w1375416246 is a very flat surface in the woods (with moderately dense trees). It is easily traversable by foot or mountain bicycle. I have added w1375416243 inside it even though there isn’t really a path inside, so that routing engines allow hikers to pass through it, otherwise it would be a very very long detour.
But there should be a better way to tag this. Help!
In urban areas, I am well aware of highway=pedestrian, area=yes, but I was hoping there is a way to designate any area(closed way) as traversable/not traversable by foot/bicycle/car. Is using foot=yes or bicycle=yes suitable for this purpose?
It’s tempting to think that a highway=path on an area is suitable for this. This allows all the wide range of tags already available for paths, such as mtb:scale. But the highway=path page says this is only for ways and not areas.
Unfortunately I don’t. But my question is general; the world is full of open fields that can be traveled and that aren’t paths, and I am wondering how to map this.
The same is true of a very large number of wooded areas around the world. As you said, the world is full of freely traversable open fields. The absence of e.g. a natural=cliff, etc. in the area makes this also explicitly clear.
Which is why I think this:
isn’t perhaps such a great idea.
Of course, desire paths do exist. In this particular case, when enough people take a detour along some direction, a desire path or paths will eventually emerge in those woods as well, and it/they can be tagged with e.g. highway=path + informal=yes + a trail_visibility= estimate. And a suitable mtb:scale= value, if applicable.
To me it seems clear that in a wooded area with no cliffs, one can wander in any direction one wishes. For a path line in OSM, I would expect there to actually be a path of some sort present.
We don’t. If people are usually walking across an area along a relatively consistent route, that can be mapped as a highway=* way, but we don’t tag all random fields or forests if they are traversible.
I suppose I’m finding it a bit baffling if there’s no way to distinguish a traversable area from a non traversable one in OSM, and that routing engines simply assume only paths are traversable. Isn’t this a huge missed opportunity for more pedestrian routing options?
But there is. A wooded area or a “field” without cliffs exactly maps an openly traversable area.
I think that it is relatively sensible that routers route people primarily through actual ways. Of course, routers could offer the user a possibility to draw a direct line between two waypoints if the users wish to cut through a woodland or a desert (maybe give a warning if that line crosses a cliff, etc.).
This heuristic may work in perfect mapping scenarios, but that isn’t the case in many areas around the world.
A forest without cliffs may be traversable, but maybe the cliffs are simply unmapped, or maybe it’s not traversable for other reasons such as dense bushes. I assure you that a large portion of the places marked as natural=forest around that area aren’t traversable.
I do not have the means to map all the cliffs, and it seems traversability needs to be opt-in and not opt-out. So no wonder routing engines do not make this assumption about no-cliffs implying traversable forest. It’s inaccurate, and likely computationally complicated.
The problem with this is that traversability of forests is on gradient.
There are forests where my grandmother can walk, there are forests which I can traverse with my father, there are forests that I could traverse with my father before his injury, there are forests that I can traverse, there are forests that I cannot traverse but others have no problem.
And just rain or snow will change exact delineations.
I would map barriers like rivers, cliffs, wetlands, fences instead.
I agree, and even for fields where the terrain itself is not difficult, there may well be unmapped fences or hedges especially if they are private property. I don’t think we can rely on those barriers that don’t cross ways being mapped, as they don’t generally have much impact.
For Ireland, for example, where the majority of the land is privately owned and there is no right to roam, my default assumption for any piece of land without a specific path/track would be that it is either illegal or impractical to cross, as the exceptions are a minority. Even where a marked trail traverses a field, permission is normally restricted to that single line.
Note that even for this relatively well-defined case, routing across areas is technically not trivial and has generally not been well supported by mainstream routers. Perhaps this gives some context for your surprise about the lack of tagging of traversable fields and forests. If routing support is limited even for a clear case like a public paved square, there is not much motivation for mappers to work on routability across wider areas.
That’s nice! I think the reason one isn’t forming here is that there are so many possible routes between the trees from the track to the paved road, and people do not consistently take the same one. I also think it’s infrequently hiked.
This is also true for regular paths! Keys such as Key:sac_scale attempt to record the difficulty level. One could argue that in an ideal world all those tags paths already enjoy should be applicable to areas.
Summarizing the discussion so far:
The world is full of traversable areas, and there is currently no way to explicitly mark a traversable area except in specific cases such as highway=pedestrian, area=yes, or by adding a “pseudo-path”. There seem to be differing opinions on the latter practice.
