Talk:Proposed features/boundary=forestry( compartment) relations/Archive 1

From OpenStreetMap Wiki
Jump to navigation Jump to search

Manager tag

manager=* appears in text. Maybe existing operator=* would fit? Mateusz Konieczny (talk) 15:27, 12 February 2021 (UTC)

Well spotted; I corrected. Thanks! Penegal (talk) 16:03, 12 February 2021 (UTC)

Description of what is being proposed

This proposal lack a description of exactly what is being proposed, specifically:

  • Which tags will be approved
  • Which tags will be deprecated (if any)
  • What new definitions are being adopted

This needs to be stated explicitly in the proposal section. The "goals" of the proposal can go in a separate section, but the proposal needs to state exactly what is being voted on. --ZeLonewolf (talk) 18:54, 30 December 2020 (UTC)

Good remark; I addressed it in the first subsection of the rationale. Penegal (talk) 07:29, 31 December 2020 (UTC)
The proposal section still does not state what is being proposed. Is it "to adopt boundary=forest"? Is it "to adopt definitions for forest and wood"? Please, I encourage you to use very plain language in the proposal section as to what is being proposed. The "rationale" section is for you to state the reasons for which the proposed things should be approved. --ZeLonewolf (talk) 02:55, 1 January 2021 (UTC)
@ZeLonewolf: is it OK, now? --Penegal (talk) 14:58, 1 January 2021 (UTC)
@Penegal: better, I moved the "tagging digest" (not quite sure what a digest is, I think you mean "tagging changes") into the proposal section, so there is now (I think) more of a summary of the proposed changes --ZeLonewolf (talk) 15:14, 1 January 2021 (UTC)
@ZeLonewolf: yes, I had "changes" in mind; I corrected the title accordingly. Thanks for your modifications, that sounds clearer now. --Penegal (talk) 15:25, 1 January 2021 (UTC)
Resolved: Penegal (talk) 08:09, 14 February 2021 (UTC)

What is called a forest inside this proposal?

The proposal calls "forest" a wooded land managed to produce wood for human consumption and uses, as stated at the beginnig of the rationale. It is not meant to apply to what is generally called forest, that is to say a wide wooded area, managed or not by humankind. The proposal is designed only for managed wooded areas. Nevertheless, if there are non-managed wooded lands which are divided in compartments and would benefit of the proposal, I can amend it to precise that it can apply to any wooded land, as long as it is generally considered as a single forest, divided or not in compartments. Penegal (talk) 07:09, 31 December 2020 (UTC)

That is not really the common meaning of the word "forest" in English, though it might be the meaning of some cognates in other Indoeuropean languages. In English a forest is either: "a large area of land covered with trees and plants, usually larger than a wood, or the trees and plants themselves" or "a large area full of trees, usually wild". ( Or a large area of land that is thickly covered with trees ( In North American English especially, a forest is a large area (dozens of kilometers across, usually) which is semi-natural, as opposed to a "wood" which is usually a smaller area of trees, such as might be found between patches of farmland. However, there was also an older definition in Britain: "[historical] An area, typically owned by the sovereign and partly wooded, kept for hunting and having its own laws" and similarly "[in place names] Denoting an area that was formerly a royal forest". This is similar to the use with places like "The Black Forest" in Germany, which includes villages and farmland as well as woodland. Contrast with wood: "An area of land, smaller than a forest, that is covered with growing trees" -
So when you are talking about a "forest" boundary in this proposal it can be confusing, because many of these areas are small "woods", not large "forests", and a "forest" is more likely to be "wild" than a "wood" (in English). --Jeisenbe (talk) 07:51, 1 January 2021 (UTC)
@Jeisenbe: OK, I see the confusion here; I was using forest and wood according to the managed/unmanaged opposition I used when contributing (I interpred natural=wood as unmanaged, and landuse=forest as managed, hence the confusion). I clarified the proposal accordingly by specifying, at the beginning of the rationale, that I use forest and wood along the proposal to represent the opposition about management. I hope it is clear enough now, as I wouldn't like to replace them by their full meaning all along the proposal, that would make it very verbose. --Penegal (talk) 15:10, 1 January 2021 (UTC)
Per the suggestion below in the next section, please change the mentions of "forest" to talk more about "forestry". --Jeisenbe (talk) 21:08, 1 January 2021 (UTC)
Resolved: Penegal (talk) 08:10, 14 February 2021 (UTC)

Naming suggestion: "forestry"

Given the existing confusion of landuse=forest versus natural=wood, I would suggest that if this proposal moves forward related to land-management for sometime-production of forest products, that the tagging use the term "forestry" rather than "forest". It is far too easy to see "forest" and think "trees" whereas "forestry" is more unambiguously the actions and practices surrounding human use of land. -- Adamfranco (talk) 15:59, 1 January 2021 (UTC)

I agree strongly with this sentiment. In US English, the term "forest" is used for any wooded area, not just ones used for harvesting trees. Where as forestry is specifically about wood or paper production. --ZeLonewolf (talk) 20:41, 1 January 2021 (UTC)
But the compartment tagging already in use is boundary=forest_compartment; the proposed boundary relation should have a name based on the same word, "forest", at the risk of confusing the logical link between them. I could also propose renaming boundary=forest_compartment for boundary=forest_compartment, but is it OK to propose this, as these relations are already in use? --Penegal (talk) 07:53, 2 January 2021 (UTC)
Yes, it's certainly ok to propose changing a tag which is not very common, especially when there are 2 similar tags in use. In this case boundary=forest_compartment has only been used a little over 7000 thousand times, and most areas have a number of these features right next to each other, so there are only hundreds of places where the tag has been used. And the tag boundary=forestry_compartment is almost as common: it has been used over 5000 times: --Jeisenbe (talk) 21:48, 2 January 2021 (UTC)
I've now documented boundary=forestry_compartment at Tag:boundary=forestry_compartment. --Jeisenbe (talk)

I took a look at how boundary=forestry_compartment is currently used. Take a look at this grid of 252 forestry compartments in Russia!

Forestry grid.png

--ZeLonewolf (talk) 04:13, 3 January 2021 (UTC)

So be it. I just amended the proposal; now it proposes the formalisation of boundary=forestry_compartment and the deprecation of boundary=forest_compartment in favor of the first one. Penegal (talk) 15:10, 3 January 2021 (UTC)
Thanks, @Penegal:. It looks like the proposal should be reviewed in light of this change to also update the text to change boundary=forest to boundary=forestry as well. --Adamfranco (talk) 15:38, 3 January 2021 (UTC)
I'm unsure about this one: although I see why you're asking that (mostly consistency), I thing boundary=forestry would not be self-describing enough. The proposed boundary=forest is about what most people would consider as a forest/wood in its whole, either used for wood production or not (as stated in the Proposed_features/boundary=forest(_compartment)_relations#Some_definitions subsection); I fear renaming it boundary=forestry would be interpreted as This relation is only for wood-production forests, whereas it is not. Though compartments are clearly related to forestry, forest management and wood production, the forest as defined for the scope of the proposal, is not necessarily related. Penegal (talk) 15:59, 3 January 2021 (UTC)
My original comment in this section was intended to suggest tagging both the outer boundary as boundary=forestry and the components as boundary=forestry_compartment. It does make sense to have the tags consistent between the whole/part relationship. There are a few reasons for my preference for "forestry" over "forest":
# At least in American English, the word "forest" does not have a strong implication of "management area" and often connotes simply a "medium/large area of trees" (Wikipedia). In contrast "forestry" (Wikipedia) is the management of forests/wood-lands. To avoid this new boundary being confused as simply a "named area with trees", boundary=forestry would be more clear. That said, if I perceived the intent wrong and the intension is to just have named area with trees with no implication of management, then the naming boundary=forest makes more sense, but indicates that this proposal has other problems.
# There are many areas that are covered in trees that are protected by law from cutting of trees and many other human activities. These areas are often designated as something like "wilderness areas" or "nature reserves" and are tagged in OSM with boundary=protected_area+protect_class=1,1a,1b. Using the term "forest" may add to the confusion of how to tag these areas as they may have "forest" in the name like the Hammond Pond Wild Forest. In contrast, using the term "forestry" denotes to mappers that this is an area for forest management and sometime-harvesting of forest resources. "Forestry" doesn't necessarily imply wood/pulp production as wild mushrooms, ginseng, or other edible fruits/plants might be the target harvested from "managed forests", though wood/pulp is often the biggest use-case for forestry management. --Adamfranco (talk) 16:56, 3 January 2021 (UTC)
@Adamfranco: OK, I was afraid that "forestry" implied productivity, as it does in French. Still, could anyone give a third-party advice on the suggestion of Adamfranco and my remark about the non-intuitivity and lack of self-description I feel about boundary=forestry? I'm still in doubt about how boundary=forestry would be understood by mappers. Penegal (talk) 17:25, 3 January 2021 (UTC)
Resolved: Doubts removed through discussions; proposed changes applied. Penegal (talk) 08:11, 14 February 2021 (UTC)

 Zone d'exploitation contrôlée

@ZeLonewolf: from what I got by Web searches, it seams that there is no significant forestry in  Zone d'exploitation contrôlée; am I wrong? Penegal (talk) 06:26, 7 January 2021 (UTC)

You are probably right. The name "controlled harvesting zone" seems like a misnomer in that case. --ZeLonewolf (talk) 06:32, 7 January 2021 (UTC)
@ZeLonewolf: OK then I'll ignore it. Penegal (talk) 06:53, 7 January 2021 (UTC)
@ZeLonewolf: I think the harvesting is about fauna, not wood. Penegal (talk) 10:50, 7 January 2021 (UTC)
Resolved: Penegal (talk) 08:11, 14 February 2021 (UTC)

Proposal ordering suggestion: boundary=forestry before boundary=forestry_compartment

Having followed this proposal (which is shaping up nicely!) I am aware of its original focus on boundary=forestry_compartment. That said, reading it with fresh eyes, I think it would be more understandable by people reading it for the first time if the sections describing boundary=forestry came before the descriptions of boundary=forestry_compartment. Basically, describe the whole first, then its sub-areas. This also helps because boundary=forestry may stand alone, but boundary=forestry_compartment must be within a boundary=forestry. I don't suggest changing any of the terminology or content, just how the proposal is structured. The two places I noticed this mis-ordering are the "Proposed tagging changes" and "Tagging" sections. --Adamfranco (talk) 04:36, 27 January 2021 (UTC)

@Adamfranco: good point, I just corrected that. Penegal (talk) 06:36, 27 January 2021 (UTC)
Resolved: marking section that seem solved, for archival Mateusz Konieczny (talk) 18:19, 13 February 2021 (UTC)


I would suggest that the rendering section could be sharply reduced, or even dropped entirely. Since this proposal is about tagging, I would think "how it renders" is a separate discussion for each of the rendering styles to potentially consider. The comments made by @Stevea: on the mailing list tend to make me think that setting rendering aside and focusing on tagging will increase the probability of success. --ZeLonewolf (talk) 23:07, 9 February 2021 (UTC)

@ZeLonewolf: sounds relevant to me, so be it. Penegal (talk) 18:25, 10 February 2021 (UTC)
Resolved: Penegal (talk) 08:13, 14 February 2021 (UTC)

Distinguishing important forestry compartments from unimportant

As I understand, this proposal is motivated by fact that forestry compartments are important in some regions.

But in some, for example in Poland, there is basically no trace of them on the ground, are of basically zero importance.

As result rendering them would be a mistake, especially in a general purpose global map style.

It may be worth including something allowing to distinguish them... Mateusz Konieczny (talk) 07:35, 12 February 2021 (UTC)

@Mateusz Konieczny: it is one of the rationales, yes. That being said, if they are not materialized on the ground, and are not supposed to be, then they should not be mapped, as they lack physical existence. I will precise this. Penegal (talk) 15:01, 12 February 2021 (UTC)
Resolved: Penegal (talk) 08:14, 14 February 2021 (UTC)

Items to be discussed

Role admin_centre or centroid for rendering?

