15D.3 Results From The 2013 HFIP Realtime Demo GFS Ensemble

Friday, 4 April 2014: 8:30 AM
Regency Ballroom (Town and Country Resort )
Philip Pegion, CIRES/Univ. of Colorado, Boulder, CO; and J. S. Whitaker and T. Hamill

In support of the HFIP realtime demo for the summer of 2013, an experimental version of the GFS analysis and forecast system was run at NOAA/ESRL/PSD. This system differes from the operational system run at NCEP in several ways. The two key differences is that the HFIP system was run with the semi-Lagragian version of the GFS, with and ensemble resolution of T574L64, and the deterministic model run at T1148L64 ( versus the Eulerian T254L42/T574L64 operational configuration). In addition, the ensemble run at ESRL included a collection new stochastic physics schemes. There are 3 stochastic parameterization implemented in the realtime demo, stochastic kinetic-energy backscatter (SKEB), perturbed boundary-layer humidity (SHUM) and stochastically perturbed physics tendencies (SPPT). SKEB is meant to represent the effect of unresolved or poorly resolved scales of motion on the resolved scales, SPPT is mean to represent structural uncertainty in the physical parameterizations, and SHUM is meant to represent the effect of sub-grid scale variations of humidity on convection

The result of the ensemble with new stochastic physics results in a larger, and more realistic ensemble spread to the tropical cyclone track forecast. The increased spread results in a more reliable forecast at longer lead (days 3 through 7) even the deterministic skill of the tropical cyclone forecasts did not perform as well as the operational GEFS.

- Indicates paper has been withdrawn from meeting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