Routing an area is technically challenging and often isn’t done even for highway=pedestrian, area=yes. Basically, only path routing is widely supported.
If people are usually walking across an area along a relatively consistent route, some people map that as highway=*
One can guess if an area is traversable by its other content (such as cliffs) but this is very unreliable in areas not comprehensively mapped.
I am getting the impression that leaving the pseudo-path is actually the best solution for now. It kind of falls into “If people are usually walking across an area along a relatively consistent route”, and removing it causes a massive detour.
There is perhaps room for future tagging improvement:
some tagging scheme for whether an area traversable: This seems complicated and it’s plausable that even then routers wouldn’t support it due to the complexity involved.
A simpler implementation of this would be hijacking sac_scale and mtb_scale, allowing their use on any area and not just ways. Their absence means traversability is unknown. Their values allow indicating explicitly whether it can be traveled or not, and how difficult that is.
Some tagging scheme for marking that a path is a desire paths. Off the top of my head:
desire=invisible: Indicates a route people are known to travel by but isn’t physically visible.
desire=yes: Indicates a partially or fully formed desire path.
What do you think? desire=invisible is in some sense a violation of Map what’s on the ground and wouldn’t rank well on verifiability, but so are country borders, and it might be a reasonable tradeoff. The practice seems present and necessary and this would only record it.
If the fild splits up two otherwise visible pathr, in my opinion you should tag it as highway=footway footway=link informal=yes and any additional access or surface tags that are applicable.
This is exactly the scenario. Two visible paths split by a field. I was not aware of Key:informal or footway=link! This seems to be the best solution so far.
Edit: The best solution in this particular case appears to be path=link, highway=path.
It looks like highway=footway, footway=link is the closest, but I would also like to indicate that this is accessible for mountain bicycles. Is there an equivalent for highway=path? e.g. path=link? That is not so common on TagInfo.
I occasionally mapped highway=path with surface=grass or surface=sand where given short specific section is very clear segment of path but without signs on the ground.
Here is an ESRI Imagery screenshot with the flat forested patch marked. Note the path on the north and the path on the south, cut abruptly by the patch.
I’ve looked around at path=link usage in the real world, and it seems it’s used just like footway=link but for paths. Here is w1445461012 as an example (highlighted red)
Links are indeed used for very short hops between two explicit roads or ways, mainly to aid routers. In your example above, the link is a few meters long, and crosses a kerb, for example. I think what you’re discussing (a forest that one can wander through) is a bit different scenario.
My problem with your approach is that to me it seems to be a solution looking for a problem.
In short: I don’t think the points above are something that a router should solve. As I intimated above, I think it’s fairly reasonable that routers guide poeple through and along actual paths and ways. Any sain person is free to take a shortcut that the router wouldn’t suggest—if they so decide.
There are two reasons why I think “traversable” areas (woods, fields, deserts, etc.) should not be tagged with explicitly traversable ‘imaginary’ ways (links or whatever).
On the one hand, routers point out explicit and existing ways that I can travel on. Obvioulsy, if I see from a glance of the map that I can take a shortcut through a wood or a field, I can take it if I want. I don’t need a router for that. I’m left wondering, what is the downside that a router won’t suggest it to me. That I arrive early? Oh, the horror! The same applies for a stupid router that suggests I traverse an open square following its edges. Only a complete moron would follow such instructions to a tee, everyone else of course crosses the square in the middle.
On the other hand, I would never want a router to suggest that I take a shortcut through a field or a forest. I want to decide that, and—again—I don’t need the router to tell me that. For one thing, a suggestion of “Walk four hundred meters through the wood on heading 210” isn’t very helpful advice. What if it’s dark and I don’t have a torch? What if I’m wearing my leather-soled oxfords and don’t want to get them damp? What if I know that there are ticks in the shrubbery and don’t want to spend my evening searching for those mofos in my groins?
I used words like ‘sain’, ‘stupid’ and ‘moron’ above. I realize that we live in a world where such qualifiers don’t always apply. There are, unfortunately, people who blindly follow a router and nearly die on a alpinist route, blindly trust Google and drive to a desert. Or paralegals who let ChapGPT to hallucinate legal discovery. I think people who do those things instead of using their mark one eyeball and a few brain cells deserve what’s coming for them. No amount or routing will help them.
