Well, all the models are somewhat accurate, and all the models have inaccuracies. A track 1 day out is usually pretty good, but as we can see with Katrina, even one day away we are having pretty big divergence. I wouldn't discount any model for this storm - none are real outliers.

In fact, of all the models and tracks out there, the one the NHC is predicting is the one that least matches any model concensus.

http://www.sfwmd.gov/org/omd/ops/weather/plots/storm_12.gif

And NOGAPS (scroll down for the track): http://moe.met.fsu.edu/cyclonephase/ngp/fcst/archive/05082500/29.html

And GFDL is looking like a really strong system: http://moe.met.fsu.edu/cyclonephase/gfdl/katrina12l/fcst/archive/05082506/1.html

------

Looking at the Miami radar the system has resumed a westerly motion. It is as someone said earlier: stairstepping to the west southwest. I'd say landfall looks like it will be somewhere between Ft. Lauderdale and Miami.

http://radar.weather.gov/radar/loop/DS.p19r0/si.kamx.shtml

--RC