On 22/08/2009, at 13.14, John Smith wrote: >> When subtracting two positions from each other, the >> absolute positioning error will disappear. >> >> In addition, for many traces there will be multiple >> measurements, which will give a much better determination of >> the gradient. > > I disagree and I urge you to test this speculation out with 6-10 GPS > devices and a hill and see what the results are, I doubt you would > get any where near as accurate as you are assuming.
What kind of accuracy do you want? We are talking about people wanting to tag incline={up, down}. GPS elevation differences are certainly good enough to make that binary classification, especially if the incline is one that matters. Should every sorry GPS trace be used? Perhaps as a quick-and-dirty start, but you are right, better and more accurate data is needed. People who care about these things in a particular area could certainly arrange to get accurate gpx data with an altimeter equipped GPS. In addition, we have access to topographical data many places, a fact which has not been mentioned in this thread. This data can also be used to derive elevation gradients of roads. I realize it wont be possible to compute elevation gradients tomorrow, but why not plan ahead? Who would've thought a few years ago that the project would be so far advanced? Wouldn't you rather have nodes on a way with "incline=7%" than "incline=up"? Wouldn't it be better to get this data from a database lookup than from manual tagging? Cheers, Morten _______________________________________________ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk