GNGTS 2013 - Atti del 32° Convegno Nazionale

Ambient noise measurements carried out in 2007 at these two sites provided quite consistent results in terms of identification of site response directivity and resonance frequency (Del Gaudio et al. , 2008), even though measurement repetitions have pointed out variations on H/V ratio amplitude, possibly related to seasonal change of site conditions (Del Gaudio et al. , 2013). The same noise recordings have been now re-analysed using the approach described in the previous section: in particular data acquired in July 2007, May 2011 and July 2013 were examined. Noise recordings, acquired by Tromino tromographs with a sampling frequency of 128 Hz, had different duration for different campaigns and sites, thus the analysed time series lengths are variable from a minimum of 15.5 minutes to a maximum of 1 hour. The analysis of instantaneous polarisation was conducted both on the complete unfiltered signals and on time series obtained by applying Gaussian filters: these were centred on 64 frequencies (spaced by 0.25 Hz from 0.5 to 16.25 Hz) and shaped adopting for the Gaussian curve a standard deviation equal to 10% of the central frequency. For the identification of Rayleigh and SH-type wave packets, after some tests carried out on synthetic signals, the criterion was adopted of selecting series of not less than 5 consecutive data samples that Fig. 3 – As Fig. 2 for the site CAR5. 231 GNGTS 2013 S essione 2.2