HF took care of most of the question (very well, I might add), but I'd also like to add a thing or two.

The pressures being higher on average won't affect the models picking up the storm, just that a higher pressure than normal inside the storm can result in higher wind speeds than one might expect from climatology. This goes into the PGF stuff HF mentioned - the gradient wind.

Models can't initialize a tropical cyclone well at all, though they tend to be able to pick up on a vortex once it is well-defined or if it is large enough. The resolution in many of these global models is not very high - 12 to 40km - meaning that smaller systems may have additional troubles being captured in the model analyses.

The buoy system is very useful, but there aren't terribly many buoys out there. With these small systems, it's tough to find many impacts unless the center passes near or over a buoy. Bonnie passed very near buoy 42001 in the northern Gulf and some impacts were measured there; I believe the NHC used it as justification for some intensity changes.

GFDL 12z on Earl did not really even capture the storm, so I'd like to see some more run-to-run consistency with capturing the storm before latching on to where it may head. I did see the 18z to the Yucutan, however, and it certainly is one possibility out there...can't really discount anything right now, unfortunately.