|
Good afternoon. I am working with OpenStreetMaps and I would like to know what is the tolerance for measuring distances between coordinates with OSM. Is there any article or website where I could find this tolerance specification? Thanks in advance. |
|
There are numerous published studies going back over 10 years (for instance, https://www.ucl.ac.uk/~ucfamha/OSM%20data%20analysis%20070808_web_orig.pdf) comparing OSM road networks with alternative (usually official) sources. For Western Europe it's probably reasonable to assume highways are within 10 m of their true location. It's entirely possible that most may be within 5 m or better. I don't have a feel for other parts of the world: imagery offsets in places like Africa may result in quite large discrepancies. Less attention has been paid to other data such as buildings. I did look at one set of boundaries http://sk53-osm.blogspot.com/2015/12/how-accurately-have-townlands-in.html using a fairly simple approach. Broadly OSM prizes topological accuracy above spatial accuracy, and for users of consumer grade GPS 5-10m discrepancies are not usually noticeable. |
|
Generally it's whatever the contributor feels like, refined iteratively by subsequent contributors. There is no "specification" as such - you can't specify what volunteers might or might not want to do. |
