GNGTS 2023 - Atti del 41° Convegno Nazionale

Session 3.1 GNGTS 2023 Ambient noise cross-correlations using local seismic monitoring networks I. Barone 1 , V. Cascone 2,3 , A. Brovelli 2,3 , D. Dupuy 3,4 , J.M. Lavanchy 5 , G. Cassiani 1 1 Dipartimento di Geoscienze, Università degli Studi di Padova, Padova, Italy 2 Isamgeo Italia srl, Gallarate (VA), Italy 3 Seismoring Sàrl, Yverdon-les-Bains, Switzerland 4 Geo2x, Yverdon-les-Bains, Switzerland 5 CSD Ingenieurs SA, Lausanne, Switzerland Monitoring both the natural and possibly induced seismicity in geothermal or gas storage reservoirs is essential to guarantee safe industrial operations, especially in densely populated areas. Microseismic monitoring networks are composed of several three-component seismic stations, continuously recording. These data are used to identify and locate microseismic events originating in the vicinity of the reservoir (Wang et al. 2016). The location process makes use of seismic velocity models, both for compressional (P) and shear (S) waves. While reflection seismic could help constraining the P-wave velocities, S-wave velocities are difficult to measure, especially in an intermediate depth range (between the near surface and the engineering bedrock; Cascone et al. 2022). We propose to use the continuous ambient noise data recorded by microseismicity monitoring networks, composed by short-period seismometers, to derive a shear-wave velocity model of the subsurface, successively used to improve the location of seismic events. When the ambient noise azimuthal distribution is sufficiently homogeneous, the Green’s function between each station couple can be retrieved through cross-correlation analysis (ambient noise interferometry; Wapenaar et al. 2006, Curtis et al. 2006). Depending on the interstation distances and on the frequencies characterizing the noise, the depth of the obtainable shear-wave velocity model could span between a few hundred and a few thousand meters (Barone et al. 2022). Two viable procedures to analyse ambient noise are here proposed. The first consists in the retrieval of the time-domain cross-correlograms, from which the average group- and/or phase-velocity dispersion curves are extracted (Bensen et al. 2007). The second approach consists in the analysis of the zero-crossings of the real part of the cross-correlation spectrum for each station pair, to retrieve the phase velocity dispersion curve (Kästle et al. 2016). This approach is capable of highlighting lateral velocity variations, but its drawback is the need for a sufficiently accurate starting velocity model, such as a regional model. Finally, the obtained dispersion curves need to be inverted to find the shear-wave velocity structure.