GNGTS 2014 - Atti del 33° Convegno Nazionale

GNGTS 2014 S essione 2.2 255 the top of an elongated ridge with an elevation of 300 m from the valley and about 1 km wide. The topography isolines are drawn in the top-right inset. Also ambient noise polarization (red rose diagram) and earthquake polarization (blue rose diagram) at station CERT are compared, the contour map of H/V spectral ratios being shown as well. On the Cerreto Laziale hill, predominant polarization is oriented N80°, there is a strict consistency between ambient noise and earthquake polarization, and the predominant azimuth is transversal to the hill major axis. During earthquakes, horizontal motion polarization can have a strong influence on the response of engineered structures, their rigidity too depending on azimuth. Measures of intensity shaking were proposed in the past to take into account the ground motion variability versus azimuth. Boore et al. (2006) defined GMRotDnn as response spectra obtained for period-dependent rotation angle, where nn is the fractile of the geometric means for rotation angles 0° < θ < 180° sorted by amplitudes (e.g. GMRotD50 is the median value and GMRotD100 is the largest geometric mean over all rotation angles). Response spectra were computed for rotation angles from 10° to 180° with increments by 10°. In Fig. 1 we also compare the maximum amplitude azimuth of response spectra (cyan rose diagram) with the horizontal polarization calculated by using seismic events (blue rose diagram). We found a good consistency between response spectra and polarization analysis in terms of polarization direction. Analogously, in the middle and bottom panels results of stations ILLI (Lipari) and SGTA (Sant’Agata di Puglia) are drawn, respectively. At these stations the directional amplification effect is evident even though the directional amplification effect occurs in a direction that is not orthogonal to the topography elongation. Pischiutta et al. (2011) also calculated the observed response spectra by rotating the horizontal components, finding largely different amplitudes for different directions of motion. They deduced the potential amplification from the comparison of the observed response spectra for a number of earthquakes and ground motion prediction model, finding a general (but not systematic) tendency of GMPEs to underestimate the observed amplification levels. More details can be found in Pischiutta et al. (2011) and Burjanek et al. (2014b). The main finding of Pischiutta et al. (2011) statistical study was that surprisingly at least 30% of stations of the Italian seismic network are unexpectedly affected by directional amplification and horizontal polarization. This observation was consistently found using ambient noise and earthquake recordings. We stress that these stations were installed in a rocky environment (supposedly stiff rock) to exclude as much as possible the contribution of the site, and so no site amplification would be expected. The conclusion by Rovelli et al. (2011) was that the orthogonal relation between directional amplification and hill elongation was found for only 25% of total stations. This means that the remaining 75% have a variable geometrical relation with topography. Also Burjanek et al. (2014a) performed a systematic study using 25 stations with pronounced topography of Swiss CHNet and Japanese KiK-net sites. The advantage of using these sites was that a detailed site characterization was available, including measured S-wave velocity profiles down to 30-100 m. They found that many stations on rock sites (EC8 class A) did not exhibit any systematic amplification even if installed in pronounced topography conditions. On the other hand, the rest of the sites (non EC8 class A) presented systematic frequency dependent amplification, ground motion vibrating along site-specific directions. This feature was observed on the both ambient vibration and earthquake recordings, the effect being source independent. Burjanek et al. (2014a) finally stressed that some of sites identified as outcropping rock sites looking at the borehole lithology log, were characterized by Vs30 values which are usually measured in sediment sites (see Fig. 2). All the described studies concluded that the strong systematic amplification observed at sites with pronounced topography is controlled by subsurface velocity structure, rather than the shape of the topography. Thus, although the effect of geometry is present, it cannot be simply decoupled from the site response, as assessed by Burjanek et al. (2014b).