Greenery in sidlisko areas in Bratislava

Posted by ScarlettB on 1 June 2026 in English. Last updated on 2 June 2026.

This is a summary of the problem and a proposal regarding inconsistent treatment of greenery in sidlisko areas in Bratislava, Slovakia. Some arguments stem from my own experience with mapping these, some echo those I saw in various changeset comments. The following text is likely not exhaustive and I invite anyone concerned with the matter or area in question to join in discussion.

The subject

Greenery areas around blocks of flats on Bratislava’s sidlisko areas are specific. They’re typically grass, but often also with trees, benches, bins and play structures within them, all of it somewhat regularly maintained. I’ll call these areas panelak greenery for the remainder of this text.

The motivation

I began micromapping Karlova Ves and simply found that when adding detail up to splitting landuse on roads, using leisure=park becomes problematic. While leaving it mapped as huge blobs of leisure=park is an option, it’s a very rough approximation which doesn’t describe these areas in detail, which then defies the point of micromapping.

 The problem

The subject as described above implies usage identical to that of a park, and indeed, leads one to consider that leisure=park is technically the right tag for such areas. It is also apparent that these areas are not parks in the traditional sense of being compact areas dedicated to leisure and intended for the wide public. Since this is a bit of an instinctive classification, I’ll try to break it down into particularities. What differentiates these areas from parks follows:

  • Parks imply use by wide public. While panelak greenery areas aren’t normally gated and access to them is not restricted, it is rare that people from further away than a few houses visit these areas for leisure. This is unlike traditional parks. The extreme example would be Sad Janka Krala, which gets visited by people from all over the city, but even smaller parks see visitors from a wider radius. When panelak greenery is utilised for leisure, it’s usually by the closest panelaks’ inhabitants. Others usually just pass through, and the inhabitants, in fact, also mostly use these areas to get to their houses rather than for leisure. A good example is the neat areas between the single panelaks on Lackova. Locals also often decorate and improve these areas on their own, which may deter unrelated people to use them.
  • Some of panelak greenery areas are too small to be considered a park in any sense. When micromapping, this creates a mess of parks and not-parks (grass, shrubbery) on the map, while in reality these areas do form compact wholes with the rest of the greenery. For example these triangles near Kuklovska.
  • Some of panelak greenery is in close proximity with major roads and so even if it is park-like, it doesn’t provide any leisure and is not used in such a way. For example this area near Tilgnerova.

These are mostly problems which stem from tagging these areas as parks. Other problems may arise when tagging these areas as landuse=grass, which is the other closest match:

  • Grass implies a simple grass surface. A park is expected to contain trees, shrubs and some amenities. Grass is not, so if one does not map trees separately within grass areas, they’re missing from the map rather than being present implicitly.
  • If one does decide to map trees on these grass areas, a problem arises when there are multiple trees growing in disorganised fashion, making it impossible to place them as single nodes without an aerial map shot in winter, which still isn’t ideal, or GNSS with RTK correction, which is scarce among mappers. Tagging these as landuse=forest is not correct, since a forest retains a somewhat natural floor. Panelak greenery areas don’t, the trees are surrounded by mowed grass and anything that drops from the trees, be it branches or leaves, is also collected by the municipal services.
  • The wiki, while it is descriptive and certainly not a holy grail, states grass use beyond that associated with roads is controversial. Though I’m not convinced of this, I’m including it because the wiki is an important resource.

The proposal

My proposal is simple.

  • Since tagging these areas as parks is problematic, I’d tag these areas primarily as landuse=grass, because that’s a primary characteristic common for pretty much all of them. Even when not micromapping, using leisure=park should be avoided to keep the distinction between actual parks and panelak greenery.
  • An attempt to map trees should be made, as these are a major visual and functional element and it’s desirable to know of their (lack of) presence for both orientation and planning, when using the map. Individual trees and tree rows should be used. I’d personally speak for using natural=tree_group for groups of trees, even though the tag only has about 8000 uses. It is, however, documented on the wiki and usage is rising. This tag perfectly solves the problem with individual trees being too difficult to place.
  • Smaller details like shrubbery, flowerbeds, benches and bins can be left to the micromappers. While implied in parks, they’re likewise often present on roadside greenery, so little information is lost when not including them by implication.

