in: PBricker; PBricker > 2006-12-20;
| # Posted 2006-12-21 02:53:40 | |
| GlenT: | Forerunner spec is for 1 meter immersion for 30 minutes, so you''re still going to have to accept the data for future runs, even if you don't like it. :-) Unless you flushed. |
| # Posted 2006-12-22 05:16:59 | |
| PBricker: | I'm hoping that the Forerunner doesn't track all the little twists and turns on this run, though I doubt that could add more than a tenth or two. Professionally I like to theorize without any data, but I think I'll try the following experiment. Go to a football field and do switchbacks, staying on a 45 deg slant from one end of the field to the other, first switching every 30 yards, then every 20, then 10, then 5. The questions is: when will the distance measured by Forerunner start to drift down from 141 yards (always the true distance) to 100 yards. Any predictions? |
| # Posted 2006-12-23 05:45:28 | |
| GlenT: | Interesting problem. The accuracy of the measurement will depend on how fast you go. The measured distance depends on taking enough points to fully define the actual path you are trying to measure. If you only collect a few points over the time it takes to cover each "zig" or "zag", the path seen by the Forerunner will be one which "cuts the switchbacks", rather than the path you are actually taking. The distance is then calculated as the sum of the lengths of the segments connecting the points the Forerunner collects, which will be less than the "true" length. The more points you collect on each zig or zag, the better defined the path shape will be and the closer the measured distance will be to the theoretical length.
The Forerunner takes a point once a second and I figure that about 30 points per zig or zag should be enough to measure with about 1% accuracy (if I calculated correctly). So when you do your actual experiment, you need to take 30 seconds for each segment (if you're in a hurry, just take 10 seconds and you'll still be at about 3% accuracy). Of course to maintain the accuracy, as your switchback frequency increases, you'll take longer to complete each length of the football field. And in the limit, it'll literally take forever. Of course one assumption in all of this is that the measurement errors (due to random noise) all cancel each other, which they seem to do based on my experience with the Forerunner. Also, if the Forerunner loses track of the satellites occasionally, you may get some bogus interpolated points (which is more likely to happen when you're in the woods). The experiment you really need to do is to use the Forerunner on a course where you actually *know* the distance (i.e. measured by some other technique) so you can determine the accuracy of your particular Forerunner unit. So far you've only demonstrated that your personal distance estimation technique gives a different answer than the Forerunner. For most of my local road routes that I've measured with both the Forerunner and a bicycle odometer, I get agreement to within less than 1% of the total distance. I agree that on courses with lots of twists and turns the Forerunner will probably under-measure the distance, but probably not by 10%. |
| # Posted 2006-12-24 20:12:50 | |
| PBricker: | Yes, everything you say was borne out in the test reported in my log.
You say: "The experiment you really need to do is to use the Forerunner on a course where you actually *know* the distance (i.e. measured by some other technique) so you can determine the accuracy of your particular Forerunner unit." The problem, of course, with respect to trails is that the Forerunner itself is likey the best measurement technique, since bicycles (and wheels generally) are also not super-accurate over rocky or rooty or muddy trails. But as to my particular Forerunner unit, it was very accurate running two laps (800 m) on the track. And, looking at it in map mode, it was a perfect oval, overstruck twice with data points. |
You must be logged in to add a message