Should there be an Role admin_centre for telling the renderer where to render the forest name, or should the renderer simply search for the centroid, and adapt itself if the centroid is out of the forest? Penegal (talk) 16:22, 30 December 2020 (UTC)

Regardless if there were such a node, it shouldn't be a role tagged Role admin_center, as that is already in use for a separate purpose. Another name should be used, one which doesn't have "admin" in its name. Stevea (talk) 18:38, 30 December 2020 (UTC)
Absolutely not. This idea is far too related to a particular notion of what a forest is, which I think is not applicable in most countries. You could use a label role for pretty much achieve the same objective without introducing inapplicable concepts. Labelling on Carto-OSM already places labels within polygons. SK53 (talk) 19:59, 30 December 2020 (UTC)

Real-world examples

Very interesting proposal! This is a topic I'm quite interested in as I do a lot of work with protected areas and boundaries. It would be really helpful if you could supply a few real-world examples that we could examine to better understand the class of object that is being proposed. --ZeLonewolf (talk) 18:34, 30 December 2020 (UTC)

Well, for instance, you have this forest: It spreads on hundreds of hectares of wooded land but, to be modelled as a single entity, one needs to pretend that the whole forest is needleleaved, whereas it is not: there are broadleaved areas and mixed areas, which do not necessarily follow compartment or forest boundaries. One cannot draw the needleleaved, broadleaved and mixed areas of the forest or parcels, then merge them in a single entity representing the forest or parcels: the site relations is designed for separate node elements, not for polygons; the multipolygon relations is not what's needed either, as it is not for adjacent polygons. Furthermore, even if I could describe compartment boundaries with one of these, how to represent their belonging to the whole forest?
Besides, this proposal is not supposed to apply everywhere, but only for forests corresponding to the assumed characteristics: a mainly wooded, managed area, possibly divided in compartments, and with several areas of different characteristics. If the forest you're working on is uniform and comes in a single, non-fragmented area, you probably don't need what is proposed here; the current tagging practices are enough. That is the case of many small forests; this proposal is aimed for bigger ones. Penegal (talk) 06:53, 31 December 2020 (UTC)
Re: the example - why is the rather oddly shaped area called "Forêt domaniale de Moyenmoutier" while there is another tree-covered area inside of it called "Forêt communale de Moyenmoutier". And I see that there are areas of conifers immediately next to way 42133340 which are not part of "Forêt domaniale de Moyenmoutier" - why not? Is there some sort of physical boundary like a fence? Or is this representing who owns or manages the land: the local community versus the State perhaps? --Jeisenbe (talk) 08:35, 31 December 2020 (UTC)
In France, the forêts communales are owned by townships, whereas forêts domaniales are directly owned by the central state; both have the same manager, but the rules they are subject to differs. The odd shapes often have historic causes; besides, the forest of a township is not necessarily within its borders and may be within another city boundaries or even across such boundaries. The example polygon here has this odd shape because it follows the borders of the forêt domaniale de Moyenmoutier, which are themselves odd; without a better way to represent the forest boundaries, I assume the contributor resigned himself to draw it as a unique polygon but, having been there, I assure you this unique-polygon solution removes many details regarding, for instance, leaf_type variations across the forêt domaniale. One could say it has a mixed leaf type, but that would only minorly increase the level of details, while still being far from reality. Penegal (talk) 08:55, 31 December 2020 (UTC)
In that case what is being mapped is land ownership and operator, not actually the name of a woodland. Do people near these villages actually say things like "I found these mushrooms on the southern tip of forêt domaniale de Moyenmoutier. Tomorrow I am going to search for some more in the northern part of forêts communale de Moyenmoutier." I'm skeptical - those don't sound like common names people would use. And I am opposed to mapping land owndership and operation in OpenStreetMap, because this information can only be confirmed by checking a government document or database, and we can never hope to duplicate this information accurately in OpenStreetMap, since a new user can accidentally delete a way or move a node at any time, so users who want accurate land ownership information should get this directly from the official source, not from us. --Jeisenbe (talk) 07:41, 1 January 2021 (UTC)
@Jeisenbe: these names are bouth displayed on site and used by local people; trust me, I manage such public forests in the area. The names of French forests are routinely displayed, with the communale/domaniale/whatever adjective, along their borders and on compartment signs, so, as strange as it could sound, these are the used names. And people usually don’t tell where they find mushrooms, that is a secret. ;-) Of course, the name may be abbreviated when spoken out, generally saying "in the domaniale de Moyenmoutier", but the communale/domaniale/whatever distinction is important, used and publicly displayed, so yes, it is the name of the forest, as strange as it can sound.
Regarding ownership and operator, in France, the first is given by the communale/domaniale/whatever adjective of the name, which, as told before, is displayed onsite (domaniale means it is owned by the central state, communale means the forest is named after the township owning it); regarding the operator, in France, public forests are managed by a single operator, the Office National des Forêts. It's the law, this organism is the only one authorized by the law to manage public forests, unless a law specifically excluded a forest from its authority. --Penegal (talk) 14:55, 1 January 2021 (UTC)
I can confirm whatPenegal says, as it matches my experience in various parts of France. These names are pretty prominent on national topo maps (IGN Top25) as, in selected areas, are the forest compartments & numbering. Virtually always these are signed on the ground, and most commune owned and national owned forests will have been in existence for a long time. Famously there is a national forest of oaks planted in the late 1600s to ensure a supply of timber to build warships in the 20th century. SK53 (talk) 10:01, 4 January 2021 (UTC)

landuse=forest clarification must come before this proposal

This proposal further muddies (confuses) what is already a highly confused issue: that of landuse=forest. Currently, we denote no fewer than SIX different semantics for forest. This has been an ongoing issue of confusion in many parts of the world for many years in OSM. Until this is resolved, I believe this proposal is not only unhelpful, but it actually cannot be properly introduced, as it adds further specificity to the semantics of "forest" which are not clear themselves. For this reason, I cannot support any further development of the proposal. Perhaps the author can untangle how OSM defines, understands and applies natural=wood and landuse=forest, as this remains a difficult, even fundamental, unsolved problem. Stevea (talk) 18:38, 30 December 2020 (UTC)

Completely agree. This makes very minor changes refinement to forest_compartment (which has been in use for a long time (and reason why it is not better used is that they can be quite awkward to survey unless very regular or open data is available) whereas proposing a massive series of changes in relation to landuse=forest. It is the latter which needs to be discussed. For instance I agree with the rationale, but I think this is still a landuse & not a boundary issue for the most part. At various stages there has been a tentative discussion about using landuse=forestry to cover the area dedicated to forestry (clear-cut, new plantings, fire-damaged areas, wind-breaks, other non-wooded land etc) & it seems much of the suggestion is aimed at solving similar problems. US National & State Forests are slightly different & would benefit from a boundary approach as these are effectively a kind of admin entity. SK53 (talk) 20:04, 30 December 2020 (UTC)
There is a confusion here, I think: I talked about landuse=forest as the tag for managed wooded areas all along the proposal, because this is how I use this tag. I know there are different interpretations for this tag and natural=wood. I may have not been precise enough: the purpose of this proposal is only to give a way to describe forest boundaries without natural=wood and landuse=forest which, in my mind, should describe the trees population. The goal of this proposal is not to solve the issue of the meaning of these tags. They still have the role for describing the effective, physical wooded area; the proposal only aims at giving a way to describe forest management partitionning, not a new way to describe the managed lands. It is only about the administrative, boundaries issues, not land description. Additionnally, I must say that, even in a managed forest, there can be areas which are voluntarily left unmanaged, to keep some "wild" areas, so, even if a managed forest have its boundaries modelled following the proposal, a tag for drawing an unmanaged wooded land, as natural=wood is for many people, is fully adapted and still useful. Penegal (talk) 06:40, 31 December 2020 (UTC)
Yes, there IS confusion here. In addition to the request (below) to be more specific about what is proposed, I'm curious why simply a large number of properly tagged polygons (perhaps a complex multipolygon tagged landuse=forest) can't do this. I think it can, but I also think that the author does not find the rendering sufficient. That in and of itself is not enough to make a proposal like this. In the section answering what SPECIFICALLY is being proposed, I'd like to see a use case where existing tagging makes it impossible to do this. It seems that tagging a set of polygons / multipolygon with the desired "partitioning" CAN be done. So until I can see how this can NOT be done, I assert it can. You might not be getting the desired rendering, but that's not OSM, that's something to take up with a particular renderer. Stevea (talk) 07:21, 31 December 2020 (UTC)
Amplifying my previous: I still think (in a generic sense) ANY polygon can be "broken up into units" (whether administrative or for another reason, like "sub-management") and turned into "the same polygon expressed as a collection of contiguous connected smaller polygons." If you have a whole pie, you can cut it into six pieces, number them 1 through 6 and you have the same amount of pie, simply divided into 6, numbered accordingly. As with pies, so with forests in OSM. Stevea (talk) 07:26, 31 December 2020 (UTC)
You are right, but the problem resides in the logical link between these parts: a human will immediately see that the pieces form a single pie, but a software would need to be told that he must see the pieces as constituing a single pie, and that this merged object is to be called "a pie" and is a whole. The problem with forests is the same: suppose I have a single forest with a needleleaved area and a broadleaved area; you would create one entity for each area, I think, but how to put in OSM that these entities are treated as a whole by the manager or that they are commonly known as a single and unique forest, with a single name? Putting the name on both entities would not represent the oneness of the forest; you would need a relation for that, but which relation? Multipolygons and site relations are not meant for that, and there is, AFAIK, no currently standard way to do this. Putting the name in both entities would lead the renderer to render the name twice, which is technically right, but semantically wrong: there is not two different forests on the ground, only one, but the renderer cannot know, as it finds two different, unlinked entities. Using a boundary relation allows to emancipate of such cases, by acknowledging that boundaries in forests are generally not a physical division, which different entities would allow to model, but an administrative one, which is what boundary relation were made to model. Penegal (talk) 08:24, 31 December 2020 (UTC)
I disagree with making this prerequisite for that proposal. Similarly, leaf_cycle=* could be introduced without untangling this tagging issue Mateusz Konieczny (talk) 07:33, 12 February 2021 (UTC)
As it stands, this proposal deprecates landuse=forest} that is quite bold solution for that problem Mateusz Konieczny (talk) 19:18, 13 February 2021 (UTC)

Tagging list thread

Discussion initiated by author of proposal begins here. Stevea (talk) 02:27, 31 December 2020 (UTC)

Please comment here, it is easier to keep comments on a single page instead of having some on the mailing list too. Penegal (talk) 06:58, 31 December 2020 (UTC)

Confusion about the meaning of "parcel"

I may have not stated this clear enough: what I mean by "parcel" or "compartment" is not a cadastral parcel or a land lot, but a piece of a single forest, which is designated, in this forest, with a single reference, for instance "Forest of X, parcel #12". I may have not been clear enough as, in French, my mothertongue, "parcelle" can designate a land lot, or a forest compartment, hence the precision at the beginning of the proposal: "Parcel and compartment are used indifferently to designate a single piece of a given forest which used as a whole by the forest manager and passers-by, and which is given a reference, often a number, which is unique inside the given forest.". Penegal (talk) 07:02, 31 December 2020 (UTC)

Please also note that forest compartments not always have their limits marked with boundary stones. In my region of France, they are mostly materialized by paint and a small sign. Their can be boundary stones, but they are rare along the boundaries, and are frequently uprooted by forestry machines; embankments are more used, but, in most cases, the limit is only materialized by cutlines and paint, nothing definitive. Penegal (talk) 07:14, 31 December 2020 (UTC)
Are these cadastral parcels, that is, are they areas with a recorded land ownership? That is the common meaning of "parcel" when talking about land in English. --Jeisenbe (talk) 07:42, 1 January 2021 (UTC)
No, the proposal is not about cadastral parcels, but physical and signed divisions of a forest, which are unrelated; cadastral parcels are routinely merged or divided in compartments, as these are 2 very different things. Forest parcels, or compartments in the proposal, are only management-related divisions of the managed wooded area.

--Penegal (talk) 15:06, 1 January 2021 (UTC)

Please add some images illustrating this on the forest compartment page. I've tried to illustrate a range of examples from different countries, but I don't think there are any from France. I'd really like one from Fontainebleau if possible as from memory these are quite elaborate compared with what you describe. SK53 (talk) 10:05, 4 January 2021 (UTC)
@SK53: I'm not around Paris, but I know places around me that can serve as example. I'll try to take pictures and add them in the following days. Penegal (talk) 11:00, 4 January 2021 (UTC)
@Penegal: Thanks, I may just have something from Forêt d'Huelgoat too, but my visits to Fontainebleau from Paris were far too long ago. SK53 (talk) 10:13, 5 January 2021 (UTC)
@SK53: done. Penegal (talk) 07:01, 6 January 2021 (UTC)
Why are you suggestion mapping of compartments? At best it appears a node with the compartment number could be added but unless we have access to geospatial data the mapper would be unable to add the compartment polygons. If added what purpose do they serve? Glassman (talk) 00:01, 27 January 2021 (UTC)
@Glassman: because, as told in the proposal, these are frequently displayed on the ground and ease orientation in the forestry area when given. Of course, open data would greatly ease their modelling, but you can also simply follow their limits by foot or bicycle; regarding mapping them as polygons, well, it's what they are: polygons, not nodes. Why using a node for modelling a polygon? Penegal (talk) 06:26, 27 January 2021 (UTC)

Why not just use landuse=forest for compartments?

I just documented how boundary=forestry_compartment is used. But I wonder if it's really necessary to map these compartments as individual areas.

Why not just map them as areas of landuse=forest and add the ref=*? I understand that this doesn't prove that it is a managed forestry area, but that can be specified with a operator=* tag.

If a compartment contains 2 different types of trees, they you can map each area as a separate landuse=forest area, with the appropriate leaf_type=* and leaf_cycle=* etc, and just add the same ref=* to each one. And the marker=post features which show the boundaries or corners can certainly be mapped as individual nodes, since those are helpful for orientation and navigation.

