GNGTS 2016 - Atti del 35° Convegno Nazionale

264 GNGTS 2016 S essione 1.3 The source location of volcanic tremor is affected by uncertainty much larger than high frequency VT earthquakes due to the lack of impulsive phases recognizable at the local network stations. A careful inspection of the signal emergent onset and signal envelope at the many stations available at Vesuvius permits to locate the tremor epicenter in the crater area, but determining the source depth with a sufficient precision is much more difficult. We pursued such aim through a comparison of volcanic tremor with LFVT events and regular VT earthquakes. Acareful visual inspection of the three component seismograms of the stations characterized by the highest signal to noise ratio, and particularly those of the array stacked signals, shows the presence in the tremor signals of many P-S wave pairs with constant Ts-Tp. This feature has been observed for many of the analyzed tremors in the array stacked signals, and sometimes it is seen also at the summit stations when the signal to noise ratio is high enough. As an example, Fig. 2c shows the beginning of the tremor 20120329 at VAS array. The array stacked signals contain many pulses on the vertical component (P waves) followed after about 1.35 s by a corresponding pulse on at least one of the horizontal components (S waves). The constant Ts- Tp suggest that such P-S wave pairs are small VT earthquakes of low frequency, or in other words they are LFVTs as those described above. The estimated Ts-Tp time of such LFVT events in the volcanic tremor is in the range 1.25 s <= Ts-Tp <= 1.6 s at the summit stations. Since it is not possible to recognize the same phase at many stations due to the chaotic nature of the wavefield and the low signal to noise ratio, we can not locate the source of individual P-S wave pairs through the inversion of time picking at the network stations. Nonetheless the estimated Ts-Tp time is the most important information to constrain the depth of tremor source. The Ts-Tp of volcanic tremor estimated in the range 1.25 s <= Ts-Tp <= 1.6 s at the summit stations corresponds to a source depth between 5 km and 6.5 km below sea level (La Rocca and Galluzzo, 2016). LFVTs have two very important features in common with volcanic tremor: the same Ts-Tp time and the same frequency contents. A first straightforward conclusion inferred from this consideration is that the tremor is simply a sequence of LFVTs. We carried out a detailed analysis of these features and compared these events with regular high frequency VT earthquakes with the aim of gaining some insight about their source properties. We did this comparative analysis by taking into account regular VT located as near as possible to the same depth of tremor and LFVTs. We estimated the corner frequency of displacement spectra for all events characterized by Ts-Tp > 1.0 s at the summit stations, those near the epicenter. The results, shown in Fig. 3, depict two different groups of sources that do not merge to each other. VT earthquakes with Ts- Tp < 1.2s have corner frequency between 15 Hz and 22 Hz, while volcanic tremor and LFVTs all have Ts-Tp > 1.2 s and corner frequency smaller than 6 Hz. We searched the data recorded during the last 25 years looking for VT earthquakes with Ts-Tp > 1.2s and corner frequency greater than 6 Hz, but could not find any. We also searched for LFVTs and tremor characterized by Ts-Tp < 1.2 s, but could not find any. This result suggest the existence of a transition zone between shallow seismicity characterized by the typical brittle failure that produces high frequency VT earthquakes, and a deeper volume where shear failure radiates energy at much lower frequency. Such transition zone is located at depth corresponding to Ts-Tp = 1.2 s at the summit stations, which is estimated to be about 5 km bsl. Discussion and conclusions. Vesuvius is one of the few cases where bursts of volcanic tremor are observed at a closed conduit, quiescent volcano, without any apparent relationships with other volcanic phenomena. The presence of P-S wave pairs suggests LF shear failure as source model, rather than the interaction of fluids with surrounding rock. On the other hand, LFVTs and tremor form a group well separated from high frequency VTs with regard to corner frequency and source depth (Fig. 3), thus we conclude that their striking difference must be related with the rock properties at depth where they are located. Our results indicate a significant change of the medium mechanical properties with depth, reasonably in terms of stiffness, temperature, and perhaps the presence of nearly melt material. A significant change of these