GNGTS 2013 - Atti del 32° Convegno Nazionale

The HVSR technique was made popular at the end of the 80s by Nakamura (1989). They discovered that the relationship between the horizontal components H and the vertical compo- nents V of the ground motion, i.e. HVSR (Horizontal to Vertical Spectral Ratio), highlights the resonance frequencies of soil. Afterwards, theoretic explanation of the phenomenon given by the first authors, in which the resonance was caused only by the body waves, was abandoned in favor of theories that surface waves (Rayleigh and Love waves, and to a lesser shear waves) plays a key role in the phenomenon (Lachet and Bard, 1994; Konno and Ohmachi, 1998). Many researchers (e.g. Ohmachi et al. , 1991; Lermo et al. , 1992; Field and Jacob, 1993, 1995) showed that H/V ratio of noise can be used to identify the fundamental resonant frequency of a sed- imentary layer. Besides the amplification factors are more realistic than those obtained from sediment to rock site ratios (Nakamura, 2000). In order to define a preliminary subsurface model for the first level of microzonation, we performed 23 HVSR measurements, quite uniformly distributed in the area with a minimum spacing of 250 m. For each measurement 46 min of noise was registered using a sampling frequency of 256 Hz and processed using a cluster analysis technique (D’Alessandro et al. , 2013), in order to identify all the significant peaks of the H/V spectra in the frequency range 0.1-20 Hz. Their attribution to resonance phenomena of buried structures has been validated in agreement with SESAME criteria (2005), by analyzing standard deviation of the spectral coefficients and independence from the azimuth. The HVSR measurements have revealed the presence of amplification of ground motion probably due to resonance phenomena over a large part of the urban area. A simplified interpretation of the HVSR techniques can be made by assuming that the fundamental resonance frequency f of the overburden depends on its thickness h and on average shear-wave velocity V as follow: f = V / (4h). This interpretation can be made only in cases of homogeneous two-layered media. It also requires during the acquisition and the analysis to comply with certain criteria listed in the guidelines of the European SESAME project (2005). Using an appropriate inversion method the relationship between f and v s permits to estimate the value of v s versus depth from a H/V curve provided the inversion is constrained by information on a depth of a layer whose effect is recognizable in the H/V curve (Fäh et al. 2002; Bonnefoy-Claudet et al. , 2006). The limits of this procedure reside in the still not clear and debated composition of microtremor. Furthermore, the amplitude of the H/V peaks depends on numerous factors and it is not always possible to identify the constraints on the H/V curve, especially in complex cases of velocity inversion. Unfortunately, the limited number of boreholes and laboratory tests, useful to characterize in detail the stratigraphy and geometry of the bodies buried in the plain of Oliveri, did not allow to derive a detailed model of the subsurface. Consequently only a single borehole carried out to the eastern boundary of the municipality was used to constrain the HVSR inversions. Peak frequencymaps. Peak frequency maps, obtained from seismic microtremor measures, were performed considering the multiplicity of causes of HVSR amplification, in an attempt to discriminate peaks caused by different phenomena (seismic interfaces at various depths, high topographic gradients, anthropogenic causes). In this way, different frequency maps can be developed, each of them grouping and separately representing amplifications probably caused by the same source. Below are described the criteria used and the processing steps. In each graph of HVSR spectral ratio all the significant peaks were identified and characterized by their center frequency and amplitude. The whole dataset of HVSR amplitude versus peak frequency was used to statistically identify clusters of points probably having a homogeneous source (D’Alessandro et al. , 2013). Those clusters related to sets of points with significant coverage of the investigated zone can be used for the construction of a frequency map. This implies that for each area of investigation can be built from zero (e.g.. absence of peaks for substrate outcropping) to three maps, considering that the detection of a greater 237 GNGTS 2013 S essione 2.2