[Tagging] Deprecation of landuse=forest (was: Feature Proposal - RFC - boundary=forest(_compartment) relations)

Martin Koppenhoefer dieterdreist at gmail.com
Wed Apr 14 22:34:00 UTC 2021

Am Mi., 14. Apr. 2021 um 18:20 Uhr schrieb Bert -Araali- Van Opstal <
> bert.araali.afritastic at gmail.com>:Very true, happy you mention and
> confirm it. However, that is not how it is used by many mappers. We see
> whole villages and cities mapped as landuse=residential, not a big deal
> since you can easily define inner areas with different landuse, perfectly
> viable and correct although complex due to growing number of inner areas
> when we go into more detail.

while I personally don't do it and believe it is not a helpful approach in
general if you want progress in landuse mapping, it can be seen as rough
and approximate way of landuse mapping that can be later refined. Adding
more complexity by adding inners is one possibility, but I have seen that
it leads to problems because as people add more detail it becomes ever more
complex, and ultimately you end up with the constructs either broken, or at
least many less experienced mappers are shying away because of the

My suggestion is to do it the other way round: start puzzling many small,
self contained landuse pieces together, leave the roads out (stop a
property lines), and you will receive an easy to maintain, improve and
refine, stable (geometry-error wise) structure with low complexity. If
drawing the single blocks seems too much work, at least start by leaving
arterial roads, railroads and waterways out.

It becomes different however when the landuse tag is used on a macro scale
> to map areas which refer to the "management" of the area, not the actual
> landuse. This is the case for f.i. refugee camps (very large in many cases
> with a variety of landuses inside them), mapped and tagged as
> landuse=residential,

while they could be seen as residential landuse, there should be a feature
tag for a refugee camp (I think there is already). Just landuse=residential
clearly isn't sufficient.

> and for our case here landuse=forest. Creating inner areas with different
> landuse explicitly excludes them from the management philosophy. The only
> viable solution for that seems to me a boundary.
> Correct me if I am wrong, but also in countries where the landuse=forest
> is heavily used, is it mapped on areas that are no longer covered with
> trees, the trees are cut or as in the proposal areas with ponds, some
> agriculural use, some small villages ?

typically landuse=forest is added to (any) area of trees, but not when they
are cut or when there are different features like ponds, lakes, meadows.
These are not included.

> All of those are managed by the forestry department, is it not to harvest
> and economically use the forest, then to protect (which is a management
> strategy) the natural remaining tree stands, open areas, water features,
> wildlife grazing or feeding grass covered or scrub land etc... as they are
> all part of the forest ecosystem and necessary to sustain the forest, make
> it economically viable on the long term.

landuse=forest has no such implications about forest ecosystems. If you
want to find true forests, you will have to analyze the size and shape of
the landuse=forest polygons together with their  neighbours, to estimate
wether it is a forest or just some trees.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tagging/attachments/20210415/5d00b698/attachment.htm>

More information about the Tagging mailing list