I’d like to hear your opinion on this, but I’m also prepared to simply harmonise landuse in Karlova Ves in this fashion if no protest is made in a longer period of time.

Location: Rovnice, Bratislava, District of Bratislava IV, Bratislava, Region of Bratislava, 841 04, Slovakia

Discussion

Comment from ticho2 on 1 June 2026 at 10:42

I’m all for replacing most of leisure=park with landuse=grass. It’s been a thorn in my eye for a while now, mostly while mapping around Karlova Ves, exactly for reasons you state - they’re not really parks.

Using natural=tree_group seems like a good idea, I was not aware of this tag.

Comment from ScarlettB on 1 June 2026 at 12:07

Glad to see I’m not alone.

Yeah, tree group it’s one of the less used ones. Carto doesn’t render it, but nor does it show natural=shrubbery (despite multiple proposals and almost 100k uses), but fortunately, we don’t tag for the renderer. Both tree group and shrubbery are important building blocks in micromapping urban green areas.

One thing to be solved if we do start using tree group is, do we just plop it onto the grass, or do we make multipolygons out of it so that they don’t overlap? I think leaving them overlapping better represents reality, since there’s grass usually under the trees themselves. Just like we don’t make holes around tree nodes and tree rows.

Comment from Filip009 on 1 June 2026 at 19:40

I recommend to ask most active mapper in Bratislava, who will likely have something to say about this topic, so you won’t get cooked in comments. osm.org/user/aceman444

Comment from ScarlettB on 2 June 2026 at 06:02

I wonder if we can tag people here? @aceman444 I summon thee

Comment from ScarlettB on 2 June 2026 at 06:03

Nope, I’ll try in the note.

Comment from Filip009 on 2 June 2026 at 06:15

Share this topic in forum https://groups.google.com/g/osm_sk, or ask him in direct message. But it would be nice to at least inform community about mass edit.

Comment from ScarlettB on 2 June 2026 at 06:22

I’ve tried to get into the mailing group before, but it doesn’t seem to play nice without a google account. I’ll just try a direct message here.

I wasn’t planning on doing this beyond Karlova Ves at the moment, so it wouldn’t be that much of a mass edit, but yeah, if we can get the local community to agree on this, perhaps it could be done en masse for consistency, too.

Comment from Filip009 on 2 June 2026 at 07:44

I think you can write e-mail here osm_sk@googlegroups.com, and then it needs to be approved by an administrator.

Comment from Marcos Dione on 2 June 2026 at 08:45

In Argentina we have a similar concept, “la plaza”, and we map them as parks. They’re usually whole blocks, with more or less green, and depending on many factors, can only be used by locals or by a wider audience. Granted, in our case they’re not at densely located as your panelak greeneries; there could be many hundred meters between them; see the green rectangles in f.i. osm.org/relation/5167559#map=16/-31.42253/-64.21605

I don’t think how many people use them or where they come from should make a difference; the fact is that I’m passing by and decide to use it, I’m free to do it.

But then, I’m not a local, just an opinion from an are that has similar amenities.

Comment from ScarlettB on 2 June 2026 at 08:57

Thanks for the input. Even if we ignore how people use them, I think the most important thing is, they’re not meant to be parks. You’re free to use the spaces, but they don’t always conform to the idea of a park. Looking at the examples on your link, it seems yours are much more park-y.

Here they’re really areas that fill the spaces between blocks and within blocks, which contain some park amenities to make them pleasant and lend them the function of a park, but not always - and that’s what either creates massive inconsistencies, or really stretches the definition of park in some places.

It’s easier to consider them case by case if they’re as disjointed and scarce as in your example and so I’d totally go with park if they were like that.

Comment from Tomas_J on 8 June 2026 at 04:47

Hi, I tag these “panelak greenery” spaces as landuse=grass and add all the amenities (benches, playgrounds, fitness stations etc.) if there are any. If there is a wider area of trees or shrubs, I make a multipolygon into the landuse=grass with respective other landuse (e.g. landuse=forest). The suggested tree_group area sounds also ok although the wiki is quite vague in it’s definition and I assume that there are only a few/no renderers which recognize it.. If there are some solitary big trees, I add them as nodes. I agree with not using leisure=park as these places are more liminal than leisure ones.

I suggest to link or copy your post also to the Slovak osm Google group (https://groups.google.com/g/osm_sk) to attract wider local audience.

Comment from SeverinGeo on 14 June 2026 at 23:10

Why not leisure=garden, with either garden:type=community or garden:type=private?

Comment from ScarlettB on 15 June 2026 at 08:24

@Tomas_J Yea, nothing renders it, but it’s the same as with shrubbery. As an area equivalent of tree (node) and tree row (way), there’s no other tag which fulfills its purpose, so I’d like to play the long game here and slowly work on increasing its usage, so that it may get noticed by renderer developers in the future. The wiki doesn’t define it precisely, but usage is what shapes young and vaguely defined tags the best.

I considered sending an email to the group, but so far everyone I spoke to seems to be in favour of not mapping these as parks, so I’m inclined to just go ahead with fixing it. It might be worth discussing for Slovakia as a whole, but my intention was really to allow me to progress with mapping in Karlova Ves. I’d like to go ahead with that, as my patience for mapping is rather episodic.

I might broach the topic again with the group later, when I’m somewhat done with my project, but if anyone feels like leading this discussion on a country-wide level sooner, feel free to!

@SeverinGeo They’re not at all gardens, they’re really park-like areas, but sometimes small community gardens are embedded in them and we map those separately.

Comment from ticho2 on 16 June 2026 at 14:46

Once you decide to go ahead, let me know, I’ll help, there are several “parks” near me that I’ve been itching to change to grass areas. :)

Comment from ScarlettB on 18 June 2026 at 12:49

@ticho2 I wrote to @aceman444, he said he needed a couple of days but would check it out, but it’s been two weeks now and some of his changeset comments I saw earlier suggest he wouldn’t be completely against this, so I suppose we can go ahead with it, and if the need arises, finetune the details later.

I’ll start at the Karlova Ves tram terminus and work my way along the track. I’ll leave Dlhe Diely for later, or to you, since I don’t go there much and I’m currently sick anyway, so I’ll stick to areas I know well enough without having to go out.

As for the trees, I’ll still try to primarily map them as single trees or tree rows (Mapbox was shot in winter or late autumn here, very helpful), but I won’t shy away from tree_group when I can’t place the trees precisely enough.

Comment from aceman444 on 21 June 2026 at 20:14

Hi. Thanks for bringing this up. I support the idea. It is a complicated problem. These same green areas appear in Bratislava-Petrzalka and need the same solution.

First, mapping these areas as leisure=park seems technically correct according to the wiki, they have managed grass, managed trees, footways, benches, playgrounds. They are used for recreation/leisure activities.

So if they are somehow different, from other parks in the city, it may be only in that the city authorities and people do not call them parks. Also it is a good hint that they are only used by the local residents of neighbouring apartment houses. That could be a distinction we could settle on.

We can also tag all these natural areas (grass, treegroups, shrubs) in areas that are encompased with a leisure=park tag, for areas, that you consider real parks.

I was using natural=wood for groups of trees, but natural=tree_groups seems to be good idea. The wiki implies we should not tag all random individual trees as natural=tree (even though some people do it, see e.g. Slnecnice in Petrzalka), only the important or major or notable ones. So tree_group would solve this too.

We can make the distinction: natural=tree_group - a few trees where there is underlying grassbed (managed). natural=wood - trees where the soil under the trees is not managed (there are many forest or patches of forest in Karlova Ves or Petrzalka that would meet this too). landuse=forest is the same, but it also sets the landuse (while between the houses that we talk here the landuse is still ‘residential’, it just has some trees on it). Then, tree_group could overlap grass areas and not be cut out of it (via multipolygons and such).

As for why landuse=grass may be controversial on the wiki, it may be because it indicates the land use. That part of land is destined to be only covered by grass (e.g. because it delineates a major road) and will never be covered by houses. On the other hand, the ‘parks’ we talk here are in the middle of landuse=residential (which is correct, anytime a new house can be built there), there just happens to be some grass grown right now. So landuse=grass stacked on landuse=residential is problematic, if we consider data purity, that landuses (or any other same main keys), should not overlap. But there is no better tag for now. It should be natural=grass, or landcover=grass, but it seems those aren’t very supported.

In some towns people use landuse=village_green for any random green area similar to what we talk here, and that is wrong too. I also assume you don’t want to use osm.wiki/Tag:landuse%3Dgreenery as that is too generic, when we can clearly distinguish grass, trees, shrubs,flowerbeds from imagery and map them accordingly.

Comment from ScarlettB on 21 June 2026 at 20:47

Lovely, I think we’re in full agreement here, except I went and mapped quite a few individual trees on some of these grass areas, oops. It wouldn’t cross my mind to try and map individual trees in forests, but here it’s often doable, even with groups of, say, 15 trees, if they’re spread out. How would you choose when to map them as individual trees or groups? So far I went mostly by density, mapping dense clusters as groups an clearly distinguishable trees as nodes.

And yes, I’d prefer not using greenery for precisely the reason you mention - it defies the point of micromapping.

Comment from aceman444 on 21 June 2026 at 20:54

It seems people now commontly enough map any random tree so that is noted on the wiki too osm.wiki/Tag:natural%3Dtree#Evolution_of_use :)

Yes, when the individual trees cannot be distinguished from imagery then I would use tree_group.

Comment from aceman444 on 22 June 2026 at 18:50

Before we start micromapping the patches of various types of plants in these greenery areas, we should first make sure to be clear about some of the more problematic plant types. So we have:

landuse=grass - managed/mown grass: osm.wiki/File:Traffic_Island,Coulby_Newham-geograph.org.uk-_91699.jpg

natural=tree_group - a few trees that cannot be distingushed individually, but that may have managed grass/surface beneath them: https://pixabay.com/photos/trees-tree-group-meadow-nature-795138/

natural=wood/landuse=forest - big group of trees/forest with unmanaged surface on the ground: https://preview.redd.it/a-group-of-trees-called-a-forest-v0-5npxf5nvqt941.jpg?auto=webp&s=58778b43ecaea91f3710d7311a0f368ef62bb753

Now, what exactly can we consider flowerbed, shrubbery, grassland, heath? I always assumed natural=grassland is some large wild grass like savanna in Africa, or puszta in Hungary. But people use it for random patches of widl grass at road side in Slovakia. Similarly, natural=heath is said to be some small woody scrubland, but that is sometimes hard to distinguish even if you are at the place (see e.g. image 6 below).

Which tag are these areas? 1. https://www.bratislavak.sk/assets/images/ba/articles/img-1/1724735929-foto-bratislava-sa-pochvalila-zelenymi-ostrovcekmi-treba-sa-o-ne-aj-starat-a-kosit-ich-odkazuju-vallovi-obyvatelia.jpg 2. https://cloudia.banoviny.sk/rx640/4d1d1506-1ca6-44b3-b4d4-ffa5c0054ce5 3. https://www.yimba.sk/upload/Verejn%C3%BD%20priestor/Bratislava%20-%20kl%C3%ADma/priklad_vegetacneho_ostrovceka_na_dostojevskeho_rade._foto_velcek.jpg 4. https://cloudia.banoviny.sk/rx640/81a7009f-f2a4-46fc-92ab-fb9270ec36c5 5. https://www.zahrada.sk/wp-content/uploads/2017/06/00_ThinkstockPhotos-167494748.jpg 6. https://www.zoznamrealit.sk/fotky/800x600x2/202/1008341_c0aad2bd894bcdb4a97ac58e1cc1da4f.jpg

Comment from ticho2 on 22 June 2026 at 19:47

The closest for 1 to 5 is probably landuse=flowerbed, although I struggle how to tag similar decorative areas which have bushes instead of flowers.

5 seems like a perfect fit for natural=meadow, while 6 is something betweeen natural=grassland and natural=heath (it’s still just grass-type plants, although overgrown, but no bushes)

Comment from ticho2 on 22 June 2026 at 19:48

I meant “closest for 1 to 4” at the beginning of my previous post, sorry.

