Determination of Site Amplification from Regional Seismicity: Application to the Swiss National Seismic Networks

Abstract

Online Material: Supplemental figures and tables. The propagation of energy released by an earthquake through the uppermost crust has a significant impact on the ground motion that is observed at the surface. Knowledge of this site amplification effect can significantly reduce the uncertainty involved in the determination and application of stochastic or empirical predictive ground‐motion equations and therefore reduce the uncertainty in subsequent hazard calculations. However, due to the heterogeneity of the upper crust, the site amplification effect is highly variable over scales of kilometers or less (e.g., Boore, 2004). It is important to characterize site amplification at seismic instrument locations in order to correct or account for data from several recording sites, such that a common reference (e.g., Poggi et al. , 2011) is defined in ground‐motion prediction equations (GMPEs). State of the art GMPEs take advantage of site characterization based on broad classes, such as National Earthquake Hazards Reduction Program (NEHRP) soil class (BSSC, 2003; Akkar and Bommer, 2010), or the average shear‐wave velocity of the uppermost 30 m ( V S 30; e.g., Abrahamson and Silva, 2008). However, despite the use of this simple characterization, recent work has shown the importance of considering site‐to‐site amplification variability in the estimation of ground‐motion prediction uncertainty (Atkinson, 2006; Al Atik et al. , 2010). This highlights that the strong variability of site amplification, even within a single NEHRP or V S 30 class, contributes significantly to the uncertainty of ground‐motion prediction. Developers of GMPEs typically make an assumption of ergodicity; that is, that the variability over space is treated as an uncertainty in time. Spatial variability (beyond simple site classification) is therefore assumed to be zero. In order to minimize the influence of site‐to‐site variability in single‐site hazard analyses, Atkinson (2006) introduced the concept of single‐site sigma (uncertainty). Its use …