Scholarlink

<p>Reliable assessment of spatial correlations in strong ground motions is crucial for assessing seismic risk across distributed assets. Traditionally, these correlations have been modeled based on stationary and isotropic assumptions. However, recent studies have highlighted the limitations of these assumptions, particularly for sites affected by&#xa0;near-field earthquakes, where correlations of earthquake intensity measures exhibit significant spatial variations due to source, path, and site effects. Addressing the nonstationary patterns of spatial correlation is essential for improving regional seismic hazard and risk&#xa0;evaluations. In this paper, we apply a clustering-based approach that subdivides the study region into subregions and develops individual stationary correlation models for each to simulate ground motion residuals. This method captures the nonstationary structure of spatial ground motion correlations while leveraging the simplicity of stationary models. Additionally, we evaluate its performance against traditional stationary models. Our results demonstrate that this approach outperforms traditional stationary correlation models and provides more reliable predictions of observed ground motions. By accounting for spatial variations in correlation structures, this study can enhance the precision of regional seismic risk assessments.</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

Clustering-based analysis to address nonstationary spatial ground motion correlations using physics-based simulated data

  • 期刊论文

  • 发表日期：

    2025年3月25日

查看全文
  • M. R. Zolfaghari,
  • M. Forghani

摘要

Reliable assessment of spatial correlations in strong ground motions is crucial for assessing seismic risk across distributed assets. Traditionally, these correlations have been modeled based on stationary and isotropic assumptions. However, recent studies have highlighted the limitations of these assumptions, particularly for sites affected by near-field earthquakes, where correlations of earthquake intensity measures exhibit significant spatial variations due to source, path, and site effects. Addressing the nonstationary patterns of spatial correlation is essential for improving regional seismic hazard and risk evaluations. In this paper, we apply a clustering-based approach that subdivides the study region into subregions and develops individual stationary correlation models for each to simulate ground motion residuals. This method captures the nonstationary structure of spatial ground motion correlations while leveraging the simplicity of stationary models. Additionally, we evaluate its performance against traditional stationary models. Our results demonstrate that this approach outperforms traditional stationary correlation models and provides more reliable predictions of observed ground motions. By accounting for spatial variations in correlation structures, this study can enhance the precision of regional seismic risk assessments.