Mark's Blog

Now they seem like too little :-)

Note that he pretty much says GPS is useless for serious applications because it can be so inaccurate. I've noticed that on hikes, between the errors on the maps and in the reception, you really don't know where you are. It seems like there could be software correction, for example if you tell the GPS you're going to stay on roads, it should be able to correct itself and place you on a road.

Maybe it depends both on the terrain and also up the satellite's contact with you. If you go under a bridge, doesn't it get worse?

Yes, by "errors...in the reception" I really meant "accuracy of the measurement." The way I understand GPS, it either receives a signal from a satellite or it doesn't. The number of satellites it can receive determines the accuracy of the triangulation. The GPS displays the accuracy, for example +/- 80 feet. So you know when readings in forests, canyons, and probably cities like NY are less accurate.

When you're recording a track (a series of readings), accuracy isn't a big deal because eventhough a point may be inaccurate, the next point is inaccurate in the same way but their distance apart is still generally accurate and true.

The problem comes in the exact case of going under a bridge. The GPS has no way of knowing where you'll come out, so when you do, it has to find its satellites again. In the process you've moved, and it records an inaccurate reading 100 feet off. After a few seconds it gets more accurate and records a point back on your real track. But the changing accuracy has caused it to add 200 feet of distance to your recorded track.

It's this "flutter" that I wish the GPS makers would remove with software analysis. When I download a track of a hike, I often have to clean up such inaccurate points to make the path smoother. However, I bet they avoid doing it so they can avoid the liability of telling somebody where they might actually be. As it is, they just hide behind the fact that it's raw data from government satellites.