Proposal talk:Boundary=forestry( compartment) relations

First voting: rejected

As the first vote quickly showed that the proposal, in the state it had at the time, could not reach consensus, I prematurately closed the vote and contacted some contributors to understand exactly why no consensus could be found. The issues will be corrected in the following days, and hopefully allow to restart the vote. As I resetted the voting, here come the votes of this first try for the record.

Voting closed

Voting on this proposal has been closed.

It was rejected with 7 votes for, 8 votes against and 2 abstentions.

Remaining issues prevented consensus; the proposal will be reworked and, if possible, resubmitted to the vote.

• I oppose this proposal. Unfortunately I have to recommend that this proposal not be approved. The current proposal allows mapping non-verifiable areas as boundary=forestry based an import from an external database or the mapper’s own opinion, even when there is nothing marked on the ground which shows that the areas is as forestry feature. I would support this proposal if it were changed to be limited to areas which are clearly marked or signed. --Jeisenbe (talk) 06:34, 16 February 2021 (UTC)

• I have comments but abstain from voting on this proposal. I think the consequences of deprecation of landuse=forest and the choice for natural=wood to have only the meaning "covered by trees" have not been properly addressed. Natural=wood and landuse=forest are very widely used with a different meaning. To the bulk of the OSM-world, this proposal is yet another idea about forests and trees, coined by a relatively small group of mappers. I think you need to present an implementation plan for a tagging change of this magnitude. --Peter Elderson (talk) 08:06, 16 February 2021 (UTC)

"implementation plan for a tagging change of this magnitude" can you be more exact what you mean by that? Mateusz Konieczny (talk) 08:55, 16 February 2021 (UTC)

I agree that an "implementation plan" is required here. Without wanting to "put words in the mouth of" (speak in replacement for) Peter Elderson, I believe a table-oriented "old tags" -> "new tags" including wood, forest, landcover and related tags (even if the results are "no changes") is what Peter Elderson has in mind. This would truly help my understanding of what is expected of Contributors / authors going forward should this become Approved. Stevea (talk) 20:52, 16 February 2021 (UTC)

What I mean is: how will you go about making sure the suggested tagging, including deprecations of tags, will be adopted by mappers, renderers and data users, and how will the backlog of possibly different meanings be tackled? Because without that, all the history remains, mappers will stilll use the old ways and plans, and there will just be another scheme on top of the 4 or 5 already there. If there is no such plan and the idea is to let the mappers and data users sort it out, I think this proposal should simply not talk about meaning of landuse/natural/landcover tags, deprecation and such. NB. A common aspect "there are trees on this area" does not mean it's all the same thing. --Peter Elderson (talk) 09:07, 17 February 2021 (UTC)

• I oppose this proposal. although I agree with the basic statement (landuse=forest should be depreciated in favor of unambiguous tags), this proposal introduces more inconsistency than it solves. for example, landuse=* of a dedicated forestry area should simply be landuse=forestry (keep as simple as possible !), in the same way that landuse=residential and not boundary=residential. landcover=* is also quite orthogonal to forestry. Marc marc (talk) 08:30, 16 February 2021 (UTC)

• I oppose this proposal. There is an ongoing discussion on talk page that is about major problem that I see with that proposal (I will answer there!) - I worry about importing unverifiable boundary data. In case of Poland it would be basically land ownership registry getting imported as boundary=forestry. I have no trouble with landuse=forest deprecation. I disagree with "Natural=wood and landuse=forest are very widely used with a different meaning", typically meaning is the same and if meaning is different - various mappers use different distinctions, making any differences inconsistent and not usable anyway. I really hope that you will not be scared away by results of this vote Mateusz Konieczny (talk) 08:53, 16 February 2021 (UTC)

