|
I've noticed on uk OSM derived maps that the sumit location of hills is often off, this is based upon me downloading the Database of British & Irish Hills form here http://www.hills-database.co.uk/index.html & adding bits of it into a couple of different apps (PC & Android) that use maps based upon OSM, is there a reason that the whole databse is not incorporated into the OSM as it's under an open licence. |
|
They list their license as CC-BY 4.0 which is not considered compatible with the OSM license without additional waivers. 2
See also https://lists.openstreetmap.org/pipermail/talk-gb/2019-February/022515.html and following messages in that thread.
(25 Jul '20, 18:36)
SomeoneElse ♦
|
Do you have specific examples that you consider to be off?
Example as requested, the pic of data is what my Android app pulls up from the imported part of the the Database (DoBIH) & the map is an OpenAndroMap with the blue pin the waypoint (from the DoBIH) overlayed on the map. Which indicates that there is about 200ft error
So this has been discused before & I think I get that the database has a differnt openlicence & possible copyright of the database as a whole but that alows free use of its data, but not the wholesale import without credit & presumably there's no way to credit it. Having skimmed through the link provided it would take me a chuck of time to truly understand the fine detail of the liscenceing & copyright issues, so I'll leave that to those with greater understanding.
I do personally use the maps here https://www.walklakes.co.uk/maps/ that incorporate OSM & different data sets but the "about link" in the muneu gives credit & the footer copyright link gives more detail on sources. It seems a pity that something similar can't be done with the source OSM.
To be clear - you're saying that there is a horizontal offset between a feature in DoBIH and a feature in OSM.
Arguably Crib Goch should be in OSM as a linear feature rather than a node as it's a ridge! In that sense, both DoBIH and OSM (which apparently here gets its information in this example from out-of-copyright OS maps: https://www.openstreetmap.org/node/3686236093/history ) are correct - they're both just points on the ridge.
A wholesale import from DoBIH has actually been attempted "under the radar" by at least one nefarious user in the past and reverted, partly because of licensing issues and partly because some of the information there is historical or was badly coded on import ("peaks" at the bottom of quarries and at the top of bridges were examples of those). Should the licence issues get resolved and any import happen, there would need to be conflation of data in OSM already to see which one was correct, and the data imported would need to be coded properly (most DoBIH data by volume are just spot heights in OSM terms).
A quick peak at the Crib Goch node shows it was added from old OS 7 series imagery which may suffer from rectification errors. The new Bing imagery (which is probably well orthorectified) & OSM GPS traces clearly show that the path as marked on OSM does not follow the ridge line. Improving location of both would reduce the discrepancy. Many peaks were added from old OS maps with varying quality of rectification as well as any location issues because the maps are generalised. Even with imagery highly accidented terrain is often poorly orthorectified, so locational precision is necessarily reduced. For most use cases this imprecision is acceptable (within precision of consumer grade GPS receivers), and attempts to improve it can be counter-productive (using nice, but poorly-rectified imagery for example).