Earthquake S waves can become trapped, or resonate, between the free surface and high-impedance basal layers, strongly contributing to site response at specific frequencies. Strong S-wave resonances have been observed in the central and eastern U.S., where many sites sit on unlithified sediments underlain by stiff bedrock. To evaluate S-wave resonances in this region, we calculated 1D linear site-responses at 89 seismic stations with developed S-wave velocity profiles into bedrock. We found that S-wave resonances at the fundamental and strongest (peak) modes occur across large ranges of frequencies, each spanning more than two orders of magnitude — 0.21–54.0 Hz and 0.29–71.5 Hz, respectively. Amplifications of ∼5 and ∼6 are common at the fundamental frequency and peak modes, respectively; the largest amplification calculated was 13.2. Using simple regression analyses, we evaluated the skills of six proxies derived from the S-wave velocity profiles to predict the frequencies and corresponding amplifications of the fundamental and peak modes. We found that the depths to the 1.0 km/s and 2.5 km/s horizons, consistent with other studies, and to the maximum impedance contrasts strongly correlate with the resonance frequencies and that the fundamental-mode and peak amplifications correlate with the maximum impedance ratios. Correlations improved for data subsets based on the number and magnitude of impedance ratios underlying the sites and are the strongest at sites underlain by a single impedance ratio of 3.0 or greater. Finally, we calculated the S-wave horizontal-to-vertical spectral ratios (HVSR) at each possible seismic station and found, consistent with other studies, that the first peak can be used to estimate fundamental-mode frequencies and the corresponding amplifications. Thus, S-wave HVSR, can provide useful estimates of the fundamental-mode linear site response parameters at sites lacking S-wave velocity profiles. Furthermore, S-wave HVSR curves appear to be useful to broadly categorize impedance-ratio profiles.
Read full abstract