Do we need more than this? --Jeisenbe (talk) 22:19, 2 January 2021 (UTC)

In my mind, landuse=forest (using the definition "land used for forestry") is completely separate from the land cover tag natural=wood. You could have a landuse=forest, but then within that have separate natural=wood+leaf_type=*/leaf_cycle=* areas, or even small natural=water or natural=wetland areas, perhaps some waterway=stream running through, maybe a bit of natural=bare_rock, all within the larger landuse=forest. Let's separate in our minds the notion of "this parcel of land is used from forestry" from "every square meter of this land has trees", because that isn't the same thing. A large forestry area may still have incidental areas of other land cover even though the area "as a whole" is used for forestry.
If we are expressing "this land is used for XYZ", we should think landuse=*
If we are expressing "this is what is physically observable here, we should thing natural=*
If we are expressing "a special law or status applies within this polygon", we should think boundary=*
So the question in my mind is -- are forest(ry) compartments describing some type of legal status or meaning? Or is it just land used for timber production? --ZeLonewolf (talk) 04:50, 3 January 2021 (UTC)
Re: "You could have a landuse=forest, but then within that have separate natural=wood+leaf_type=*/leaf_cycle=* areas, or even small natural=water or natural=wetland areas - that is not how this tag is normally used by most mappers. It is mostly used for areas covered by trees, not for marshes or areas of water or even for swamps (which are tagged as wetland=swamp + natural=wetland). See Forest for more info. --Jeisenbe (talk) 05:24, 3 January 2021 (UTC)
@Jeisenbe: how would you model different leaf types or other variations in the compartment? I can think of several reasons for that, a tornado causing a local plantation for instance; using a simple landuse=forest polygon would not allow to map that, but a relation could. Of course, using a polygon per compartment would still be an improvement when compared to current, kilometers-wide polygons, but it could be useful to allow a better resolution than compartments, as they can include hectares. Moreover, that would leave the problem with water, screes, glades… unsolved, whereas a relation would allow these features to be modelled with natural tags. Finally, forest and compartment boundaries, at least here in France, are frequently materialized all along, by cutlines, repeated painted marks on the trees and, if present, boundary stones. They are continuous features, so tagging them with nodes doesn't seem applicable.
Regarding landuse=forest, adding a new definition doesn't sound a good idea: there is currently no consensus for this tag, and at least 6 definitions. I thing adding a new one would simply increase the trouble about this tag, whereas a relation simply bypasses and clarifies its usage when both the proposed relation AND landuse=forest are used.
@ZeLonewolf: a compartment, AFAIK, is simply a subdivision of forests, used to ease their management by fragmenting it in smaller pieces. They are, de facto, used for orientation, but the root idea is management by forestry. Penegal (talk) 14:46, 3 January 2021 (UTC)
Why is landuse=forest + name=Name of Compartment insufficient to describe a compartment? --ZeLonewolf (talk) 14:51, 3 January 2021 (UTC)
@ZeLonewolf: Typically, you would not name a compartment, but give it a ref. Apart from that, please read the answer I just gave to Jeisenbe, it also answers yours. Penegal (talk) 15:04, 3 January 2021 (UTC)
Thanks for the response. Can you also address whether or not a forest compartments are cadastral data? --ZeLonewolf (talk) 15:22, 3 January 2021 (UTC)
This is addressed in the Some definitions subsection. Penegal (talk) 15:49, 3 January 2021 (UTC)
"Why not just map them as areas of landuse=forest and add the ref=*" - many of them include also areas that are not tree covered (in some cases it includes urban areas, industrial areas like wood storage areas and sawmills, not forested wetland, roads - up to motorways) and so on Mateusz Konieczny (talk) 07:37, 12 February 2021 (UTC)
Also, since landuse=forest lacks a universal meaning, this would be problemmatic --ZeLonewolf (talk) 06:07, 13 February 2021 (UTC)

Impact to forest(ry) areas in other countries

The descriptions of forest and forest compartments seems grounded in country-specific organizations of forest lands. If this proposal is approved, can you address whether the following lands would be tagged with the new boundary=* tags, and how they should be modeled?

Also, please address how the current tagging scheme in United States/Public lands#U.S. Forest Service on Wikipedia should be changed. --ZeLonewolf (talk) 15:28, 3 January 2021 (UTC)

In particular, the U.S. Forest Service manages "National Forests" for sustainable resource extraction, which neatly fits the tagging of boundary=protected_area+protect_class=6. It is not clear from this proposal if these and similar areas should be changed to this new boundary=forest/ry tagging. Thinking through possibilities, the same ways could be used as part of two relations one with the boundary=protected_area+protect_class=6 tagging and one with the boundary=forest/ry tagging, but that is duplicate effort. An alternate possible distinction is if the area is privately managed, then it could use the boundary=forest/ry tagging, while places protected by law could use boundary=protected_area+protect_class=6. None of these possibilities feels very clear cut (ha!) to me, so I think this is an open question: "What is the distinction between this new tagging and boundary=protected_area+protect_class=6?" --Adamfranco (talk) 17:10, 3 January 2021 (UTC)
@Adamfranco and ZeLonewolf: good question here. What I had in mind was managed woodlands, which can include a equal care for productivity and ecological protection. It is the French conception of "forest multifunctionality": public forests (unless in the case of established natural reserves or specifically protected areas inside, or if specific criteria are matched) are to be managed in order to preserve both ecosystems, productivity and public access, and they are not considered protected areas as such, at least not under class VI of the IUCN according to its French comittee, though the French laws governing them could match class VI protection. By what I read about forests managed by the USFS, they seem to be managed the same way, that is to say following a compromise between productivity, ecological protection and public access, but are considered protected areas. If so, that means that what is considered UICN class VI in the USA is not considered such in France?
If the situation is as I understood it, I can propose the following distinction for forests: a managed wooded area is considered a protected area if the IUCN considers its legal status makes it a protected area (for France, a list of protected area statuses acknowledged by the IUCN is available here), or if local authorities or laws consider its status makes it a protected area; unless, it is not a protected area. With this distinction, USFS forests would be mapped boundary=protected_area+protect_class=6, as they are considered protected areas by US law, whereas French public forests would generally be mapped boundary=forest(ry) as, although they have a specific legal status which includes ecosystem protection, they are not considered protected area by French law and the IUCN does not consider them as such.
Would this distinction be OK? Penegal (talk) 18:22, 3 January 2021 (UTC)
I'm going to go out on a limb here (haha) and make some suggestions. However, this really should get vetted by folks that understand forestry lands better than I do. First, background:
Given all that I'd offer the following suggestions:
  1. boundary=forestry is used as the primary tag for areas that are predominantly used for forestry. This would (probably) include USFS lands as well as Canadian controlled harvesting zones. It may or may not include US state forests, depending on whether those lands are actually used for forestry or whether they're only for conservation of forests.
  2. If a forestry area (tagged with boundary=forestry) is categorized with an IUCN Category, then it could also get an appropriate protect_class=* tag added as well. Thus, the Colorado examples above would be boundary=forestry+protect_class=5. Neat and elegant.
  3. The tag boundary=forestry can optionally be combined with protection_title=* to add a protection title, e.g. "National Forest"
  4. The tag boundary=forestry can optionally be combined with admin_level=* to indicate association with a government hierarchy, e.g. boundary=forestry+admin_level=2 would indicate a forestry area managed owned by the national government.
  5. The tag related_law=* can optionally be added in order to indicate the law which establishes or regulates the forestry area.
  6. The use of boundary=protected_area to tag forestry areas should be deprecated. Some text would be added to that page that says something like "For forestry areas, use boundary=forestry instead"
And to reiterate - we really need to change boundary=forest to boundary=forestry and the text "forest" to "forestry area" or folks in North America will be really confused. --ZeLonewolf (talk) 20:32, 3 January 2021 (UTC)
@ZeLonewolf: your suggestions seem logical and consistent to me, so I'll include them in the proposal. Just a notice: admin_level would be for the forest owner, not its manager; in France, the manager of public forests, the Office national des forêts, is state-wide, whereas the owner can be virtually at any admin level, from central state to a portion of a township/village. Penegal (talk) 09:16, 4 January 2021 (UTC)
@ZeLonewolf: it will take me some days to merge your suggestions in the proposal, I'll warn you when it will be done. And why are you thanking me for my editions when they follow your suggestions? Is that some sort of "Attaboy"? grin Penegal (talk) 17:36, 4 January 2021 (UTC)
Keep up the good work :) --ZeLonewolf (talk) 18:07, 4 January 2021 (UTC)
Thanks for gaming out all of these overlaps with boundary=protected_area, @ZeLonewolf:. I'd support replacing boundary=protected_area with boundary=forestry for U.S. National Forests and BLM lands where forestry is a significant activity. There might be some cases where the lands are managed by the BLM or USFS, but are not primarily tree-covered and/or where the primary resource activity is mining, petroleum extraction, and/or grazing. One thing this highlights to me is that there may not need to be a rendering difference between these two types of boundaries depending on the goals of a given renderer. For something like the OSM main rendering the same green border used for protected areas might be sufficient. For another renderer that distinguishes between wilderness and working lands, a distinction might be made. We seem to have encountered a sort of Ven diagram -- not all protected areas are for forestry and not all forestry areas are protected. --Adamfranco (talk) 19:39, 4 January 2021 (UTC)
@Adamfranco: so the green outline render that you see on those areas is caused by either boundary=protected_area+protect_class=1a,1b,2-6 OR leisure=nature_reserve. Therefore, as a transition strategy, any national forest that already has leisure=nature_reserve can get immediately converted from boundary=protected_area to boundary=forestry with absolutely no change in render behavior. That will get a tagging base established while render support can be sought in Carto (and I'm pretty sure this is a very simple change to the Carto style sheet from what I have looked at). --ZeLonewolf (talk) 19:59, 4 January 2021 (UTC)

I am opposed to this. For the USFS lands, boundary=protected_area is sufficient and better than this new tag. . Adding this tag will create more confusion. --Bradrh (talk) 03:55, 27 January 2021 (UTC)

@Bradrh: could you explain what's wrong with the proposal, in your opinion? Is it in the notions definitions, the rationale, or the suggested tagging scheme for applying these notions to US National Forests while retaining their protection status? How do you think the proposal should be adapted, and why? Penegal (talk) 16:27, 27 January 2021 (UTC)
I think what is wrong with it is that the current recommended tag of protected_area works. It doesn't need to be fixed. USFS manages lands for multiple uses. One tag for the USFS boundary is simple and works. There might be cases where -forestry, or -forestry_compartment is reasonable inside a -protected_area. Perhaps the proposal could be adapted by abandoning boundary=forestry, and making boundary=forestry_compartment a subdivision inside a protected_area. Perhaps forestry could be an additional tag added to a protected_area. --Bradrh (talk) 20:42, 27 January 2021 (UTC)
@Bradrh: the proposed tagging was not designed for US National Forests, but for any forestry area. But it happens that US National Forests are both protected areas and forestry areas, so the proposal must take them into account in a way that allows modelling them both as a forestry area and as a protected area, notions which use the same boundary=* key. A possibility would be to use another key for forestry areas, but the described notions fit in the boundary=* tag; that being said, do you see another key which would be better fitted for the proposal?
The goal of the proposal is not to fix National Forests tagging, though; as you said, there is nothing to be fixed. That being said, forestry areas are not necessarily, and by far, protected areas in the OSM/IUCN meaning, so it doesn't seem reasonable to propose the forestry tagging as an additional tag of protected areas. Should the proposed tagging be approved, as there are very likely more forestry areas than protected areas, it seems reasonable, when a forestry area is also a protected area, to use boundary for the most frequent notion, that is to say use boundary=forestry as the main tag, and protect_class and related tags to model the protection status of the area. The proposal includes changes of tagging for National Forests because they currently use another boundary value than the one proposed, and, because there are virtually more forestry areas than protected areas, it seems relevant to prioritize forestry over protection in the boundary=* tag.
Please also note that the reasoning of the proposal must be valid for the whole planet and different notions of forestry, not just follow USA tagging and definitions. Consequently, though I must take into account multiple countries and their customs, I cannot prioritize any, and that can, of course, bring on changing tagging practices for the USA, as long as the proposal doesn't ask inconsistent tagging practices. Please note this is not a rebuff, just following the logic to the end; of course, feel free to prove me wrong. If I can improve the proposal while keeping its overall consistency, please explain how. Penegal (talk) 07:09, 28 January 2021 (UTC)
Yes, I understand that it must be valid for the whole planet. It may be a valid tag for private forests, or forests in some other countries. My point is that it is not a good tag for USFS, protected_area is much better. Perhaps if you can narrow the scope, make it clear that the tag is for areas that are managed exclusively as forestry areas, not multiple uses as the USFS is. --Bradrh (talk) 15:49, 28 January 2021 (UTC)
@Bradrh: I think I understand the problem: forestry, in essence, is not a goal; specifically, it not a synonym of wood extraction, unlike what people often think. Forestry is a mean, the whole methods for management of (mainly) wooded areas, whatever the goal; in that meaning, it applies to US National Forests AFAIK. Of course, people often think that it is only in the goal of wood extraction, because it is its most visible application, but it is not necessary.
Besides, the goal of forestry is not always clear; stating in the proposal that the goal must be resource extraction and nothing else would render it useless for many forestry areas, where wood extraction is one of the many goals of forestry, for instance environmental protection. Even in private forestry areas, there is often some level of environmental protection, even if it is only the very minimum required by local law or a side effect of forestry-related decisions; for instance, the forbidding to use heavy machinery when the soil is too wet to prevent soil compession is a wood production concern (prevent damages to tree roots), but also an environmental one (prevent damages to soil life, which include tree roots). As you can see, the goals are often intricated, and not easily determined. To be used following a narrower scope as the one you suggest, such tagging would need the area be verifiably used only for wood extraction, so you would need to be able to justify that no other goal, whatever it is, is pursued. As explained, it will likely be very difficult to prove that; on the other hand, the proposed definitions are simple enough to be used on the ground by general mappers. What would be the use of a forestry proposal that would be useless for a significant part of forestry areas around the world?
Lastly, please note that what is considered protected in the USA (National Forests) may not be considered such elsewhere; in essence, French public forests are managed with the same goals in mind (wood production, general public access and environmental protection as three, equally important goals of forestry) but are not considered protected. Culturally, this french management is considered by nationals as the normal, multifunctionnal management of forests, and even private forests are required to have a (very) basic level of environmental protection. Maybe our disagreement is simply the reflect of this cultural difference. Penegal (talk) 06:57, 29 January 2021 (UTC)

Description of forestry compartment

Would it be accurate to say something like:

"A forest compartment (tagged with boundary=forestry_compartment) is an administrative sub-division of a forestry area (boundary=forestry)"?

Is it true that a boundary=forestry_compartment must always be enclosed by a boundary=forestry?

