[Tagging] The endless debate about "landcover" as a top-level tag

Marc Gemis marc.gemis at gmail.com
Wed Jun 13 07:44:41 UTC 2018

Private vs. public does not matter.
Private gardens with some trees in it are not a forest.
A tree row in a field is not a forest for me.

According to Wikipedia [1] there are hundreds of different definitions
of forest. Not only that, some people are only interested in mapping
"groups of trees" so the map is nicely coloured.
For me, a forest is more than just a group of trees. I don't have a
ready to use definition so there are many grey areas left. But think
about a park but with less human shaped areas (grass, ponds, benches,
waste bins), but still maintained by humans. It includes trees but can
also include bushes, plants, flowers, grass, ponds, streams and
paths/roads. Typically the paths are not as well maintained as in
parks. Please note that is how I see parks/forests in Belgium, other
countries can have different definitions.

For me, the current data is only fine to see some colour on the map. I
wonder how people that really work with this type of landuse/landcover
data think about the OSM-data. I know about one scientific article
that even recommends OSM for landuse/landcover.(don't have the URL
So perhaps it is not so bad, and it's only my idea of a neatly
structured landuse/landcover separation that finds the data useless
for other purposes.

As to Peter, I have written my rough ideas on landuse/landcover 2
years ago or so on the Belgian mailing list. While there are others
that share the landcover vision, we have not done anything with it.
There are still many open issues:
landuse/landcover/leisure/surface/natural all overlap in some ways.
Even the building tag overlaps with the landcover tag. If you define a
different system, where do you draw the line? What do you want to map
in another way ?
And even if we agree, it would take too long to revisit all place and
redo the armchair work that created the current landuse mapping in
Belgium at this moment.

But I haven't seen any reply on how I should map

* trees in gardens or in a meadow or any other area whose primary
function is not "trees" ( (using overlapping landuse ?)
* where the name of the forest has to be placed when one uses a
multi-polygon to exclude ponds from the tree covered areas.

So Mateusz, Andy, how do you solve those problems ?

[1] https://en.wikipedia.org/wiki/Forest
On Wed, Jun 13, 2018 at 9:21 AM Mateusz Konieczny
<matkoniecz at tutanota.com> wrote:
> 13. Jun 2018 07:47 by marc.gemis at gmail.com:
> won't work, see e.g.
> https://www.openstreetmap.org/search?query=waasmunster#map=16/51.1215/4.0932&layers=N
> that's not a forest, that are a lot of private gardens with trees in it.
> I opened https://www.openstreetmap.org/note/1424854 - somebody included also areas
> without trees.
> A forest is a place where you can walk, ride, cycle. Not someones
> private backyard.
> So you want to count tree-covered areas that are not private? (including ones in private backyards).
> I think that it is not doable with current OSM data (access=private is extremely rarely
> tagged on private areas) though excluding areas tagged also as landuse=residential
> may be a good approximation.
> Our government talks often about there is so many square meter of
> forest in Belgium.
> It's not sufficient to subtract all small areas, you need to subtract
> somehow everything that is not a forest (see above)
> So do you want to exclude private forests or something else?
> _______________________________________________
> Tagging mailing list
> Tagging at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/tagging

More information about the Tagging mailing list