To answer a couple of other questions...

BillD re: radar fix in RECCO -- it's another factor of the storm being over land. They can't fix the center with a dropsonde if part of it is over land, particularly Cuba, so the best they are able to do is radar. Since an updated position was requested/required, they were tasked to use radar to provide the NHC with such a fix.

Steve -- it's interesting to note, though, that the ETA and many other regional/mesoscale models receive their initial/boundary conditions from the GFS. Whatever the GFS sees along the periphery, so does the Eta. It is true that the GFS occasionally suffers from some convective feedback problems, but I would rather go with a global dynamical model rather than a simplistic dynamical model such as the BAM series...or even a mesoscale model such as the Eta. Personal preference, though, but one that has seemingly proven over time to best handle tropical cyclones. With a complex flow pattern involving Ivan, I think the global model is going to have the better handle on it.