对流尺度集合预报初值扰动技术及系统应用研究进展
CSTR:
作者:
作者单位:

作者简介:

通讯作者:

中图分类号:

P456.7

基金项目:

国家自然科学基金资助项目(42205166;42475167);中国气象局气象能力联合提升研究专项(24NLTSQ010);中国气象局青年创新团队项目(CMA2024QN05)


Advances in initial perturbation techniques and systemic application research for convective-scale ensemble prediction
Author:
Affiliation:

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 文章评论
    摘要:

    对流尺度集合预报系统(Convective-scale Ensemble Prediction System, CEPS)具有时空高分辨率,能提供概率预报,对于提高强对流天气的可预报性具有重要意义。目前,CEPS已经成为业务中心建设的重点,而初值扰动方法是攻关技术难点之一,目前业务上主要的初值扰动方案有:动力降尺度、集合资料同化及奇异向量等,也有学者将中尺度集合预报的初值扰动方法如增长模繁殖法方案及混合扰动方法沿用到对流尺度集合预报中。然而现有方案尚未考虑快速增长的非线性扰动,开发能够描述初始场非线性增长信息的初值扰动方法,是CEPS的一个发展方向。国际上一些先进的数值预报中心和机构已成功地将CEPS应用到业务天气预报中,从现有的CEPS对强降水天气事件预报可见,CEPS能够为大量级降水提供概率预报,对暴雨的极端性有较好的参考意义,而集合平均的参考性并不强,其对强降水的预报仍然存在改进空间。人工智能技术的进步为集合预报的发展带来了新契机,首先集合预报能够为人工智能模型训练提供优质的数据基础;其次,构建基于人工智能的集合预报模型避免了传统集合预报的高耗能问题;此外,利用人工智能技术对集合预报结果进行后处理,可提升集合预报的使用效率。随着数值天气预报逐步实现精细化,CEPS与人工智能相结合也将是未来发展的一个重要方向。

    Abstract:

    Convective-scale Ensemble Prediction System (CEPS), characterized by its high spatiotemporal resolution and capability to provide probabilistic forecasts, plays a critical role in improving the predictability of severe convective weather. Currently, CEPS has become a key focus in operational center development, with initial perturbation methods being one of the challenging technical issues. Primary operational initial perturbation schemes include dynamic downscaling, ensemble data assimilation, and singular vectors. Some researchers have also adapted mesoscale ensemble forecast perturbation methods such as the BGM(Breeding of Growing Modes) scheme and hybrid perturbation approaches to CEPS. However, existing methodologies have yet to account for rapidly amplified nonlinear perturbations. The development of initial perturbation methods capable of characterizing nonlinear growth dynamics in initial fields represents a pivotal direction for CEPS innovation. Some international numerical weather prediction centers and institutions have successfully implemented CEPS in operational weather forecasting. Evaluations of CEPS performance in forecasting heavy precipitation events reveal its effectiveness in providing probabilistic predictions for extreme rainfall magnitudes, offering valuable insights into the extremity of storm events. However, ensemble mean forecasts demonstrate limited utility, indicating substantial room for improvement in predicting intense precipitation. Advancements in Artificial Intelligence (AI) technology present new opportunities for advancing ensemble forecasting. Firstly, ensemble forecasts provide high-quality datasets for training AI models. Secondly, AI-driven ensemble forecasting frameworks could circumvent the high energy consumption inherent in traditional ensemble methods. Furthermore, AI-based post-processing of ensemble forecast outputs could enhance operational efficiency. As numerical weather prediction progresses toward finer precision, the integration of CEPS with AI technologies is poised to become a vital future development direction.

    参考文献
    相似文献
    引证文献
引用本文

王静,张楠,李红祺,杨晓君,韩雨盟,王婧卓,刘昕.对流尺度集合预报初值扰动技术及系统应用研究进展.气象科学,2025,45(4):525-534 WANG Jing, ZHANG Nan, LI Hongqi, YANG Xiaojun, HAN Yumeng, WANG Jingzhuo, LIU Xin. Advances in initial perturbation techniques and systemic application research for convective-scale ensemble prediction. Journal of the Meteorological Sciences,2025,45(4):525-534

复制
分享
相关视频

文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:2024-09-19
  • 最后修改日期:2025-01-07
  • 录用日期:
  • 在线发布日期: 2025-11-29
  • 出版日期:
文章二维码
您是第位访问者
气象科学 ® 2026 版权所有
技术支持:北京勤云科技发展有限公司