Scholarlink

<p>The escalating intensity and frequency of heavy rainstorms pose critical challenges to natural ecosystems and human living environments. Understanding the intra-event pattern (IEP) of heavy rainfall events is crucial, as it significantly influences hydrological processes such as infiltration and runoff, which in turn govern the timing and scale of water-related disasters. This study proposes a machine learning framework that combines fuzzy c-means clustering with a real-coded genetic algorithm to characterize IEPs across seven engineering climate zones in China. Based on 12-year hourly rainfall data, the analysis revealed four key findings: (1) a temporal restructuring of rainfall events, marked by the growing dominance of short-duration, high-intensity storms and the decline of long-duration events in specific inland regions, indicating shifts in climatological dynamics; (2) a quadripartite clustering structure consistently emerging across all study zones, identifying four IEPs with distinct peak timings; (3) the parametric representation of normalized IEP curves and hyetographs using a two-parameter beta distribution, ensuring precision and flexibility; (4) detection of double-peak rainfall events through membership degree analysis, revealing multi-peak structures often overlooked in conventional models. These findings highlight the necessity for region-specific stormwater management strategies and climate-resilient infrastructure design, particularly in high-risk alluvial regions vulnerable to flash floods and waterlogging. By integrating IEPs into a systematic spatiotemporal characterization framework, this research contributes meaningfully to advancing urban climate resilience, flood risk mitigation, and sustainable stormwater management under evolving climate conditions.</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

Characterization of intra-event patterns of heavy rainfall events in China during 2008–2019 by machine learning

  • 期刊论文

  • 发表日期：

    2025年6月14日

查看全文
  • Weizhen Jiang,
  • Yong Tan

摘要

The escalating intensity and frequency of heavy rainstorms pose critical challenges to natural ecosystems and human living environments. Understanding the intra-event pattern (IEP) of heavy rainfall events is crucial, as it significantly influences hydrological processes such as infiltration and runoff, which in turn govern the timing and scale of water-related disasters. This study proposes a machine learning framework that combines fuzzy c-means clustering with a real-coded genetic algorithm to characterize IEPs across seven engineering climate zones in China. Based on 12-year hourly rainfall data, the analysis revealed four key findings: (1) a temporal restructuring of rainfall events, marked by the growing dominance of short-duration, high-intensity storms and the decline of long-duration events in specific inland regions, indicating shifts in climatological dynamics; (2) a quadripartite clustering structure consistently emerging across all study zones, identifying four IEPs with distinct peak timings; (3) the parametric representation of normalized IEP curves and hyetographs using a two-parameter beta distribution, ensuring precision and flexibility; (4) detection of double-peak rainfall events through membership degree analysis, revealing multi-peak structures often overlooked in conventional models. These findings highlight the necessity for region-specific stormwater management strategies and climate-resilient infrastructure design, particularly in high-risk alluvial regions vulnerable to flash floods and waterlogging. By integrating IEPs into a systematic spatiotemporal characterization framework, this research contributes meaningfully to advancing urban climate resilience, flood risk mitigation, and sustainable stormwater management under evolving climate conditions.