Yes, though it will depend on how well the curve is filled out with efforts that were truly maximal for each duration (or, at least, the durations that the GC algorithm uses for estimation). It's a clever algorithm but it's not clever enough to know when you're dogging it.

The three-parameter fit is often a little bit better than the two-parameter fit (you still have fill out the curve with maximal efforts). Either will tend to overestimate the power you can sustain for an hour-long TT but the three-parameter fit usually overestimates less. If you really really need to know, if you're training for something very specific, you'll probably want to do an actual hour-long maximal effort rather than use these shortcuts. However, for training purposes these estimates are usually okay -- that's because general training isn't a terribly demanding application for a power meter.