OpenStreetMap
I've been wandering around Saltaire recently, and I've noticed that the jitter in the GPS traces seems much worse than in Leicester. The streets in Saltaire are narrower, with long rows of stone terraces, and a poorer view of the sky. I've had to do some 'best fit' ways, or ignoring certain outlying spikes when I know for certain that the road is straight...
Discussion
Comment from tenzarelli on 18 October 2008 at 10:56
That's cool, are you using your knowledge, aerial photos or a better GPS?
Comment from Chris Elston on 18 October 2008 at 11:15
There don't seem to be aerial photos of the areas I'm mapping, which is a shame as I'd really like some confirmation :) So, for example, I went out and specifically walked Victoria Road in Saltaire (which is actually one of the wider ones) - keeping to one side of the road, and generally trying to maintain a 'straight' trace (here @Chris%20Elston/traces/229059) . I also made a point of sighting down the length of the road, and it's damn near straight (Saltaire was built in the late 1800s by the local mill owner Titus Salt as a model village - I'm pretty sure he'd have specified a straight road ;) ) Yet the trace shows quite a big deviation. I don't know why. So in that case, I've taken the start and end points.
Comment from RichardB on 19 October 2008 at 11:35
You can get deviations if you are near to large buildings - sometimes you can get a "multipath" error - where the signal from one or more satellites has reached you via 'bouncing' off the building. This makes the unit think you are further from that satellite - because the signal has taken a longer route to reach you - hence you'll get some error in your position which can make a straight route not straight on your trace. If you actually can get down the middle of the road safely - then this will make sure you are further from any buildings that may cause these errors. And sometimes you'll just need to travel down a route more than once to get a decent trace. The more independent traces you have - the closer the "average" trace is likely to be to the correct route.
E.g. see here http://en.wikipedia.org/wiki/GPS#Error_sources