Correct use of 'relation' on group of Woods

[jamwilco]

I’ll preface with saying I’m not interested whether this is something I should be spending time mapping, rather whether it’s correct (not to mean useful) to be done at all. This is something that’s bothering me about the map in my area, so I’d like to address it.

Near my village sits a large wood. At present, it’s mapped as a single wood: “Lower Fillington Wood”. By official account (OS Six Inch, 1888-1915 | 2nd Land Utilisation Svy., 1:10k ca. 1958-74), it is mapped incorrectly. “Lower Fillington Wood” is much smaller than mapped and is neighboured by the larger “Upper Fillington Wood” and nearby “Fillington Wood”.

Now, the simple solution is to split Thirds Wood and Lower Fillington Wood to create the missing named woods, fix the area boundaries, and call it a day. However, I’m interested in knowing whether all the features labelled any form of “Fillington Wood”, in this specific scenario, belong in a ‘relation’ together? In geographic terms, do they all form part of the same wood (regardless of whether they are currently used for the same purpose or not; forest vs. wood)? Or, were they given their names only for more semantic reason; basically, are they all “Fillington Wood”, or are they individual wooded entities close enough together that their supposedly-structured names are nothing more than convenient?

I have a poor understanding of Relations right now, so the ‘how’ is also a challenge. I’ve tried mapping an example but fail to get JOSM’s Validation Results to happily report that nothing’s wrong with my technique.

In essence, I’ve got a Relation with name=Jim’s Wood, type=multipolygon, natural=wood. A second multipolygon for “Upper Jim’s Wood” and the attached “Jim’s Field”. Then, “Lower Jim’s Wood” (forest) and “Jim’s Wood” (wood). No matter what combination of tags I set for either the Relation or Members, JOSM won’t pass it. Perhaps type=multipolygon is the wrong type of Relation…

Either, there is a false-positive 'warning’ to let slip when it is perfectly mapped, or I have such a fundamental misunderstanding of Relations that what I’m trying to do is simply never going to work.

Remember, I’m only concerned about this very specific wood and I’m not interested in whether this is a good use of mapping time — only whether it is geographically correct. This is a learning opportunity and I’m curious. Thank you for your considered answers!

[Duja]

@jamwilco

However, I’m interested in knowing whether all the features labelled any form of “Fillington Wood”, in this specific scenario, belong in a ‘relation’ together?

There’s no hard rule when you need to create a relation distinct map elements. Well, one such criterion would be if they are the same “thing”, i.e. have the same name – but “Fillington Wood”, “Lower Fillington Wood” and “Upper Fillington Wood” are distinct enough to be mapped just as three separate areas.

Having seen your source, all you need to do is to split the current wood polygons differently, and assign them appropriate names. Just as you recently did with the nearby Watercroft Wood, which is adjacent to Dell’s Wood, Harpershill Wood, and Withy Copse.

On second look, you will need to keep that scrub inside of the future Upper Fillington Wood (that’s what you tried to emulate on that JOSM screenshot). JOSM complains that you tagged both the outer member of “Upper Jim Wood” and the area/multipolygon itself with the same tags. See our fine documentation how it should be done – tags go to the multipolygon relation, but not to its members. And you have some suspicious multipolygon (wood), 3 members in there, that should possibly be deleted.

[Janjko]

If a wood doesn’t have holes, it doesn’t have to be a multipolygon. It can just be a way, with the same tags except type=multipolygon.

[jamwilco]

Wonderful answers, thank you both very much, I really appreciate you looking into this case! I shall read up on the documentation linked but cast out the idea that features such as these need to be part of a Relation at all.

[SomeoneElse (Andy Townsend)]

Another possibility - use boundary=forest for the named forestry area, and then map the internal patches of wood with names or without.

Here’s an example. A renderer can then show the larger forest area as lighter green, with the individual tree areas as darker green:

(link)
Those tree areas inside aren’t named in this example, but they could be.

Edit: See also here, with names inside, which looks like this.