GNGTS 2024 - Atti del 42° Convegno Nazionale

Session 3.2 GNGTS 2024 From the beginning of April 2023 the audio sample frequency was reduced to 16 kHz, and the vertcal frequency axis scale was set logarithmic covering the interval between 10 Hz and 8 kHz. In this way, spectrograms become comparable with the ELF ones for the electric feld, and the spectral content in the lower part of the audio sound becomes more intelligible. During the Mefte staton maintenance, it was possible to verify the good state of the microphone afer several months of audio recording despite its contnuous exposure to the corrosive emissions at Mefte. Indeed, the device was protected by a hood that was put on to shield it from gusts of wind for land audio recordings. While, the copper wires, although covered by sheath, were corroded perhaps due to small cracks in the sheath. Thus, a further comparison was momentarily not possible between electric and acoustc signals. The spectral content of the lower part of the audio band, now visible by logarithmic scale, immediately evidenced a rich variety of eigenfrequencies. Furthermore, their amplitudes presented non-constant intensites both in a collectve sense and in a singular sense. Some acoustc resonances appeared sporadically during the intense increase of the Mefte’s blow. The Fourier analysis was partcularly investgated at the tme of small seismic events that occurred a few tenths of km away from the site. An example is shown in Figure 2, where more than 30 hours of audio recording have been reported startng from the day before the Mirabella M=2.1 earthquake occurred at the beginning of April 19, 2023. Hypocentre depth occurred at less than 10 km at a distance of around 20 km. The seismic event was preceded by a sensible noise increase that started about 11 hours before and was maintained for about 6 hours, then slowly faded over many hours. The noise increase was in turn preceded by two puf dampening lastng approximately one hour each, one immediately before the noise increase and the other around 4 hours before it. The earthquake tme was indicated by a cyan vertcal arrow and the colour legend permited the evaluaton of a noise increase of about 10 dB during the most intense emission, and audio suppressions of up to 10 dB during the dampening. Three sudden pufs with a 10 dB noise increase were evidenced. Moreover, the frequency band between 100 and more than 1000 Hz also showed an evident increase during the maximum noise manifestaton. The main eigenfrequencies of Figure 2 are at 45 and 65 Hz. They maintain constant intensity for a long tme before the seismic event. Other less apparent horizontal bands occur at 103 Hz, and 125 Hz, other two bands at 300-400 Hz and 700-800 Hz, and many weak horizontal bands above 1000 Hz. More detailed observaton of the power spectra in Figure 2 shows the entre series of 19, 45, 65, 103, 125, and 170 Hz, which appear more clearly during the damping hour preceding the signifcant increase of the audio noise of Figure 3 middle. Instead, longer periods of audio noise modulaton whose duratons were between 5 and 10 minutes are shown in Figure 3 lef. Details of the third puf are shown in Figure 3 right, where a noise increase lastng two minutes between 10 and 40 Hz was accompanied by a well-defned audio note at 55-80 Hz with frequency increased in less than 2 minutes. It could be thought of as a resonance excitaton of the fnal conduct due to the sudden intensifcaton of the fux.