Quantitative analysis of the tremor wavefield at Etna Volcano, Italy

The properties of volcanic tremor wavefield at Mt. Etna Volcano, Italy, are investigated using data from two dense, smallaperture arrays of short-period seismometers deployed on the North and South flank of the volcano. Spectral analysis shows that most of the seismic energy is associated to several, narrow spectral peaks spanning the 1–5 Hz frequency band. Analysis of simultaneous recordings evidences that most of these peaks are common to different sites, thus suggesting a source effect as the origin of this energy. Frequency-slowness analyses evidence a complex wavefield, where body- and surface-waves alternatively dominate depending on the frequency band and component of motion taken into account. Surface waves are found to dominate at frequencies below 1 Hz and above 3 Hz. Conversely, the 0.8–2.3 Hz vertical- and radial-component wavefields at both arrays exhibit a nondispersive nature, with apparent velocities spanning the 1–2 s/km range. Particle motion analysis suggests these arrivals are associated to both P- and SV-waves inciding at shallow angles. At the northern array, back-azimuths of these waves encompass the whole summit crater area. At the southern array, back-azimuths are instead clustered around a direction pointing about 500 m east of the SE crater. At frequency around 4 Hz, the dominant direction of wave propagation at the southern site shifts about 30jW, pointing to the Bocca-Nuova/Voragine craters, and concordance of location is found with the source imaged by the northern array. The 0.8–2.3 Hz transverse-component of motion depicts velocities of about 0.5 km/s, a value which is about three times lower than those associated to the vertical and radial components. Results from polarization analyses at the two array sites depict the dominance of horizontal, linear particle motion oriented transversally with respect to the source direction. Polarization ellipsoids at the stations of the sparse network all depict a quasi-horizontal setting. With two exceptions, the direction of particle motion is always oriented tangentially to the summit volcanic edifice. The origin of the large transverse motion observed at the two array sites is thus attributed to SH waves generated by free-surface interaction of waves impinging the concave topography. The correlation method is used to derive the dispersion properties of short-period (0.5–5 Hz) Rayleigh waves, from which the shallow shear-wave velocity structures are derived for beneath the two semicircular arrays. Using a probabilistic approach, we invert slowness data measured at the two dense arrays for retrieving source location and extent. The joint inversion of slowness data from the two arrays point to different sources. This observation is interpreted in terms of ray bending associated to lateral heterogeneity and/or strong topographic effects on wave propagation. Once the propagation effects are taken into account, the most probable source locations are associated to a shallow region encompassing the summit craters and the eruptive fissures active at the time of the experiment (September 1999)