集合—变分数据同化方法的发展与应用

Development and Application of Ensemble-Variational Data Assimilation Methods

近年来,集合—变分数据同化方法已成为大气数据同化领域研究的热点问题。该方法能够综合利用集合卡尔曼滤波和变分同化的优势,是实现"集合预报和数据同化一体化"的有效途径。在分析变分同化和集合卡尔曼滤波优缺点的基础上引出集合—变分数据同化的概念;按照不同实现方式,将集合—变分同化分为协方差线性组合和增加控制变量2类,介绍了相应的研究进展,并将集合—变分同化概念拓展;然后介绍了集合—变分同化在英美两国的应用;最后回顾了集合—变分同化研究的主要问题,展望了未来的发展趋势。

In recent years, ensemble-variational Data Assimilation （DA） methods have become cutting-edge issues of atmospheric data assimilation. The ensemble-variational DA methods which adopt the advantages of en- semble Kalman filter and variational DA is an effective way to the integration of ensemble prediction system and DA system in the Numerical Weather Prediction （NWP） system. Firstly, the concept of ensemble-variational DA is in- troduced after the comparison of advantages and disadvantages between variational DA and ensemble Kalman filter. Secondly, the ensemble-variational DA methods are divided into two categories by different ways of background er- ror covariance generation. One is simple linear combination of static and ensemble covariance, and the other is aug- mentation of control variables. Moreover, the related development is introduced and the concept of ensemble-varia- tional DA is expanded. Then, the application of ensemble-variational DA in the Great Britain and the U.S. is in- troduced. Finally, the main issues of ensemble-variational DA are reviewed and the prospect of the future develop- ment trend is listed.