Sure, I know that. But this method would be simpler than relying on the magnetic field. A drone would calibrate "the compass" by itself. What's more, while flying a drone would be able to adjust "the compass" constantly.
A simple solution to fly-aways caused by erroneus GPS reading would be monitoring by the software sudden changes of drone's position. When the software sees such a sudden change it could, for instance stop, a drone and wait for next readings. Somebody mentioned such s solution already.
The compass is essential for flying indoors where GPS signals are generally not available. Another potential issue with GPS is the frequency of new position fixes. It may be different now, but years ago GPS chip sets calculated position only once per second, and this is way too slow to control yaw.
The phantom doesn't use the old fashioned single axis gyros. It uses a 3 axis MEMs gyro and a 3 axis MEMs accelerometer. But that has little if anything to do with the compass adaption. They can help but only so much.
In order to adapt the compass purely, you need to derive a heading adaption from the course over ground. If the course over ground is effected by the wind (which it is), you can only trust it so much. You can probably improve confidence if the reciprocal shows something different than a 180 difference in course over ground but it is not exact.
The real issue with sensor fusion, which is what we are talking about, is when one sensor goes rogue. Adapting compass values automatically from other sensors is fine right up until they drift or experience noise or other factors that can cause the whole thing to go wonky very quickly.
Just curious, whats on the transom of your boat? If its not a wooden paddle and its a big metal thing called a motor, don't you think being only 10' away from that magnetic thing could affect the compass? I would also think that the rocking of the boat during the calibration might also sway the calibration results.
I think any metal implanted or attached to your body for medical purposes has to be non-magnetic to be compatible with MRI scanners.
But there can be magnets that you could forget about. For example, I wear glasses that have tiny neodymium magnets in the temples that hold sunglasses. Some tablets and cellphones have magnets built-in or added as part of a mounting system, many electric cases have magnetic closures, etc.
Luckily the force of a small magnets field is inversely proportional to the cube of its. distance.