Comment from aceman444 on 22 June 2026 at 20:20

  1. landuse=meadow implies the area is used for grass production or grazing for animals. So it is closer to landuse=farmland, just that the product is grass/hay. We don’t usually have this in the residential blocks of Bratislava, but could have this on the outskirts of the city. There may be some special cases with meadow=wildflower where indeed there are areas in the city the administration specifically grows wild grasses to host insects. But yes, the image can be outside of city so it easily can be meadow.

Yes, the problem with 4. appears when those are no clear flowers, but some very small shrubs (at least they look like that), maybe 20-30 cm tall. We can probably agree https://www.familyhandyman.com/wp-content/uploads/2021/07/shrubs-GettyImages-1148591849.jpg is natural=shrubbery. But if it is very small, I am not sure. What about https://www.grafgrowers.com/wp-content/uploads/2025/03/Smith-Landscape-5-1024x678.jpg ?

Comment from ticho2 on 22 June 2026 at 21:47

I kind of feel like we’re getting off-topic. These are all niche cases, at least when it comes to Karlova Ves, or Petrzalka, and can be dealt with individually.

And frankly, some of them probably deserve a discussion in the international mapper community, not here.

Comment from aceman444 on 22 June 2026 at 22:21

Yes, these are less common in Karlova Ves and Petrzalka, but we may cover the city center too where these types of greenery are cropping up. Also in new apartment projects, like Slnecnice, Petrzalka.

Comment from ScarlettB on 22 June 2026 at 23:05

1 through 4 would be flowerbed, most of these patches do bloom and beautifully so, it’s just that most of the year they look like funky grasses. I’d use grassland for both 5 and 6. Meadow implies grazing or hay production, maybe a little counterintuitively to the word’s colloquial usage, both in English and Slovak, I suppose. So I guess that’s out of question except for areas positively known to be grazed or mown for production, or for special cases like the insects you mentioned, or for protecting rare species, like they do for orchids in Biele Karpaty.

I do know of several meadows in the city, they’re on Devinska Kobyla and they’re grazed by municipal goats kept by Devinska Nova Ves, lovely project.

Heath I would avoid, it’s very specific, it’s a narrow group of species which create the typical landscape, vresovisko, I’m told it’s called. Not uncommon in the UK, but I don’t think there’s any heath in Bratislava.

The last two images linked are, I’d say, shrubbery. They’re small, but artificially shaped, which is typical for human-managed shrubbery, and there’s shrubbery:density=* to help distinguish the big, dense impassable ones from smaller, decorative ones like these. In case there’s an area with an equal amount of flowers and trimmed shrubs, I’d say shrubbery should take priority, because it’s more visible and it acts as a barrier. Not that people should trample on flowers, of course. But if it’s mostly flowers with a few shrubs inbetween, we can map the shrubs individually within a flowerbed - unlike flowers within shrubbery, that would be uh… nanomapping?

While we’re at it, another green conundrum I came across are the green tram tracks in Karlova Ves. @Spr3 set them as landcover=sedum for the lack of a better tag. They’re pretty enough when in bloom to make me consider flowerbed, but that simply isn’t the landuse, the landuse is railway amd the flowers are the landcover or surface. I couldn’t come up with anything better than landcover=sedum, so I’m open to any suggestions if you have them. Maybe it’s good enough, but it would be the only use in the world. I’ve checked a couple of cities with sedum covered tram tracks. They either don’t have these areas mapped, and those which do have them tagged as landuse=grass, which is incorrect, both because the land is used for tram tracks, and because sedum is not a grass.

Comment from Spr3 on 10 July 2026 at 01:20

I guess I’m a bit late to the party but as someone who almost exclusively maps Karlova Ves I thought I’d weigh in. Firstly I’m really happy this thread exists and that someone is talking about this as all these are are things I was struggling with when I started mapping Karlovka. (I might be giving myself too much credit here but I’d say roughly half of the park spam is from me :p). In short my reasoning for tagging the greenery as parks was threefold:

  1. It was done in some other places such as Ruzinov so I followed (I know this defense did’t work at Nuremberg trials)
  2. A lot of the greenery and “vnutroblok” space was intended as microparks from the get-go so they actually are “parks”, even though it is not the standard “western” definition of a park
  3. landuse=grass doesn’t apply without mentioning the trees, landuse=forest doesnt fit at all (I remember quite a lot of greenery having this tag) and micromapping would be tedious.

