Lars Isaksen, ECMWF

The Ensemble of Data Assimilations (EDA) at ECMWF


The Ensemble of 4D-Var Data Assimilations (EDA) was implemented operationally at ECMWF in 2010. Its role is to provide flow-dependent estimation of the analysis uncertainty and short-range forecast uncertainty. The EDA has been a significant contributor to the improvement of the initial perturbations for the Ensemble Prediction system. It has also provided flow-dependent background error variance and covariance information to the 4D-Var analysis system. This has been the key contributor to the analysis and forecast improvements at ECMWF since 2011. The EDA has also been used for diagnosing the data assimilation system. It has been shown to characterize the wave structures in the tropics much more realistically than the static background errors used in 4D-Var. The poster will provide examples of the versatile use of the EDA. It will also show the benefit of increasing the resolution of the EDA and how the representation of model error and handling of bias correction is important for the quality of the analysis and background error estimates.