Scholarlink

<p>Extreme precipitation events are critical natural phenomena that often lead to riverine flooding and geological disasters. However, the mechanisms driving the spatiotemporal variability of Extreme precipitation events remain insufficiently understood. To address this knowledge gap, this study focuses on the Hekou-Longmen segment (HL). Using daily precipitation data, extreme precipitation indices (EPIs) are evaluated across three temporal scales—annual, seasonal, and monthly—covering the period from 1961 to 2022. Additionally, nine large-scale climate circulation indices (LCCIs) are incorporated into the analysis. The XGBoost algorithm, combined with the Shapley Additive Explanations (SHAP) method, is applied to quantitatively assess the global and localized impacts of the CCIs on extreme precipitation indices (EPIs) at the monthly scale. The analysis ultimately elucidates the mechanisms by which LCCIs influence EPIs variability. The results indicate several key findings: (1) In the HL, although there is a an insignifican increase in the intensity, quantity, and duration of EPIs, no statistically significant inter-annual declining trend is observed. Seasonal and monthly analyses reveal that EPIs are most pronounced during summer, with July consistently recording the highest values. (2) Long-term spatial variability in EPIs exhibit either stability or insignificant improvement. Regions with slight significant improvement and significant improvement are identified for indices such as R95p, RX1day, RX5day, and SDII. Remarkable regional variability is observed, with increasing trends in the middle and upper reaches, and decreasing trends in the lower zones. (3) Key factors influencing EPI variations are pinpointed, including solar flux index (X1), Western Pacific Subtropical High (X4), Atlantic Multidecadal Oscillation (X7), and El Niño (X9), with X4 exerting the most substantial influence. (4) SHAP analysis reveals that X4 is negatively associated with CCD but positively correlated with the remaining 10 EPIs. These findings provide important insights into the management of climate extremes and offer strategic guidance for mitigating the impacts of extreme weather in the HL and comparable climatic regions.</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

Insights into teleconnection mechanism of extreme precipitation events based on the SHAP-XGBoost model: evidence from Hekou-Longmen section in China

  • 期刊论文

  • 发表日期：

    2025年1月2日

查看全文
  • Chun Deng,
  • Xiaohui Jiang,
  • Chentian Jiang,
  • Tong Nie,
  • Yuxin Lei,
  • Anle Yang

摘要

Extreme precipitation events are critical natural phenomena that often lead to riverine flooding and geological disasters. However, the mechanisms driving the spatiotemporal variability of Extreme precipitation events remain insufficiently understood. To address this knowledge gap, this study focuses on the Hekou-Longmen segment (HL). Using daily precipitation data, extreme precipitation indices (EPIs) are evaluated across three temporal scales—annual, seasonal, and monthly—covering the period from 1961 to 2022. Additionally, nine large-scale climate circulation indices (LCCIs) are incorporated into the analysis. The XGBoost algorithm, combined with the Shapley Additive Explanations (SHAP) method, is applied to quantitatively assess the global and localized impacts of the CCIs on extreme precipitation indices (EPIs) at the monthly scale. The analysis ultimately elucidates the mechanisms by which LCCIs influence EPIs variability. The results indicate several key findings: (1) In the HL, although there is a an insignifican increase in the intensity, quantity, and duration of EPIs, no statistically significant inter-annual declining trend is observed. Seasonal and monthly analyses reveal that EPIs are most pronounced during summer, with July consistently recording the highest values. (2) Long-term spatial variability in EPIs exhibit either stability or insignificant improvement. Regions with slight significant improvement and significant improvement are identified for indices such as R95p, RX1day, RX5day, and SDII. Remarkable regional variability is observed, with increasing trends in the middle and upper reaches, and decreasing trends in the lower zones. (3) Key factors influencing EPI variations are pinpointed, including solar flux index (X1), Western Pacific Subtropical High (X4), Atlantic Multidecadal Oscillation (X7), and El Niño (X9), with X4 exerting the most substantial influence. (4) SHAP analysis reveals that X4 is negatively associated with CCD but positively correlated with the remaining 10 EPIs. These findings provide important insights into the management of climate extremes and offer strategic guidance for mitigating the impacts of extreme weather in the HL and comparable climatic regions.