Two quick notes on the outlier GFDL model.

The last two runs of this model were outliers in respect to turning Dean toward the NW.
The 06Z run began the turn at 96hours 20.2N/ 82.0W
The 12Z run begins the turn at 90hours 19.5N/ 81.8W

(the difference is only 6 hours and that is the difference in the times of the model runs.)

So the GFDL is now the right hand side of the Cone of uncertainty...roughly.

One other note. The CLP5 model has shifted from the Bahamas at 120 hours, to a FL Straits/ N Coast of Cuba at 120 hours. In other words it seems to be following the trrend of the GFDL. GFDL shifts to the right...CLP5 shifts to the left.

http://flhurricane.com/modelanimator.php?storm=4&year=2007

Gulfstream NOAA aircraft is now airborne enroute to Dean.