GNGTS 2022 - Atti del 40° Convegno Nazionale

44 GNGTS 2022 Sessione 1.1 DETECTION OF LOW FREQUENCY EARTHQUAKES AT MT. VESUVIUS BY USING A FREQUENCY DOMAIN APPROACH R. Manzo 1,2 , D. Galluzzo 2 , M. La Rocca 3 , L. Nardone 2 , R. Di Maio 1 1 Dipartimento di Scienze della Terra, dell’Ambiente e delle Risorse, Università degli Studi di Napoli Federico II, Napoli, Italy 2 Istituto Nazionale di Geofisica e Vulcanologia, Sez. di Napoli, Osservatorio Vesuviano, Italy 3 Dipartimento di Biologia, Ecologia e Scienze della Terra, Università della Calabria, Rende, Italy Introduction. The most common and recognised typologies of seismic events occurring in volcanic environments can be classified as: high frequency ( HF ) events, often identified as volcano-tectonic earthquakes ( VT ), earthquakes associated with explosive activity (explosion quakes), low frequency ( LF), long period LP seismic events and volcanic tremor. The assessment of a volcano’s eruptive potential, and the variation of its internal dynamics, can be provided through detection, identification and analysis of the different typologies of seismicity. In particular, the onset of low frequency events ( LF or LP ) or volcanic tremor is considered as a potential indicator of the awakening of volcanic activity related to in-depth magma movement or injection of fluids. Although high volcanic risk areas are monitored by modern high-sensitivity seismic instrumentation, the background noise compromises the detection and study of low-energy earthquakes. In this framework, a new multi-methodological analysis approach is provided to overcome this critical issue. In particular, in this note we present the results of the proposed approach applied to the densely urbanized area of Mt. Vesuvius, where low amplitude seismic signals due to local seismicity are recorded by a monitoring network equipped by broad band high sensitivity seismic stations. Our aim is to detect small low frequency events into the background noise by means of methodologies that operate in the frequency domain. Methods. Nowadays, big data, not always manageable by seismologists’ experience, requires new detection methodologies. To this end, the approach we propose focuses on the analysis of seismic signal spectral content using the following two techniques jointly: - Coherence analysis : The coherence ( K xy ) between two seismic signals ( x and y ) provides an indication of the similarity of their respective spectra in a specific frequency band: where C xy (f) is the cross-spectrum and P x (f) * P y (f) is the eigenspectral product. Coherence gives values between 0 and 1. - Spectral parameters analysis : it takes into account the central frequency ( Ω ) and the form factor ( δ ) from the statistical moments of the seismic power spectrum. The first parameter represents the value around which most of the spectral energy is centered; the second one indicates the distribution of the spectral amplitudes around the central frequency. Case study: Mt. Vesuvius. As it is well known, Mt. Vesuvius (southern Italy) represents one of the highest risk volcanoes in the world. In the last 40 years, it has been characterized by: volcano-tectonic events ( VT , about 900 earthquakes per year in the last two years) of small magnitude (M D < 3), sometimes occurring in low energy swarms and at depths between 0 and 2 km b.s.l.; several low-frequency events ( LF ) characterized by deeper hypocenters (La Rocca and Galluzzo, 2016). The two techniques described above were applied to the continuous seismic signals recorded from 2019 to 2020 by the monitoring seismic network installed at Mt.Vesuvius (Fig. 1). Taking into account some benchmark stations characterized by low