Scholarlink

<p>The current generation of weather forecasting models struggles to generate extreme rainfall events (&gt;204.5 mm/day). Therefore, there is a need for a method to overcome this shortcoming and to provide a meaningful forecast, especially for extreme events. We propose two diagnostics to overcome the limitation of a fixed threshold of rainfall to identify extreme rain events. First, a percentile-based threshold instead of a fixed rainfall value is used, and second, the Extreme Forecast Index (EFI) is used to add confidence in the former. The data from a high-resolution (12.5 km) Global Forecast System (GFS) and its ensemble version (GEFS) are analysed. The 95<sup>th</sup> percentile of the model climatology as a threshold of extreme rainfall provides much better extreme rainfall forecasts, and the EFI of rainfall from GEFS further gives confidence in the occurrence of extreme rainfall events, which were missed earlier by using the traditional fixed threshold of rainfall. A gain in lead time of two days is also observed through these forecast diagnostics. A flowchart is also proposed to help monitor the forecast for extreme events.</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

Effective forecast diagnostics for predicting extreme rainfall events: A statistics-based approach

  • 期刊论文

  • 发表日期：

    2025年7月18日

查看全文
  • Snehlata Tirkey,
  • Tanmoy Goswami,
  • Malay Ganai,
  • M K R Phani,
  • Medha Deshpande,
  • Parthasarathi Mukhopadhyay,
  • D R Pattanaik

摘要

The current generation of weather forecasting models struggles to generate extreme rainfall events (>204.5 mm/day). Therefore, there is a need for a method to overcome this shortcoming and to provide a meaningful forecast, especially for extreme events. We propose two diagnostics to overcome the limitation of a fixed threshold of rainfall to identify extreme rain events. First, a percentile-based threshold instead of a fixed rainfall value is used, and second, the Extreme Forecast Index (EFI) is used to add confidence in the former. The data from a high-resolution (12.5 km) Global Forecast System (GFS) and its ensemble version (GEFS) are analysed. The 95th percentile of the model climatology as a threshold of extreme rainfall provides much better extreme rainfall forecasts, and the EFI of rainfall from GEFS further gives confidence in the occurrence of extreme rainfall events, which were missed earlier by using the traditional fixed threshold of rainfall. A gain in lead time of two days is also observed through these forecast diagnostics. A flowchart is also proposed to help monitor the forecast for extreme events.