• I approve this proposal. I see this as no different than protected areas and administrative boundaries - both are "verifiable" from government maps and GIS systems and provide immense value to the OSM community. This will prevent numerous forestry areas from being mis-tagged as protected areas, as these are not equivalent to a bird sanctuary or national park. I support creating a richer language for mappers needing to tag these areas which are not adequately described by existing tagging. In addition, I strongly support the deprecation of landuse=forest and landcover=trees as synonyms for natural=wood. I thank David for pouring his expertise into this project and giving OSM a useful set of guidelines for finally untangling forests and forestry areas! --ZeLonewolf (talk) 12:24, 16 February 2021 (UTC)

The difference is the corresponding place=* (or leisure=*) isn't deprecated in favor of boundary=*. They are different concepts. I don't agree this is "creating a richer language" when its deprecates a landcover=* (as well as landuse=*) value either, when this is an issue at large. The only problem that needs to be solved is landuse=forest being used for tree cover, and this proposal can already improve the scene by providing a tag for a additional specification. ---- Kovposch (talk) 17:28, 16 February 2021 (UTC)

• I have comments but abstain from voting on this proposal. The discussion is still ongoing and especially the boundary issue needs to be clarified, in order to be viable as a replacement for landuse=forest. I am in support of the proposal that boundary=* is the best approach. I strongly support the concept and that we should deprecate landuse=forest. This proposal is also of very high quality and crafted by an expert in the matter, it deserves a chance and due to it's high expertise support maybe a "once in a lifetime" opportunity for our OSM community to guide us to better mapping practices. If not for deprecating landuse=forest, then to lay a good and lasting foundation for decent forestry mapping. Refer to the ongoing discussion on the talk page for details. --Bert Araali (talk) 12:35, 16 February 2021 (UTC)

• I approve this proposal. The proposal clearly says that only boundaries that are verifiable on the ground or via GIS database are to be mapped, which is exactly as any other boundaries that we have in OSM. So I don't quite understand the confusion here. --Mashin (talk) 14:49, 16 February 2021 (UTC)

• I approve this proposal. This proposal would make it possible to properly clean up tons of messy forests I've come across in a really precise way. In terms of the information being verifiable, I see know difference between this and, for example, a city park. Both are outlines of the property, and yet we don't have massive verification issues with parks. We just map them as well as we can. This proposal will allow me to really cleanly map managed forests in my area where the borders are usually very clear. --LeifRasmussen (talk) 16:10, 16 February 2021 (UTC)

• I oppose this proposal. I don't see why you need to touch landcover=trees. That's a bigger issue with natural=* overall. Secondarily, landuse=forest is again not the same as a certain specific and defined boundary=forestry. For comparison, we don't need to deprecate landuse=residential, landuse=commercial etc if the place=plot or place=city_block (theoretically would be a kind of boundary=* in the first place) is available either. As an example of a functional feature, man_made=works would still exists together with landuse=industrial. You only really need to add your forestry boundary=*, clarify the definition of landuse=forest, and deprecate other usage.---- Kovposch (talk) 17:16, 16 February 2021 (UTC)
• I approve this proposal. I support this proposal. Forestry management boundaries are sufficiently orthogonal to land-cover and land-use to warrant this new approach to tagging. Boundaries say "certain rules apply within" rather than implying that certain activity or ground-cover exists evenly throughout the area as landuse=*/landcover=*/natural=* tags imply. Because so many mappers use both landuse=forest and natural=wood to mean "there are trees here", landuse=forest can never be used to accurately model forestry management boundaries. This proposal does a good job of unsticking this problem by introducing verifiable boundaries in a way that allows landcover/natural/landuse tagging to more accurately model the other types of features on the ground. --Adamfranco (talk) 19:16, 16 February 2021 (UTC)
• I approve this proposal. A much-needed cleanup of forest tagging. I have doubts about messing with landcover=trees though. Rayleigh1 (talk) 19:27, 16 February 2021 (UTC)

