Recent Developments of NCAR 4D-Relaxation Ensemble Kalman Filter System

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Tuesday, 6 January 2015: 11:30 AM
131AB (Phoenix Convention Center - West and North Buildings)
Yubao Liu, NCAR, Boulder, CO; and Y. Wu, L. Pan, A. Bourgeois, G. Roux, J. Knievel, J. Hacker, J. Pace, F. W. Gallagher, and S. F. Halvorson

A Four-Dimensional Relaxation Ensemble Kalman Filter (4D-REKF) mesoscale analysis and forecasting technology has been developed jointly by NCAR and ATEC (Amy Test and Evaluation Command) in the last three years. A real-time operational 4D-REKF system has been implemented recently at the Army Dugway Proving Ground, which provides ensemble weather analyses and forecasts to support the routine test and evaluation activities at four Army ranges located in UTA, Arizona, and New Mexico respectively. The system is based on a multi-model (MM5 and WRF), multi-approach (perturbations), and multi-scale (nested-grid) ensemble design. With 4D-REKF, each observation is assimilated into WRF model equations with spatial and temporal weights (i.e. Newtonian relaxation), and a data-confidence factor. Ensemble-based Kalman gains are computed and used in the place of spatial weights in the relaxation (“nudging”) data assimilation terms of the WRF model equations. The ensemble Kalman Gains for assimilating observations are computed based on the forecasts of the multiple-approach ensemble forecast cycles. Thus, 4D-REKF takes the advantage of ensemble-based flow-dependent (anisotropic) correlation structures when nudging observations into WRF. Furthermore, unlike the traditional “observation-nudging”, 4D-REKF allows to assimilate indirect observations because the observation operators are included in the Kalman gain computations. In this paper, we will introduce the recent progresses with 4D-REKF. At the mesoscale, ensemble forecasts often leads to formation of sporadic, unrepresentative local structures in the Kalman gains. These structures often introduce strong noises and harm the effectiveness of data assimilation. Empirical algorithms for mitigating this issue have been studied which improves 4D-REKF analysis and prediction. Another progress is to expand and test 4D-REKF for assimilation of radar radial winds. The results of these new developments will be presented at the conference.