Expanding the GSI-based hybrid ensemble-variational system to include more flexible parameter settings

The ability to incorporate flow-dependent, ensemble-based estimates of background error covariances has recently been developed for use in the GSI, the current operational 3DVAR system at NCEP, by utilizing the extended control variable method. The initial implementation of the hybrid method allowed for only minimal flexibility in choice of parameters. The only tunable parameters included single global values for specifying localization distances as well as weighting parameters to determine just how much of the ensemble based estimate to incorporated.

In order to make better use of the static and ensemble estimates for the background error, the hybrid method is being expanded to allow for spatially (and perhaps temporally) varying parameters. Some preliminary results from experiments utilizing these more flexible parameters will be presented, both in terms of impact on analysis (fitting of observations, incremental balance) and subsequent forecast skill in the context of the NCEP global forecast system. Plans for allowing these parameters to be fully adaptable (based on information from observations or ensemble forecasts) will be discussed.