• I oppose this proposal. I agree that a great deal of effort has gone into this proposal, for good purposes (to "better delineate" forests in certain parts of the world where this is necessary or even critical to map), by highly qualified people who have truly listened and responded to input (criticism, solid suggestions...), making numerous improvements as this has unfolded. (I profusely thank David for his efforts). However, similar to previous comments in this vote, I believe the vote was called for too early, before full, wide understanding of how this proposal would affect existing tagging — especially of landuse=forest and landuse=wood, which remain exceedingly confused in ways which are difficult to even express, let alone "capture" or clear up. I'm all for a deprecation of confusing tags, and while this proposal does seem to go a great distance in that direction, it doesn't complete the task (for me, and apparently others). While as ZeLonewolf states, "this will prevent numerous forestry areas from being mis-tagged as protected areas," two things emerge in my mind: 1) I don't know how much this particular kind of mis-tagging occurs (it might be a minor problem, I doubt it is major), and 2) if there is such mis-tagging, part (most?) of the blame for that is on the author of the error. Yes, "better tagging schemes" (what this proposes to be) can help such mis-tagging, reducing or eliminating the root cause of it (poor specification of syntax to achieve desired semantics), so I continue support both refinement to this, as well as its inclusion in a "grander scheme" to better clarify tagging of forests and woods to reduce current confusion extant between them. We shouldn't be multiplying confusion, creating "confusion-squared," although this proposal's thrust to deprecate landuse=forest is appreciated as a step in the right direction (though, "standardization" of what we mean by landuse=forest and natural=wood could achieve the same thing). Additionally (and unfortunately), the (contentious? not-very-widely-used?) tag of landcover=* as a specific replacement for tags like natural=* which attempt to alleviate confusion, actually creates more confusion. (By "confusion" here, I mean lack of knowledge of what best tagging strategies might be because of ambiguity or lack of clarity between "competing" tags to achieve similar or identical semantics). THEN there are the rendering issues, which I have already mentioned are not necessarily a component of a tagging proposal, but which do play a part in why they come into existence (let's be candid about that). In short (too late), "good work so far, but incomplete in how it will FULLY address existing tagging (practices), at least in my mind, and put to a vote far too early." There is no rush to complete this: it is far, far more important to get clarity of understanding (especially about what happens with existing tagging) than it is to get this done "sooner" rather than later. Later is fine, as long as we have clarity of understanding that leads to consensus. Unfortunately, we aren't there yet.

Additionally (I must say it, though I'm not sure this is the best place), if we could "standardize" on landuse=forest meaning (roughly) "managed area for forestry" and natural=wood meaning (roughly) "trees are here, with no specification as to whether they are or aren't managed," I think we might not need to deprecate these, avoiding a great deal of necessary retagging. Yes, there would need to be re-tagging of those areas tagged with those two tags which do NOT meet those definitions, that is a smaller task. --Stevea (talk) 19:51, 16 February 2021 (UTC)

• I approve this proposal. I second Mashin's reasoning, and I think this proposal is a much-needed standardization of forestry tagging. --Pwhite119 (talk) 19:55, 16 February 2021 (UTC)

• I oppose this proposal. I don't think this should be a boundary tag. The overlap between protected_area and the proposed forestry tag is confusing. I would be in favor of clarifying that landuse=forest is for managed forest. I could be in favor of landuse=forestry, and deprecating landuse=forest. --Bradrh (talk) 23:55, 16 February 2021 (UTC)
• I approve this proposal. --Carnildo (talk) 07:49, 17 February 2021 (UTC)
• I oppose this proposal. The basic suggestion, to have forestry compartments in the data, is fine. The side effects suggested, like deprecating landcover=trees, is something I oppose. --Mueschel (talk) 13:32, 17 February 2021 (UTC)
• I oppose this proposal. deprecating landcover=trees, is also something I oppose Adavidson (talk) 10:11, 18 February 2021 (UTC)

A boundary does not make trees a managed forest

You write: 'A wild wooded area can be considered "managed" when its limits are clearly materialized, for instance with paint, boundary markers, or cutlines. Marking boundaries, even simplistically, is the very first act for forest management:'

Not always. There is a gorge a few miles from me. Trees grow on the slopes of the gorge, which are too steep to permit forestry. Sheep farmers adjacent to the trees have put up fences to stop their sheep getting into difficulties on the slopes and being hard to recover. There is a clear boundary, but it is "My agricultural land stops here" rather than "Managed woodland starts here."

