We update ground‐motion prediction equations (GMPEs) for eastern North America (ENA) using the referenced empirical approach of Atkinson (2008). The technique is based on the use of residual analysis that models differences between regional ground‐motion observations and a reference GMPE developed for a data‐rich region. The update is timely because the Next Generation Attenuation‐West 2 GMPEs for shallow crustal earthquakes in active tectonic regions enable a significant improvement in the implementation of this model, relative to previous work (e.g., Atkinson and Boore, 2011). The predicted ground‐motion amplitudes of the ENA referenced empirical model are very similar to the equivalent California values of Boore et al. (2014; hereafter BSSA14) at close distances (R≤50 km), at low‐to‐moderate frequencies (f≤5 Hz). At regional distances (R>50 km) and at high frequencies (f>5 Hz), the ENA data suggest higher ground‐motion amplitudes than the BSSA14 reference model, presumably due to lower attenuation and higher stress for ENA events relative to those in active tectonic regions. We also show that the referenced empirical approach predicts ground motions that are consistent with those that would be produced by the hybrid empirical approach (Campbell, 2003), considering recent equivalent point‐source models that match both ENA and California ground‐motion databases.
Comparison between the referenced empirical GMPE of this study (HA14) and the stochastic GMPE of Atkinson and Boore (2006, 2011; denoted AB06′) shows that both models imply similar attenuation shape at all frequencies. For M 7 at R≤50 km, the HA14 model predicts relatively smaller ground‐motion amplitudes than does the AB06′ model, likely because of the greater saturation effects in the empirical BSSA14 reference model.