New supercomputing power at NCEP will finally yield an upgraded ensemble prediction system similar to the ECMWF EPS. This is excellent news for us at WeatherBELL as a properly calibrated GEFS can be used in concert with the ECMWF system to provide blended forecasts e.g. 1-5, 6-10, 11-15 day averaged temperature anomalies.
1 high resolution control = GFS deterministic at 13-km
1 low resolution control = GEFS control at 27-km
20 perturbed members = GEFS at 27-km
The ECMWF system has 52 ensembles where the HRES or deterministic is at 14-km while the ensembles are around 25-km.
I'm (Ryan Maue) pulling in the parallel feed of data from NCEP. The model completes and then is posted to the FTP server about 3-hours after the operational GEFS run. Of course, once the upgrade is completed, the data will post at its normal time. You can read about the changes in the NWS TIN here: http://www.nws.noaa.gov/om/notification/tin15-43gefs.htm
I (Ryan Maue) expect much better 2-meter temperatures, tropical cyclones, and everything else as the overall prediction system will have higher skill than the much older (5-7 years old) system.
Quote selectie