--ZeLonewolf (talk) 13:01, 4 January 2021 (UTC)

@ZeLonewolf: a compartment, at least here in France, is basically a management, forestry subdivision, which is in fact used for designation of forest pieces in administrative documents, but, first, forestry commands compartments creation.
AFAIK, compartments are always enclosed in a managed forest; it is the first act of management for wide forests, as it eases this management by cutting the forest in smaller pieces which are easier to manage than a single, big forest. I can't think of a compartment outside a bigger forest, as, by definition, the goal of compartments is to divide a forest. Without a forest to divide in pieces, no pieces. Penegal (talk) 13:20, 4 January 2021 (UTC)

Lands managed by the  United States Forest Service: different of National Forests?

@ZeLonewolf: Are there other forestry areas managed by the USFS than just National Forests? I found none. Penegal (talk) 06:53, 7 January 2021 (UTC)

Here's what we have documented: US Public Lands - US Forest Service. --ZeLonewolf (talk) 15:08, 7 January 2021 (UTC)
@ZeLonewolf: well, unless you know a USFS-managed land which is not a National Forest and still used for logging, I would be happy to know, as I wasn't able to find one. Penegal (talk) 16:51, 7 January 2021 (UTC)

Yeah, I think what you have is fine. It'll be up to the US community to figure out what is a forestry area and what isn't. --ZeLonewolf (talk) 16:58, 7 January 2021 (UTC)

I do not think this proposal will work if it is going to include all US National Forests. The lands managed by the Forest Service include wilderness areas where no mechanical devices, development or forestry are allowed (e.g. Mount Hood Wilderness). They also include areas which are managed for conservation and recreation (e.g. the Columbia Gorge National Scenic Area - and ski areas: and many other areas near to cities). If this tag is going to have a meaningful definition it needs to be limited to specific areas of forestry land use only. --Jeisenbe (talk) 02:17, 8 January 2021 (UTC)
@Jeisenbe: a boundary=forestry may contain areas with stricter protection against public access or resource extraction, which would model the areas you describe. In such cases, the stricter protection would win. Maybe you want me to specify that in the proposal? Penegal (talk) 10:04, 8 January 2021 (UTC)
The problem is that these areas are definitely part of the National Forest - they have the same owner and managers and are legally included in the boundary. But they allow basically no forestry activities: "human activities are limited to non-motorized recreation, scientific research, and other non-invasive activities". There is no cutting of live or dead trees for wood or forestry products. You can camp and walk in the wilderness areas, but you can't even ride a bicycle. You can't use a chain saw. So these areas could not be considered "areas used for forestry", even though they are part of the "National Forest" boundary. --Jeisenbe (talk) 00:11, 10 January 2021 (UTC)
@Jeisenbe: what I meant is that, if the National Forest contains areas with stricter protection, then you can simply model them with a boundary=protected_area relation inside the boundary=forestry area, in which case the higher level of protection wins. Penegal (talk) 10:39, 10 January 2021 (UTC)
But the boundary=forestry polygon would mean "this whole area is used for forestry" while the part of it which is Federally Wilderness is "no human activities except recreation and research", that is, no forestry allowed, so you have 2 contradictory ideas represented by the 2 overplapping areas. --23:55, 10 January 2021 (UTC)
@Jeisenbe: boundary=forestry would essentialy mean that the area, as a whole, is used for forestry; it does not mean that each single hectare of land of it is used for wood extraction. On one hand, forestry is the human management of forests, which does not necessarily lead to wood extraction — even if it is often the case —; on the other hand, even in forests with severe wood extraction, there may be areas voluntarily preserved of logging, which are still considered as part of the forestry area — voluntarily preventing wood extraction of a forestry area is a decision of management, a negative one, but one nevertheless . Consequently, having a protected area in a boundary=forestry is perfectly legit: it simply explicits that, in this specific boundary=protected_area, no wood is extracted, but it is still part of the forest management area. Penegal (talk) 10:11, 11 January 2021 (UTC)
I will vote against that definition of boundary=forestry, because it is even less clear than landuse=forest, more like place=region or boundary=place which are not verifiable and do not represent anything in particular in the real world. It sounds like under your proposed definition, the entire region of "The Black Forest" in Germany (êt-Noire) - which is nearly as large as the old French administrative division "Alsace" - could be mapped as boundary=forestry even though it includes protected areas, towns, farmland, industry, etc. --Jeisenbe (talk) 04:58, 12 January 2021 (UTC)
@Jeisenbe: hmm, you seem right; said that way, the proposed definition seems too large. What I had in mind was something verifiable, which allows the modelling of forests even if they are not used for wood extraction, which happens; that being said, with the current wording, that would not prevent something as excentric as modelling the whole Black Forest. I have to rethink that, I guess. Penegal (talk) 06:17, 12 January 2021 (UTC)

Definition of "forestry" needed

The comments made by @Jeisenbe: in the prior section should give us serious pause. How do we evaluate an area to decide that it is a "forestry" area? I took a look at the WP article on  Forestry. Is it required that an area is used for harvesting timber in order for it to be forestry? Or is it just a matter of the area being "managed" in some way? What is considered "managed"?

Across the world there is a wide variety of nomenclatures and management practices. Definitions of "forestry" and probably "managed" are needed so that mappers can evaluate their wooded/forest area categorizations in order to determine whether an area should be tagged as a boundary=forestry or a boundary=protected_area. Depending on how you read the definition of forestry, virtually any wooded protected area might be considered a forestry area! Clear criteria are needed.

Note the current definition:

    1. significantly managed by humans, that is to say it matches one or more of the following criteria:
      • it is divided in compartments,
      • it has a specific name which designates it as a whole,
      • it is used to produce wood for human consumption and uses;

By this definition "has a name" is sufficient to call it a forestry area, and producing wood is optional. --ZeLonewolf (talk) 02:51, 8 January 2021 (UTC)

@ZeLonewolf: you are right, these can be improved. I can propose this:
a forestry area is a mostly wooded land which is significantly managed by humans, that is to say it matches one or more of the following criteria:
  • its boundaries are materialized on the ground (boundary stones, cutlines, paint, signs…);
  • it is used to produce lumber, firewood, bark, cork, sap, and related forest products like leaves, mushrooms, wild berries or herbs for human consumption and uses;
  • the number, species or shape of trees are altered by human by any of the following, with or without extracting wood for human needs: trees undergo a selection for biodiversity, phytopathologies, wood quality or to regulate the respective proportions of present species; they can be planted, selected, fallen or pruned; the area undergoes forestry works such as mulching, enrichment, protection against damages from animals, coppicing, thinning…
Does that sound clear enough? Penegal (talk) 09:07, 8 January 2021 (UTC)
So any wooded area with materialized boundaries, is considered a forestry area? --ZeLonewolf (talk) 20:36, 8 January 2021 (UTC)
@ZeLonewolf: in this definition, yes. I think it is necessary to include that in order to model some forests which may not be used for wood extraction, but are still designated with a name and have known boundaries. I can tell at least one forest, in France, which is considered managed (as it is a public forest, forest management is mandatory), has compartments and materialized boundaries, but is neither a protected area nor used for wood extraction. I don't know the exact reasons for such a uncertain situation, but such cases should be allowed as well, else some other forests in this situation, which could benefit of the proposed tagging, would be left without standardized tagging scheme, and no way to be modelled as a whole despite leaf_type and other physical variations; as it is one of the goals of this proposal, I feel useful to allow the proposed tagging scheme to be used for such forests, which could not be accurately modelled in their entirety without that. Penegal (talk) 14:50, 9 January 2021 (UTC)
I think you will have trouble convincing the community that any wooded area with a boundary marker is a "forestry area". I would ask that you dig a little deeper in to what you mean by "managed", and define that. It's not really enough that the definition works for France, there needs to be criteria that can be applied globally to answer "is this forestry?". --ZeLonewolf (talk) 15:48, 9 January 2021 (UTC)
@ZeLonewolf: I'll try to explain. An area is "managed" when its existence, as a physical area with defined, precise boundaries, necessitate a significant human work, with perceptible effects; management is about the land itself, its trees, its glades… but also about its limits. Forestry, in my mind, is the management of forests; "managed" would then apply on a wild forest when its limits are cleary materialized with paint, boundary stones, cutlines… These show that humans have the will to consider this area as managed, at least at the very basic level of marking its boundaries; else, nobody would make the effort of materializing the boundaries. Marking boundaries, even simplistic ones, is the very first act for land management, hence I consider this to be enough to mark the wooded area as "managed", as a "forestry area". Penegal (talk) 17:56, 9 January 2021 (UTC)
I would define forestry as "the production of forestry products from woodland" - e.g. trees are used to produce lumber, firewood, bark, cork, sap, and related forest products like leaves, mushrooms and perhaps wild berries or herbs. In forestry trees are grown to produce these things for human use. If the trees are for food production it is probably an landuse=orchard. If the trees may never be cut and firewood can't be gathered, due to being a protected area perhaps, then it's not being used for forestry.
Unfortunately in my experience it is very difficult to look at an area of trees on the ground (even worse in aerial imagery) and determine that it is NOT being managed for forestry. And even if it looks like a tree plantation, with one type of tree all in rows of the same age, it's possible that the trees are now abandoned and growing semi-wild, or that the land has been converted recently to a protected area. --Jeisenbe (talk) 01:03, 11 January 2021 (UTC)
@Jeisenbe: it is partly because forestry effects on wooded lands may be difficult to see, that I defined boundary materialization as an act of management which allows using boundary=forestry: most people won't be able to tell if the trees themselves are currently managed, but virtually anyone will be able to tell if boundaries materialization are maintained or not; if they are, the area is still managed. As I told, materializing boundaries is an act of management; doing it on a forest is managing the forest, hence forestry. That being said, thank you for remembering me that forest products may also be berries or mushrooms; I just amendend the above definition. Penegal (talk) 10:15, 11 January 2021 (UTC)


@Jeisenbe and ZeLonewolf: as you both have relevant corrections for my definitions, which I have to regularly adapt, I propose the following: I'll propose you a draft of the definitions, you'll comment that, then propose a new draft, that you'll comment, and so on until we can agree on the definitions. I think that would ease their processing and reading.

Round 1

A given, limited area is "managed" when its existence, as a physical area with defined, precise boundaries, necessitate a significant human work, with perceptible effects; management is about the land itself, its trees, its glades… but also about its limits. Please note that this only works for delimited, verifiable areas: the boundaries must be verifiable, either on the ground or using open data; if the work is based on open data, you should tag the features with the relevant source.

"Forestry" is the management of forests; a wild forest would be considered "managed" when its limits are cleary materialized with paint, boundary stones, cutlines… These show that humans have the will to consider this area as managed, at least at the very basic level of marking its boundaries; else, nobody would make the effort of materializing the boundaries. Marking boundaries, even simplistic ones, is the very first act for land management, hence I consider this to be enough to mark the wooded area as "managed", as a "forestry area". Forestry area boundaries must be distinguishable of land lot boundaries: if the forestry area is only delimited with land lot boundary stones, without signs or paint highlighting the difference between its boundaries, as a forestry area, and a mere land lot boundary, the area should not be modelled, as its boundaries are not verifiable.

Forestry is also, of course, the management of the trees. The number, species or shape of trees are altered by human by any of the following, with or without extracting wood for human needs: trees undergo a selection for biodiversity, phytopathologies, wood quality or to regulate the respective proportions of present species; they can be planted, selected, fallen or pruned; the area undergoes forestry works such as mulching, enrichment, protection against damages from animals, coppicing, thinning…

Consequently, a given "forestry area" is a mostly wooded land, with verifiable, distinct and precise boundaries, with a distinct existence (that is to say, it is considered, as a whole, one and only one); it must be in one of these two cases:

  • significantly used to produce lumber, firewood, bark, cork, sap, and related forest products like leaves, mushrooms, wild berries or herbs for human consumption and uses; this may be one of the official goals of the modelled area, or common knowledge; the extraction of resources may not take place everywhere on the area: some places within the area may be subject of extraction prohibition, or host protected areas; if these areas have a distinct, verifiable existence, they may be modelled at such with a boundary=protected_area;
  • subject of alteration of number, species or shape of trees by human by any of the following, with or without extracting wood for human needs: trees undergo a selection for biodiversity, phytopathologies, wood quality or to regulate the respective proportions of present species; they can be planted, selected, fallen or pruned; the area undergoes forestry works such as mulching, enrichment, protection against damages from animals, coppicing, thinning… By exception, if the management applied to the wooded area is totally and verifiably unrelated to resource extraction (for instance its purpose is only maintaining the biotope of protected species), the area is to be tagged as boundary=protected_area, not boundary=forestry, as the first more accurately describes the area.

A forestry area itself may also be a protected area if it is considered such by local authorities and laws; then, the boundary=forestry entity should be accordingly tagged with protect_class and other relevant tags.

Applying these definitions would lead to the following modelling:

  • a mostly wooded area used for mushrooms collection, without verifiable and distinct existence and boundaries: no boundary=forestry (unverifiable boundaries);
  • a mostly wooded area used for mushrooms collection, with verifiable and distinct existence and boundaries: boundary=forestry;
  • a US National Forest with distinguishable management of trees, with official open data for its boundaries, with or without areas where resource extraction is prohibited: boundary=forestry (the area existence and boundaries are verifiable, and trees are managed; add protect_class and relevant tags if applicable); protected areas inside the National Forest are to be modelled by themselves and, within their boundaries, their stricter protection is modelled by their boundary=protected_area;
  • a US National Forest with distinguishable management of trees or wood extraction, with verifiable and distinct boundaries, where resource extraction is prohibited and the prohibition explicit: boundary=protected_area (the management is explicitly unrelated to resource extraction; add protect_class and relevant tags if applicable);
  • a wooded area used for logging, which is common knowledge but without materialized boundaries nor open data sources for its boundaries: no boundary=forestry (unverifiability);
  • several distinct, mostly wooded areas which are treated by their manager, the french Office National des Forêts, as a single, fragmented forest, with verifiable and distinct existence and physical boundaries, which is not a protected area, with or without wood extraction: boundary=forestry (verifiable existence, unity and boundaries);
  • a wooded area with verifiable and distinct existence and boundaries, used for wood extraction, which contain screes and glades: boundary=forestry;
  • the Black Forest, a managed, mostly wooded land, commonly known as a single forest, but without verifiable and distinct boundaries: no boundary=forestry (unverifiable boundaries).

