Originally Posted by TheRCF
If so, then I assume that the distance is based on a straight line between those two track points.
Exactly. Like measuring the circumference of a circle with a straight edge ruler.
Originally Posted by TheRCF
If you are going around curves, this would obviously give the wrong distance and if you are doing sharp turns like switchbacks on a mountain, it would be way off.
And that's if you get a good signal around a mountain switchback at all. Riding along a mountain side will generally give you a bad reception as it is, adding more error to the equation.
Originally Posted by TheRCF
mile straight course, your would still be within 1 percent. Doing the test periodically and averaging the distances would increase accuracy considerably.
If you could find a straight line for say, 1km. You could take most good GPSrs and mark a waypoint at each end of the distance and "average" the start and end point markers. With most GPSrs, they'll allow you to get a good avg of 1-2' of accuracy if the GPSr sits in place for 15 minutes or so. Then you can get a good solid reading for that 1 km distance, hop on the bike and ride from one painted line on the ground to the next and you'll know how close your cyclometer is. Once you put the GPSr on the bike with you and you start moving, it's going to take readings every couple seconds and see how "straight" it was from the last reading and determine whether or not it should plot another point on its route. A two point line can be accurate to the low accuracy described above. Once you start moving, every point along the route will be "placed" with the accuracy at its location - likely anywhere from 10-50feet. So on a route for 2300 track points, each of the 2300 trackpoints will have a 10-50 foot (sometimes more, very rarely less) accuracy.
The assumption is that the accuracy is always off in the same direction, so that the accuracy readings would offset each other, which is an assumption that just can't be real. But at the end of the day, I'm pretty pleased with a 1.6km difference between my cyclometer and GPSr over 160km. It at least attests that all of the "inaccuracies" in measurements of our simple wheel rotation measurement or high-speed GPSrs all equal out to very little - however they are doing it.