数值天气预报中集合-变分混合资料同化及其研究进展  

PROGRESS ON HYBRID ENSEMBLE-VARIATIONAL DATA ASSIMILATION IN NUMERICAL WEATHER PREDICTION

作　　者：马旭林[1] 陆续[1] 于月明[1] 朱金焕[1] 陈静[2] 

机构地区：[1]气象灾害教育部重点实验室/南京信息工程大学,江苏南京210044 [2]中国气象局数值预报中心,北京100081

出　　处：《热带气象学报》2014年第6期1188-1195,共8页Journal of Tropical Meteorology

基　　金：国家自然科学基金资助项目(41275111);重大研究计划培育项目(91437113)共同资助

摘　　要：数值天气预报中混合资料同化是结合集合预报与变分同化方法各自优势研究发展的一种新型资料同化方案。它在变分资料同化框架的基础上,利用集合预报扰动场信息的优点,构造具有流依赖属性的背景误差协方差结构,克服变分同化中固定、均匀及各向同性的背景误差协方差的缺陷,改善对不同天气系统预报误差协方差的表达能力,从而改善分析和预报的质量。近年来,基于集合与变分方法的混合资料同化快速成为数值预报资料同化发展的新趋势,具有良好的业务应用潜力。对集合与变分混合资料同化的研究进展进行综述,在简要介绍数值天气预报中资料同化发展趋势的基础上,分析混合资料同化方案的理论框架和优势,总结混合资料同化的发展和研究成果,并讨论我国研究和发展混合资料同化的基础和前景及面临的问题,指出混合资料同化发展中亟需研究的关键科学问题。As a new advanced data assimilation method, the hybrid variational-ensemble data assimilation system in numerical weather prediction （NWP） is developed based on the merits of the ensemble forecasting and the variational data assimilation method. By taking advantage of the information on perturbations in ensemble forecasting, the hybrid variational-ensemble method improves the ability of the background error variance to represent different weather conditions through the construction of flow-dependent background error covarianee, avoiding the deficiencies of the static, homogeneous, isotropic background error covariance from the variational method, and thus improving the quality of the analysis and prediction. In recent years, the method has rapidly become a new way of data assimilation that has the excellent potential for operational application. This paper gives an overview of the research development of the method. Firstly, the developments of the data assimilation methods in NWP are briefly described. Secondly, the theoretical framework and advantages of the hybrid variational-ensemble method are analyzed with a summary of its latest development and research results. Finally, a discussion is provided about the basis, prospect, and difficulty during the development and application of the method in China with highlights of key scientific problems that need to be studied in the development of the method.

关 键 词：数值天气预报 变分资料同化 混合资料同化 

分 类 号：P456.7[天文地球—大气科学及气象学]