同化技术：数值天气预报突破的关键——以欧洲中期天气预报中心同化技术演进为例
Data Assimilation: A Key Technology for NWP—— Technical Review of Data Assimilation in ECMWF
中文摘要：
国际先进业务数值预报中心未来几年的全球资料同化技术发展，采用变分与集合混合的新同化技术方向更加明确，遥感资料得到更广泛的使用，特别是有云区卫星遥感资料的有效应用成为重点。欧洲中期天气预报中心作为资料同化技术的领跑中心，其技术发展具有代表性。尝试从资料同化框架和卫星资料同化应用两个方面回顾欧洲中期天气预报中心的资料同化技术进展情况及对数值预报总体进步的贡献。介绍了其在长同化时间窗弱约束四维变分、集合资料同化、湿度与水凝物同化分析、线性化湿物理过程、晴空高光谱卫星资料、云雨条件下红外与微波资料同化等的关键技术进展。同时本文指出这些技术发展对我国GRAPES模式的资料同化系统发展的借鉴意义：面向4DVAR业务化应用进行关键技术的优选，解决关键技术问题。在发展3/4DVAR的同时，提前考虑混合资料同化技术的发展问题，做好技术准备和铺垫。
中文关键词：
ECMWF，资料同化，技术进展，变分框架，卫星观测，云雨条件
KeyWords：
ECMWF, data assimilation, technical advance, variational data assimilation system, satellite observation, cloud and precipitation conditions
Abstract：
Global data assimilation techniques developed in the world leading numerical weather prediction centers in the years to come are more clearly seen in the direction of combining variational data assimilation with ensemble information in model background error covariance, or hybrid data assimilation. The remote sense observations are used more efficiently. The effort will focus on the satellite radiance data assimilation in the area of cloud and rain covered regions. As the first leading numerical weather prediction center on data assimilation technique development, European Center for Medium-Range Weather Forecasts (ECMWF) technique development has representativeness. The progress on data assimilation technique in ECMWF is reviewed in this paper, specially focusing on the fields of variational data assimilation system development and satellite observation application. The key technical advances are introduced, including long-time window and weak-constraint four-dimensional variational data assimilation, ensemble data assimilation, moisture and cloud condensation analysis, linearizing moisture physics processes, assimilation of satellite infrared hyper-spectral observation in clear-sky, satellite infrared hyper-spectral and microwave observations under cloud and precipitation conditions. The technical review can be seen as a guidance for the development of Global and Regional Analysis and Prediction System (GRAPES) numerical weather model: priority should be given to the key technique selection of GRAPES 3/4DVAR operational application, and hybrid data assimilation technique should be considered and prepared in advance when the 4DVAR technique is being developed.