[r9etb]

As far as GPS is concerned, they have picked a limited selection of spheroids and insist you use them.

I'm talking more about the broadcast message and how it's built.

The message is constructed by a series of least squares fits over a long-term prediction of the satellite trajectory. If the gravity model is significantly flawed the prediction may be off by meters per day.

In real life, some of is mitigated by the operators, who keep an eye on the accuracy of the broadcast message; however, it increases workloads, and if there are any delays in sending new orbit information to the satellites, the world will see additional errors in their solution.

Not too big a deal for most, but probably rather upsetting for a GPS-guided bomb.

[RightWhale]

Each satellite in the GPS constellation has to be updated constantly. Whether it is every day I don't know, but if they start drifting the receivers will just follow. I assume they are updated every time there would be enough error to measure. For differential stations, that would be of the order of a foot.

[RossA]

Why do you think variations in Earth's gravity are important to GPS. For an example of the important effects on the satellites' orbits, see:

http://www.ngs.noaa.gov/GPS/noaa_acn.html