GNGTS 2015 - Atti del 34° Convegno Nazionale

OPTIMIZATION OF ANALYSIS OF AMBIENT NOISE INSTANTANEOUS POLARIZATION FOR SITE RESPONSE INVESTIGATION V. Del Gaudio Dipartimento di Scienze della Terra e Geoambientali, Università degli Studi “Aldo Moro”, Bari, Italy Introduction. The most commonly used technique employing ambient noise to investigate site response is the Nakamura’s method (Nogoshi and Igarashi,1971; Nakamura, 1989), also known as HVSR (Horizontal-to-Vertical Spectral Ratio) or HVNR (Horizontal-to-Vertical Noise Ratio). According to this technique, site resonance properties can be inferred calculating the mean ratios H/V between spectral amplitudes of horizontal and vertical components of ambient noise recordings. There is a general agreement that, at least in simple geological conditions, this method allows a reliable identification of site resonance frequencies. Furthermore, if site response presents directional variations, an analysis of azimuthal variations of H/V ratios can reveal resonance directivity and maximum amplification directions (Del Gaudio et al. , 2013). However, amplification factors appears more problematic to be derived. In principle, one could invert curves of H/V as function of frequency in terms of subsoil models (see Castellaro and Mulargia, 2009), which, in turn, can be used to calculate amplification factors. However, H/ V curve interpretation requires the definition of ambient noise composition among different types of waves (body waves, Rayleigh, Love) that, according to site conditions and source characteristics, can differently contribute to the observed signals (cf. Bonnefoy-Claudet et al. , 2006). In general, analyzing ambient noise to characterize site response, it would be desirable to isolate Rayleigh waves within the recorded signal, since they can potentially provide more information, i.e. resonance frequency, amplification directivity and (indirectly, from particle motion ellipticity) amplification factors. Indeed, the presence of a significant or even predominant proportion of Love waves in the horizontal component of ambient noise (cf. Bonnefoy-Claudet et al. , 2008) can considerably alters the H/V ratios compared to what would be observed for Rayleigh waves only. Recently, I proposed a method to derive site response properties from a different kind of ambient noise processing, based on analysis, instant by instant, of ground motion polarization to identify Rayleigh wave packets over which to calculate mean H/V ratios (Del Gaudio, 2013). The implementation of this method requires the definition of a series of procedural and parametric choices. In this study, in order to define some guidelines in such choices, a series of tests were carried out on synthetic signals. These were also used to compare the performance of the new method with that of the classical Nakamura’s technique. Methodology. Instantaneous polarisation properties of an ambient noise recording u(t) can be obtained from its analytic representation, given by (1) where j is the imaginary unit and û ( t ) is the Hilbert transformof u ( t ).Applying this transformation to all the signal components, ground motion can be represented as the real part of a complex vector, which describes a time-variant elliptical trajectory. Morozov and Smithson (1996) provided a simple method to calculate the semi-axes a →  ( t ) and b →  ( t ) of these instantaneous ellipses. Their vectorial product p →  ( t ) (named planarity vector: see Schimmel and Gallart, 2003), allows defining the attitude of the trajectory plane, whereas the quantity (2) GNGTS 2015 S essione 2.2 75