Site relation for All Trails in a National Forest?

Is it good practice to have a site relation for all trails in a given US National Forest, e.g.

I am not sure what purpose such a relation would serve, but I don’t want to delete it without consulting others.

2 Likes

It looks not very useful, though it would make sense to have the individual trails as a route relation.

1 Like

Relations are not categories - OpenStreetMap Wiki may be relevant.

People sometimes create relations for easier access - they want to have one relation that groups all the cycleways of one city, so they can easily download the cycle network. That however is a misuse of the relation data type.

In the case at hand, the forest trails can probably be easily identified by the fact that they are in a specific area, and a relation is not needed. Perhaps (if there are various types of trails in the area) some of them could receive an “operator” tag or something that makes it clear they are part of some network. A relation would only be considered if the same stretch of trail can somehow be part of the “Arapaho Roosevelt Forest Trail” and also another “Forest Trail”.

5 Likes

Given how few highway=track members there are in that relation, I’d be surprised if it is anything close to being complete.

There was a discussion on OSMUS Slack a while ago about whether all USFS roads and trails could be uniquely associated with a specific National Forest, and I think the thread ended with some agreement that that there were edge cases that are not easily verified and that the concept was not necessarily needed.

Some of the edge cases are roads or trails that USFS claims to maintain but that are outside of the nominal forest boundaries. There are also cases where roads or trails cross from one National Forest to another but are not fully connected to the other forest’s transport network, and so they might be in Forest A but maintained by Forest B.

In general, I agree that this type of relation is not likely to be useful, but a discussion with the mapper(s) responsible for it might be helpful. It might be interesting to know what the intent was and if it’s possible to help those mappers satisfy it another way.

1 Like

I have seen a number of cases where all the trails in an area are in a relation. My impression is that it is because in the past trails in a relation showed up on Waymarked Trails (WMT). However it now looks like WMT is only showing the valid long distance trails so their algorithm for selecting trails to display has apparently changed.

It looks like most of the trails in the Arapaho Roosevelt forest area linked were added a couple of years ago. Back when I was noticing that WMT was displaying trails in that type of relation. It could be that the newer editors are simply following the practice they see as existing.

1 Like

This is a loose categorization. In some places, a “local bunch of bridleway paths” are tagged network=lhn though no “formal” local networking scheme (numbering protocol, for example) is known.

I could go on, but there is a lot which remains unspoken and I am not sure many parties included here are aware of all the (tagging, rendering, networking,…) issues.

Let’s be really, really “wide-area aware” here. @n76 (and I, to be candid) go way, way back. And we have long, wide, deep perspectives on things.

WMT is good, and faithfully represents tagging in OSM. It’s pretty awesome (thank you, Sarah).

Let’s “go slow” and talk to and listen to each other here.

I am not proposing anything rash here (or anything at all yet), but it is something that caught my eye that didn’t fit with the principle that “relations are not categories.” I do like the idea suggested by @woodpeck of using the operator tag to identify those trails managed by a particular US National Forest.

Yes, but we shouldn’t be building relations just so trails show up in it in a certain manner in a given website, that is akin to tagging for the renderer. In any event, as suggested by @n76, this doesn’t seem to be necessary now.

I will try to reach out to the mapper that originally created the relation.

3 Likes

I did not mean to suggest that route relations for hiking trails weren’t necessary. Only that in the past the WMT website showed any trail that was in a route relation even if that relation was strictly a local collection of trails in a park or similar area.

My observation a couple of years back was that there were mappers who apparently wanted their local trails to show up on WMT (and/or some other apps or renderers that I am not familiar with) and so would misuse a route relation as a site collection relation, basically “tagging for the renderer”.

It appears that for my part of the US, WMT is now only showing long distance hiking routes. So that form of “tagging for the renderer” will no longer render. At least not at WMT.

But trail route relations are still very much needed to deal with long distance routes like the Pacific Crest Trail.

If you are interested in getting into the weeds, I wrote a blog post about how I ended up handling trail relations when rendering my maps.

Good points here. @tekim please delete the relation when you are able. I’ve commented on the original changeset with a note about how this sort of relation is not supported and listed some alternative methods.

1 Like