Directional amplification of horizontal ground motion at rock sites

Abstract

Directional amplifications of horizontal ground motion along site-dependent azimuths are often unexpectedly found in rocky environments and in stiff-rock conditions. Directional amplification has been studied by several authors at sites with fractured rocks across fault damage zones (e.g. Pischiutta et al., 2013), or close to gravitational instabilities (e.g. Burjanek et al, 2012). In the majority of the cases, a transversal relation between the maximum amplification and the orientation of the predominant fracture field is recognized, interpreted as the effect of the stiffness anisotropy. Cracks in fault damage zones also cause the reduction of rock velocity (especially near-surface Vs). Studies performed on high number of rocky sites revealed that directional amplification effects are much more diffuse than expected. A systematic study involving 258 seismological stations of the Italian Seismic Network points out that 56 of stations (20%) are clearly affected by directional site amplification effects. This station sample was studied by analyzing the surface topography in order to unravel the role of topographic irregularities on ground motion amplifications. Models proposed in literature predict amplification at wavelengths comparable to the mountain width (e.g. Géli et al, 1988) and the scatter of wavefield that is polarized in the direction orthogonal to the relief elongation (e.g. Spudich et al., 1996). Since some bias exists in the objective quantification morphological parameters, we use an original methodology that combines morphometric analysis of digital elevation models and principal component analysis to define the dimension and elongation orientation of pronounced ridges. This study showed that the expected relations between the shape/dimension of relief and ground motion were found only at 5 stations (out of 56) , suggesting a major of local morphology, geological complexities and subsurface properties on directional amplification, consistently with Burjanek et al. (2014). Thus, recent findings have suggested that models proposed in literature in many cases are not able to reproduce observations even for damaging earthquakes (Avallone et al., 2014). Much effort has to be devoted in the future to model the amplification mechanism and take it into account in seismic hazard assessment