SHIPS is actually one of the best-performing intensity models out there. Out intensity forecast errors are generally fairly substantial despite rapidly-improving track forecasts, unfortunately.

A note about the oceanic heat content...
Compare the following two images:
Pre-Katrina: http://www.aoml.noaa.gov/phod/dataphod1/work/HHP/NEW/2005238go.jpg
Current: http://www.aoml.noaa.gov/phod/dataphod1/work/HHP/NEW/2005262go.jpg

The threshold value of 60 kJ/cm^2 is denoted with the black outline. While conditions around Katrina were highly favorable in terms of energy, they aren't as much with Rita. However, they are still above the threshold values for significant storms and except for in pockets extend across the entire Gulf. The surface temperatures have recovered but the sub-surface ones haven't. Ultimately, we have tools such as the one above, but we aren't certain what is going to happen. I'd put it about equally likely that Rita maintains as a minimal cat. 3 all the way to landfall or that Rita deepens to a minimal category 5 storm over the central Gulf, with the most likely scenario lying somewhere between there.

Ultimately, the NHC is generally conservative with their intensity forecasts. Storms of high-end category 4 or low-end category 5 intensity are rare in all basins; it is somewhat like predicting record temperatures over land -- tough to do because, after all, they are records. You'll see them mention the possibility while keeping the intensity forecast more conservative. They may well be doing that here, though I think the landfall intensity of 115kt sounds about right (this is in the range I noted last night, toward the high end).