Will the Future of Data Assimilation be 4D-Var or EnKF?

Eugenia Kalnay
Department of Atmospheric Sciences, University of Maryland, College Park

We consider the advantages and disadvantages of EnKF and 4D-Var, in view of simple experiments with the Lorenz (1963) model, with the SPEEDY primitive equations model (using both perfect model and reanalysis “observations”), and in view of recent results with both perfect models and real observations. We point out some advantages of the Local Ensemble Transform Kalman Filter, and its extension to 4 dimensions, which brings to EnKF the main advantage of 4D-Var. A table summarizing the pros and cons of the two methods by Lorenc (2004) is adapted with additions and comments.

Last Update: October 3, 2005