Single-station standard deviation analysis of 2010–2012 strong-motion data from the Canterbury region, New Zealand

Abstract

A systematic analysis was conducted of the different variability components that affect the prediction of $\text{ log }_{10}(PSA)$ (i.e., Pseudo-Spectral Acceleration) ordinates on (mostly) deep sedimentary soil sites using a sizable set of strong motion data recorded in the strong earthquake sequences of 2010 and 2012 in the Canterbury region of New Zealand. Following recent, well established approaches of residual analysis of ground motion predictions, as well as recent GMPEs based on a global dataset, it was found that the event-corrected single-station standard deviation (“sigma”) is strongly decreased, for all selected stations, with respect to the uncorrected sigma. Likewise, the event-corrected intraevent sigma estimated for the entire dataset is significantly reduced compared to the standard deviation associated to ground motion prediction models, i.e. the “ergodic” sigma, for all spectral periods. The event-corrected sigma values for the present dataset are surprisingly consistent with those recently derived using KiK-Net strong motion data from Japan and those by Boore and Atkinson (Earthq Spectra 34(1):99–138, 2008) GMPE, and remain fairly constant with respect to the spectral period at about $0.15\sim 0.2$. An interpretation was provided of the physical meaning of the site correction term (${\delta }S2S)_{s}$ indicating a plausible correlation with prevailing geological conditions in the site area.