Descombes, Gael, and AulignŽ, Thomas, National Center
for Atmospheric Research
The specification of
model background error statistics is a key component to data assimilation since
it conditions the impact observations will have on the analysis. While Ensemble
Kalman Filter (EnKF) algorithms rely on an ensemble of model forecasts to
implicitly convey the background error covariances, variational methods
represent them through a statistical model. The model for the WRF data
assimilation estimates separately the variances, vertical auto-correlations via
Empirical Orthogonal Functions (EOFs), horizontal auto-correlations via
Recursive Filters and cross-covariances via linear regressions. Recent work to
include new variables in the analysis such as cloud parameters and chemical
species have required to fundamentally restructure the software interface to
allow for a simpler, flexible, robust, community oriented framework. We will
present the advantages of this new design for the data assimilation community
and demonstrate some of the new features on data assimilation experiments.