High Resolution ensemble Background Error Covariances for 4D Ensemble-Variational Data Assimilation

Location: FL2-1001
Speaker:
Joël Bèdard, Environnement and Climate Change Canada
Description:

Higher model resolution model implies a higher number of degrees of freedom and a need for dense observation networks (e.g. satellite, radar and surface observations) to constrain the model initial state. Like in many other NWP centers, only a small fraction of the available observations is being used in ECCC operational systems. The horizontal thinning for all assimilated radiances is 150 km; radar observations are not yet assimilated operationally; and the screen level wind observations are not yet operationally assimilated over land. Although data assimilation for convective scale NWP has been the object of intense research lately, the resolution and the quality of background error covariances remain factors limiting the assimilation of dense observations.

The data assimilation component for a new short-term convective-scale numerical weather prediction (NWP) system covering most of Canada at2.5 km resolution is currently being developed. It is based on a fully cycling deterministic 4DEnVar scheme with analysis increments initially computed at 10 km resolution. Several practical approaches have been evaluated and compared for generating ensembles of short-term forecasts for specifying the required background-error covariances. This includes ensembles from an EnKF and also from much simpler approaches. The new system is evaluated and compared with using Environment and Climate Change Canada's currently operational regional data assimilation system (with increments computed at 50 km resolution) for initializing forecasts from the identically configured atmospheric model.