Scholarlink

<p>A novel method based on a Gaussian mixture particle filter is proposed for optimal scenario selection, aiming to objectively identify the most probable scenario among four typhoon forecasts. This method integrates ensemble clustering with an optimal-scenario selection approach that incorporates analysis data through a Bayesian framework. Each ensemble member and the analysis data were initially projected onto the phase space spanned by the two leading principal components from an empirical orthogonal function analysis of the ensemble forecast. The realization probability of each scenario was computed using a particle filter-based Bayesian approach that evaluates the similarity between the forecasts and the analysis within the phase space. The scenario with the highest probability was then selected as the optimal scenario. A statistical analysis of 34 typhoon forecast cases using the regional ensemble prediction system of the Japan Meteorological Agency (JMA) demonstrated that the proposed method can identify the optimal scenario up to six hours in advance of the verification time (24-h forecast). A case study of Typhoon Hagibis (2019), an extreme event, was also conducted. The selected scenario successfully predicted a mesoscale front and associated coastal heavy rainfall that exceeded 100&#xa0;mm per 3&#xa0;h centered on Marumori Town, Miyagi Prefecture, Japan, which JMA’s deterministic mesoscale model failed to forecast. Notably, the optimal scenario was identified prior to the onset of heavy rainfall. These results suggest that the proposed method offers operational forecasters more reliable scenario-based guidance for significant typhoon-related weather events in advance.</p> Graphical abstract <p></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

Optimal typhoon scenario selection using ensemble clustering and Bayesian estimation of realization probability

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年1月13日

查看全文
  • Junpei Yamaguchi,
  • Kosuke Ono

摘要

A novel method based on a Gaussian mixture particle filter is proposed for optimal scenario selection, aiming to objectively identify the most probable scenario among four typhoon forecasts. This method integrates ensemble clustering with an optimal-scenario selection approach that incorporates analysis data through a Bayesian framework. Each ensemble member and the analysis data were initially projected onto the phase space spanned by the two leading principal components from an empirical orthogonal function analysis of the ensemble forecast. The realization probability of each scenario was computed using a particle filter-based Bayesian approach that evaluates the similarity between the forecasts and the analysis within the phase space. The scenario with the highest probability was then selected as the optimal scenario. A statistical analysis of 34 typhoon forecast cases using the regional ensemble prediction system of the Japan Meteorological Agency (JMA) demonstrated that the proposed method can identify the optimal scenario up to six hours in advance of the verification time (24-h forecast). A case study of Typhoon Hagibis (2019), an extreme event, was also conducted. The selected scenario successfully predicted a mesoscale front and associated coastal heavy rainfall that exceeded 100 mm per 3 h centered on Marumori Town, Miyagi Prefecture, Japan, which JMA’s deterministic mesoscale model failed to forecast. Notably, the optimal scenario was identified prior to the onset of heavy rainfall. These results suggest that the proposed method offers operational forecasters more reliable scenario-based guidance for significant typhoon-related weather events in advance.

Graphical abstract