natural=wood and landuse=forest should render using the same style #647
Comments
|
Aren't they two different tags for a reason? |
|
It is an example of broken tagging scheme, everybody has his/her own definition. There is no point in differentiating between these two tags. |
|
I know the arguments. My point was people use these two different tags to mean something (to them). There may not be consensus on what that means. Making the rendering the same would be similar to saying "the two tags mean the same thing and my opinion counts more as I'm in control of the render rules". My preference would be to push for more widely accepted tagging guidance. |
|
2014-06-20 1:24 GMT+02:00 RobJN [email protected]:
+1. My guess is with rendering support for landcover=trees we could --> should go to [tagging] |
|
As confused as the tagging may be (even though my definition is the best and will eventually win :p), it tries to distinguish cases which are IMHO important to distinguish, and should be rendered differently (if only to help mappers who care to spot inconsistently-mapped areas). There may be a lot of tagging arguments, but nearly all of them look at the "amount of management" to decide between forest and wood, so at least we're arguing around the same criteria. That said, I agree that the various renderings of natural=wood, landuse=forest, landcover=trees, natural=tree, natural=tree_row, and natural=scrub are too different. I'd like to see all of these share the same tint of green for the area (except for natural=tree which could do without an area) plus some tree pictograms, varying in density (sparse for scrub, exactly-aligned for forest, etc) and perhaps kind (deciduous/evergreen/etc) if we have the info. |
|
2014-06-23 14:58 GMT+02:00 vincentdephily [email protected]:
no, the usual distinction is size (woods=small, forest=large) and density Besides this, there is also argumentation with the keys: landuse=how the |
Not in my experience! But in any case, this is something well beyond the scope of the issue tracker. I am happy to render them differently, I'm open to changing the rendering of each, and I'd encourage everyone to get together (elsewhere) to nail down the differences and come up with clear documentation across the wiki, editor and elsewhere. |
|
2014-06-23 15:07 GMT+02:00 Andy Allan [email protected]:
probably have been unclear, this was deliberately refering to common |
|
On this note, does the leaf_type render still? http://wiki.openstreetmap.org/wiki/Tag:landuse%3Dforest#Rendering |
|
I would also be in favour of rendering them the same, as long as there is no meaningful difference in semantics. |
|
@math1985 To many people there is a meaningful difference (the problem is that each country seems to have their own tagging rules). I will raise on tagging mailing list. |
|
I would agree with the OP, natural=wood and landuse=forest are indistinguishable across the world: they may mean specific things to each user but one cannot do anything with that as a data consumer. Ideally their use would be orthogonal (allowing landuse=forest for new plantations or felled areas) and natural=wood for anywhere with trees. In practice using a single rendering style would mute the problem and would allow OSM to move on. We have a range of newer tags, such as leaf_type which aim to move away from the semantic confusion associated with the wood/forest dichotomy (which AFAIK is partially a UK-specific thing). In particular we would avoid the default of rendering forest with coniferous symbology which looks absurd in many places. This problem has existed for the whole time I've been contributing and I see no prospect of solving the semantic confusion: at least the rendering can avoid adding to it. Support for leaf-type and/or wood=broadleaved would improve matters considerably in the right direction. |
|
The definitions natural=wood and landuse=forest greatly overlap, and would have been better treated with tags such as managed=* The original tag to distinguish between different vegetation types (wood=) was too generic for many types of maps resulting in either use of wrong (often generic) values, or creation of non-supported values. I understand that the new tagging scheme with leaf_type= and leaf_cycle=* is meant to clear up this in a scientific correct way, and to avoid disambiguate. For the general (road) map, there is no need to specify tree-types, but for specialist maps, such as nautical and aviation maps, this distinction can be of great use to the navigator, therefor having a tagging scheme giving a consensus in how to tag the difference between deciduous wood that shed their leaves and evergreen deciduous, or even areas with eucolypt wood. In some countries such as Norway, distinguishing between coniferous and deciduous forest on general maps is common, while most other countries I know don't do that type of specific rendering. It certainly doesn't make sense rendering natural=wood and landuse=forest differently, in my opinion, landuse=forest should be replaced with natural=wood + managed=yes |
|
The orthogonality of natural=wood and landuse=forest @SK53 pointed out could be emphasized in rendering by drawing only natural=wood as a solid color area and distinguish landuse=forest with a different overlay pattern indicating the use for forestry (some trees+piles of logs symbolism maybe) |
I agree that orthogonal usage would be preferable. Natural=wood (or forest) could be used for named entities (could also be nested), landuse for landuse (I.e. land where trees are grown, this would include also forest areas which have recently been cut down but where new trees will grow in the future), and landcover=trees for all areas where trees are growing. The distinction of wood and forest could be about size/density (maybe woodland would be better than wood then) |
