CMA-GFS全球预报模式东亚地区中期预报性能极端下降原因研究

Investigation into the Causes of Occasional poor Midium-Range Weather Forecast of CMA-GFS over the East Asia Region

中文关键词:  预报性能极端下降,误差关键区，松弛, 误差回溯法, 集合敏感度

英文关键词:Forecast drop-out or forecast busts, Key error origin, Relaxation, Error tracking, Ensemble sensitivity

中文摘要:

      在过去的十几年里，尽管中国气象局CMA-GFS全球模式的预报性能有了很大提高，但个别预报仍会出现预报误差非常大的情况，这类事件通常被定义为“预报性能极端下降”。本文主要针对中国气象局CMA_GFS在东亚地区中期预报性能极端下降事件，利用误差溯源、集合敏感度和松弛技术对2021年12月30日的一次极端下降个例进行研究和分析，通过松弛误差敏感区域，对模式预报误差进行溯源，研究分析模式预报性能突降的原因，为进一步探讨这些预报下降个例奠定基础。

英文摘要:

      Although the performance of operational global model CMA-GFS in CMA has increased significantly over the decades, individual forecasts still occasionally experience very large errors. And these events are often defined as ‘drop-outs’ or ‘forecast busts’. This paper mainly focuses on such a forecast bust of CMA-GFS global model in East Asia by using error tracking, ensemble error sensitivity and relaxation techniques to study the case occurred on December 30, 2021. The main aim of this study is to diagnose the origin of the forecast errors through relaxation of the key error region, to understand the causes lead to the sudden-drop of the forecasts and to form a basis for further exploring of these cases. By relaxing individual model prognostic variable, we can track and analyze the contribution of each variable to the total forecast errors and its impact on downstream, which is very helpful to better understand the origin and propagation of forecast errors, and further investigation into the sudden-drop of forecast skills scientifically. The results indicates that this bust case are mainly originated from the propagation eastward of 24h perturbed forecast errors in the east of Greenland, and finally result in the drop-outs in East Asia. By relaxation of individual model prognostic variable, we find that the forecast error of potential temperature θ, then v component lead to the perturbed forecast errors in the key error region. It is also found that the reduction of large errors in the east of Greenland leads to the improvement of the performance of CMA-GFS model in East Asia and the Northern Hemisphere considerably. The methodology used in this study can be applied to understand future bust cases and to enhance our knowledge of the origin and propagation of forecast errors.