GNGTS 2019 - Atti del 38° Convegno Nazionale

546 GNGTS 2019 S essione 3.1 because seismic ambient noise is mostly originated by coupling between the solid earth and the oceans, it is reasonable to expect that this is best achieved by deploying instruments on an island. At the end of June 2016, our team has deployed an array of broad-band seismic stations (Trillium Nanometrics 120s posthole broad-band seismometers) across Sardinia (Fig. 1). This temporary deployment was complemented by three permanent stations belonging to the Italian MN and IV networks. Most stations recorded continuously for 24 months. The relationship we have determined involves a proportionality factor, relating normalized cross-correlation and the theoretical Green’s function, that had been neglected in previous studies (e.g., Prieto et al. , 2009; Lawrence and Prieto, 2011). The need to account for such a factor was pointed out theoretically by Tsai (2011), while Harmon et al. (2010) and Weemstra et al. (2013) introduced it as a free Fig. 1 - (a) Geographical locations of receivers (black triangles with station names) over the island of Sardinia. (b) Distribution of interstation distances for all station pairs in our deployment; the mean and median of the distribution are 114.45, 104.95 km, respectively. Acronyms starting with the letters UT identify stations deployed by our team. Fig. 2 - Proportionality factor relating ambient-noise cross correlation and the (attenuated) Green’s function (eq. (30) of Boschi et al. , 2019). Its numerical value is evaluated based on observed dispersion curves, and estimates for α obtained by minimization of three differently cost functions C 1 , C 2 , C 3 , as explained in detail by Boschi et al. (2019). Panels (a) through (d) correspond to station pairs UT.006–UT.009, UT.002–UT.004, IV.AGLI–IV.DGI and UT.002– UT.003, respectively (see Fig. 1).