Ground motion model for reference rock sites in Italy

To assess site-specific ground motion it is common practice to calculate seismic hazard at bedrock and then multiply it by a deterministic site-amplification factor typically computed from 1D numerical simulation. For this reason, the ground motion at bedrock should be free from amplification phenomena and its site response flat. Ground Motion Prediction Equations are generally calibrated using records at stations classified as rock that, however, can be affected by site-effects, caused by peculiar morphological/stratigraphic features. In this work, we propose six proxies based on geological, topographical and geophysical data to identify reference rock sites. We apply these proxies to the same set of recording stations used to derive the most recent ground-motion attenuation model for Italy [6] - ITA10. We find that about half of the analyzed sites, classified as rock on the basis of VS,30 or geological conditions, are unaffected by amplifications and can be actually considered as reference rock sites. Then, we re-calibrate the ITA10 prediction equations for horizontal peak ground acceleration at 20 spectral ordinates in the period range 0.04–2s, accounting for sites that we identify as references rock sites. The resulting reference median values are, on average, 35–40% lower than those calculated by Bindi et al. (2011) model for rock sites. Conversely, the ground motion variability is not significantly changed, even if we introduce a new site soil category to describe the reference rock stations. •Rock sites can be affected by site amplification.•Identification of reference rock site has implications in the calibration of GMPEs and in site-specific hazard assessment.•VS,30 value is not able to discriminate between reference and generic rock sites.