Gruppo di lavoro scuotibilità, 1979), Gruppo Nazionale per la Difesa dai Terremoti (GNDT, Slejko et al., 1998), Servizio Sismico Nazionale (SSN, Romeo and Pugliese, 2000), SSN and GNDT (Albarello et al., 2000). Next, to those developed between 2000 and 2021) in the frame of Istituto Nazionale di Geofisica e Vulcanologia (INGV; Stucchi et al., 2011; Meletti et al., 2021). With the exception of the PFG one, all models were in some way funded by Italian Civil Protection. Although such perspective has been positive because it has provided funding and goals, it has also constrained the model’s development to be immediately suitable for supporting the building codes. In addition, it has sometimes determined a sort of “mingling” among customers, developers and final users which has partially affected the useful multidisciplinary collaboration. To develop new models as soon as new data, approaches and requests emerge is part of the current scientific procedure. Traditionally, the most recent one is considered the “best one”, because input data, approaches and software are the most updated. Moreover, one may presume that developers know how previous models have been developed. On the other hand, to evaluate the quality of the models and compare them is difficult and, definitely, not so straightforward. Even the often invoked “validation” of the models against data from later earthquakes encounters practical and conceptual difficulties; actually, it is possible only perform tests of the model against data/observables as a reliable consistency check of the models. In this paper we will first compare seismic hazard values from the above-mentioned models at some localities. Then we analyse how some of the model’s components, the computation approaches and the hazard descriptors have changed along time.