From OpenStreetMap Wiki
Jump to navigation Jump to search

royal hunting and Robin Hood

Just a small point and easy to overlook in British English (en-uk) a forest had a varient meaning as a speacial type of landuse with special laws set aside originaly for royal hunting and leasure purposes so included areas well beyound trees such as heathlands etc.

By the late 20th centry most people associated forests with the like of Robin Hood so the idea about the non-wooded bit began to fade from most peoples minds - unless you lived in a place that was a Named Forest such as the New Forest were there are still special laws appyling in areas beyound the trees.

In the same vain British English also has Copse {for trees regulerly havested above ground level so they regrow from the stumps [action of cutting to copice] & where cattle graze cutting is made higher up and called pollarding - this was also becomeing less common as lots of people gave up wordburning as a major fuel choice] there is also the term spinney for a small group of trees on there own [sometimes encoraged amongst grassland to shade cattle or mark a dip of the land closer to the water table so a spring or well may be there with a cattle water trough.]

--Govanus (talk) 19:26, 15 June 2015 (UTC)

Similarly to the situation in the UK, in Denmark an area might be officially part of a forest, while not being covered by trees (and never intended to be). This can potentially lead to different ways of mapping the "forests" of Denmark, depending on whether you map the physical or the administrative forests. I just checked some of my "favourite" forests, and it seems indeed like both ways are used.
NisJørgensen (talk) 09:08, 1 August 2015 (UTC)

Scheme for wooded areas

My attempt to create clear scheme framefork for tagging wooded areas --BushmanK (talk) 03:09, 19 August 2015 (UTC)


The present confusion could be reduced if mappers were to use the tag landcover=forest. It does not identify the use nor origin of the forest .. simply that a forest is present. Warin61 (talk) 23:23, 18 December 2015 (UTC)

In data modelling sense yes. if mappers were to use... as you say. But the reality would likely be ...just another competing tag.
I see people have made the same point in the wider discussion here: Proposed features/landcover. I wouldn't rule it out completely, but not sure how the transition could happen
-- Harry Wood (talk) 15:38, 3 January 2017 (UTC)

rename this page as "Forest / woodland"

I think maybe we should rename this page as "Forest / woodland" to make it clearer from the outset. Otherwise the assumption is that this page is for documenting a forest tag (whereas in fact it is about both landuse=forest and natural=wood tags presented evenhandedly).

This is pretty similar to a page I created called "Gym / Fitness centre" to give pointers to different tags there.

I suppose another title which might work would be "Wooded areas" (the term used by BushmanK above)

-- Harry Wood (talk) 15:23, 3 January 2017 (UTC)

Or maybe "tree covered areas"? Mateusz Konieczny (talk) 08:34, 13 August 2019 (UTC)

Regularly clear-cut forests

Parallel to wrangling on the issue of natural forests and managed forests, there is the issue of tree farms. In parts of the US (and other parts of the world, I'm sure), quick-growing trees such as pine are grown on large farms and regularly clear-cut. The clear-cut land is then reseeded in rows with new trees. In areas where this is common practice, this is often distinct from a "managed" forest, where the forest remains with selective culling. While my initial instinct is to attempt to tag these locations as managed forests using a tagging scheme from this page, many of these areas are in fact devoid of all trees, or consist of large areas of saplings, which doesn't fit neatly into either category. The farmland tag doesn't quite seem appropriate, either Thoughts?

Benjamin Muller (talk)

Map the areas of trees on the "tree farm" with landuse=forest. Any areas that have been cut and replanted with saplings can be mapped with natural=scrub since they will probably be bush-sized. If the saplings are still tiny, I would just leave it blank. Some people double-tag the replanted areas with natural=scrub + landuse=forest, but this ambiguous to many database users - some will think it's covered by full-grown trees, so I would not recommend this. --Jeisenbe (talk) 08:29, 13 August 2019 (UTC)

Confusing land cover with land use, How to describe tree-covered areas like forest or woodland.

