GNGTS 2015 - Atti del 34° Convegno Nazionale

“minimum roughness regularization” strategy, so to obtain smooth transitions with depth but still maintaining the capability of capturing the major impedance contrasts. The HVSR (Horizontal-to-Vertical Spectral Ratio) technique, first proposed in 1970 by Nogoshi and Igarashi, based on the initial study of Kanai and Tanaka (1961), and today popular thanks to Nakamura (1989), is a “passive” method, which uses three-component recordings of ambient seismic noise to evaluate the site fundamental resonance frequency(ies), by estimating the horizontal-to-vertical ratio of the spectral amplitudes of motion. The measurements of the seismic noise were performed using a 3-component short-period seismometer (fc 2 Hz) for time intervals variable between 30 and 50 minutes. The final HVSR curves as a function of frequency are given by the average of the H/V ratio computed for each window (window size 60 s). The curves were computed by averaging the horizontal spectra with the quadratic average and dividing it for the vertical spectrum. Such spectra were smoothed following the filter proposed by Konno and Omachi (1986) using a constant b value of 40. Moreover, seismic noise source directionality was evaluated for all the measurements. Results: pseudo-2D sections. The discrete information of all 1D models was interpolated with a minimum curvature algorithm in order to obtain the pseudo-2D velocity section. The resulting Vs profile is shown in Fig. 2. Although 1D models locally reached higher depths, the section we show reports the distribution of Vs down to an average depth of about 160 m b.g.l. (considering that the elevation of the sites ranges between 0 and 4.5 m a.s.l.). The shear wave velocity ranges between 100-150 m/s, just below ground surface, and locally reaches 600 m/s at the maximum investigation depths. The vertical gradient of Vs is stronger between sites 03-13 and between sites 21-26, while in the southernmost and central portions of the profile, between sites 01-02 and 14-20 the gradient is weaker and the Vs at 160 m b.s.l. is ca. 400-450 m/s. If we compare the Vs profile with the Structural Model of Italy (Bigi et al. , 1992) we observe that the strongest Vs gradients are located above the sets of thrust faults respectively pertaining to the Argenta and Ferrara anticlines; conversely, the sites where the gradient is weaker are located above a tectonically “depressed” area bounded by a reverse fault to the north. The corresponding frequency section, obtained using the HVSRprofile routine (Herak et al. , 2010), which performs a side-by-side assembly of the observed HVSR-spectra, is based on the 26 single station measurements of seismic noise, elaborated following the HVSR method. Considering the characteristics of the seismometer and the influence of weather-climate conditions for frequencies below 0.5 Hz (SESAME, 2004), the analysis was limited to the frequency band between 0.5 and 5 Hz. The HVSR amplitudes, depending on the impedance contrast at the discontinuity surface, are color-coded and these are greater in the southern portion of the profile with respect to the northern one. The fundamental frequency varies along the profile from a minimum of 0.55 Hz up to a maximum of 1.6 Hz (Fig. 3a). Assuming the Vs pattern and the fundamental resonance frequency variations to be determined by lateral lithological variations, especially in terms of differential compaction (i.e. age) of the sediments, the two pseudo-2D sections (Figs. 2, 3a) allow hypothesizing the Fig. 2 – Pseudo-2D shear-wave velocity section reconstructed by the interpolation of several 1D shear-wave models obtained from the inversion of the ESAC seismic noise data. GNGTS 2015 S essione 1.2 117