When hiking in Nepal and now editing OSM based on my notes, I noticed there are a lot of
mountain_pass=yes
name=Pass
They are scattered inacurately and mostly wrongly (if anything, they tend to be on peaks, not in passes). When they manage to be in a pass, it tends to be a very minor pass (often next to a much more prominent pass that is not tagged). I am not even mentioning the wrong use of the name tag.
When I investigated, it seems these are all made by Peter Van Geit | OpenStreetMap . The first changeset in the series is Changeset: 162071281 | OpenStreetMap
and the last is Changeset: 162232684 | OpenStreetMap
(not sure if there is an easy way to list them all and link to them, they are all named Nepal Mountain Passes. There even seem to be a blog about it: Mapping Nepal – UltraJourneys To me it seems the passes were determined programatically and then imported in batches. Not sure if there was much human oversight and no idea about the copyright of the source. For me, the most important consideration is that the data is wrong and confusing for a map user. I messeged Peter Van Geit a week about it but from this blog, he tends to go offline for months at a time, so not sure how long to wait for a reply.
When you overlay the introduced nodes in JOSM over something that shows countour lines like OpenCycleMap, you will see that most of these are no mountain passes.
I propose to delete the passes thus imported. If there is agreement about that, what would be the best workflow for it? It is about 30 changesets, if I got it correctly.
I just did. As I said, I already contacted the user, but from the info on his OSM page, it is clear he goes travelling for months at a time, where he might be offline for a long time (these “ultra” things, whatever exactly they are).
I’m not so sure. The source is described as “georeferenced legacy topo maps”, and from reading the blog it sounds more likely they were added manually by tracing the points.
It doesn’t necessarily make a lot of difference, except that comments about not following automated editing procedures may be out of place.
It would be interesting to know the source of the issue, e.g. do the “passes” in the topo map not match the concept of a pass in OSM, or are they genuine passes displaced for some reason (perhaps due to issues with georeferencing the originals).
It seems I was indeed wrong about the automation, sorry.
However, the source map was copyrighted in 1997. If I understand it correctly, this makes it legally incompatible with OSM: Copyright - OpenStreetMap Wiki ?
That would save us the debate about quality of the data :-).
However, even then, there are issues:
Name=pass is wrong.
mountain_pass=yes “should be the highest point of the way, so it’s a single node on that way”, however these are mapped without ways, they are single independent nodes. It is unclear if there is a way or not. I would argue that first a way should be mapped and then mountain_pass=yes should be added, not the other way around.
I think there is a misunderstanding about what a mountain_pass is
Where ever a trail marked on the topo sheets hits the ridgeline (highest point) [whether on a saddle / peak / anywhere on ridge] is a mountain pass (refer wiki osm mountain_pass definition).
Whereas the definition is
marks the highest point of a mountain road, railway or path as it passes over a crest.
So there should only be at most one mountain_pass per road. And it should only be there when it crosses (=passes over) a crest/ridge, not when it follows it!
