I know that the model consensus has shifted dramatically to the right, but it's worth looking at the WRF-ARW hurricane model. This model will eventually replace the GFDL. During the 2005 hurricane season, the WRF-ARW 12-km hurricane model was tested during Katrina, Maria, Ophelia, Rita, and Wilma. It was the most accurate model at 48h and beyond for intensity forecasts and at 72h and beyond for track forecasts.
At 48h, the 12z WRF-ARW has Ernesto at 998mb very near the Cuban coastline. This model seems to pick up on the more northerly component of Ernesto's motion that is happening right now. However, at 120h, it shows Ernesto as a 968mb hurricane moving into the Western Gulf. In this forecast, Ernesto is not even at 25N when it passes the longitude of New Orleans.
