It's not the pages that are broken or bad, it's the GFDL output. The fine-mesh grids are only available for a subset of variables nowadays, not the full suite that used to be available. Sea level pressure is the main one and it always works; potential vorticity is usually not available or calculated. Note that the sites are only current to the data they get in, usually an issue on the national centers' side and not FSU/PSU.

There's nothing wrong with the radar imagery from Melbourne/Miami, either. The storm is far enough away from Miami's radar that the radar beam is overshooting most of the precipitation and not picking up on the low-level banding features quite as well. The Melbourne radar shows the structure better. Remember that this isn't a Katrina-like deep storm that will always show up well on radar; it's a weak storm that won't show up that well on radar due to the limitations Ed mentioned earlier today.