Scholarlink

<p>Accurate and reliable subseasonal precipitation forecasts are critical for disaster prevention and mitigation, particularly in densely populated regions like East Asia. However, substantial gaps remain between the reliability and accuracy of dynamical model forecasts and societal demands. This study proposes a machine learning-based adaptive bias correction (ABC) method to postprocess forecasts from the Climate Forecast System version 2 (CFS) and the European Centre for Medium-Range Weather Forecasts Integrated Forecasting System (EC). Results indicate that ABC effectively reduces systematic errors in 3–6&#xa0;weeks lead forecasts. The method improved the precipitation prediction skills (based on uncentered anomaly correlation) of the CFS model over East Asia during 2018–2021 by 70% for 3–4&#xa0;weeks lead times and 78% for 5–6&#xa0;weeks lead times (with the most significant enhancements observed in autumn), while increasing the EC model’s skills by 26% (3–4 w) and 22% (5–6 w) under the same temporal averaging. Spatially, ABC substantially reduced subseasonal biases in the Indian Peninsula, tropical regions, Lake Baikal, and Southwest China for both models. Compared to quantile mapping (QM) and locally estimated scatterplot smoothing (LOESS), ABC demonstrated superior bias correction performance, with QM exhibiting the weakest efficacy for dynamical models. Thus, the ABC method may shed light on improving the uncentered-anomaly-correlation-based skill of subseasonal forecasts.</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

Subseasonal precipitation prediction over East Asia using an adaptive bias correction model: methodology and evaluation

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年1月19日

查看全文
  • Haonan Ji,
  • Chuhan Lu,
  • Dingan Huang

摘要

Accurate and reliable subseasonal precipitation forecasts are critical for disaster prevention and mitigation, particularly in densely populated regions like East Asia. However, substantial gaps remain between the reliability and accuracy of dynamical model forecasts and societal demands. This study proposes a machine learning-based adaptive bias correction (ABC) method to postprocess forecasts from the Climate Forecast System version 2 (CFS) and the European Centre for Medium-Range Weather Forecasts Integrated Forecasting System (EC). Results indicate that ABC effectively reduces systematic errors in 3–6 weeks lead forecasts. The method improved the precipitation prediction skills (based on uncentered anomaly correlation) of the CFS model over East Asia during 2018–2021 by 70% for 3–4 weeks lead times and 78% for 5–6 weeks lead times (with the most significant enhancements observed in autumn), while increasing the EC model’s skills by 26% (3–4 w) and 22% (5–6 w) under the same temporal averaging. Spatially, ABC substantially reduced subseasonal biases in the Indian Peninsula, tropical regions, Lake Baikal, and Southwest China for both models. Compared to quantile mapping (QM) and locally estimated scatterplot smoothing (LOESS), ABC demonstrated superior bias correction performance, with QM exhibiting the weakest efficacy for dynamical models. Thus, the ABC method may shed light on improving the uncentered-anomaly-correlation-based skill of subseasonal forecasts.