Two-sided railway platforms

Let’s say there is a railway platform number three. On each side of it it has tracks. And on it has signs saying 3A and 3B. So if you get on on the 3A side you’re going north, and if you get on on the 3B side you’re going south. Should I split it into two platforms?

From someone who has done somewhere between a fair bit and a lot of rail mapping (including platforms and the sometimes-difficulties that can be associated with tagging them in OSM):

I would say “yes, use two platforms.” (I wouldn’t say “split,” as that implies “something is there that must be chopped up” and maybe that’s not an accurate way to state how things are and should be — though in your case it sounds like that may be true).

“Around here,” (let’s say passenger rail platforms in the Western USA — I’m in California) such things would be numbered “platforms 3 and 4” or “platforms 5 and 6” (and there would be 1 and 2 at that station, and may be even more, like 7 and 8, 9 and 10…). But if the way “you do things around you” (or “your country-or-region names / numbers rail platforms”) uses the convention of 3A and 3B, well, then, that’s two platforms (one north, one south). And I think it appropriate you use name=3A and name=3B.

I mean, even if that convention is a “one-off” (at that one, single, particular railway station), I’d still “tag what is.” (Using two platforms, tagged 3A and 3B).

1 Like

What’s your thought on railway=platform_edge???

1 Like

This is caused by the difference in terminology of platform (island) vs track (side). I personally don’t think an island platform should be split into half at a arbitrary middle. Your example clearly signposts the “platform” as ref=3, while it has 2 sides of ref= =3A and and =3B. So =platform_edge has both a strong case in reality and theoretical basis.

If you are asking me, I have very little opinion about it. I’ve read our wiki about it, I don’t find its level of detail (needed or wanted) to be particularly applicable to “our” present state of rail tagging (I’m USA-based in California), though I could see it being useful here if the level of detail so denoted in our wiki becomes something more Contributors want to map. Right now (at least in the USA), “not so much.” A very, very little bit, to which I have no objection, but “not very much at all.” And that’s fine.

I believe there are only 22 instances of this in the USA, a few in my Bay Area (California), Denver (Rocky Mountains urban area), Austin (capital of Texas) and Greater Boston (East Coast urban center).

As usual, “read wiki, apply what’s in the wiki to your mapping if it is applicable.” Lather, rinse, repeat.

BTW, rail mapping (between Europe — and VERY much in Germany — and USA, especially) can have significant tagging differences. These both appear to be rather widely documented (in wiki, as discussed among their “local” users / taggers / mappers) as well as understood to “work in their respective localities” without significantly harming things like routing or rendering of rail (for example, with ORM).

1 Like

I want to add clarification to my “solved” reply, as it appears that

is also a workable solution that could work for the original questioner. Dan, if you are someplace and your platforms are really “one named 3” and two “platform edges” where 3A and 3B are a more locally-preferred option (by you and your rail mapping community), please feel free to do it that way (too).

There is sometimes (often?) more than one way to do things in OSM. We might want to keep that to a minimum to reduce ambiguity and/or parser efforts, but that can be hard when crowdsourcing and “any tag you like” happen. We’re getting better at it, but regional difference simply do exist and appear they will continue to exist.

So, “more than one solution” to this. You could even add to that “a locally-preferred, arguably more-correct solution compared to the one stevea said” and I’d nod my head and say “yeah, I not only agree with you, I learned something today.”

This attitude is important to how our map grows (especially in quality): good map, good discussion, listening, open minds, better map!

Taiwan is influenced by Japan, which navigates by track number. “Platform” is a separate concept. So that’s part of my reason for it.

Tag:railway=platform_edge - OpenStreetMap Wiki is a wonderful idea! OK, I’ll use that.

I sure wish Tag:railway=platform - OpenStreetMap Wiki would have mentioned it.

The former indeed does mention the latter, but it wasn’t hyperlinked. So I tried to make it a hyperlink using a cellphone. But I ended up wrecking the page. Perhaps somebody could repair it for me. Thank you. And while repairing it, perhaps somebody could mention the former on the latter. Thank you.

Completed.

1 Like

Tag:railway=platform_edge - OpenStreetMap Wiki is a wonderful idea! OK, I’ll use that.

I sure wish Tag:railway=platform - OpenStreetMap Wiki would have mentioned it.

there is “local_ref” which is used for railway numbers in a station or similar (on the rails AFAIK): https://wiki.openstreetmap.org/wiki/Key%3Alocal_ref
I think it is used for situations like the one you are mapping

no idea what is the difference to “loc_ref” or why a new key was used

I sure wish somebody will mention local_ref or loc_ref on Tag:railway=platform - OpenStreetMap Wiki .

Seems like a great idea for those bus stop positions. And yes in Vespucci it is very easy to split a long thin train platform into two even thinner long platforms, with same ref 3, and different local_refs A and B, or 3A, and 3B… But should I? Was it designed for long platform edges, and not just bus stop boarding positions?

1 Like

Completed.

1 Like

Bummer, now I’ve run to run into a lot of platforms where the mapper has just used a single line, an unclosed way, to map the platform. I can’t tag both “sides” of a line.

Tag:railway=platform - OpenStreetMap Wiki should mention that creating closed way platforms is required, to save others a lot of possible work later, even if one doesn’t see the need yet.

Further to Talk:Tag:railway=platform - OpenStreetMap Wiki explanation, I can indeed find exactly 1 railway=platform + ref:left= ref:right=. Way: ‪Wertheim‬ (‪44808748‬) | OpenStreetMap

3 Likes

Wow @Kovoschiz that’s smart. I get it. It doesn’t have anything to do with the direction the trains are going in, it’s the direction the mapper used internally in the way.

Yes, I agree. I believe that “using the direction of the way” also can simplify / eliminate such confusing complications as [type=route + route=tracks] data structures to need to be built in OSM. In the USA, we have essentially conflated rail relations from “three into two:” we don’t use the (often used in Germany) convention of [route=tracks, route=railway, route=train], we have eliminated route=tracks (with careful and cautious paying-attention-to of the direction of the way tagged railway=rail) down to only [route=railway, route=train] relations.

Using “sub-tags” like ref:left and such with railway=platform is, I agree again, a bit of genius.

1 Like

OK, so far for e.g., island platform 8A/8B I am using

  • Two railway=platform_edge’s if the platform is a closed way.
  • And if it is an open way, then I am directly editing the railway=platform’s tags, adding ref:left and ref:right.

So far so good. But…

Two different methods to describe the same thing? Bad.

And what if one day someone casually helpfully makes that one dimensional platform into a two dimensional one, closing the way?

Then ref:right and ref:left become referring to the inner and outer
sides of the now closed way. Disaster.

OK, how about instead we always use railway=platform_edge?

I.e., for open ways too we now also create a single
railway=platform_edge overlaid on the platform, and add the ref:left and
ref:right tags to railway=platform_edge, never messing with
railway=platform’s tags at all? Good idea?

There has always been 2 different levels of detail by using a line or an area. Not a unknown situation. The who is changing a line into an area is responsible for converting all the *:left= & *:right etc linear attributes.
A both-sided =platform_edge would be invalid by definition.

1 Like