GNGTS 2021 - Atti del 39° Convegno Nazionale

GNGTS 2021 S essione 2.1 234 A CONTRIBUTION TO THE PROBABILISTIC EARLY WARNING AND FORECASTING FOR EARTHQUAKE-GENERATED TSUNAMIS M. Zanetti 1 , A. Armigliato 1 , S. Tinti 1 , E. Baglione 2,3 , F. Zaniboni 1 , G. Gallotti 1 , C. Angeli 1 1 Alma Mater Studiorum – Università di Bologna, Dipartimento di Fisica e Astronomia “Augusto Righi”, Bologna, Italy 2 Istituto Nazionale di Geofisica e Vulcanologia (INGV), Sezione Roma 1, Roma, Italy 3 Istituto Nazionale di Oceanografia e di Geofisica Sperimentale (OGS) - Trieste, Italy The main duty of a generic tsunami warning centre (TWC) is to issue, timely and with the minimum possible level of uncertainty, proper information, advisory or watch/alert messages after an earthquake with the potential to generate a tsunami occurs in the TWC’s jurisdiction area. Decision matrices (DMs) are the tools that have been used so far and are still in use by the majority of the worldwide TWCs, to select the proper level of warning to be issued. DMs classify the tsunamigenic potential of a given earthquake based on a combined evaluation of its hypocentral depth, of its distance from the nearest coastlines and of its magnitude. Although very rough and not connected to the details of the earthquake process and of the potential tsunami generation mechanism, DMs still represent a reasonable and easy-to-apply compromise between the requirement to provide timely information to local authorities and population and the scientific treatment of the process. Nonetheless, significant efforts are ongoing in the global scientific community to address the problem in more rigorous terms, i.e. by devising modelling and computational strategies, mainly based on probabilistic approaches, allowing to put the focus on the details of the rupture process taking place on an earthquake fault, as well as on the deformation of the seafloor that represents the initial condition for the tsunami. The strategies must be able to quantify at least the essential physical and geometrical properties of the earthquake fault and the main observables related to the ensuing tsunami, there including at least the time of propagation to different coastal segments and the geographic distribution of the expected (maximum) inundation levels. The aforementioned outputs must be provided with a proper and clearly defined level of uncertainty attached, and in a time short enough to be useful in an early warning perspective. The problem is dramatic especially for the coastlines close to the tsunamigenic earthquake fault: this is the typical situation in the Mediterranean, where the time between the earthquake occurrence and the first tsunami arrival can be as short as few minutes. The present study represents a contribution toward the probabilistic assessment in near-real time of the potential tsunami impact on a given coastal region based only on the early estimates of the parent earthquake’s location and magnitude provided by seismic monitoring networks. The described approach is inspired by the observation that the main features of the tsunami impact on the nearest coastlines is strongly related to the earthquake focal mechanism, to the position of the fault with respect to the coastline, to the pattern of the on-fault slip heterogeneity. The classical inversion procedures that are usually employed to retrieve the so-called finite-fault models (FFMs) take too long to be useful in a tsunami early warning approach. The strategy we propose can be summarized in the following steps: 1. fault geometry and on-fault slip distribution with associated uncertainties are computed through available regression laws against the magnitude provided by seismic monitoring networks; 2. the relative position of the hypocenter and of the fault surface, which is an intrinsically unpredictable information, is derived by fitting FFMs contained in global finite-fault databases (SRCMOD, USGS) with suitable probability density functions; a similar approach can be used to find possible relations between the position of the hypocenter and, for instance, the maximum of the slip pattern;