Grass versus grassland

Hi there. I am contributing to OSM here in Scotland and would like to ask some advice. I have added some areas of grass where the grass is cut regularly, for example in housing estates there are often small areas of grass which I would not consider to be a park. There are also other areas of grass which are not cut and which grow more wild, as well as actual grasslands, for example on the Ochil Hills. Currently on OSM ‘grass’ and ‘grassland’ render the same. I would like to be able to differentiate between them, particularly where the areas are adjacent. Should I use ‘park’ instead of ‘grass’?

It’s a good question.

People typically tag small areas of grassland (verges etc.) as “landuse=grass”. Quite a lot of generic upland has been tagged as “natural=heath” even when it’s not really heathland at all but really grass (and some people object to “landuse=grass” as it’s not really a “land use”). You can use to search for tags by name, and click the “overpass turbo” link from there to find uses of a tag in an area.

Coming at the question from the other side, there are lots of different maps using OSM data - 4 are directly linked from alone. If you want to learn about the way that they render certain sorts of data you can look at e.g. (and especially the issues list in there). If you’d like to see what the impact of rendering different sorts of grass differently is, I’d be happy to try and change so that different grassy things like were slightly different shades of green or differentiated in some other way.

Best Regards,


For smaller patches within residential areas etc, it leads to overlapping landuses, which is part of the frustration.
On the other hand, what would be a suitable match would be to use landcover = grass, which is a more accurate description (or so I feel), but that one… doesn’t render in the default style.

landuse = village_green is an often abused intermediate choice I guess, that one does render…

That was what I was doing, until I discovered natural=grassland. It also hadn’t registered in my brain that grass was actually a landuse tag. I’m unsure how landuse, natural, leisure etc are all meant to be used. In theory everywhere should have a landuse, and then the actual features be defined on top of that. It doesn’t seem to work that way on OSM though.

(apologies for the somewhat offtopic rant that follows…)

I do get a bit frustrated when people talk as though the OSM Carto style is somehow “standard” or even the most viewed one. I’d suggest that whatever Facebook/Instagram are using in their mobile apps probably takes the “most viewed” accolade.

Various styles do render landcover - I know that does, for example, because I control that one. I think some of the mobile app styles do too. I wouldn’t be put off using a particular tag just because it doesn’t get rendered (yet).

You’ve also go the option of using a lower-level tag, perhaps “landuse=grass; grass=”. Taginfo shows a few examples where people have tried to clarify things like “type of grassland”.

… one more thing

The tagging list’s archives have a fair bit of discussion about this sort of thing. You might find searching those helpful.

… if I can chip in too:

I use landuse=grass for the sort of things you mention in towns. These are definitely not parks and tagging them that way creates many more problems. They fall into a generic class usually described by councils and the horticultural trade as amenity grassland, so when I remember I usually add grass=emenity_grassland in the (vain?) hope that at some time we’ll improve the tagging. Another reason is that in some places people have used landuse=grass to map rural grassy areas.

Don’t worry too much about overlapping landuses (or no landuse at all). I experimented [some years ago](some years ago) with how landuse was mapped on OSM in urban areas and compared it with maps produced by the European Environment Agency. We didn’t do too badly at all, and I would think our mapping is much more complete now. I used (simple) algorithms to determine which landuse to select for any given area: they more-or-less approximated what the Carto-CSS style does when it renders the map.

Upland grassland, perhaps best termed rough grazing, is one of the biggest holes in our tagging scheme. As SomeoneElse says many of these areas have been tagged natural=heath. Grassland gets rendered in a nice green colour which I think will put many off using such a tag for these areas. Also upland grassland can have many wet flushes and I don’t think we have a good way of capturing that aspect either. The only interim solution I have at present is to add vegetation tags (UK Phase 1, see the wiki). Most upland grassland is unimproved acid grassland.

I’m also very sceptical about the value of landcover particularly for landcover=trees & landcover=grass. There are massive differences between a lawn, a playing field, a low-land pasture, a meadow and various upland grasslands. Not forgetting that reedbeds are areas covered by grass and in various parts of the world there are forests made of grasses (although I don’t intend to embark on super pedantic mapping). I suspect that to most contributors grass really means a limited spectrum of grasses of high agricultural or horticultural value in NW Europe (notably Perennial Rye-grass), but this is not adequate either ecologically in Britain or as a concept which is applicable world-wide. Equally cogent, is that landcover doesn’t really solve the problem just moves it somewhere else, and it’s inherently fissiparous. If you do use landcover it’s is still necessary to discriminate what kind of grass the area contains, so two tags should be needed.