This would be a good idea for a new tag, but not for one with established use and used more than two million times. Orthogonal usage is a good idea, but now it is too late to introduce it. People that want to tag whatever forest is managed should invent new tag and stop pretending that landuse=forest and natural=wood have any meaningful difference in OSM data. |
|
@mkoniecz Broadly I agree with this position. Recovering landuse=forest to mean that timber (trees) & wood (small tree parts) are harvested from a given area is now too difficult. The problem is that landuse is the obvious key to use (personally I would have named all the landuse values as activities not nouns: something like forestry, farming, selling etc, in order to make the distinction clear) and landuse=forestry the obvious value. Will think more about this as some kind of migration/mitigation strategy needed too. |
|
@mkoniecz - changing the meaning of widely used tags is problematic, i agree here. I have no problem with using a new tag for this distinction, this would however be pointless without clear documentation about its purpose, clarification that natural=wood and landuse=forest are synonymous and rendering the new tag from the start. In general we have to be bold to achieve even a minimal level of consistency. OSM has grown so large meanwhile that there is a very high inertia making even sensible changes quite difficult. In this case rendering natural=wood and landuse=forest identically would hamper the introduction of a new tag i think since two widely used tags with a clearly implied difference in meaning will make many mappers assume a distinction can be made by distinct use of these tags and no other tag is needed to make this distinction even if they are rendered identically. The bold move would be introducing a new tag and stopping rendering landuse=forest when starting to render the new tag. Note the situation with use of landuse=forest is not as bad as the bare numbers might imply, the vast majority of features with that tag are in central Europe and the US and the majority of those actually serve a forestry purpose to some extent. In terms of hypothetical bot edits - automatically changing all landuse=forest to natural=wood would probably do a lot more damage (in terms of loss of meaningful data) than automatically changing all landuse=forest to a strictly defined landuse=forestry/managed=yes (in terms of incorrect data). Still neither of these would be practical of course. |
Landuse in geography usually refers to human landuse, so this is not a good choice for a feature which also occurs without any human intervention like forests and other woody areas |
|
@dieterdreist: I am referring to the human abstraction of timber from wooded areas usually known as forestry & certainly a landuse! |
I am aware of this, but then we'd still be keeping at lot of emphasis on the forestry / management aspect for a feature (forest) which does occur also by nature. My plea is to focus on the trees and put the human intervention into a subtag like "managed". Another important aspect is that of named entities. Typically forests have names, and several smaller pieces (with names) form bigger forest entities (again with names) and so on (nesting). This is something that should be catered for as well, but landuse is not very suitable for nested areas, while "natural" IMHO is. |
|
@dieterdreist you have to remember that in Britain virtually all forestry is carried out in plantations. This is why the issue arose in the first place: why wood & forest existed. Forestry, Agriculture, Fishing & Mineral Extraction are very old human activities add have major landscape impact. A tag to mark all areas which are managed for forestry including nursery plantations, clear-felled areas etc. separate from the woodland would be very useful. I can use landuse=forest at the moment because that would be very misleading. If it was just the woods and their management, resource usage was just attributive your suggestion would be fine, but as I stated at the outset the relationship is not that straightforward. |
|
I suspect for most use cases, it really doesn't matter whether an area of trees is managed or not. "Here be trees" is all that needs to be known. Therefore, using something like landcover=trees would cover "here be trees", and an additional tag like managed=yes could be used to indicate resource management. Those data users who simply need to know that there are trees can ignore the managed tag, those who are interested can use the managed tag to render forestry plantations differently, and there's no conflicting/duplicate tags. Over time, mappers would hopefully transition to the new, less ambiguous tagging system and the old ambiguous ones would die off. |
I completely agree. FWIW I have been adding landcover=trees for years to all areas where trees are growing (a lot of smaller ones definitely won't qualify as forests anyway). My suggestion is to add the landcover tag to the rendering db in the future, it is also useful for other values like sand or rock. |
|
General note: I suggest to move this discussion to tagging mailing list or wiki. It is certainly no longer about rendering current OSM data in this style. And than, later, once situation changed lets open a new ticket. This ticket is closed by lead maintainer and further discussion how tagging should be improved is not going to change the fact that it is closed as WONTFIX. |
Is it intentional that there is a significant difference between natural=wood and landuse=forest?
See http://www.openstreetmap.org/?mlat=50.07725&mlon=19.88101#map=19/50.07725/19.88101
I would suggest to render natural=wood using exactly the same rules as landuse=forest
Top left - natural, bottom right - landuse.
The text was updated successfully, but these errors were encountered: