Overview

On May 9, 2011, NCEP implemented a "bugfix" to the GFS model that had a significant negative impact on the GFS MOS wind speed forecasts; mainly in the Western CONUS, within GFS vegetation types 7, 8, 9, and 11 (see map). Once a sufficient sample was collected, in this case two years, we were able to redevelop wind equations. The final development sample is comprised of two years of warm season (April - September) GFS model data. This sample consists of a mix of pre v9.0.0 model data; pre-bugfix model data (GFS v9.0.0); and post-bugfix model data (GFS v9.0.1). The following list shows the break down of months and which model was used:

This equates to roughly a 50/50 mix of pre and post "bugfix" model data being used. Verification statistics (to be shown in the near future) of several different independent test systems have shown that an equally mixed development sample proved to be the more accurate/skillful system.

Impact on Gridded MOS

We have set up a real-time comparison webpage of the experimental 2.5km CONUS Gridded MOS. The new wind guidance will be used in the operational 5-km CONUS Gridded MOS and Hawaii GMOS. In the Alaska region, we have retained the previous station-based GFS MOS wind equations for all Alaska METAR sites and for a few Canadian METAR sites that would influence the Alaska Gridded MOS analysis.

Official NWS Notification

The official NWS Technical Implementation Notice (TIN) has been announced. Click here.

Presentations