So I kinda accepted the park spam as a tradeoff, with the intention of micromapping it some time later. Where applicable (no trees) I used landuse=grass and that was that.

So overall I’m quite happy someone is finally up to the task of mapping the trees more accuratelly and I’m giving my full support.

Maybe one suggestion would be to still tag the whole area in-between the square types of blocks as parks as this was something intentionally meant by the architects and the municipality continues to refer to them as such till this day. [1] [2]

As with the sedum tram tracks, I went through the painful search for good tag a year or two back and came up empty handed as well (this thread has been so validating for me as everything mentioned is something I’ve scoured the wiki for ages to find something fitting). Anyhow I might make a wiki page for sedum and see if it catches on, as it seems there is a hole in the tags for this kind of thing.

TLDR: the feeble american minds that created OSM could not have comprehended not living in asphalt desert or suberb

Comment from ScarlettB on 10 July 2026 at 07:20

First off, thanks for the work you’ve put into KV in the past. Thanks to that I mostly only have to touch up the shape and alignment (bing really was wild sometimes) and add detail/harmonise the tags.

I’m not even against having the vnutroblok areas set as parks per se, It have considered it when writing up the diary entry, but then I struggled to find a meaningful difference between these, and open areas around panelak clusters, like on Kuklovska (48.162232, 17.045639). And then, it’s only a short way to argue that, if Kuklovska kas a park, then why not every other street with a piece of green? So that’s why I went for the “nuclear” option in the end. But I think with the trees and shrubs mapped, it’s a good tradeoff.

If we could find a consistent way to discern park from not park, I’m all for dragging some extra park polygons over them, I’ve just failed to do that myself. I’m not going to stop micromapping until KV is all pretty and detailed, so whether they’ll be contained in a leisure=park area or not, the patches of greenery will be on the map and it’s just a matter of settling on a definition of park (which apparently is the hard part T_T).

Concerning sedum, documenting how we use it would be great, sometimes there’s no other choice but to invent a new tag or value when nothing else really fits. If you end up doing it, feel free to link to this thread in the wiki discussion page or wherever it’s fitting, it might attract someone with the same problem and maybe get us some validation or other ideas.

(also, I thought OSM was British, but yeah, neither has had that much urban planning of the Eastern European flavour, unlike say France)

Comment from imagico on 30 July 2026 at 07:45

I have been asked to comment here so:

I will leave out the matter of what a leisure=park is (as well as the conceptually similar tags leisure/landuse=recreation_ground and landuse=village_green) and focus on the physical geography mapping.

The most well established way to map this kind of situation is to map the individual trees and to map whatever is under the trees with various established tags for that. This is least confusing for either mappers or data users, least disputed among mappers and provides the most detailed information while often being not more work to map well than an attempt at polygon based mapping. The trees can be detailed with leaf_type, leaf_cycle, species/taxon, diameter_crown, height etc. Lack of positional precision should not be too much of a concern because that is no different than for other features.

The second fairly well established mapping is using polygons tagged natural=wood/landuse=forest. While for either of the two tags there is a small but loud minority among mappers that the tag should not be used in this kind of situation there is no consensus among mappers that either of the two tags is wrong. And both are fairly well established for this kind of situation. Like here:

osm.org/#map=18/55.695556/12.563985

osm.org/#map=18/48.150193/11.588878

osm.org/#map=18/48.000914/7.805805

The concern is valid that this method does not provide a well established way to map what is underneath the trees. Mapping secondary vegetation layers (or non-vegetation ground) underneath taller vegetation has been an occasional subject of discussion several times but no consensus was reached. Overlapping polygons have been suggested, but the semantics of that are not clear.

