GNGTS 2024 - Atti del 42° Convegno Nazionale

Session 1.1 GNGTS 2024 Detect and characterize swarm-like seismicity L. Passarelli 1 , S. Cesca 2 , L. Mizrahi 3 , G. Petersen 2 1 INGV - Isttuto Nazionale di Geofsica e Vulcanologia, sezione di Bologna, Bologna, Italy. 2 GFZ – German Research Centre for Geosciences, Potsdam, Germany 3 SED – Swiss Seismological Service, ETH Zürich, 8092 Zürich, Switzerland Tectonic earthquake swarms do not follow the classical temporal and spatal patern observed for mainshock-afershock sequences. Unlike the typical occurrence of a largest mainshock that initates the sequence and triggers afershocks following an Omori-Utsu decay and productvity scaling with the mainshock magnitude, earthquake swarms show a distnctve increase of seismic actvity without a clear mainshock. Typically, the largest earthquake(s) occur(s) later in a swarm sequence that ofen consists of multple earthquake bursts that show spatal migraton. This erratc clustering behavior of earthquake swarms comes from the interplay between the long-term accumulaton of tectonic strain and short-term transient forces driving swarm-like seismicity. The detecton and investgaton of earthquake swarms challenges the community and ideally requires an unsupervised approach, and indeed numerous algorithms have emerged in the last decades for earthquake swarm identfcaton. In this study, we comprehensively review commonly used techniques for detectng earthquake clusters. We frst applied those techniques to thousands synthetc earthquake catalogs produced with state-of-the-art ETAS model, featuring a tme-dependent background rate to simulate realistc swarm-like sequences. Our approach enables us to identfy boundaries in some parameters commonly used to distnguish earthquake clusters, says mainshock-afershocks versus swarm sequences. The insights gained from synthetc data contribute to a more accurate classifcaton of seismicity clusters in real earthquake catalogs. We therefore apply the same algorithm to real cases of seismicity where earthquake swarms have already been identfed, i.e. the 2010-2014 Pollino seismic sequence; the Húsavík-Flatey transform fault seismicity and the regional catalog of Utah. However, applying these fndings to real cases is contngent upon the clustering algorithm used, the statstcal completeness of catalogs, and the spatal and temporal distributon of earthquakes. Unfortunately, full automaton of swarm detecton and characterizaton remains difcult to atain, necessitatng manual verifcaton and investgaton of individual swarm-like sequences. Corresponding author:  luigi.passarelli@ingv.it