Analysis of Single-Station Standard Deviation Based on Empirical Data Set from the Central and Eastern United States

ABSTRACT In this study, we use an empirical set of residuals of the response spectral values based on ground-motion models (GMMs) developed by previous studies for areas inside and outside the Coastal Plain to identify repeatable site effects and estimate the single-station standard deviation by removing the ergodic assumption on site response for the Central and Eastern United States. The data set used in this study combines three distinct strong motion data sets from published data. It comprises peak horizontal ground acceleration and pseudospectral acceleration at periods ranging from 0.01 to 10 s. The total residuals were calculated as the differences between the natural logarithm of the observed data and the values predicted by GMMs. The mixed-effects regression approach was used to partition the total residuals into between-event and within-event components. Then, the within-event residuals were further partitioned into repeatable site terms and remaining residuals. The results indicate that single-station standard deviations are lower than those obtained using the ergodic assumption. The results of this study can be used in nonergodic (partially nonergodic) probabilistic seismic hazard analysis to constrain standard deviations.