J13.4 Comparison of RTMA and ensemble Kalman filter surface analyses

Wednesday, 26 January 2011: 11:00 AM
2A (Washington State Convention Center)
Brian C. Ancell, Texas Tech University, Lubbock, TX; and C. F. Mass, P. Regulski, B. Colman, and L. K. Cook
Manuscript (371.2 kB)

Real-time mesoscale surface analyses are very important to National Weather Service (NWS) forecast operations. Routine, gridded analyses provide forecasters with a spatially and temporally consistent characterization of the current weather situation at high resolution. These analyses are also used to verify other operational models within the NWS National Digital Forecast Database (NDFD) as well as the NDFD forecasts themselves. A leading NWS method to produce mesoscale (5 km and 2.5 km) surface analyses is the real time mesoscale analysis system (RTMA). The RTMA is a two-dimensional variational data assimilation system that assimilates surface observations using a prior estimate from the coarser (13km) Rapid Update Cycle (RUC) model. The RTMA does not use flow-dependent error covariances during data assimilation, a technique that has been suggested by recent studies to improve analyses. The purpose of this study is to compare surface analyses produced by an ensemble Kalman filter (EnKF), a data assimilation system that produces and uses flow-dependent covariances, to those of the RTMA.

Surface analyses from the Gridpoint Statistical Interpolation (GSI) system, which is used to produce RTMA analyses, are first compared against analyses from the University of Washington Pacific Northwest WRF-model EnKF. This comparison involves both the 4- and 12-km EnKF and uses the same EnKF prior estimate to isolate the effects of the data assimilation system and the resolution of the background error covariances used during assimilation. Both the fit to sets of unassimilated observations and the analysis increment fields are examined here to assess the two systems.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner