Hi, when I drove through Letterkenny, County Donegal, on the Irish north coast a few days ago, I noticed that GPS appeared to have a bit of a problem, because my track was consistently a few metres north north east of the map. Just now I wanted to enter a few things I had found missing when I was there (using Potlatch 2), and saw that my track fits perfectly not just for Bing, but also for Mapbox and Maxar Premium, so it must surely be the map that is "off". (How) can this be fixed? asked 15 Aug '19, 23:36 Marabu_Too |
Interestingly at https://www.openstreetmap.org/#map=19/54.94726/-7.73106 there's a visible offset between various imagery resources. at that roundabout, OSM data is offset to the southwest compared to both Bing imagery and GPS traces. ESRI is offset a few meters to the northwest, Maxar premium and standard slightly to the south. The offset of different imagery sources may vary as you move around; where were you looking?
@SomeoneElse I panned the map to Donegal, zoomed in onto that roundabout, because around it a number of shops have changed since someone put them on the map, opened Potlatch 2, saw the offset to Bing, loaded my track with its annotations, saw that it fitted Bing, tried some of the other satellite imagery.
I loaded some older tracks from journey to that Retail Park and saw that all were nearly identical around the roundabout, at at any Letterkenny street that extends more or less east/west, all ran a bit to the north of thos streets.
In general I wouldn't bother. It's really tedious realigning objects like this, and there is a possibility of making things worse. We mapped a lot of Nottingham from Yahoo imagery years ago which turned out to be displaced about 10 m N of true (accurately located street lamp data helped show this). The problem is that a mass realignment is unlikely to be correct (adjacent map tiles will have different offsets & different zoom levels will have different offsets), and partially corrected stuff just looks awful. If you do want to correct something choose an area with 'give' at the edges (major roads, parks etc) so that the differences are not too jarring.
In this particular case I agree that there is a correction-worthy offset, but just to prevent over-eager correction: Be careful to not trust a single GPS trace, even if it shows a consistent offset compared to OSM data. GPS receiver errors can be self-consistent and last for hours. Only trust a cloud of dozens of GPS traces.