One other thought: 3 meters seems like a fairly large tolerance for location, but the GPS receiver spec sheet lists the speed accuracy at 0.1 knot. So while you might not want to place buoys based on GPS readings, it's accuracy for determining moving speed seems pretty good. I assume this is because it takes 5 readings per second, and the 3m margin of error gets quickly averaged out across the multiple readings and the speed calculation derived from them.