Scholarlink

<p>Surface-to-borehole spectral ratios (SBSRs) observed at vertical arrays often exhibit substantial event-to-event variability, even for weak motions within the nominally linear range. This variability matters, because one-dimensional (1D) site-response analysis assumes a stable transfer function for a given site, whereas empirical SBSRs often depend on event and arrival direction. If the origin of this directional variability is not understood, some of the variability commonly absorbed into ergodic ground-motion model (GMM) residuals may reflect unresolved but repeatable site behavior. Here we investigate whether the azimuthal dependence of SBSRs in Japan’s Kiban–Kyoshin network (KiK-net) records is more consistent with shared regional path effects or with local site behavior. Using 152,352 records from 6973 earthquakes at 693 KiK-net stations, we first summarize directional SBSR variability with frequency-dependent statistics, Gaussian mixture clustering of SBSR peaks, and multivariate analysis of variance, and show that back-azimuth organizes the variability more strongly than geometric incidence angle. We then reformulate the twofold spectral ratio (TSR) framework for nearby vertical-array station pairs and use this reformulation to define a dual-twofold spectral ratio (DTR), which links the operational comparison of surface TSRs and borehole TSRs to the physical comparison of interevent SBSR fluctuations between stations. Because the main inference is based on the DTR framework and on the agreement between surface TSR and borehole TSR, rather than on direct interpretation of individual spectral peaks, the analysis is less sensitive to noise, pseudoresonance, and other unstable event-specific spectral features. After total-least-squares-based quality control, surface TSRs and borehole TSRs remain positively correlated over 1–10&#xa0;Hz for 283 station pairs, whereas the interstation correlation of SBSR fluctuations remains weak for both earthquake-pair datasets. Case studies further show that strongly negative correlations arise mainly when the assumptions of the TSR framework break down. These results suggest that, within the analyzed nearby-pair set, the dominant azimuthal SBSR variability is chiefly local and site-specific at the station scale rather than a shared regional path effect. The implications for non-ergodic GMM development are discussed as motivation for future residual-based studies.</p> Graphical Abstract <p></p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Investigating the origin of azimuthal site-response variability using KiK-net data: a twofold spectral ratio approach

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月22日

查看全文
  • Masashi Aoki,
  • Yu Yamamoto,
  • Ryoichi Tokumitsu,
  • Yasuo Uchiyama,
  • Nobuo Takai

摘要

Surface-to-borehole spectral ratios (SBSRs) observed at vertical arrays often exhibit substantial event-to-event variability, even for weak motions within the nominally linear range. This variability matters, because one-dimensional (1D) site-response analysis assumes a stable transfer function for a given site, whereas empirical SBSRs often depend on event and arrival direction. If the origin of this directional variability is not understood, some of the variability commonly absorbed into ergodic ground-motion model (GMM) residuals may reflect unresolved but repeatable site behavior. Here we investigate whether the azimuthal dependence of SBSRs in Japan’s Kiban–Kyoshin network (KiK-net) records is more consistent with shared regional path effects or with local site behavior. Using 152,352 records from 6973 earthquakes at 693 KiK-net stations, we first summarize directional SBSR variability with frequency-dependent statistics, Gaussian mixture clustering of SBSR peaks, and multivariate analysis of variance, and show that back-azimuth organizes the variability more strongly than geometric incidence angle. We then reformulate the twofold spectral ratio (TSR) framework for nearby vertical-array station pairs and use this reformulation to define a dual-twofold spectral ratio (DTR), which links the operational comparison of surface TSRs and borehole TSRs to the physical comparison of interevent SBSR fluctuations between stations. Because the main inference is based on the DTR framework and on the agreement between surface TSR and borehole TSR, rather than on direct interpretation of individual spectral peaks, the analysis is less sensitive to noise, pseudoresonance, and other unstable event-specific spectral features. After total-least-squares-based quality control, surface TSRs and borehole TSRs remain positively correlated over 1–10 Hz for 283 station pairs, whereas the interstation correlation of SBSR fluctuations remains weak for both earthquake-pair datasets. Case studies further show that strongly negative correlations arise mainly when the assumptions of the TSR framework break down. These results suggest that, within the analyzed nearby-pair set, the dominant azimuthal SBSR variability is chiefly local and site-specific at the station scale rather than a shared regional path effect. The implications for non-ergodic GMM development are discussed as motivation for future residual-based studies.

Graphical Abstract