What are your comments? Would a decision tree be preferred, given the complexity of the decision making process? Penegal (talk) 07:46, 12 January 2021 (UTC)

@Jeisenbe and ZeLonewolf: no comment? I'll then assume this version is OK to you; I'll merge it in the proposal, maybe wait a few more for other comments, and launch the vote process. Penegal (talk) 05:50, 19 January 2021 (UTC)
The text above is quite long. Consider that if this feature is added to iD or JOSM it will have to use a short name, and short description which clearly describes it. But I find the two different suggested possiblities for US National Forests to be quite confusing. All US National Forests have some areas used for production of timber / lumber, and most have some areas which higher levels of protection. All have some protection, since you can't develop the land for an industrial facility or housing estate. --Jeisenbe (talk) 06:14, 19 January 2021 (UTC)
@Jeisenbe:: the length of the proposal mainly comes from the variety of forestry types — some are far more extraction-oriented than others, some even exclude wood extraction — and from the existing features, which include areas with ressource extraction and considered as protected areas. Forestry is a wide field, with many different varieties, so, if the proposal is to allow modelling of all varieties, it must be detailed to accomodate these variations; else, the proposal will be partial and far less relevant. By the way, if all US National Forest have wood extraction, then there will be no two modelling schemes for them, only one: boundary=forestry+protect_class=.
That being said, the short label and description do not need to encompass the detailed explaination needed to accomodate neighbour notions; they can simply be "Managed forest: mainly wooded, defined area managed by humans (border marks, wood extraction, management of forest stand…)". As for the notion of a given forestry area encompassing areas with wood extraction and areas without, it can be the case of any forestry area; as told before, wood extraction in a forestry area does not necessarily occurs on any single hectare of land, some may be left out of wood extraction. Moreover, that doesn't prevent stricter protected areas, which can be modelled as such; in this case, their stricter level of protection applies within their boundaries. There again, explaining that necessarily lenghthen the proposal, but people will not necessarily have to read it all; parts are only for border cases.
If you find it relevant, I can precise that "In many cases, one of the goals of the management applied to the area is to keep the area wooded, and prevent some land uses such as industrial or residential.". It is the case, AFAIK, and would make the proposal more explicit. Penegal (talk) 11:25, 19 January 2021 (UTC)
I agree with Jeisenbe's comments. I remain concerned that the forestry definition may be overbroad and lead to any forested/wooded area being tagged as a forest boundary. I am not a forestry expert so I'm probably at the limit of my usefulness here. --ZeLonewolf (talk) 12:59, 19 January 2021 (UTC)
@ZeLonewolf: I'm a professional forest manager, which is why I designed this proposal, and may explain the differences between my views and yours. I explained that, in this new definition set, a forestry area must either have its boundaries physically materialized in a different way than plain land lots, or have its boundaries available in an open data set that explicits its forestry nature. That prevents using the relation for any wooded area: if one wants to make it a forestry area, the very first act is to materialize its borders in a way which explicits the relation between the area and forestry, first because it tells neighbours that here starts an area potentially used for wood extraction (with related dangers and property rights), secondly because it allows the forest workers to know where their work starts and stops. Given the time and money it takes to maintain the materialization (some hundreds dollars/euros per km at each pass), one does not simply do that for pleasure, it is a voluntary, management act. I also allowed open data to accomodate large forestry areas in very remote places, where, given the boundary length and the remoteness of the forest, boundary materialization may not be maintained.
I understand your doubts here, but I must take into account wooded areas which, while managed, simply have their boundaries physically marked without any significant work on its trees. I know such forests exist, such as the "forêt domaniale de Bannes": the goal of this is to have some sort of protected area (no wood extraction) without having to apply the related laws; even if you wanted to distinguish between such an area and wood-extraction areas, you must be a professional to tell the difference between these, and even then you could have trouble to tell. Besides, the simple act of materializing the borders is, technically, forestry; any forest professional could tell you he uses, and needs, the border materialization in its work. Penegal (talk) 15:34, 19 January 2021 (UTC)
I would encourage you to declare your credentials as an expert in forestry as that is helpful context (i.e. that you actually know what you're talking about in the proposal). But to be clear - I'm not the one you need to convince - it is all of the layman mappers who will (a) vote and (b) need to apply these definitions to different categories of land that they may not be expert in. Given the changes made here, I would recommend a second announcement to the tagging last. I suspect that the definition of "forestry area" will be a problem for folks, because it still says that any wooded area that has a name is considered a forestry area, while most laymen will likely associate forestry with tree harvesting. It might be helpful, as an expert, to analyze a few examples outside of France and show how your definitions would be applied ("THIS is a forestry area, but THAT is not a forestry area) and explain why certain areas are or are not forestry areas under your definitions. Good luck! --ZeLonewolf (talk) 22:26, 19 January 2021 (UTC)
@ZeLonewolf: I added some such examples with the above definition set, but I can add others following your remarks and Jeisenbe's ones. Regarding my job, do you suggest I add it in the rationale? In the mail addressed to the tagging ML, in the form "This proposal was designed by a forest profesionnal; please check its understandability for general public."? Penegal (talk) 07:40, 20 January 2021 (UTC)
Here's an example of what a US National Forest can look like. This is just part, on a map which shows which areas are open for cutting firewood for personal use. Note the 3 wilderness areas which are partially within this National Forest, and many areas of privately-owned land, as well as areas where firewood cutting is restricted (due to recent forest fires): - forestery in the western USA is quite different than your experiences in western Europe. And I found Indonesia to be more different than either: is a trade-oriented overview, but note that 73% of land clearing in Indonesia is illegal: - and the "ban" on deforestation is toothless: --Jeisenbe (talk) 05:16, 20 January 2021 (UTC)
@Jeisenbe: in the case of this US National Forest, the proposal would apply to the outer limit of the National Forest. Your remark reminds me of an element I didn't talk about, but could further explain our disagreement: in a given forestry area, as told before, there often are specific areas that are voluntarily excluded of wood extraction. That being said, such excluded areas are not necessarily permanent nor materialized: they are often left at the discretion of the forest manager, as it is a pure management issue. There may be regular changes of these areas, to allow each piece of land to rest some years without wood extraction; they may also be simply described in the management documents, for instance by telling "all surfaces with a tilt beyond 20% are excluded of logging". That really is a management issue, which likely changes with time, and this data may not be available to general public mappers; I cannot take such areas into account, as they will probably lack stability in time, or verifiability, or both.
In the case of Indonesia, forests with forestry concessions are in the scope of the proposal: even if they are subject to illegal logging, they are wooded, and forestry is the goal of the concession which also explicits the limits of the area; of course, once the area is totally, permanently deforested, and the concession expired, the area is no longer a forestry area. Deforestation by farmers, though, is out of the scope of the proposal: area limits are likely not materialized nor published as open data, and the area does not keep its wooded state for more than a year. Such area lacks precise boundaries and wooded cover stability over time, and are consequently out of proposal scope, and probably out of OSM scope, which needs precision and sufficient feature stability over time. Penegal (talk) 07:40, 20 January 2021 (UTC)
I do not think this works for US National forests. Re: "such excluded areas are not necessarily permanent nor materialized: they are often left at the discretion of the forest manager, as it is a pure management issue". In the USA the wilderness areas are established by federal laws, approved by the legislature. I was mostly posting the map above to show how fragments a US National Forest boundary is: often it has a checkerboard pattern of 1 mile squares, with only half included in the management area. I don't think your proposal works for these.
Re: forestry concession... "even if they are subject to illegal logging, they are wooded, and forestry is the goal of the concession which also explicits the limits of the area". A forestry concession is completely different than a US National forests, or the original examples which you showed in France. --Jeisenbe (talk) 04:27, 27 January 2021 (UTC)
(Edited) @Jeisenbe: wait a minute: it seems that wilderness areas and the areas you described, where firewood gathering is prohibited, are different notions? Or is it the same? Anyway, are all forestry operations, logging or not, forbidden in these, or only firewood gathering? Are they wilderness areas like protect_class=1b? Do areas with firewood area forbidden also considered protected areas?
Regarding forestry concessions, I was talking about those you pointed in indonesian state forests, which do not, AFAIK, suppose that the area is a legal protected area; does US National Forests, which are protected area have concessions? If so, with which restrictions to forestry operations? Penegal (talk) 06:32, 27 January 2021 (UTC)

US National Forests: help needed

@Jeisenbe and Bradrh: I just made some research to understand the problems with the applicability of the proposal to US National Forests, and would value your comments about it.

As I understood it, the multifunctionality of a National Forest is based on a spatial division of it: the USFS tends to segregate different uses/protections on different areas, instead of favoring compromises to allow several concurrent uses of a same area as we do in France. In US National Forests, many types of protected areas coexist, which prevent or drastically limit forestry works, with or without wood extraction: research natural areas, wilderness areas, wild rivers, special interest areas… These may overlap and tend to cover big portions of national forests. This situation explain the reluctances around the applicablity of the proposal to US National Forests: saying that the whole National Forest is a forestry area would be exaggerated, as significant portions are covered by these overlapping protected areas, leaving only small parts of National Forests opened to forestry. Is my understanding correct?

About the remaining parts, inside a National Forest and outside of any protected area, does the USFS manage the tree stand and the land itself, or does the USFS concede the land under a "property stewardship contract", which, when designed for forestry operations, are basically forestry concessions, in which the USFS expects specific counterparts from the company in exchange of wood extraction? May both situations coexist in a given National Forest? If the forestry area is subject of a property stewardship contract, is the contract applicability area published or materialized in a way considered verifiable, allowing its modelling in OSM? Penegal (talk) 15:47, 30 January 2021 (UTC)

I don't believe there are verifiable limits to the areas used for forestry within the US National Forests. Even the limits of the National Forest are barely verifiable, since they are inconsistently marked on the ground, and often are only available by looking at official maps. That's why I recommend that this tag not be designed for use in this way:
Approximately 73 percent of the 191 million acres of national forests are considered forested. Of that forested land, 35 percent is available for regularly scheduled timber harvest and about ½ of 1 percent of those trees are harvested in any 1 year. The remaining 65 percent of the forested land is designated for non-timber uses, such as wilderness and other areas set aside for recreation, or cannot be harvested due to environmental conditions, such as steep slopes and fragile soils.
The US Forest Service does not contract out management of the forest, but rather it sells areas of timber (standing trees) at regular intervals, with private companies bidding to buy the right to harvest the trees. This is an example of sale which is happening soon: - there is an overall sale area boundary, and then specific cutting areas inside. This is the official advertisement (which appears to be a scan of a printed document!) - it appears the Forest Service will mark certain trees with orange (paint or plastic) if they are not to be cut. Here in example of a recent auction, where the high bidder won the right to harvest the timber: --Jeisenbe (talk) 22:48, 30 January 2021 (UTC)
@Jeisenbe: are you sure about that? It looks like there's lots of public domain data about exactly what the US Forest Service is doing in their lands! I'm not smart enough to know what all of those GIS data layers means, or how they intersect with this proposal, but it seems like data isn't the problem here. --ZeLonewolf (talk) 23:01, 30 January 2021 (UTC)
@Jeisenbe and ZeLonewolf: it seems that several of these data match the proposal scope, for instance Activity Silviculture Timber Stand Improvement (areas where the timber stand is subject to forestry for improvement of overall quality) or Timber Harvests. Of course, that seems to confirm that a whole National Forest can't be considered as a forestry area, only some parts of it; as for which areas could be modelled following the proposed tagging, these data sets seem to at least partly answer the question. Penegal (talk) 09:43, 31 January 2021 (UTC)
For a feature to be verifiable, it should have an existence in reality, not just on a government document or database. I am opposed to "approving" any more tags for non-verifiable features. --Jeisenbe (talk) 00:52, 1 February 2021 (UTC)
@Jeisenbe: the proposal states that there must be forestry-related work inside the area; the database would only be used to verify its limits, when they are not physically materialized. Penegal (talk) 10:31, 1 February 2021 (UTC)
@Jeisenbe and ZeLonewolf: I edited the proposal to remove the buggy tagging scheme for US National Forests. Now, the proposal states that, typically, a US National Forest merely contain forestry areas, but is not a forestry area by itself. Is the new version OK to you? Penegal (talk) 15:41, 8 February 2021 (UTC)
That seems OK to me, but I am by no means an expert, so I would want others to weigh in. --ZeLonewolf (talk) 15:47, 8 February 2021 (UTC)

Unclear "per se"

"such areas are currently mapped per se" - sorry, but it is unclear to me. I found that "per se" means "by or in itself or themselves; intrinsically." what does not help me.

Current tagging section is really, really unclear to me.

Having "Rendering example of a forest including a grassland" - that at least in my opinion is a clearly incorrect tagging - is making it even more confusing.

"Some forestry areas may even be something else than natural=wood or landuse=forest: for instance, they can be a swamp whose trees are felled for human consumption. If so, they are tagged with natural=wetland and landuse=forest, confusing their rendering. " - so it is both wetland and forest, right? Then why "be something else than natural=wood or landuse=forest" if that is a forest? (and in addition wetland)?

