I believe Andy makes an important point here. All accuracy is relative. On a global scale the positional accuracy of BING or a GPS trace is going to be perhaps 10m or more. More GPS traces does not necessarily improve the accuracy in all cases because the shape of the terrain and existing features can influence what part of the sky has satellites visible or effects the reported position because of multi path effects.
At the local level if you use the same source (BING, GPS or something else) on the same day to map a set of features you can be reasonably sure their relative accuracy is better than 10m and with careful work could be as close as 1m. At a later date you may take the view that globally the day of editing was off a bit and adjust. This tends to degrade relative accuracy unless we move all objects at the same time. Anyway, the bottom line is that unless you are using survey grade instrumentation the location of objects in OSM is never going to me much better than the 10m globally referenced accuracy and could easily be worse in places. But then who actually needs to know that one object is within 10m of an object 100km away. It's just not that important. Relative accuracy is of far greater interest in my opinion. So, my suggestion is to ignore the apparent discrepancies between imagery versions and GPS etc. Just pick one (perhaps the best one for our needs) and work with that for your whole area. I, like the others here in Birmingham, mostly use BING. I know it's off by a small amount (perhaps less than 5m generally luckily as we are not so hilly, unlike The Lakes) but is by far the most accurate resource for tracing. As for Streetview, I would never trust wholeheartedly the work of others, including the OS. They have had errors in their mapping ever since they started more than 200 years ago and they are still correcting things as far as I am aware against their current reference frame. And of course the most important point is that if it's not in OSM then it's of no benefit to anyone. So mapping objects, even poor quality mapping of objects, is arguably much better than not mapping them at all. Cheers Andy > -----Original Message----- > From: SomeoneElse [mailto:li...@mail.atownsend.org.uk] > Sent: 22 October 2012 09:44 > To: talk-gb@openstreetmap.org > Subject: Re: [Talk-GB] Lake District Satellite Imagery > > Henry Gomersall wrote: > > I noticed the new imagery has all the walls visible. This is really > > great info for walking, and a real pain to map. Is the imagery good > > enough to trace for walls and suchlike? > > Once it's been aligned against a few GPS traces, yes. Until that's been done a > wall mapped just from Bing imagery might be a few meters out. > However, even if a wall is mapped a few meters wrong, the position relative > to other features will still be useful - although once GPS traces are added it'll > need to be realigned, of course. > > Cheers, > Andy > > > _______________________________________________ > Talk-GB mailing list > Talk-GB@openstreetmap.org > http://lists.openstreetmap.org/listinfo/talk-gb > > ----- > No virus found in this message. > Checked by AVG - www.avg.com > Version: 2013.0.2741 / Virus Database: 2616/5846 - Release Date: 10/21/12 _______________________________________________ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
