An hour to go until the 5p advisory, though all of the particulars are probably getting announced to the various NWS offices on the standard conference call as I speak. Here's what I see as the main question: the forward speed of both Dean and the upper low near Florida. The GFDL has been showing the fastest forward speed of all of the models plus the slowest retrogression of the upper low. This isn't counterintuitive, as they are being steered by two different ridge centers right now. The other models are a tad slower with Dean and a tad faster with moving the upper low out of the picture, resulting in a more westward track. Which is going to win out, or as is the case with so many things in the tropics, does the true answer lie somewhere in between?

Right now, the consensus models as well as most of the globals are clustered fairly tightly over the northeastern portion of the Yucatan peninsula. The GFDL, some of its variations, and many of the GFS ensemble members are the ones that are significantly further north of that area. Regarding the ensembles, one has to ask the question of if the perturbation method used to create the ensemble initial conditions is leading to a scenario where the storm interacts with a slower-moving or more intense upper level feature in the Gulf? Comparing the 12z early guidance to the 12z ensembles, there's not a ton of speed difference between either cluster of models, making me think that the other scenario of more upper level interaction is more likely than anything related to how they represent the storm.

It's unnerving to be 3 or so days away and not have a better idea of what is going to happen, but I think we'll know by tomorrow morning whether the GFDL solution is out to lunch or if it is a very viable possibility. How this upper low and storm start to evolve today is going to be key, I think, moreso than any potential land interaction that Dean could have until it nears the Yucatan. I hope the comments above highlight this possibility and the thinking that goes into the track forecast making process when it comes to evaluating model guidance as a whole.

For now, however, the consensus track is the 'option of least consternation' -- and the track that both I and the NHC are going with right now -- but that could change as we get into tomorrow morning.