--Brian de Ford (talk) 14:57, 17 March 2021 (UTC)

Surely not EVERY boundary makes a managed forest, but the other way works fine: a managed forest does have some sort of boundary - whether it be man-made (fence, markers), natural (gorge, river) or just purely administrative (area on land use maps in a town hall somewhere). As opposed to an unmanaged clump of trees. - SinusPi (talk) 15:55, 17 March 2021 (UTC)

@Brian de Ford: please also note that the whole boundary of the area must be materialized. It is not necessarily the case in your example. In addition, should a mapper ask himself if he should map the area as a forestry area, it is up to him to decide if, according to the local laws and customs, what he sees is a legit way to mark a forestry area. Based on what you say, it won't be the case in your example, as barbed wire are typically not used for forestry area boundaries. Penegal (talk) 17:27, 17 March 2021 (UTC)

Constrain the meaning to having an official boundary, regardless of use or features.

My take on this tag would be that its meaning should only refer to whether the forest area is in some capacity official (someone owns it, someone manages it, it's marked on official maps, maybe there are fences, markers, plaques, etc.) - regardless of whether it's being actively managed, logged, hunted in, or used for leisure... or whether it's actually wooded. All those features should be assigned (or not) by way of additional tags, so as not to "bloat" the meaning of the boundary tag. Hence, I'd:

• REMOVE part of the Proposal section that reads: "which has at least one of the following characteristics: (.....)".

- SinusPi (talk) 16:00, 17 March 2021 (UTC)

@SinusPi: forestry is a complex subject. I agree with you when you state that there can be many operations in it, but, in my experience, I cannot expect general audience to be able to distinguish these operations, you need an expert eye for that. Please also note that, below your suggested deletion, the proposal state that materialized, unambiguous border marking of a wooded area is enough to consider it as a forestry area, as it explicits a will to manage the area and is typically the very first step to use forestry in it. Penegal (talk) 17:33, 17 March 2021 (UTC)

@Penegal: The "general audience" probably shouldn't really mark "forestry" areas without really knowing who manages it and why - they should rather stick to adding "forest-like" areas they see, without having to worry about administrative aspects. Note that, with this proposal accepted, we'd finally move towards having rather rare "true forests", with common "foresty" areas just being natural=wood.

As for the deletion - sure, a boundary may be enough to assume a "forest", that's why I suggested to remove the requirement of "logged or managed": one might want to clearly mark boundaries but keep the forest pristine, never touch a single fallen log (with the purpose of creating a weald, perhaps). The important part would be that there IS an owner or manager. Also, we might want to keep in mind that a boundary could be assumed (the area is clearly defined by surrounding roads, or "everyone knows").

@SinusPi: you are right, mappers are supposed to know what they are doing when creating entities using the proposed tagging. That being said, the proposal merely give the possibility to unambiguously state that an area is managed through forestry; it is not mandatory, and people should not use it if they are unable to tell if the area is managed or not; that is the case with current practices and will remain. As for the section you proposed to edit, it is merely a summary of the proposed changes, not the definitive definition of a forestry area which is below this section. The possibility of simply using materialized boundaries, IMO, needs too long explainations to fit in this summary, which is why this possibility is mentioned further down. Penegal (talk) 18:54, 17 March 2021 (UTC)

Transition Of Existing Tagging

I think you need at least another line in the table under 'Transition of existing tagging' for some National Forests. Current Tagging: boundary:protected area; New Tagging: boundary:forestry (in some cases). --Bradrh (talk) 03:45, 18 March 2021 (UTC)