If you are close enough to upland grassland to visit it regularly then you are in a better position than those of us who have been concerned about this for a while. There is nothing like being able to visit a place as one refines how to map it. One other resource, which may be overly technical unless you have an ecological background, is data from Scottish Natural Heritage. I’ve looked at some areas I know on Kintyre and it was not particularly useful: everything seemed to be mosaic of different habitats.

Thank you for these replies. Based on these I think my first thought of changing how natural=grassland is rendered would be my preferred option. Where does that get decided?

I know it’s been discussed in-depth, and I’m not trying to point fingers. I’m merely stating my perspective, as an OSM-user and contributor, as it was something I’ve struggled with myself.

The subtag approach is definately an option, but still clashes a bit with how I’d personally perceive the ‘landuse’ concept, making it counter-intuïtive from my point. I’ve come to terms with accepting not everything here is ‘as I would do it’, or ‘compatible with other sources’, I’ve even written a diary post on one such topic in particular (road categorisation - ). Doesn’t mean I can’t reference the situation and it’s multiple sides as they currently are, or at least how I perceive them.

While it’s not good practise, we cannot deny the fact there’s (still) a shitload of ‘tagging for the OSM-carto render’ still going on.

And I’ll assume there’s a ton of valid arguments and sensible discussion going on in that mailing list. I however have to admit I’ve not read it all. I’m one of the ‘late arrivals’ I guess, mapping for a bit over a year now. I’ve read up on a TON of things before I got into actively mapping, to avoid making a ‘faux-pas’, but let’s just say there’s quite the backlog to try and catch up on the historical discussions.
At some point, you have to rely on the knowledge and expertise by others, and on certain topics you’d otherwise get nearly as many opinions as there are people involved, which makes it hard to get anywhere. So for the majority of topics, I’ll just sit back and let the older, wiser and more experienced folks call the shots.

It is only in a few rare cases, where the consensus feels counter-intuitive to me. This just happens to be one of them (which is why I pitched in here), but it’s also one where retroactively trying to fix all tagging deemed ‘wrong’ by the consensus is hardly an option at all.

This is most definitely true of landuse/landcover tagging, and undoubtedly means that to solve the issues probably requires one-or-more:

  • subtagging or orthogonal tagging (former is not always as easy as one thinks, for instance to me a big split in bogs is whether they are raised or blanket, but the subtag bog contains basically information about how the bog is used).
  • additional values in landuse/natural: one which has light usage is landuse=forestry which allows coverage of new plantings, recently felled areas etc.
  • additional rendering approaches (SomeoneElse has done lots for UK specific stuff already). The Phase 1 stuff I mentioned earlier benefits from QGIS style files to render data according to the scheme.
  • a series of little steps breaking down each tag: for instance is grassland: rough, unimproved, semi-improved, improved or amenity.
  • good test areas which are familiar to several mappers (in the UK well-known locations in National Parks probably fit)
  • ideally, external data to help evaluate edge cases

Having looked quite a bit at how different grass related tags are used i would like to add one important suggestion:

Don’t try to instill a more specific meaning into one of the widely used grass tags (landuse=grass, landuse=meadow and natural=grassland). Even if you make such a difference consistently in your area of mapping there are other mappers who do not and there are even other mappers who make a completely different distinction. And given the volume of use of these tags this is unlikely to change. Unfortunately none of these tags was ever established with having a specific, well defined and practically verifiable meaning other than there is grass growing here.

As others have mentioned one possible approach is creating supplemental tags to more precisely specify what kind of grass area you have. You should try to make such supplemental tags as precise as possible in meaning so mappers can determine easily where they apply and where not. The most widespread tag in that regard so far is probably meadow=agricultural - which is used to indicate grass areas that are used agriculturally either for cutting hay or as pasture. Most use of this tag is however from an import in Czechia and use elsewhere is very patchy.