Mateusz Konieczny (talk) 07:46, 12 February 2021 (UTC)

@Mateusz Konieczny: what I meant by "mapped per se" is that they are mapped as a separate feature, without taking into account their logical link with the surrounding forest. In my explications, I will use indifferently "forest" and "forestry area", as I explain the problems within the current situation, which also mix these notions.
Non-wooded areas such as glades are routinely included in the limits of the forestry areas surrounding them. That is counterintuitive to general public, but I will explain. It may sound as an argument from authority, and I'm sorry for this, but I can't explain otherwise why I have this reasoning: I'm a professional forest manager, and took into account the situations I saw happening on the ground. I explained in the proposal why such situations may exist: Such [non-wooded] areas are still often considered by foresters or general public as part of the wooded area, as they may be:
  • managed by the manager of the surrounding forest;
  • under reforestation;
  • lands with a distinct, wood-related ecosystem (glades for instance, which essentially exist because they are in a wooded area, else they would be called grasslands);
  • legally considered part of the surrounding wooded land, i.e. they are subject to the same subset of laws, rights and obligations ( Régime forestier in France, for instance).
Given this definition of what a forestry area may include, if you use landuse=forest to tag it, then the "Rendering example of a forest including a grassland" is technically correct tagging: the glade is considered part and parcel of the surrounding forest, hence it must not be modelled as an inner limit of the surrounding forest, hence it provokes the example buggy rendering, which makes the viewer think that the area is wooded, even if it is not. The problem with the current practices is: whether you exclude the glade of the surrounding forest, and this does not reflect the reality (glade considered part and parcel of the forest), or you include the glade in the surrounding forest (i.e. you do not map it as an inner limit of the forest), and you have a buggy rendering.
As to your last paragraph (swamp being a forestry area), the problem is that a forest is currently mostly thought about with landuse=forest in mind, that is to say (and the rendering enforces this illusion) a forestry area is always a wooded area, with no special characteristics such as being also a wet area. People like simplicity, and have trouble to think about a forest being also a swamp, but that happens. The renderers have the same problem: such corner case provokes the buggy rendering I gave in example, as the renderer understands the dual landuse=forest+natural=wetland tagging as display both textures. Displaying both textures, in such cases, troubles viewers. As much as this behaviour works well for, say, wooded residential areas, for forestry area it simply fails to describe the reality in an understandable way; it only troubles the viewer. The point is, the current tagging does not allow a correct rendering; the proposed one, by modelling the forestry area with boundary, removes the buggy renderings ("buggy" understood as "the rendering gives an erroneous understanding of the reality") by removing the modelling of forestry areas with landuse=forest. As the modelling of forestry areas will no longer rely on a landuse tag, it will no longer induce the use of a texture for rendering, essentially removing the buggy rendering on such cases.
I'm sorry for the verbose reply, and I may have repeated myself, but I'm working on this since a month and a half, and the proposal logic is crystal clear in my mind: to me, it is consistent both with OSM rationale and customs, and with existing forestry customs. Consequently, I may have trouble explaining it to others in an understandable way. That being said, if I failed to render the proposal consistent with OSM customs, please explain me, but also elaborate; I'd be happy to further improve the proposal, but may have trouble understanding your reasoning. Penegal (talk) 15:59, 12 February 2021 (UTC)
"I may have trouble explaining it to others in an understandable way" - I really understand this, I run into it some times with people being opposed and my complete failure to explain that objection is not relevant at all. Hopefully my potentially misinformed comments will be useful - even if coming from person with some opinions here and defending rendering they implemented (I am trying to distinguish personal opinion from general consensus, but I may fail - please do not assume that I a unbiased) Mateusz Konieczny (talk) 18:37, 13 February 2021 (UTC)

buggy rendering

  • Looking at Proposed_features/boundary=forestry(_compartment)_relations#Land_cover again, after significant improvements:
    • I would say that root problem is that currently there is no tagging for "forestry area". We have at least three tags for "there is area of trees here", one with "landuse" key (see Counterintuitive key names). None of them is for "forestry area". I would not complain here about "buggy rendering". If someone misuses one of natural=wood or landuse=forest then problem is with tagging, not with rendering. landuse=forest is not for forestry areas (except ones that are tree-covered, potentially with patches of trees cut down and reforested at rate faster than mappable).
    • If rendering would be buggy and landuse=forest would be standard for forestry area, then it should be changed by modifying renderer. It is changed by inventing new tagging as there is currently no accepted way to mark this
Mateusz Konieczny (talk) 18:37, 13 February 2021 (UTC)
I am not removing parts about buggy rendering on my own, due to glaring lack of objectivity - I am one of people behind releasing it (implemented final part of it, though most of the work was done by others) 18:38, 13 February 2021 (UTC)
@Mateusz Konieczny: I just amended that part to closely describe what you told, as it seemed consistent to me; is it OK now?
"mappers mostly use landuse=forest for forestry areas" - are you sure? From what I know from local area and helf-remembered research done long time ago, dominating global use is for forested/tree covered areas, not for forestry areas. Is it possible that use for forestry is some local decision (some locally active mapper or some import)? Mateusz Konieczny (talk) 08:56, 14 February 2021 (UTC)
@Mateusz Konieczny: hmm, you're probably right; I rephrased that. Penegal (talk) 09:21, 14 February 2021 (UTC)


Note that this proposed tagging seems to suffer from being unverifiable - in many cases official plot/boundary data will be sole source of mapping data Mateusz Konieczny (talk) 07:48, 12 February 2021 (UTC)

@Mateusz Konieczny: why is it unverifiable? If the compartments and forest limits are materialized, then they are verifiable; if they come from official forestry open SIG data, they are verifiable. Of course, this is for verifying the limits of the area; the area being a forestry area is deduced from forestry works taking place on the area. As said in the proposal, it may be alteration of number, species or shape of trees by humans by any mean, including the following, with or without extracting wood for human needs:
  • trees undergo a selection for biodiversity, phytopathologies, wood quality, maintaining the wooded state of the area or to regulate the respective proportions of present species,
  • they can be planted, selected, fallen or pruned; the area undergoes forestry works such as mulching, enrichment, protection against damages from animals, coppicing, thinning…
Such works leave visible traces (prunned trees, stubs, forestry machinery tracks, mulched areas…). Their presence AND the presence of materialized limits (or official forestry open SIG data) allow the area to be considered a forestry area: it is announced as a forestry area AND forestry takes place. If it is still not clear to you, could you elaborate what you feel is a problem? Penegal (talk) 15:09, 12 February 2021 (UTC)
See verifiability - in this context it does not mean that "there is official data confirming this", but "another mapper should be able to come to the same place and collect the same data ("verify" the data you have entered).". If official data is sole source of them, and it is impossible to verify it in terrain then I think that mapping them is dubious. On the other hand we map protected boundaries, low level of administrative boundaries, in some places people map parish boundaries that suffer from the same issues Mateusz Konieczny (talk) 15:25, 12 February 2021 (UTC)
@Mateusz Konieczny: the SIG would only be used for the verification of the limits of the forestry area; the fact that the area is a forestry area would need to be confirmed by physical traces or effects of forestry works. Penegal (talk) 16:02, 12 February 2021 (UTC)
Since this these tags are being placed in the boundary=* key, it would seem to me that they are able to meet the same standard of verifiability that we hold other boundaries to. And, in the case of compartments, it would seem from the description that the boundary is even more verifiable given the presence of border markers and cut lines --ZeLonewolf (talk) 06:09, 13 February 2021 (UTC)

I don't think the proposed mapping ideas based on materialized boundaries are at odds with verifiability in principle. That however is of course no guarantee that mappers would use the tags practically in a verifiable fashion. What i however object to is: Please note that this only works for delimited, verifiable areas: the boundaries must be verifiable, either on the ground or using open data - because something that is verifiable only with the help of some external data source deemed authoritative is not verifiable at all according to OpenStreetMap's concept of verifiability. External data sources are only relevant and useful in OSM as far as they represent independently verifiable information about the geographic reality. They are not considered part of the mappable geographic reality themselves. Verifiability in OSM is always local verifiability on the ground. --Imagico (talk) 12:18, 13 February 2021 (UTC)

@Imagico: I agree with what you say, which is why I precised that this would only allow to give precise limits of the forestry area, not prove that the area is indeed a forestry area; to prove that the area is indeed a forestry area, there must be traces of forestry works in it. I mean, the open data would be used to increase the precision level of the knowledge about the forestry area at a level which is sufficient to allow mapping it. I don't understand the basic difference between this usage, and using open data to tag or precise limits of, say, protected areas: often, their limits are not materialized on the ground; you may have signs explaining the protected area, but the limits themselves are often not clearly materialized and in such cases, open data are routinely used to increase precision, improve tagging… Please, what is wrong with applying this reasoning with forestry areas? Penegal (talk) 16:37, 13 February 2021 (UTC)
I am confused now - you base the proposal on materialized boundaries:
  • to standardize a mapping for forest management (forestry) areas and their management-related, physically materialized divisions
  • A wild wooded area can be considered "managed" when its limits are cleary materialized, for instance with paint, boundary stones, cutlines…
  • if the forestry area is only delimited with land lot boundary stones, without signs or paint highlighting the difference between its boundaries, as a forestry area, and a mere land lot boundary, the area should not be modelled as a forestry area, as its boundaries are not verifiable
but now you are saying but the limits themselves are often not clearly materialized and in such cases, open data are routinely used. Which is it? --Imagico (talk) 14:40, 14 February 2021 (UTC)
@Imagico: I may have not explained it clearly, or maybe got lost by the many times I explained that in different ways (note that I'm not being haughty here, I'm just trying to explain my difficulty to elaborate on this matter by the number or times my brain processed it). The point is, some may be tempted to model forestry area limits by using mere land lot/cadastral parcels boundaries. When I told in the proposal that A wild wooded area can be considered "managed" when its limits are cleary materialized, for instance with paint, boundary stones, cutlines…, I also explained, two paragraphs after, that Forestry area boundaries, when materialized, must be distinguishable of standard land lot boundaries. The materialization must be made in a way which is clearly distinct of land lots; you may find examples of what I mean here. Without this precision, we would likely have people starting to create one forestry area by wooded land lot, saying "But it has materialized borders". I don't think it's OK; in addition, when, near a real forestry area, the land lots are very small and there are many boundary stones or land lot markers, that could make the determination of the forestry area limits very difficult.
Is it clearer to you now? Again, I'm not mean, but I'm trying to explain notions I use every day in my job, so I may have trouble making it understandable to general audience, because it's crystal clear in my mind. Penegal (talk) 17:04, 14 February 2021 (UTC)
Sorry, but i still see a contradiction. On the one hand as you explained you insist on there being clearly materialized boundaries, as you say even beyond the level of standard land lot boundaries. On the other hand you write but the limits themselves are often not clearly materialized and in such cases, open data are routinely used. To me that does not match. When materialized boundaries are a prerequisite for the mapping concept you propose you cannot suggest other non-verifiable data sources instead of that. --Imagico (talk) 10:57, 15 February 2021 (UTC)
EDITED: @Imagico: ouch, it seems I lost myself in my explainations. Let's go back to the proposal: it states that materialized boundaries are enough to mark a wooded area as "managed", as a "forestry area". Please note that the materialization is not a sine qua non condition: in the case of remote or large forestry areas, the materialization may be prohibitive. In such cases, the use of open GIS data for the modelling of such areas is accepted, as long as it is verifiable and other forestry works happen in it. […] Forestry area boundaries, when materialized, must be distinguishable of standard land lot boundaries. I just re-read the proposal, and it doesn't seem to contain the contradiction you pointed out in my explainations.
The point is, specific boundary materialization can be expected whenever practical, but it may not be done, especially when impractical (notably if it is too costly for the expected advantages); in such cases, official open data can be used for finding the area boundaries. The difference between specifically materialized boundaries and open data is that the specific materialization is, by itself, technically a forestry work, and so is enough to map the area as a forestry area. On the contrary, official open data can declare a zone as a forestry area, but be wrong about it being about forestry; you may still use the open data to get the area limits, but you must, in this case, have traces of forestry work on the ground to prove that the area, whose limits the open data gave you, is indeed a forestry area and not some arbitrary area, declared as a forestry area. If the official open data describes an area saying "This area is a forestry area", and you indeed find traces of forestry works in it, then the described area is indeed a forestry area and can be mapped accordingly. You can call that a duck test, or simply an application of the Occam's razor: if it is called a forestry area, and it hosts forestry works, it must be a forestry area and should be mapped accordingly, unless it is proved it is not a forestry area. Penegal (talk) 13:16, 15 February 2021 (UTC)
Ok, i think i get it - despite parts of the proposal suggesting that materialized boundaries are a prerequisite for the suggested mapping, in substance they are not and on the ground verifiable information is only required for the existence of a forestry compartment but not for the geometry. If that is the case i am inclined to concur with Mateusz that this proposal seems to promote non-verifiable mapping. As said i think this would be avoidable by making the materialized boundaries a prerequisite for the suggested mapping concept. --Imagico (talk) 15:07, 15 February 2021 (UTC)

Limits verifiability

@Imagico and Mateusz Konieczny: I understand your reluctance about non-materialized boundaries, but I don't understand what is the problem: if there are traces of forestry works in the area which the open data describes as a forestry area, it is logical to infer that the area is indeed a forestry area with the limits given by open data. What do you think is unverifiable? The link between the traces of forestry works and the area being described by open data as a forestry area? I don't know if it is a relevant argument, but protected areas or administrative entities such as townships don't always have materialized limits on the ground, and still we map them, as we can prove they exist, and use open data to map their limits. I don't understand how this is different of the proposal. Could you rephrase what you mean, so I can again try to understand? Penegal (talk) 17:05, 15 February 2021 (UTC)

