I work for a surveyor & I don't typically go out with a gps crew so I'm not an expert on this. The reason for the accuracy of surveying gpsr is timing and time. The longer you sit there & average a coordinate (typically 30+ minutes for our control points with an adjacent local base station for 8 or so hours aka rapid static) the better the accuracy, also we use software that compares & reprocceses that data to correlate to multiply D.O.T. GPS base stations. That I know of most civialian gpsr don't average coordinates, but that may be changing... There is also the problem that most gps units don't have atomic clocks in them, so the timing between measured signals and actual signals causes error. Using my own gpsr (Garmin eTrex, typically accuracy running 3-4 meters and a good 110 degree spread of 10 or so satellites, but rearly reading waas) I found that the coorinate never match exactly, but if you draw a circle around you with the radius being what your gps says accuracy is running, then the point will be in that circle, but the elevation is usually 2 to 3 times that error. If your coordinates are less that 2 meters off, then you just got lucky. I asked one of the field crew guys how accurate his GPS was & he said usually .03 to .1 feet.
As for Lat/Lon changing with elevation, that is possible because "straight above" to us usually means opposite of gravity, whereas gps coordinates are based on a mathmatical formula that isn't a perfect model of the earth. If you change the elevation of your gpsr you may actually shift Lat/Lon because the center of Lat/Lon may not always be the center of gravity. Shouldn't affect our gpsr though, their not near that accurate. So occasionally you'll find places where gps says water runs up hill, even if your watching it go down...
--Pyre Flos
when in doubt or danger... ATTACK!
