A Strategy for Data Assimilation

This paper reviews progress made in Data Assimilation methodology at ECMWF since the 2011 SAC Data Assimilation Vision document (ECMWF/SAC/40(11)7). It then describes ECMWF’s current overall strategy for Data Assimilation. Recent research and development activities in data assimilation methodology have been centred on the introduction of an ensemble data assimilation component for error estimation and cycling and the development of a software framework for more flexible code design through the Object Oriented Prediction System (OOPS). In addition, investigation of ensemble Kalman filter methods, the use of weak constraint 4D-Var and the sensitivity to assimilation window length are active areas of research. A new aspect discussed in this paper relates to the challenges posed by the goal to develop a data assimilation system able to provide physically consistent initial estimates for all the modelled Earth system components.