[shield]

A local old timer states:

"I removed the BAMD/BAMM/BAMS/XTRAP/LBAR/ETA models because they're pretty much worthless in the subtropics (LBAR is worthless everywhere, XTRAP is a simple extrapolation). So what you see there is the new GFDL (in red) taking Rita on a WSW jog near western Cuba then back to the WNW-NW toward the TX coast. The UKMET is the model taking Rita inland near Brownsville at sunrise Friday. NOGAPS (pink track) still insists on a BoC impact to the south. The GFS (AVN) is insistent on a landfall near Corpus Christi next Satuday. The ECMWF (not plotted) takes it inland between Tampico and Brownsville late Friday or on Saturday."

[Dog Gone]

You're right to chuck those models that don't perform well in the tropics. On the other hand, though, none of the models are very good five days out and we're basically just speculating at this point.

Texas definitely has a high pressure ridge over it for the first part of the week, but that's not the critical time period. Whether it moves or holds will make all the difference.

The worst case scenario, of course, is that Rita makes landfall just southwest of Galveston as Cat 4 or 5 storm. That would push the storm surge into Galveston Bay and knock out a huge percentage of the country's refining capacity. The best case scenario is that she slams into Mexico, although we get a lot of our oil imports from the Bay of Campeche, so even the best case isn't good.

I have a feeling it's going to be a long week, with a very active hurricane thread here at FR.