@Bradrh: I'm unsure about that; if you're talking about US National Forests, it seems that they will virtually never be themselves forestry areas: only parts of them would be eligible. In this case, mentioning such virtually inexistant case could be interpreted as a need to retag National Forests, where there would be no such need, and that would only create stir as this did before the first vote. That being said, if you're talking about National Forests in other countries, which would be themselves considered, as a whole, as a forestry area, then I would happily give them as a tagging transition example. Penegal (talk) 07:40, 18 March 2021 (UTC)

header image

"A forestry compartment within a forestry area." description in https://wiki.openstreetmap.org/wiki/Proposed_features/boundary%3Dforestry(_compartment)_relations#Proposal - is it intended to suggest that cut boundary is also boundary of a forest compartment? Mateusz Konieczny (talk) 08:06, 7 April 2021 (UTC)

@Mateusz Konieczny: in many cases, yes, compartment boundaries are marked with cutlines, but cutlines may serve other purposes; that is why you can see on the picture cutlines drawing wood stripes. Along a forestry boundary, the cutline just helps to see the forestry boundary marks (paint, signs…), which themselves are the forestry boundary references, but boundary cutlines are only good practice, nothing mandatory. Penegal (talk) 10:10, 7 April 2021 (UTC)

Country specific

Hello!

It seems this time proposal is going to be rejected - as I can see, most "no" votes are from German mappers. As I understand, they try to match proposal content with their local forestry practice and it doesn't match very well. On the other hand, many mappers for others countries ready to fully support proposal (for example all mappers from Russia, whom I know from telegram chat, voted "yes", because our forestry management ideally matches boundary scheme).

May be it makes sense in the next iteration try to make proposal not global, but "almost global"? And directly point that applying new tagging scheme for Germany (for example) is out of scope. Probably, that it will be very hard to combine all over world forestry details in one universal scheme, because forestry management is highly dependent on local laws and regulations. But as for me - if we can cover 90% of cases with common scheme (and continue discussing for other 10% further), it will be a good point to improve current situation with "6 ways to tag forests".

AnakinNN (talk) 09:30, 12 April 2021 (UTC)

@AnakinNN: we can agree on the fact that the german community did not receive well the proposal. Part of the problem seems to be that the proposal title leaded them to discard the proposal as "something that doesn't concern me", until they understood that it implied the deprecation of landuse=forest.

Discussion in their community indicate that, though there is agreement on the problems with landuse=forest, the proposal mixes "what is managed through forestry" and "what is called AND physically delimitated as a forest", which they consider bad. These are neighbour issues, but it seems that mixing them in a single tagging scheme is an error in their opinion.

Note that this issue is likely more meaningful in countries like Germany or France (my case), which are overall densely populated and where significant portion of forests have physical boundaries. Currently, in the event of a vote failure, I'm considering the possibility of splitting the proposal in two different parts: 1) one regarding what is declared and delimitated on the ground as a forest, whether managed or not, and 2) one regarding the landuse=forest/natural=wood. The simple fact of splitting the proposal would ease consensus on each part, as people tend to vote no even if they only disagree with 10% of the proposal. I'm thinking, I'm thinking… But I'll probably need some insight on national forestry practices, so if you can provide some insight or know someone I could ask and would volunteer to help me on this issue (English would be mandatory, as I don't speak Russian), it could prove useful for a potential future rewrite. Penegal (talk) 10:28, 12 April 2021 (UTC)

The thread on the German forum is light on valid concerns. But there are. I can speak a little about how this is in Austria: The owners of some piece of land do not just own some indifferent piece, but they own farmland, land that can be used for construction, and also a wood. When they own a wood, that means they cannot just do with the land as they please, they have to meet some obligations and have to consult with the local authority on some of their doings and not-doings. As the lots are typically small and some of the owners either not farmers or living a distance away, some lots can decay quite horribly. Too small to call a forestry team for a single unit, the commune must consult with 50 or so landowners to afford to book a logger. Occasionally close farmers get some firewood out. Still the blazes on the trees, that point to the stone where the boundary runs, in recent years received quite some polish. These are on the ground features, but actually the do not say anything about forestry, they only mark the land register. On the ground a scrub can be legally a wood, no way to tell from mere looking at it. How many owners will join the next logging campaign, nobody knows in advance. I must confess, I have no idea how to map that to boundary=forestry. This from a layman's perspective. --Hungerburg (talk) 21:05, 12 April 2021 (UTC)

