We demonstrate that the interperiod correlation of epsilon (ρε) is an essential component of ground motions for capturing the variability of structural response that is needed in seismic fragility and seismic risk studies. To perform this demonstration, we generate large suites of scenario ground‐motion simulations using the point‐source (PS) stochastic method. Two compatible suites of simulations are developed; one suite without any imposed interperiod correlation, and one with Fourier amplitude ε sampled from a multivariate normal distribution with covariance specified by our empirical model. We illustrate how the effect of ρε propagates through the structural response and into seismic risk calculations. Without the adequate interperiod correlation of ground motions, variability in the structural response may be underestimated. This leads to structural fragilities that are too steep (underestimated dispersion parameter β) and propagates to nonconservative estimates of seismic risk.

To assess the current state of multiple existing ground‐motion simulation methods, we evaluate their interfrequency correlations and compare with empirical models. None of the six finite‐fault simulation methods tested adequately capture the interperiod correlations over the entire frequency range evaluated, although several of the methods show promise, especially at low frequencies. Using the correlation of the Fourier spectra provides the developers of the simulation methods better feedback in terms of how they can modify their methods, which is unclear when using response spectra comparisons. Based on the relative differences in the correlations of the Song (2016) source method, it appears that changes to the rupture generator may be the most promising approach to modifying the long‐period interperiod correlations.

You do not currently have access to this article.