Spectral Inversion for Seismic Site Response in Central Oklahoma: Low-Frequency Resonances from the Great Unconformity

Abstract

ABSTRACT We investigate seismic site response by inverting seismic ground-motion spectra for site and source spectral properties, in a region of central Oklahoma, where previous ground-motion studies have indicated discrepancies between observations and ground-motion models (GMMs). The inversion is constrained by a source spectral model, which we computed from regional seismic records, using aftershocks as empirical Green’s functions to deconvolve site and path effects. Site spectra across the region exhibit multiple, strong, low-frequency (f<2  Hz) resonances. Modeling of vertically propagating SH waves reproduces the mean amplitudes and frequencies of the site spectra and requires a deep (∼1–2  km) impedance contrast. Comparison of regional seismic velocity models and geologic profiles indicates that the seismic impedance contrast is, or is in proximity to, the Great Unconformity, which marks the interface between Precambrian basement rocks and overlying Paleozoic sedimentary rocks. Depth to Precambrian basement increases to the southwest across the study region (∼1500–4500  m), and the fundamental frequencies of the site spectra are anticorrelated with basement depth. The first higher-mode resonance also exhibits dependence on basement depth; although modeling suggests that the second higher mode should depend on basement depth, site spectra do not support this. The low-frequency resonances in central Oklahoma are not represented in the GMMs used in current seismic hazard analyses for tectonic earthquakes, though approaches to account for such features are under consideration in other regions of the central and eastern United States. Given the broad spatial extent of the Great Unconformity underlying eastern North America, it is likely that similar effects on seismic site response also occur in other areas. This study highlights the impact of regional geologic structure on earthquake ground motions and reiterates the need for modeling regional effects to improve ground-motion predictions and seismic hazard assessments.