There is nothing wrong with considering inventing a new tag either, but you need to consider that for polygon mapping of trees collectively we have the two elephants in the room (natural=wood/landuse=forest) and it would be extremely hard to carve out anything semantically from their turf (under the smallest common denominator meaning here trees grow). You would most likely not be successful at all or end up with xkcd.com/927. A secondary tag to apply to both natural=wood and landuse=forest would have a better chance - but you would have to think about what verifiable fact this is actually supposed to document.

Comment from ScarlettB on 30 July 2026 at 08:59

Thanks for your insight. I think the verifiable fact I’m trying to document, or rather differentiate, is pretty clear. As you said, neither natural=wood nor landuse=forest carries any information on what’s under the trees. I find this lacking, because the difference between a couple of trees in urban greenery with mown grass underneath and a forest, managed or not, is strikingly obvious.

The reason I’m not fond of using those two elephants (mammoths, really) is that they trample all over the neat grass. It could be argued that landuse=forest already implies management of the forest floor in a way fitting to the way the trees are managed and that a forest in some context is to be understood as trees on grass. But that requires a lot of imagination on the end user’s part and is necessarily ambiguous. A secondary tag could be applied, but I can’t not see that as a bit of an attempt to shoehorn information into an existing namespace when we already have a more fitting framework with a piece missing, a piece that’s the perfect shape of what I’m trying to document.

We have a node tree, which implies nothing about the landuse beneath. We have a linear tree (row), which again implies nothing about landuse. It only makes sense to have a spatial variant too. Just like shrubbery, though I know you don’t like hearing about that one. We already know how to use this, it’s the same as using natural=tree and natural=tree_row. The landuse is grass. We do use landuseless tree elements on it. Trees within a forest define the forest. Urban trees on trimmed grass don’t. They’re elements that shape the area, yes, but they’re not what defines the landuse, single trees which we don’t map in forests at all unless they’re notable in some other way. I see no reason why we shouldn’t have a separate tag for this.

A secondary tag to forests would just make the structure convoluted, the result less clear and the mapping more difficult. It would create many unnecessary multipolygons, to describe what can be done easily with overlapping ones, in a way no different than drawing a sandbox within a playground, or a flowerbed within a park, or a single parking space within a car park.

That xkcd is relevant, I admit, but I’m afraid I’m not deterred by it. I think we need a new standard for this because it’s something that’s not covered by any of the ones present, not because I don’t like how they look.

Since the time of writing the diary entry, the usage of natural=tree_group has more than doubled. Taginfo shows a recent very steep rise, which looks like mass edits. I’d love to know the context of that.

Comment from imagico on 30 July 2026 at 10:40

Well - i have given my advice based on what i know about mapping practice in OSM. I don’t really want to get into a discussion on what is the best direction to change tagging to because - frankly - that would IMO require, as a prerequisite, a clear picture of the status quo of the current tagging practice shared by the participants of the discussion - which, unfortunately, is almost never possible to achieve because too many project their ideas how tags should be used from the start and this way fail to develop a sober picture on how tags are used at the beginning.

Since the time of writing the diary entry, the usage of natural=tree_group has more than doubled. Taginfo shows a recent very steep rise, which looks like mass edits. I’d love to know the context of that.

As i mentioned over on the OSM-Carto issue tracker this is from organized mapping activities in Sierra Leone - 11.6k of 17.9k uses can be found there. The meaning the tag is used there with is the same as the consensus meaning of natural=wood/landuse=forest - that there are trees growing there. There does not appear to be any other shared characteristics of the areas mapped there with that tag - like common characteristics of the ground underneath the trees.

Comment from ScarlettB on 30 July 2026 at 11:07

Understandable, thanks for the input. To be honest, it was never my intention from the start, to start a global discussion. My goal was local consensus. As you say, the status quo must be known, described and that description agreed upon by the community if common ground is to be found, and it was my intention to do this locally. If any consensus is reached on this level, it might be worth talking about with the wider community, with an example in hand, but that’s far beyond my scope, or indeed knowledge at this point.

The Sierra Leone matter is a disappointment, one I think we’ve all grown too accustomed to when it comes to some organised mapping parties, sadly.

Leave a comment

Log in to leave a comment