RECURRENCE OF HORIZONTAL AMPLIFICATION AT ROCK SITES: A TEST USING H/V BASED GROUND MOTION PREDICTION EQUATIONS

Abstract

Ambient noise and earthquake records of 226 three-component seismological stations installed on stiff rock are investigated using H/V spectral ratios and horizontal polarization analysis. H/V spectral ratios of ambient noise calculated on rotated horizontal components from 0° to 180° indicate that 56% of stations is affected by a significant (>2) amplification in site-dependent frequency bands. This effect is often strongly directional. A strict criterion, based on the covariance matrix diagonalization, is then applied to select sites with a strong local tendency to polarize ground motion in the horizontal plane. Results indicate that 36% of the entire data set honors this condition. This sample (81 stations) is investigated using a DEM to characterize local topography conditions as well as the proximity to seismogenic faults (DISS Working Group, 2010). Strongly polarized stations resulted to be installed on irregular topography and/or close to faults. For these stations we repeated the same analysis using earthquake records: earthquakes and ambient noise show a quite consistent pattern for 66 stations (29%). One of the possible causes of the strong directional site effect at rock stations is stiffness anisotropy which characterizes elongated ridges and intensely fractured rocks such as fault-related cracks and tectonically deformed structures. In several cases, a combination of these effects could be responsible for large observed amplifications that are generally underpredicted by theoretical models assuming massive, isotropic rock outcrops. This study suggests that amplification at rock sites can be much larger than expected on the basis of conventional soft/stiff site classifications, at least along site-dependent directions. Effects of directionality on response spectra and the need of rotating the horizontal components to compare observations with GMPEs is discussed. The performance of the H/V-based site classification criterion recently proposed by Di Alessandro et al. (2008 and 2011) is also preliminarily investigated for the amplified rock stations.