HVSR-Based Peak Frequency Identification of S-Net Ocean-Bottom Stations for Site Characterization and Ground-Motion Prediction Models

ABSTRACT The horizontal-to-vertical spectral ratio (HVSR) technique is widely used to estimate the site resonant frequency. This technique is particularly valuable for ocean-bottom stations where shallow site information and reference sites are unavailable, providing reliable site resonant frequency and characteristics. Several critical aspects of the HVSR require careful consideration, including the selection of response spectra and Fourier spectra, as well as the choice of smoothing functions for Fourier spectra. Evaluating the reliability and superiority of various identification results is essential. In this study, weak motion earthquake records (peak ground acceleration ≤50  cm/s2) from the S-net ocean-bottom sites in Japan Trench area were analyzed to investigate the correlation between different identification results for first and highest peak frequencies and the single-station site terms of the ocean-bottom sites, thereby assessing the applicability of the HVSR for ocean-bottom site characterization. By applying various HVSR identification schemes to 150 ocean-bottom stations, the applicability and uncertainty of the H/V Fourier spectral ratio (HVFSR) and the H/V response spectral ratio (HVRSR) were explored. The results show that the highest peak frequency identified by the HVFRS smoothed with the Parzen window exhibits the strongest correlation with the amplification site term of unburied ocean-bottom sites. In addition, it shows a strong correlation with buried sites in the short-period range. Furthermore, as an effective explanatory variable for site amplification, the correlation analysis of the HVRSR with the site terms reveals that HVRSR has a relatively significant correlation with both buried and unburied ocean-bottom sites. Therefore, it is recommended to use the HVFRS smoothed with the Parzen window for identifying the highest peak frequency and constructing a site effect model based on either the highest peak frequency or the amplitude of HVRSR, providing a valuable reference for the development of offshore ground-motion models.