[S-SS04] New trends in data acquisition, analysis and interpretation of seismicity

[SSS04-04] Moment tensor inversion of the 2023 seismic swarm in the Campi Flegrei (Southern Italy) volcanic complex

*Giacomo Rapagnani¹、Simone Cesca²、Gilberto Saccorotti³、Francesca Bianco⁴、Francesco Grigoli¹ (1.University of Pisa、2.GFZ Potsdam、3.INGV, Pisa、4.INGV, Osservatorio Vesuviano)

キーワード：Volcano Seismology, Moment Tensor Inversion, Seismic Swarm, Seismic Data Analysis

Campi Flegrei (CF) is a volcanic system located in a densely populated area west of Naples, Southern Italy. CF is the largest caldera system in Europe, whose evolution is marked by a long history of eruptions. The most prominent feature of CF activity is ground deformation (so-called ‘bradyseism’) consisting of rapid uplift (usually with seismicity) followed by slow aseismic subsidence phases. The most recent awakening episode started almost 20 years ago, climaxing during late 2023 with uplift and seismicity rates of up to 15 mm and 1000 events per month, respectively. More than 30 earthquakes had magnitudes greater than 2.5. The largest events had a magnitude greater than 4.0 and were largely felt throughout the region, causing significant concern among the population. In this work we analyze the seismic catalogue from the Osservatorio Vesuviano, a branch of the Italian National Institute for Geophysics and Volcanology (INGV), for the last 10 years. We select events with a Magnitude larger than 2.5 and extract seismic waveforms from INGV’s national seismic network that in the area has about 50 stations in the 50 km range from the caldera center. We then investigate the source processes of this data using full-waveform Moment Tensor (MT) inversion. Analysis of the CF recordings is challenging due to low Signal-to-Noise-Ratio (SNR) conditions and uneven geometry of the seismic monitoring network. We thus perform MT inversion using an advanced approach based on a probabilistic framework, implemented in the open-source software code Grond, part of the Pyrocko environment. Grond’s inversion approach searches for the source-time-function and MT components which minimize the misfit between observed and synthetic seismograms. Green’s functions are calculated in a realistic, 1-D velocity model used for routine monitoring operations. The inversion is conducted both in the time domain, including the waveforms from the P- and S-wave arrivals, and in the frequency domain, to get a better constraint on localizations and magnitudes. The inversion is also conducted using the cross-correlation between the synthetic and real seismograms, in order to properly account for uncertainties in signal amplitude and phase reflecting unmodeled features of the velocity structure. We present here the preliminary results of this analysis.