|Title||Ensemble of data assimilations at ECMWF|
|Year of Publication||2010|
|Authors||Isaksen, L, Bonavita, M, Buizza, R, Fisher, M, Haseler, J, Leutbecher, M, Raynaud, L|
|Secondary Title||Technical Memorandum|
|Type of Work||Technical Memorandum|
The Ensemble of Data Assimilations (EDA) implemented in June 2010 is an important new component of ECMWF's operational suite. It provides valuable estimates of analysis and short-range forecast uncertainty. This paper describes the EDA method and its applications. The rationale for developing an EDA is discussed and the considerations that went into deciding the operational configuration, including resolution, ensemble size and model error formulation, are presented. Since its implementation, EDA based perturbations have been used to improve the initial perturbations for the Ensemble Prediction System (EPS): the methodology used to combine EDA-based perturbations and initial-time singular vectors and the impact of replacing the evolved singular vectors with EDA-based perturbations on the EPS performance is presented. Another important application of the EDA is to provide flow-dependent background error estimates for the deterministic 4D-Var system: the method under development and testing is described, and the significant progress that has been made in the past year is discussed.