Scholarlink

<p>Under the combined pressures of extreme rainfall and rapid urbanization, urban waterlogging has become increasingly severe. As high-density and low-income settlements, urban villages are particularly vulnerable. Taking the “9·7” extreme rainstorm event in Shenzhen as a case, this study analyzes 623 urban-village study units across the Longgang River and Shenzhen River basins to identify urban-form-related factors and empirical thresholds associated with waterlogging risk under an extreme rainfall scenario. Specifically, we focus on the dominant role of the topographic low-lying effect and the amplifying effect of street-canyon characteristics on the vulnerability of urban villages. By integrating multi-source data and constructing a multi-scale buffer analysis framework, we quantify the spatial contrasts between urban villages and their surrounding environments. An XGBoost model, combined with SHAP analysis and partial dependence plots (PDPs), is used to interpret the contribution of each factor,&#xa0;nonlinear thresholds,&#xa0;and interaction effects. The results show that: (1) the topographic low-lying effect exhibits the strongest explanatory contribution and presents a clear threshold-switch pattern, with absolute elevation as the leading factor; (2) within a 0–400&#xa0;m range, urban villages form significant morphological contrasts with their surroundings, constituting a risk “vulnerability ring”; (3) street-canyon indicators are more informative than traditional density indicators, highlighting the critical role of micro-scale urban form in regulating runoff pathways; and (4) maintaining the sky view factor (SVF) within the range of 0.3–0.55&#xa0;is associated with lower waterlogging risk. Overall, this study identifies urban-form-related factors, nonlinear response patterns, and empirical risk thresholds of urban-village waterlogging, providing a scientific basis for targeted resilience enhancement.</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

Multi-scale buffer analysis and explainable machine learning identify urban form factors and thresholds for waterlogging in Shenzhen urban villages during 9·7 rainstorm

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月20日

查看全文
  • Yinglong Lv,
  • Doreen Heng Liu,
  • Yu Yan,
  • Xinghan Gong,
  • Caicai Xu

摘要

Under the combined pressures of extreme rainfall and rapid urbanization, urban waterlogging has become increasingly severe. As high-density and low-income settlements, urban villages are particularly vulnerable. Taking the “9·7” extreme rainstorm event in Shenzhen as a case, this study analyzes 623 urban-village study units across the Longgang River and Shenzhen River basins to identify urban-form-related factors and empirical thresholds associated with waterlogging risk under an extreme rainfall scenario. Specifically, we focus on the dominant role of the topographic low-lying effect and the amplifying effect of street-canyon characteristics on the vulnerability of urban villages. By integrating multi-source data and constructing a multi-scale buffer analysis framework, we quantify the spatial contrasts between urban villages and their surrounding environments. An XGBoost model, combined with SHAP analysis and partial dependence plots (PDPs), is used to interpret the contribution of each factor, nonlinear thresholds, and interaction effects. The results show that: (1) the topographic low-lying effect exhibits the strongest explanatory contribution and presents a clear threshold-switch pattern, with absolute elevation as the leading factor; (2) within a 0–400 m range, urban villages form significant morphological contrasts with their surroundings, constituting a risk “vulnerability ring”; (3) street-canyon indicators are more informative than traditional density indicators, highlighting the critical role of micro-scale urban form in regulating runoff pathways; and (4) maintaining the sky view factor (SVF) within the range of 0.3–0.55 is associated with lower waterlogging risk. Overall, this study identifies urban-form-related factors, nonlinear response patterns, and empirical risk thresholds of urban-village waterlogging, providing a scientific basis for targeted resilience enhancement.