| Sure as dirt, the UK model bounced back to the gulf coast of Florida in the 00 run. It has an odd "kink" in the track between Cuba and Hispanola that may give us a hint of the importance it places on those land masses in determining track. I'd have to think about that; maaybe just a coincidence of location and trough timing. At any rate, the UK is back west with the usual NW track and putting pressure back on the NHC forecast track. Both the GFDL and HWRF bounced east a little on the 00 run but regained the west track and a bit more by 06:00. The HWRF actually shows an inland part of the track in the Carolinas. Why does it look like these two are SO sensitive to initialization or seem to ignore it if the data is not fresh? Even the BAM models show an unusual consistency among the three, but it appears to be the western outlier right now and probably the result of the current westerly direction of Emily. With some continued bounce in the short-run track, it seems that the models are coming around to the Haiti consensus but it doesn't take much of a turn or track extension to change in either direction from Emily's approach point to Hispanola. To me, it all means that NHC might be a little exposed with their forecast re: the models. This one has all the indications of a "brush" and those are the worst for them. You can't forecast a "hit" for the whole coast from Daytona to DelMarVa 3-6 days out, that's for sure. Nor can you appear to ignore those three model lines for very long and be the Eastern outlier for entire forecast track. I think they are really counting on the weakness of this storm having it self-destruct over Hispanola and reforming remnants to be stalled in the Bahamas long enough to be swept away. Even if it was dead on the HWRF track, a "weak" Emily minimizes the effect of being very late in forecasting it on the coast. Right now, they are really vague about the 3-5 day thing and blaming it on the models. I don't think any of us would see it any differently from the NHC perspective. Stronger storms are easier to predict and its much easier to depend on a model or two...Not the case here. |