Frequency-dependent site factors for the Icelandic strong-motion array from a Bayesian hierarchical model of the spatial distribution of spectral accelerations

Abstract

The earthquake ground motions of over 1700 earthquakes recorded on a small-aperture strong-motion array in south Iceland (ICEARRAY I) that is situated on a relatively uniform site condition characterized as rock, exhibit a statistically significant spatial variation of ground-motion amplitudes across the array. Both earthquake and microseismic horizontal-to-vertical spectral ratios (HVSR) have been shown to exhibit distinct and in some cases, bimodal peaks in amplification, indicating site resonance at periods of 0.1–0.3 s, a phenomenon that has been attributed to a surface layer of lava rock lying above a sedimentary layer, a structure that is then repeated with depth under the array. In this study, we implement a Bayesian hierarchical model (BHM) of the seismic ground motions that partitions the model residuals into earthquake event term, station term, and event–station term. We analyzed and compared peak ground acceleration (PGA) with the 5% damped pseudo-acceleration response spectrum (PSA) at oscillator periods of T = 0.05–1.0 s. The results show that the event terms, dominate the total variability of the ground-motion amplitudes over the array. However, the station terms are shown to increase in the period range of 0.1–0.3 s on most stations and to different extents, leading to an increase in the overall variability of ground motions at those periods, captured by a larger inter-station standard deviation. As the station terms are a measure of how much the ground motions at those stations deviate from the array average, they act as proxies for localized site effects and amplification factors. These results, improve our understanding of the key factors that affect the variation of seismic ground motions across the relatively small area of ICEARRAY I. This approach can help to improve the accuracy of earthquake hazard assessments on local scales, which in turn could contribute to more refined seismic risk assessments and engineering decision-making.