Scholarlink

<p>Modeling far-field basin ground motions remains challenging due to strong geological heterogeneity and the limited availability of site-specific seismic records. This study introduces an efficient and transferable hybrid simulation framework that integrates ambient-noise-derived empirical Green’s functions (GFs) with physics-based and empirical constraints. Compared with conventional deterministic or fully empirical approaches, the method reconstructs basin-specific propagation and amplification effects by exploiting ambient-noise GFs, substantially reducing the need for new large-scale numerical simulations. Continuous ambient seismic noise is processed through an unsupervised DTW-Kernel PCA-GMM workflow to extract stable and coherent empirical GFs that characterize large-scale basin responses. These empirical GFs are incorporated into a kinematic finite-fault model, and Bayesian updating is applied to optimize rupture and slip parameters to maintain consistency with ground-motion prediction equations (GMPEs). Application to the Kanto Basin, Japan, demonstrates that the framework reproduces the main long-period response spectra and time-frequency characteristics of observed basin motions and achieves FDM-like long-period intensity-measure trends at a computational cost roughly three orders of magnitude lower than conventional 3-D numerical simulations under the tested scenario. The results demonstrate that combining ambient-noise observations with physics-based and empirical models provides a practical and physically consistent means for simulating far-field basin ground motions and for developing site-specific ground-motion inputs applicable to performance-based earthquake engineering.</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

An ambient-noise-based physics–data integrated framework for far-field basin ground motion simulation

  • 期刊论文

  • 发表日期：

    2026年7月9日

查看全文
  • Zheqian Wu,
  • Jinkun Sun,
  • Yingmin Li

摘要

Modeling far-field basin ground motions remains challenging due to strong geological heterogeneity and the limited availability of site-specific seismic records. This study introduces an efficient and transferable hybrid simulation framework that integrates ambient-noise-derived empirical Green’s functions (GFs) with physics-based and empirical constraints. Compared with conventional deterministic or fully empirical approaches, the method reconstructs basin-specific propagation and amplification effects by exploiting ambient-noise GFs, substantially reducing the need for new large-scale numerical simulations. Continuous ambient seismic noise is processed through an unsupervised DTW-Kernel PCA-GMM workflow to extract stable and coherent empirical GFs that characterize large-scale basin responses. These empirical GFs are incorporated into a kinematic finite-fault model, and Bayesian updating is applied to optimize rupture and slip parameters to maintain consistency with ground-motion prediction equations (GMPEs). Application to the Kanto Basin, Japan, demonstrates that the framework reproduces the main long-period response spectra and time-frequency characteristics of observed basin motions and achieves FDM-like long-period intensity-measure trends at a computational cost roughly three orders of magnitude lower than conventional 3-D numerical simulations under the tested scenario. The results demonstrate that combining ambient-noise observations with physics-based and empirical models provides a practical and physically consistent means for simulating far-field basin ground motions and for developing site-specific ground-motion inputs applicable to performance-based earthquake engineering.