The National Weather Service currently uses 3DVar for operational data assimilation under the framework of the Gridpoint Statistical Interpolation system (GSI). Recently, the GSI has been extended to include a hybrid ensemble-3DVar assimilation capability and is undergoing testing for operational implementation. While the GSI can currently assimilate radial velocity data from WSR-88D radars, it does not yet use a variational treatment to assimilate radar reflectivity factor. Assimilating both wind and reflectivity data, however, has been demonstrated to be beneficial for improving storm-scale analyses.

This talk will focus on the development of a hybrid-ensemble 3DVar radar data assimilation and prediction system with the NCEP GSI and the Non-hydrostatic Multiscale Model on the B-grid - the model which (at the time of writing) will replace the WRF-NMM as the North American Mesoscale model (NAM). Preliminary results from a case study will be presented.