If you don't require the geometries of the forestry compartments to be defined by something observable on the ground they are not verifiable according to OpenStreetMap's paradigm of verifiability. For verifiability a mapper tasked with drawing the boundaries of a certain forestry compartment would have to come up with the same results independently (in this case in particular without having access to the same data source deemed authoritative). Or more precisely in case of geometries: Several independent determinations of the geometry by mappers have to converge to one form. If there are other features mapped in OSM that are not verifiable is not the issue here - just like any other rule and convention in OSM verifiability is not enforced. But historically proposals for non-verifiable mapping ideas have rarely been successful. --Imagico (talk) 18:57, 15 February 2021 (UTC)
In this specific case: approving mapping of forestry boundaries not existing on the ground would encourage importing forestry compartments in Poland. It is recording of land ownership that I consider as out of OSM scope. In many places it is forming ridiculous shapes of extreme density that would harm real mapping. Basically, something like across large part of forested areas (I know that importing protected areas is accepted, administrative boundaries with no traces in terrain got accepted and mapping religious administative boundaries is happening - but I would prefer to discourage that, not encourage). Mateusz Konieczny (talk) 19:39, 15 February 2021 (UTC)
In general: I am unhappy about anything that would move OSM away from original research and mapping what is on the ground into direction of importing datasets that cannot be improved, verified or used by mappers Mateusz Konieczny (talk) 19:40, 15 February 2021 (UTC)
@Imagico and Mateusz Konieczny: according to your answers, it seems you confuse the conditions for mapping a compartment and mapping a forestry area; remember that a compartment is only valid when part of a forestry area, and that mapping a compartment which is not encompassed in a forestry area is a semantical nonsense. The matter of using open data limits is only for the forestry area, not for the compartments, and that compartments have nothing to do with cadastral data. Mateusz, according to your later remarks about this, I explicitly stated in the proposal that the compartments must be materialized on the ground; regarding the forestry area, they must have visible effects on the ground, which theoritically prevent careless dataset imports. Penegal (talk) 05:49, 16 February 2021 (UTC)
What about forestry areas with known effects on the ground but with bizarre borders, specified in detail in official sources? For example main forest area, with spikes of land owned by National Forests? So forestry has known effects on the ground, exact border is not surveyable (though approximation can be mapped) and insane detail would be importable? (this is a real case) Mateusz Konieczny (talk) 17:38, 16 February 2021 (UTC)
I don't think i have confused things here. If you want to map some units of forestry management - no matter on what level - with polygon geometries you can either do so based on on-the-ground verifiable information (in this case: materialized boundaries of some form) or it is non-verifiable mapping. And above i specifically tried to confirm to what extent the materialized boundaries are a prerequisite for the proposed mapping (that is both the forestry area and the compartments) and your answer ultimately was that they are not and that authoritative data can serve as a substitute for those (which as explained in OSM is considered non-verifiable). I did not ask specifically to what extent this applies to the different components of your proposal because i think all components should insist on mandatory verifiability. Of course practically the limits of the overall forestry area are much more likely to be verifiable on the ground than the subdivisions within that area. --Imagico (talk) 18:15, 16 February 2021 (UTC)
Oh, I'm getting lost again. You two clearly have higher standards of verifiability than me, and easily loose me in argumenting over it, so let's see how the vote proceeds. I don't know if I should follow your advice on this or not, so let's let the community decide: if the lack of mandatory on-the-ground verifiability is a problem that prevents approval, I'll merge it and restart the vote. I'm sorry, but I must focus on the overall proposal, and over this issue, I'm getting again and again lost in details. Penegal (talk) 18:59, 16 February 2021 (UTC)

This does not redefine natural=wood

The proposal states:

redefine natural=wood as describing tree-covered area, whether managed or not, limiting it to the physical description of the land

I do not agree that this is changing the definition of natural=wood in any way. I would say instead something like:

Continue the current usage of natural=wood, which describes a tree-covered area, regardless of whether that area is managed or not

--ZeLonewolf (talk) 06:13, 13 February 2021 (UTC)

I also disagree with claim that it redefines this tag. Right now, both landuse=forest and natural=wood have de facto this meaning Mateusz Konieczny (talk) 07:31, 13 February 2021 (UTC)
@ZeLonewolf and Mateusz Konieczny: hmm, you are both right. I'll correct that. EDIT: I also included the deprecation of natural=wood+managed, as saying if a wooded area is managed or not is the role of the proposed boundary=forestry relation type. Penegal (talk) 07:57, 13 February 2021 (UTC)

landcover is orthogonal

landcover=trees fully orthogonal with the landuse forest <> forestry and landcover=* is itself partly caused by the fact that the key natural=* contains natural elements and others not. I advise you to avoid tackling 2 problems at the same time, hoping to solve landuse=forest will already be huge. --Marc marc (talk) 15:55, 13 February 2021 (UTC)

I think I answered you on the ML, but have trouble to understand what you try to detail here; could you explicit yourself? Penegal (talk) 18:41, 13 February 2021 (UTC)
yes you answered me on the ml (even if I don't agree with your opinion to replace a bad landuse value=* by no precise value for each of these meanings). I have posted it here simply so that it is a list of the problems discussed with your proposal, the ML (For me) is more practical to discuss, the wiki more practical to make a summary of the different points.
about landcover : a real example: soil impermeability. i don't see why we should depreciate those which give information about soil independently of the land use. they are 2 orthogonal data even if some tags describe similar things --Marc marc (talk) 14:25, 14 February 2021 (UTC)
EDITED: If I understand correctly, you're talking about the case I exposed, a swamp being also a forestry area. If so, I'm not saying that mapping the area as a swamp is incorrect, nor do I say that the area should not be also mapped as a forestry area. What I say is, these two notions are different, and current tagging practices confuses them; in this specific case, using landuse=forest to say that the swamp is a forestry area is incorrect, as landuse=forest says that it is a managed wooded land. It is, of course, but it is also a swamp, and this dual tagging (landuse=forest+natural=wetland+wetland=swamp) leads to a buggy rendering. Having a tagging scheme clearly describing forestry area as not necessarily wooded areas removes this: the area is not to be rendered as a wooded area, leaving only the swamp pattern to fill the polygon. The swamp itself (not its management status) would be mapped as a polygon (natural=wetland+wetland=swamp), and its forestry area status would be mapped as a boundary=forestry relation.
As for landuse=forest, I don't think I give "no precise value" for mapping a wooded land or a forestry area: I merely separate these notions, with natural=wood to say "This area is a mere wooded land" (but you can use another one, like above, if relevant), and boundary=forestry to say "This is a forestry area, whether a mere wooded land or not". As for "Why a boundary and not a landuse?", I already answered that some time ago, but I think I can explain myself better, as you pointed something that wasn't clear.
A forestry area is subject to specific rules: as with countries, towns, protected areas… the encompassed area is subject to specific laws, specific decision-making process, specific bosses… Maping it as a boundary seemed relevant, though your opinion is also arguable. That being said, most other reviewers did not found this choice irrelevant nor told me that landuse would be better, so I assume this is OK; that being said, I must say I quite did not understand clearly why you oppose it, so if you can explain further, I may understand and change my mind. Did I answer your questions? Penegal (talk) 16:50, 14 February 2021 (UTC)

Especially useful comments about motivation for this proposal - especially for people who did most of their mapping in a part of the world that does not have large contiguous stretches of forest Mateusz Konieczny (talk) 18:18, 13 February 2021 (UTC)

Deprecating landuse=forest

Usually deprecating tags used millions of times would be horrific task.

But here, all data consumers can be assumed to support natural=wood already that is a de facto synonym. It may be worth mentioning that such deprecation is unusually safe. Mateusz Konieczny (talk) 18:41, 13 February 2021 (UTC)

@Mateusz Konieczny: I included your comment — very relevant, by the way. Is it OK to you now? Penegal (talk) 08:35, 14 February 2021 (UTC)
Yes Mateusz Konieczny (talk) 08:53, 14 February 2021 (UTC)
Resolved: processed Mateusz Konieczny (talk) 08:53, 14 February 2021 (UTC)
It may not be a problem for data users, but many mappers use natural=wood for "a natural wood" and landuse=forest for "a managed forest", with cutouts for significant different areas. For those mappers, natural=wood is not a "de facto synonym" for landuse=forest, even if major data users do not need to change anything because they already have to handle the (now 4? or five?) different ideas about these tags. --Peter Elderson (talk) 09:35, 16 February 2021 (UTC)
@Pelderson and Mateusz Konieczny: but the thing is, when you encounter a landuse=forest entity, you can not be sure that the mapper who did it meant that it was managed. I think this is why Mateusz spoke of a "de facto synonym": whether the entity is tagged natural=wood or landuse=forest, that does not provide other information that "the area is wooded", because you don’t really know if it is managed or not, due to there being too many different uses of these tags. Many mappers think that landuse=forest means "managed", but that is still ambiguous because of the lack of consensus about this tag. Even if the proposal was about definitively stating that landuse=forest unambiguously designates a managed forest, how could you be sure that landuse=forest entities indeed designate a managed forest? Even after proposal acceptance, there would most likely be mappers to keep using landuse=forest on any wooded area, managed or not, simply because they did not know about the proposal.
That’s why the proposal deprecates it, to replace it with an unambiguous tagging scheme: if an entity is tagged landuse=forest, it is deprecated and ambiguous about management, and its management status has to be checked; if it is inside a boundary=forestry relation, its management status has been checked and the ambiguity has been removed. Penegal (talk) 10:18, 16 February 2021 (UTC)
It's not only the "its management status has to be checked". boundary=* entails the bound, and other requirements need to be examined, which is more work in terms of both quantity and quality. As another possible solution, how about landuse=forest + managed=yes to affirm this, so that you can leave it as is? ---- Kovposch (talk) 18:24, 16 February 2021 (UTC)
That managed approach has been attempted, and is now simply one of the several approaches for mapping forests. I do not want some other new approach of the natural=wood/landuse=forest dispute which would only further confuse it. Cutting through this with a new approach that deprecates one of the two tags allows that. Besides, even if these tags could be corrected and their meaning set once and for all, remember that the Wiki has no authority over tagging practices: if the proposal states that landuse=forest is now only about managed woodlands, and someone uses it, how can I be sure he meant the new definition of this tag, or just one of the old ones? Again, creating a new tag allows to be sure about that: if someone uses the new tag, he uses the new approach, and if he uses the old landuse=forest, he uses one of the several outdated approaches. Penegal (talk) 19:08, 16 February 2021 (UTC)

Status of the proposal and chances for consensus

As much as i would have liked this proposal to develop to get broad support from mappers and this way cut the gordian knot the decade long issues with wood/forest tagging in OSM i don't think it has as is the chance to achieve consensus support from the mapper community (and that is independent of what the voting will result in - which is just a narrow cutout from the overall community opinion).

The proposal tries to connect the deprecation of landuse=forest with the introduction of a new idea of more sophisticated and more clearly defined mapping of forestry in OSM. That in itself is a good approach because it shows those who see in landuse=forest subjectively a means to map forestry while objectively and on a global level the tag does not have such a meaning a way forward for mapping this. However it explicitly tries to encourage non-verifiable mapping by adopting a Wikipedia-like concept of verifiability based on reliable sources rather than on-the-ground independent verifiability. Including that in the proposal will not allow those who value verifiability in OSM to support the proposal. This is unfortunate because the proposal already shows the way how verifiability could be ensured by clearly limiting the applicability to cases with materialized boundaries of the forestry compartments.

Ultimately in finding consensus support this proposal struggles with three main conflicting ideas that are hard to reconcile:

  • The principle of verifiability in OpenStreetMap (and that means OSM verifiability and not the generic term of verifiability including verifiability a la Wikipedia)
  • The desire to include non-verifiable information in mapping in OSM (that is information that is part of the perception of the geographic reality of mappers but that is not independently verifiable)
  • Key systematics ideas (the idea that the keys used for tags in OSM carry specific meanings, that tags in their meaning have to abide by these key systematics principles and that any change in tagging - like the deprecation of landuse=forest - can only be supported if it works towards the idea of key systematics - which is not the case here because use of natural=wood for any area covered with trees is at odds with the key systematics principles)

--Imagico (talk) 11:29, 16 February 2021 (UTC)

