Scholarlink

<p>The East Asian summer monsoon (EASM) exhibits complex and variable behavior that challenges the reliability of seasonal forecasts. In this study, we develop an optimized multi-model ensemble (MME) framework to enhance both the skill and confidence of EASM precipitation predictions. Unlike conventional MMEs that include all available models without performance filtering, the optimized MMEs are constructed by selecting only those models that demonstrate both high temporal correlation with observed principal components and a sufficiently large signal-to-total variance ratio. These criteria are quantified using the ratio of predictability components (RPC), which jointly captures real-world and model-based predictability. Results show that the optimized MMEs significantly outperform conventional MMEs in forecasting EASM rainfall: prediction skill improves by up to 13.4% and signal variance increases by up to 22.6%, particularly over East Asia. These improvements indicate not only higher forecast accuracy but also greater confidence in seasonal predictions. Our findings demonstrate that applying an RPC-based model selection method provides a robust strategy for improving seasonal climate forecasts in monsoon-affected 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

Enhancing seasonal predictability of the East Asian summer monsoon via optimized multi-model ensembles

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年7月10日

查看全文
  • OkYeon Kim,
  • Woo-Seop Lee

摘要

The East Asian summer monsoon (EASM) exhibits complex and variable behavior that challenges the reliability of seasonal forecasts. In this study, we develop an optimized multi-model ensemble (MME) framework to enhance both the skill and confidence of EASM precipitation predictions. Unlike conventional MMEs that include all available models without performance filtering, the optimized MMEs are constructed by selecting only those models that demonstrate both high temporal correlation with observed principal components and a sufficiently large signal-to-total variance ratio. These criteria are quantified using the ratio of predictability components (RPC), which jointly captures real-world and model-based predictability. Results show that the optimized MMEs significantly outperform conventional MMEs in forecasting EASM rainfall: prediction skill improves by up to 13.4% and signal variance increases by up to 22.6%, particularly over East Asia. These improvements indicate not only higher forecast accuracy but also greater confidence in seasonal predictions. Our findings demonstrate that applying an RPC-based model selection method provides a robust strategy for improving seasonal climate forecasts in monsoon-affected regions.