Session 9.1 Preliminary results of WRF model performance as a step towards the NCEP Rapid Refresh cycle

Thursday, 7 October 2004: 8:00 AM
Tatiana G. Smirnova, CIRES/Univ. of Colorado and NOAA/ERL/FSL, Boulder, CO; and J. M. Brown and S. G. Benjamin

Presentation PDF (178.6 kB)

In preparation for replacement of the Rapid Update Cycle (RUC) model by the WRF model in the "Rapid Refresh" (RR) slot at NCEP, the Eulerian Mass version of the WRF model was set up in spring 2003 at FSL to run on the CONUS domain of the operational RUC. The WRF model is initialized from the RUC 3dVAR, and runs in real-time with 20-km horizontal resolution to 48-h. Recently, the same version of the WRF model was set-up to run on CONUS with 13-km horizontal resolution, again initialized from a parallel RUC cycle at FSL running on the identical 13km grid and domain The 13-km horizontal resolution is anticipated to be the resolution of the WRF RR model. The physics options used in WRF are as close to the operational RUC configuration as currently possible. They include the Grell-Devenyi convective scheme and the RUC Land-Surface Model (LSM), whereas surface and boundary layer physics follow from the ETA model parameterizations, and NCEP 5-class scheme is used for microphysics.

The performance of the WRF model is compared against the RUC model for such variables as precipitation, cloud, surface wind, temperature and dew point, important for aviation operations. The results of these comparisons will be presented at the meeting. Also, the WRF and RUC time evolutions of the noise level were compared indicating significantly higher noise levels in WRF in the first 5-7 hours into the forecast. This result suggests that digital filters used in the RUC initialization should also be applied to the WRF model for its use in ?Rapid Refresh?.

Supplementary URL:

- Indicates paper has been withdrawn from meeting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