Hi, jeisenbe has removed the information I added to the disadvantages of landcover=trees. The sections says

  • Does not distinguish between orchards and similar tree plantations used for food (landuse=orchard), even though these are visibly different on aerial imagery and in survey.

I contend that the landcover=* should not identify the land use for food (or any other land use). Orchard is not tagged as a land cover but a land use as it also includes the different land covers of shrubs and trees. I had hoped to identify that by may addition of;

Some would argue that the land use should not be part of the landcover=* as it is about the land cover and not the land use, the land use should be tagged with a different key. landcover=trees simply indicates the presence of trees, the species of tree can be tagged with species=*.

Removal of the listed 'disadvantage' of orchard as it is not relevant being a land use applied to a land cover. -- Warin61 02:37 10 June 2020

@Warin61:, It's a disadvantage compared to the option of using landuse=forest or landuse=orchard (or natural=wood) instead, since it is almost always clear when an area covered by trees is an orchard on aerial imagery, and always obvious in person. If people use landcover=trees instead, the information of orchard vs not orchard is lost, since landcover=trees could be any of these (or perhaps a landuse=plant_nursery used to grow trees).
Your suggested edit is out of character with the surrounding section, which focuses on disadvantages. If you want to talk about this more it would need to go in the "advantages" section, though it seems that this is already mentioned: --Jeisenbe (talk) 05:54, 10 June 2020 (UTC)
Should landuse=residential be criticized because leisure=garden or building=* are better 'options'? They convey different information - they are not mutually exclusive. The same applies for landuse=* and landcover=*, they are not mutually exclusive. Indeed landuse=orchard could be a land cover of either trees or shrubs so adding the land cover information enhances the map information. Mapping both pieces of information is best, it is not a choice of one or the other. I do know of an 'orchard' that is no longer harvested due to it being remote so costs make harvesting uneconomic. I would not tag that as as land use despite how it appears on satellite views or the causal visitor, land use requires information that the land is actually used. Warin61 (talk) 00:04, 11 June 2020 (UTC)
landuse=residential have different meaning from leisure=garden or building=*, while landuse=forest and landcover=trees have de facto the same meaning (tree-covered area) Mateusz Konieczny (talk) 22:56, 11 June 2020 (UTC)
I gave that as an example of the difference that exists between landuse=* and landcover=*. The issue is the following text that attempts to say landcover=trees would be used where landuse=orchard should/could be used:
* Does not distinguish between orchards and similar tree plantations used for food (landuse=orchard), even though these are visibly different on aerial imagery and in survey.
I believe this text is deceptive at best. One could say the exact same thing about natural=wood. Warin61 (talk) 00:04, 12 June 2020 (UTC)
OK, I forgot about landuse=orchard, so landuse=forest is tree-covered area except orchards. Are there any other special cases? Mateusz Konieczny (talk) 09:33, 12 June 2020 (UTC)
Sorry, but I'll try to keep this to the topic of why landcover=trees is a disadvantage due to landuse=orchard. I contend there is no disadvantage. Just as there is no disadvantage of natural=wood due to landuse=orchard. Warin61 (talk) 23:08, 12 June 2020 (UTC)
I am not aware of any locations where natural=wood is used to tag a landuse=orchard. Can you provide an example where this has happened? In English and most languages there is a clear distinction between a "wood" and an "orchard", and tagging practice is to distinguish between these two types of tree-covered areas quite consistently. --Jeisenbe (talk) 01:38, 18 June 2020 (UTC)
In English and most languages there is a clear distinction between a land use and a land cover. There is no conflict in tagging one area with both a land use and a land cover. The land cover tag would not normally define a land use, so natural=wood and landuse=orchard may coexist on the one area, not a choice of one or the other. Can you provide an example where landcover=trees has been used in place of landuse=orchard? Warin61 (talk) 07:47, 18 June 2020 (UTC)
I disagree. In Cartography and GIS, it's quite common to see maps of "landuse" which also include categories like "scrubland" or "wetland" - which are biomes or types of vegetation but not landuses, and maps of "landcover" often include types of developed land as "landcover". In OpenStreetMap most of the features under landuse=* can be interpreted as types of "landcover" as well: an area of landuse=industrial will contain industrial buildings and services (parking, service roads, industrial railway branches, pipelines, storage tanks etc), while landuse=retail will contain retail shops and buildings, along with associated services (parking, streets, seating areas), landuse=garages is an area of roofs or buildings where cars are parked, plus service ways, etc, and landuse=orchard is covered by trees used to produce food, plus some highway=track or footways between rows of trees, and so on. All types of landuse=* in OpenStreetMap are mainly identified by the real-world features found on that area of land: we don't tag zoning regulations or areas that in theory could be used for a certain use, but we tag what is actually currently there. --Jeisenbe (talk) 17:01, 18 June 2020 (UTC) "Landuse: The purpose for which an area of land is being used, such as residential, agricultural, commercial, retail, or industrial. Also known as land cover, land utilization. [emphasis added]. --Jeisenbe (talk) 17:03, 18 June 2020 (UTC)
It would appear that the Oxford publishers can't keep their story straight. Land cover: The observed biophysical cover of the earth's surface. Strictly speaking, it describes vegetation and man-made features, and omits bare rock and water, although, in practice, these elements are often included under this term. So not land use then? I would suggest that the FAO documentation is a better introduction to the concepts of land use and land cover. Adavidson (talk) 05:07, 19 February 2021 (UTC)

