A Bayesian seismic hazard analysis for the city of Naples

In this paper we explore the feasibility of formulating the hazard assessment procedure to include the information of past earthquakes into the probabilistic seismic hazard analysis, together with the use of an ensemble modeling technique. This strategy allows, on the one hand, to enlarge the information used in the evaluation of the hazard, from alternative models for the earthquake generation process to past shaking and, on the other hand, to explicitly account for the uncertainties. The Bayesian scheme we propose is applied to evaluate the seismic hazard of Naples. The framework in which we have embedded the tools is flexible to include all types of uncertainties. Here we focus on a sensitive study of the earthquake occurrence by implementing models that span from random to cluster-type temporal behavior and models that include quasiperiodic occurrence of earthquakes on faults. We implement five different spatiotemporal models to parameterize the occurrence of earthquakes potentially dangerous for Naples. Subsequently, we combine these hazard curves with ShakeMaps of past earthquakes that have been felt in Naples since 1200 A.D. The results are posterior ensemble hazard curves for three exposure times, e.g., 5, 10, and 50 years, in a dense grid that covers the municipality of Naples, considering rocky soil and including the site amplification. Our results show the importance to include the data from past shaking since the difference between the prior and the posterior is about 8–15% for the different exposure times.