@Imagico: it seems for now that the verifiability issues you pointed several times about area limits are precisely what rebukes people from accepting the proposal. If votes keep taking this direction, I think I’ll prematurately end voting, remove this contested point of the proposal to only allow on-the-ground verificability, and restart the vote; that should do it. --Penegal (talk) 12:34, 16 February 2021 (UTC)
Dear David, and other readers, I think you end up with the same discussion if you limit the definition to ground verifiability only. But it is not the essence of this proposal. You will always have broad discussions about verifiability and what that exactly means, it's very subjective. If we keep on blocking or rejecting proposals that are an improvement because of rather subjective principles that live in OSM we will end up with a map consisting of areas represented with nodes. We end up deprecating the whole boundary tag including administrative borders, because it all depends what is to be considered a reliable source or what is ground truth when it comes to marking borders and boundaries. Even if you limit this proposal to forestry areas verifiable with ground truth, how can you determine in the field if the markings have been putted there by a "reliable" source. Land ownership and land use are the cause of many wars and deaths every year. Especially in the 'non-western' world but even in the western world I can give you plenty of examples. Ask an Indian to which country disputed areas of Kashmir belong or should be mapped and you get a completely different answer from a Pakistani. Both refer to their governments as a reliable source, and who are we to say that they are not, both have the same voting rights, are internationally recognised and represented in the UN. In OSM we do map it as a single international single border, but I wont stop anyone to map the other parties view and leave it up to a renderer to mark it as a disputed zone. In OSM we should map all boundaries, and all versions of it, disputed or not either by a reliable or not reliable ground truth or a reliable or not data source. Indirectly we should not take is as an argument to approve or object proposals in rather subjective and sensitive matters like this. You will always have this kind of disputes, it's our nature. However one of the basic foundations of OSM is that we want to map the real world, and that includes disputed or disputable borders and boundaries. When a boundary is drawn it should be accompanied with a source, even maybe a source:type. If these source references are considered not "reliable" or to vague, not specific, it should be considered as vandalism or significant misuse of OSM that can cause potential harm to peoples lives. A proposal can contain some directions of what should be considered a reliable source or ground truth in a particular application. Don't describe what is not acceptable. Anyone can propose what is a reliable source or to be considered as ground truth and reach consensus about that in the proposal phase, but we don't give examples, describe exclusions or whatever justification to exclude something or a certain opinion. We just give guidance, if someone considers something a reliable source or ground truth, as long as it has a source reference and is not vandalism or violates any-ones culture or opinion, it's fine, do it map it. Different boundaries for the same thing, resulting from different "reliable" sources should be accepted in OSM and mapped and rendered as such. Accept that a single reliable source or reliable ground truth does not exist in our real world, and that is what we are trying to map, the real world in all it's beauty but also it's shortcomings. Anyone who has a problem with that can still mark a wiki page as non-verifiable and start a separate discussion. Bert Araali (talk) 19:36, 16 February 2021 (UTC)
Just to avoid anyone unfamiliar with the topic drawing the wrong conclusions: Verifiability and what is and what is not verifiable according to this principle is of course not subjective - that is the very core of the principle. As i mentioned above many people of course consider non-verifiable things to be part of their subjective geographic reality (or even their collective reality as a certain group of people) and as a result they want these parts also to be part of OSM. We have a long time ago decided that to deliberately not include such things in OSM because of the problems this would create, especially since as a community we do not have the means to maintain such data through egalitarian cooperation across cultural divides. This is of course not the venue to discuss this in more depth so if you want to question the principle of verifiability or its application in OSM you should do so in a more suitable place.
The purpose of me mentioning this here is not to try convincing others of the importance of this principle but to explain why this proposal turns out to be controversial. --Imagico (talk) 20:38, 16 February 2021 (UTC)
Sorry, but if you addressed me or trying to say I or others are unfamiliar with the concept of "Verifiability", I need to reply. I worked for many years in the pipeline sector. There we map very distinctively, "ground truth" and all aspects that need additional verification or are considered "not tangible" because they can't be definitively concluded from ground truth, like borders or boundaries causing the most issues, we use additional verification methods, not tangible or not verifiable in the field or on the ground, which makes these subsequent procedures by definition subjective, they depend are which process or which sources you define as acceptable for verification. The verification guideline as described in OSM maybe had the good intention to address this problem but either fails or is incomplete. If you map ground truth, if that is the only allowed objective verification criteria, you map the physically present border markings only, not the boundary or border as such, without questioning if it looks like a legally border marker or not. Don't map borders at all. If it was the intention of OSM to prevent violence, disputes for cultural or other reasons, by allowing and offering anyone to map anything then you should not map these "untangible" topics. The current situation is slowly escalating, especially in the non-western world and the rising popularity here of OSM. People with conflicting interests are starting to create there own "ground truth" and then use OSM as a justification since it is "verified" to start conflicts. Moreover, since official authorities can't keep up with the rate of change in the field and the subjective subsequent verification procedures they aplly for very good reasons. If we continue or OSM is used as a tool to escalate this kind of issues all these items should be mapped as nodes with "estimated location", estimated so very subjective. Not a good move. So we don't map them at all ? Then we stop mapping all boundaries and landuses, the map will become very empty. We stop mapping many amenities since many are operating illegally, without a license, and I am talking not only about forests, it concerns health institutions, schools, shops etc... etc... and OSM gets more and more blaimed that they are closed and demolished because an enthusiastic OSM mapper verified it in the field. We end up with an empty map and OSM becomes obsolete. Sorry for the reply here but I needed to address the issue. The number of victims is growing, encroachment, border disputes, tribe conflicts, uncontrolled growing of refugee camps etc... etc... and it needs to be addressed urgently. Should we take it to a separate discussion ? I think it would be a good idea, however I don't feel comfortable to take the initiative and it is surely not my intention to stop this proposal because of this major issue. imagico ? Bert Araali (talk) 22:26, 16 February 2021 (UTC)
Let's clear up the record here. First off, despite his professional activities, Imagico does not speak on behalf the OSM community. He speaks for himself (and perhaps his company). His subjective opinions on verifiability are his own, and are not universally shared within the community. I consider his continued insistence on redefining "verifiability" to mean "observability" to be gatekeeping behavior, and his comments here are specifically intended to bully the author into abandoning the proposal -- behavior that is at odds with what we as a community should be about. I hope we will not continue to be lectured about what "OSM is or isn't" when what is really meant is "this is my opinion". This is a topic that Imagico has written about and continues to promote falsely as a community consensus. I assure you that if such a novel interpretation of verifiability were put to a vote, you would find that many members of the community think that boundary data sourced from open databases are perfectly verifiable - the mapper simply needs to compare the open data to the OSM data. Imagico's views on verifiability are so extreme that they would exclude administrative boundaries, and when pressed on this question (as he was in the linked blog article), he refuses to answer. Why? I suspect he knows that the community would quite rightly reject as preposterous the idea that administrative boundaries don't belong. So let's drop the pretense that there is some higher OSM rule that is being violated here and stick to the merits of the topic at hand. --ZeLonewolf (talk) 12:30, 17 February 2021 (UTC)
As said i don't comment here to convince anyone of the merit of the verifiability principle. If you want to reject it that is your free choice and if you want to vote in support of a tagging proposal that suggests recording non-verifiable information likewise. The merits of my arguments and reasoning - be that on the role and significance of verifiability in OSM or on the social context of this proposal as outlined above - will not be decided by how many people like or dislike these views or engage in ad-hominem attacks on me about them. If people consider my reasoning extreme or outrageous from their subjective perspective is of no concern for me.
What i am always interested in is original thoughts and ideas how OpenStreetMap could possibly function as a social project without adhering to the verifiability principle. But this is obviously and as already said a matter for a different venue (and also so far no one seems to have voiced anything in that direction here). --Imagico (talk) 13:14, 17 February 2021 (UTC)
Re: "I consider his continued insistence on redefining "verifiability" to mean "observability" to be gatekeeping behavior, and his comments here are specifically intended to bully the author into abandoning the proposal". @ZeLonewolf: please assume good will. Re: the definition of verifiaility as a Good Practice principle in OpenStreetMap, see Verifiability - or if you prefer, see the version from mid 2009: -
"Verifiability is an important concept to OpenStreetMap. OSM data should, as far as is reasonably possible, be verifiable. This is a good practice guideline covering all mapping activity, and also by virtue of common sense, this has become a policy governing choices we make about tags to use (and which tags gain acceptance). At the core, "verifiability" is that everything you do can be demonstrated to be true or false - the latter hopefully implying that there has been a change on the ground that needs mapping. We apply this not only to the mapping data itself, but also to the way in which we record it - the tags and values we use to describe the attributes of objects on the map. From a given scenario, a tag/value combination is verifiable if and only if independent users when observing the same feature would make the same observation every time." Text from 2009, bold emphasis mine --Jeisenbe (talk) 05:24, 18 February 2021 (UTC)
Just to add my bit. I personally don't care that much if forestry area would be limited to on-ground visibility or not, but the proposal definitely does not encourage to import unverifiable data. In fact for boundaries, town/state/federal GIS databases ARE the actual objective truth. You can have stakes and markings in the terrain that could have been put there by who knows whom, but they turn into nothing as soon as official state geodetic surveyors come in and determine the boundary exactly according to their database with their precise instruments.
And second, it is hard to follow your argument why natural=wood is not following semantics when marking forests. It's been used like that for a long time. (It is anyway just a placeholder and can be called abc=xyz, important is that it is documented.) --Mashin (talk) 20:13, 17 February 2021 (UTC)
This is a common misconception. It seems that many modern people now believe that government data is more reliable than what is found in the real world. If you are concerned with legal land ownership this might be true. But if you are concerned with making a current map of the real world, it is a mistake to think that government data is better than observations. In my experience in the USA and Indonesia, government datasets are often wrong when they represent features that can be seen in the real world.
More importantly, if some feature can only be said to exist by consulting a databse, then OpenStreetMap can never hope to improve on the data that is already in that database. We do a dis-service to database users by offering data that will always be worse that the official dataset (since anyone can accidentally or maliciously change it), and we waste mapper time by asking them to import or duplicate existing datasets. That is why we should encourage the community to focus on mapping things where we can add value: that is, real, current things which can be observed. For this reason we don't map historical events or vanished historical places, we don't map subjective ratings or opinions, and we should not map government or private boundaries which do not exist in some physical form in the real world. --Jeisenbe (talk) 05:30, 18 February 2021 (UTC)
What are you even talking about when you say "we waste mapper time by asking them to import or duplicate existing datasets"? We import data from public sources into OSM all the time! That is the whole point of OSM to collect data in one place instead it be spread in thousands of databases.
Is some cases state databases might not be up to date, in that case don't use the data. But the government data are imperative and you can argue that this lands is for wood production because there are signs as much as you want. The reality is that if cadaster says that land belongs to someone else then that is the objective truth, because that is the official record of ownership and no on-ground survey is going to surpass that. And the same is for country/town boundaries (which we DO map), town names or any non physical property. --Mashin (talk) 20:57, 18 February 2021 (UTC)
Re: "That is the whole point of OSM to collect data in one place instead it be spread in thousands of databases." I disagree. OpenStreetMap started as a project to create a free, open source map of the world by volunteers mappers: see the About page "OpenStreetMap is a free, editable map of the whole world that is being built by volunteers largely from scratch and released with an open-content license" ... "We are getting out from behind our computer screens, and surveying towns and countryside to create our maps. This is very much a community mapping effort."
Also see "How we map": "OpenStreetMap is a social activity: it is a teamwork effort by hundreds of thousands of people around the globe." "Contributions to OpenStreetMap should be: Truthful - means that you cannot contribute something you have invented. Legal - means that you don't copy copyrighted data without permission. Verifiable - means that others can go there and see for themselves if your data is correct. Relevant - means that you have to use tags that make clear to others how to re-use the data. When in doubt, also consider the "on the ground rule": map the world as it can be observed by someone physically there." - Mapping real features, surveying in person and making sure that other mappers can confirm and improve your data is a basic part of how this global community works.
Importing outside datasets for the convenience of database users is not the purpose of this commmunity. --Jeisenbe (talk) 23:54, 18 February 2021 (UTC)
Also see Import/Guidelines#Use_the_right_tags "Your import should use tags which are familiar to the OSM community, rather than inventing its own set of tags." ... "OSM is only interested in what is verifiable." --Jeisenbe (talk) 23:59, 18 February 2021 (UTC)
"That is the whole point of OSM to collect data in one place instead it be spread in thousands of databases." - not at all. OSM is not a free database hosting. Only tiny subsection of all geodatabases is worth importing into OSM (lets skip licensing for now). If data cannot be improved or verified after importing it into OSM, then importing it was a bad idea in the first place. Mateusz Konieczny (talk) 04:21, 19 February 2021 (UTC)
"The reality is that if cadaster says that land belongs to someone else then that is the objective truth" - I assure you that reality is far more complex. At least in my country legal proceedings to establish who actually owns land are normal and cadaster sometimes has mistakes. (and in addition, land ownership is not something mappable in OSM anyway) Mateusz Konieczny (talk) 04:23, 19 February 2021 (UTC)
"In my experience in the USA and Indonesia, government datasets are often wrong when they represent features that can be seen in the real world." - I can confirm that for Poland. Even for things that are created and setup by government itself! Reporting invalid addresses in official databases is part of making imports from the official address databases. One of the more interesting uses for OpenStreetMap data was using it to supplement official building dataset for making flood preparation plans in Silesia. Official dataset was not including illegally constructed buildings (it was based on building permission paperwork, and you are not going to tell people during flood "you are living in illegally constructed building, we will ignore that you will drown"). (official dataset was also not including recently constructed buildings, as it was acceptable to update with 5 year delay...) Mateusz Konieczny (talk) 04:31, 19 February 2021 (UTC)


A minor issue, I noticed this is planned to be introduced. Can you not use an existing tag such as official_ref=*? Or can it not be eg forestry_comapartment:ref=* to be consistent? ---- Kovposch (talk) 17:41, 16 February 2021 (UTC)

Because compartment relations already exist and can include any *_ref tag. Using an new one allows the migration of the ref of already mapped compartment using a simple logic: if the new ref tag is present, then the compartment follows the proposal terms; if not, then the data is missing, or in an old format; either way, it should be corrected if possible. As for the tag name, I chose a simple compartment_ref because it is simpler to read and understand than using another namespace as forestry_compartment:ref=*. Penegal (talk) 18:52, 16 February 2021 (UTC)

Uncertain, imprecise, or partially observed enclosures

Originally, I thought this proposal is only about introducing boundary=forestry, so I agreed with it without a doubt. Upon reading it closely, I feel reluctant to use boundary=* when I can't draw an exact bound yet (this is the lesser reason I unfortunately want to vote no to the proposal as is). I would love to see landuse=forest remain for areas which don't match the standard and definition of boundary=* yet, but still fits the meaning of boundary=forestry otherwise. If the fragmentation and diversity in landuse=forest concerns you, could we add a new value, for example landuse=forestry? ---- Kovposch (talk) 18:13, 16 February 2021 (UTC)

I don't think that this precision problem is specific to the proposal: if you use landuse=forest for a managed wooded area and natural=wood for an unmanaged adjacent wooded area, if you have no materialized limit, how do you draw the polygon? Penegal (talk) 18:31, 16 February 2021 (UTC)