Naming a fragmented forrest

Often when we map forests, we start out by mapping a big, continuous forest as a single entity. If this forest has a name, it makes sense to assign a name=* to it. At a later point, one might want to go into more detail when mapping, adding detail to individual areas of the forest. Perhaps one part is managed, and another is not. Or perhaps different parts has different leaf cycles. A dilemma then arises, for what to do with the name=* tag. Should one:

  • Subdivide the forest, tagging each individual area with the name=* of the forrest as a whole?
    • This seems wrong to me, as it implies the individual woodland parts have names in themselves.
  • Keep the original woodland area, with only globally applicable tags, but ad overlapping subdivisions?
    • This seems unideal as well, since overlapping landuse/landcover is often frowned upon.
  • Keep the original woodland area, but re-tag it with place=locality?
    • This doesn't seem to be a good fit, as place=locality should not be used for forests.
  • Perhaps there is a forest equivalent of building:part=*? Something like landuse:part=* or natural:part=*? This could be used for adding more detail to subdivisions.
    • This seems to me like an approach that would be solve the problems of the above approaches, but I don't know whether the community considers the building:part=* a good example to follow in other cases.

Another class of approach is relations:

  • Transform the full wood into a type=site relation.
    • This seems intended for man made structures, rather than natural features.
    • Also intended for spread-out features, rather than a continuous region.
  • Transform the full wood into a type=cluster, type=group, or type=multiobject relation.
    • Intended for natural features, but not approved by the community.
    • Three competing solutions. Difficult to decide which one to pick.

What would you do? Are there ideas I've missed?

multipolygon is an option - but leaf_type leaf_cycle may differ across forest parts. In general this is not really solved. I thought about place=forest for area covering entire area (which may include clearings, glades, windthrow, riverbed, lakes and so on) Mateusz Konieczny (talk) 14:23, 25 January 2022 (UTC)
Most native tree areas in Australia could not be separated by leaf_type leaf_cycle as they are intermingled. For an area named, in part, 'Forest' then place=locality could be better. I have two areas that contain the name 'Forest' however they are suburbs and so have administration boundaries. Another area named 'Orchard Grove' because that was what was there last century is now a shopping center. If some object carries some name that does not mean that name fragment must go into the OSM key or value. Warin61 (talk) 09:16, 26 January 2022 (UTC)
Realized that this problem has been discussed not so long ago on the tagging mailing list. It seems you are correct that this has yet to be solved in a satisfactory way. place=locality isn't really right either, as it is not supposed to be used for features that have other, more precise tags available (natural=wood) Added a summary of previous discussions here: --Harahu (talk) 08:58, 27 January 2022 (UTC)