Some regression to the mean kicking in. Givin the history of these kinds of events its never a good idea to completely buy into the south trend. The furthest south model run within 48hrs of the storm is much more likely to end up the laughing stock than it is to actually verify.