My take, if you see trees then map natural=wood, do not use landuse=forest. Looking a a bit of land that has trees does not mean it is used to produce timber or anything else. In some parts of Australia farmers primary use of trees is as shelter for both stock and crops. It requires more than seeing trees to know that productive forestry takes place. If that cannot be determined (Germany or elsewhere) then don't use a forestry tag, just map the presence of trees. I agree that some don't take a world view. Warin61 (talk) 08:38, 13 April 2021 (UTC)

In Australia there are 2 types of forestry areas; one grows soft wood - usually foreign pine, the other grows native hardwoods. The softwoods are easily to pick out on imagery provided you don't mix them up with dense orchards. Hardwoods being natural here means the boundaries may not be picked out from imagery. In some places OSM has legal access to these boundaries, so these are easy to map. Those without OSM access to these legal boundaries suffer the same problems as National Park boundaries where these too are not OSM legally accessible. Very hard to see a boundary on hte ground where there are trees on both sides and no track/path to divide them. Warin61 (talk) 08:38, 13 April 2021 (UTC)

Forestry practices in Australia vary from clear felling (usually on the soft woods) to selective felling (usually on the hard woods). Felling has various limits; if the ground slope is high then no felling at all, if some threatened native animal resides in some are then no felling there and so on. Warin61 (talk) 08:38, 13 April 2021 (UTC)

Well Warin, I sure like your take on this, that is why I voted yes, because I want to tag scrubs as scrubs, and not as forest, just because in some land register the piece of land is designated forest. For example, scrubs vs. woods helps when cross hiking, to tell where passing will be a pain or easy. I hope Penegal has the drive to propose something that laymen can apply (keep natural=wood, deprecate landuse=forestry) and find a scheme (boundary=forestry) for the economic/adimistrative stuff that accommodates mappers all over the world, even those from Germany. PS. with scrubs in this case I mean clearcut covered with blackberries, I think that is not too far fetched. --Hungerburg (talk) 20:35, 13 April 2021 (UTC)

Tagging compartments without forestry_area

From a mapper's point of view it is easier to map just the compartments (at least in regions where compartments are better signposted than the whole forest) because they are smaller and have simpler boundary geometry than the forestry_areas. (And then leave the mapping of forestry_area to someone who has a better overview of the region.) It that ok according to this proposal? --Lkw (talk) 20:59, 16 April 2021 (UTC)

@Lkw:: it's a case I did not take into account, but the proposal does not explicitly forbid it. It describes compartments as relative to a forestry area but does not mandate the forestry area itself to be modelled. So yes, your case is technically allowed by the current proposal. Penegal (talk) 06:21, 17 April 2021 (UTC)

Rendering of forests vs. woods

An issue raised on the OSM-Carto tracker, to show forests, which may encompass heaths, screes, wastelands, &c differently from areas, where there are tress, a.k.a woods, was turned down quickly, with the reasoning, both get used just the same by producers, so consumers must not make a distinction. --Hungerburg (talk) 22:28, 27 April 2021 (UTC)

It would be helpful to link that issue, for reference. --ZeLonewolf (talk) 22:50, 27 April 2021 (UTC)

It is https://github.com/gravitystorm/openstreetmap-carto/issues/4393 - I did open the issue, and I just gave a thumbs up to the explanation for rejection - after all, it spells out, what I observe too. At least, I learned, that in the past, forest and wood were rendered differently, but that got dropped years ago for the very same reasons. Still pondering, what the many "forestry is landuse" votes want to express: That they only want to have mapped those areas under forestal administration, where there are actually trees, a.k.a managed wood? Or is it, that they just don't like boundaries? --Hungerburg (talk) 07:36, 28 April 2021 (UTC)