Title: Strategies For The Use Of Ensemble Information in Data Assimilation

Author: Dale Barker (UK Met Office)

Ensemble prediction systems (EPSs) potentially provide a wealth of information of direct use to data assimilation through the provision of ensemble perturbations tuned to represent flow-dependent forecast error. This talk will begin with a brief summary of the theoretical and practical constraints for real-world ensemble data assimilation (EnDA), many of which are shared with more traditional variational data assimilation (VarDA) methods.

The distinction between traditional EnDA and VarDA techniques has become increasingly blurred. Numerous variational algorithms exist that utilize ensemble-based covariances to produce analyses that closely resemble those of traditional EnKF algorithms. Climatological covariance modelling, long since a core component of variational systems, can be used to ameliorate the effects of sampling error in EnDA. Variational systems have been used to do the 'donkey work' of quality control, transforming ensemble fields to observation space required to permit application of EnDA systems to real-world applications. This talk will review a number of VarDA/EnDA algorithms, and provide a few remarks on their relative strengths and weaknesses to initiate further discussion.

GMAO Head: Michele Rienecker Global Modeling and Assimilation Office NASA Goddard Space Flight Center

Curator: Nikki Privé Last Updated: March 1 2011