I don't know how many engineers lurk in this thread, so I'll chime in. I'm not an engineer, but I want to be one, I'm in school studying to be one (not too far along yet), and I understand FCC matters pretty well.
The FCC's coverage contour is basically a loose prediction of what kind of power level you can expect. While it's designed to measure coverage, it's based on some flawed data and thus is a poor estimate.
The contour plots on the FCC site should be Grade B contours on the analog side, and on the digital side are just what the FCC considers to be the minimum signal level for reliable reception (on UHF, this is the 41dBu contour). Both of these contours generally fall short of a station's actual coverage area, which is why the FCC's strict reliance on them concerns me.
The biggest shortcoming is that they don't take terrain into account, only the "height above average terrain" which means that if the HAAT is artificially low for some reason (I have first-hand experience with this) the coverage predictions will fall short.
That may not have answered any of your questions, but I tried. If you want me to attempt to clarify or restate something, let me know and I will do my best.
If you want good coverage maps, use the ones on TVFool.com instead of the FCC contours.