GNGTS 2018 - 37° Convegno Nazionale

250 GNGTS 2018 S essione 1.3 of the mobile network, three of which transmit the signal in real-time, acquire locally and are equipped with broad band, short period sensors and accelerometers. The 15 stations currently installed on the island have different technical characteristics and their positions are shown in Fig. 1. The preliminary dataset consists of ambient noise records of 10 of the 15 broad band stations between the months of June and September 2018. Array analysis. We applied the MSPAC method (Bettig et al. , 2001) to the seismic noise recorded by 10 stations (Fig. 2 top left), given that this method allows to compute the spatial autocorrelation even if the array geometry is not semi-circular. We divided the array in semi- circular sub arrays called “Rings”, whose radii are defined by the sensor’s spacing, and we calculated the spatial autocorrelation for all the possible pairs of sensors. Each Ring is the result of an appropriate balance between the number of sensor pairs per Ring (as large as possible) and the thickness of the Ring (as small as possible). Thus our geometry is composed by 6 Rings, ranging from a minimum of about 1000 m to a maximum of about 8400 m spacing of the sensors, for a total of 46 sensor pairs. Having fixed time windows and frequency band, the MSPAC method calculates the autocorrelation between the station pairs and averages over the azimuthal directions. The time windows are set to a period of T = 400 s, with a 20% overlap, for each frequency in the range 0.1-2.0 Hz, using 100 logarithmically spaced frequencies. In order to evaluate the phase velocity, a frequency-phase-velocity histogram has been derived by inverting the azimuthally averaged autocorrelation curves (Fig. 2). To validate the dispersion results of the array obtained with the MSPAC technique, we applied the Frequency-Wavenumber (f-k) technique (Lacoss et al. , 1969) to the same data-set in the same frequency range. We found that the f-k results are consistent with the MSPAC results and the additional curve is useful to deduce stable estimations and to improve the interpretation by reducing erroneous, potentially biased data. Spectral Analysis. Taking into consideration the resolution limits of the geometry of the selected stations (Fig. 2 top left), we can see that the available frequency interval for the analysis ranges from 0.1 Hz to almost 1.5 Hz. For this reason we decided to compute the horizontal to Fig. 1 - Map of Ischia Island. The red triangles indicate the permanent seismic stations in transmission. The yellow triangles indicate the mobile seismic station - local acquisition. The green squares indicate the mobile seismic station in transmission.