[Tagging] Feature Proposal - RFC - boundary=forest(_compartment) relations

Brian M. Sperlongano zelonewolf at gmail.com
Tue Feb 9 15:24:09 UTC 2021


On Tue, Feb 9, 2021 at 9:58 AM Adam Franco <adamfranco at gmail.com> wrote:

> And you say "landuse=forest" and "natural=wood" both "would be edited to
>> emphasize the difference between these polygons boundaries, describing
>> actual, physical land state, and the administrative, management-related
>> boundaries of forestry areas/compartments."  Yet, these two wiki (forest
>> and wood) are almost hopelessly (I AM hopeful!) confused right now:  you
>> propose re-writing these?!
>
>
> Current tagging of landuse=forest is "hopelessly confused" with
> natural=wood because some mappers use landuse=forest mean "this is a
> forestry area" and others use it to mean "there are trees here" (land
> cover). Common rendering of landuse=forest with tree-fill makes usage of
> landuse=forest problematic for mapping "this is a forestry area" because
> forestry areas are not necessarily all covered in trees -- those trees may
> be harvested, there will be roads, lakes, scree, and other non-tree-covered
> land-areas within a "forestry area".
>
> This proposal to add boundary=forestry and boundary=forestry_compartment
> does not directly disentangle landuse=forest from natural=wood, but it
> *does* provide new, unambiguous tagging for "this is a forestry area
> where operator, ownership, and associated rules apply" that does not also
> imply anything about land cover. Without additions like those being
> proposed, there is simply no good way to tag managed forestry areas in a
> way that won't be broken by future mappers who don't want to see trees
> covering lakes.
>
> If this proposed tagging takes off and existing landuse=forest that is
> trying to map managed forestry areas gets updated to this more precise
> tagging, then landuse=forest can stop trying to do double duty as a
> designation of managed forestry areas and just be what it is in practice --
> some version of "there are trees here, maybe the hand of man is involved".
>

I dislike the idea that we should introduce a new tag to fix a problem
tagging scheme, but not simultaneously remove (read: deprecate and
gradually remove over time) the old tagging scheme.  This is exactly the
problem we're now dealing with for landuse=reservoir.  If a new tag
replaces an old tag, we must have an approach for the removal of the old
tag, otherwise we add more needless deprecation.

I strongly appreciate the work that has gone into boundary=forest,
particularly the fact that it has definitions that come from an author that
has specific expertise in forestry. Since it's starting anew, we can be
confident that new boundary=forest tags truly describe forestry areas
rather than could-be-forestry-could-be-just trees.

The only thing that can be assumed by landuse=forest is that it's an area
mostly or completely covered by trees.  Therefore I suggest -- and I don't
do this lightly considering the magnitude of landuse=forest -- the
following approach for this proposal to move forward:

1. Deprecate landuse=forest and its 4.8 million usages.
2. Existing forestry areas tagged landuse=forest can be replaced with
boundary=forest and actual tree-covered areas can be mapped with
natural=wood consistent with actual land cover on the ground.
3. Existing non-forestry tree-covered areas tagged landuse=forest can be
re-tagged as natural=wood
4. Data consumers should continue to support de facto usages of "the six
methods to tag a forest" including continuing to render landuse=forest as a
tree-covered area until landuse=forest has been fully replaced.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tagging/attachments/20210209/57d39549/attachment.htm>


More information about the Tagging mailing list