Forest

Feature : Forest

Description

How to describe tree-covered areas like forest or woodland. There are some tagging approaches. Describing the meaning of tags natural=wood and landuse=forest.

Tags

natural=wood landuse=forest landcover=trees

A forest or woodland is a natural or semi-natural area covered by trees, which may or may not be used to produce forestry products such as wood and timber. In English, the terms "wood" and "forest" have overlapping meanings, but are clearly distinct from an orchard, which is an area of trees used to produce food (landuse=orchard).

Different tags are used to map a wood, forest or other area of trees, primarily natural=wood and landuse=forest. There are differences in the way these are used by some OpenStreetMap mappers.

The situation is complicated as different people advocate different, conflicting tagging schemes.

Depending on region there may or may not be difference between areas tagged as natural=wood and landuse=forest. Any difference may depend on who mapped the area, in some regions one of the tagging schemes below may be applied consistently but there is no consistency between different regions across the world.

As a result nearly all data consumers treat both natural=wood and landuse=forest as synonymous tags for a wooded area.

Note that, in case boundary=forest or boundary=forest_compartment entities are present, these describe the logical forest, that is to say its divisions for management by humans; in these situations, the natural=wood and landuse=forest tag are meant to map the physical forest, that is to say the physical land, which may be unwooded (scrubland, scree, pond, glades…).

The approaches are explained below.

Which tag should be used?

The differences in tagging woodland essentially result from different approaches to document human management and use of woodland areas (excluding orchards, which are consistently tagged with landuse=orchard). The following approaches are advocated by different groups:

Approach 1

• natural=wood is used to mark areas covered by trees.
• landuse=forest is used to mark areas of land managed for forestry.

Approach 2

• natural=wood + managed=yes - managed wood
• natural=wood + managed=no - wood without management of any kind
• operator=* - optional

Note that visiting location is not enough, checking whatever land is managed for forestry requires more extensive research and many people marking forests are not interested in spending time on tagging distinction between managed and unmanaged forest.

Tag managed=* is very rarely used - just 3000 uses on forests (less than 8400 instances in database [1], 5740 of them not on landuse=forest or natural=wood or landcover=trees)

Approach 3

• landuse=forest is used for maintained or managed woodland. This approach views most woodland as managed or maintained especially in areas such as Europe.
• natural=wood is used for ancient or virgin woodland, with no forestry use.

Approach 4

• wood=yes is used to mark the presence of trees. Use of wood=* is deprecated for indicating vegetation types but wood=yes is still used. It is however fairly uncommon (less than 1000 instances in database [2]).
• natural=wood is used to mark areas of unmanaged forest. It implies wood=yes
• landuse=forest is used to mark areas of managed forest. It implies wood=yes

Approach 5

• landcover=trees is used to mark the presence of trees. It does not imply the use nor origin of the trees. This can be used in combination with other approaches. Note that this tag is rarely used tag for wooded areas (100,000 instances in database [3]) compared to overall use of natural=wood and landuse=forest tags (about 5 million each).

Approach 6

• natural=wood is used to mark areas covered by trees
• landuse=forest is used to mark areas covered by trees

Does not attempt to give meaning to differences between these tags. Typically used during mapping from aerial images, or during casual survey without extensive research of a given forest. Also used by mappers not interested in distinctions discussed above or unaware about them.

Additional tags may be used to clarify purpose, forestry status etc.

Note that this approach is used by nearly all data consumers, including the Standard tile layer.

Discussion

Advantages of each approach

• Approach 1
  • Tags appear consistent – having trees on is not a "land use".
  • Does not require the tagger to make a distinction between managed and virgin woodland, which can be near impossible to make even for someone surveying the area.
  • Allows for tagging of areas of commercial forestry which are not currently wooded (landuse=forest + natural=scrub).
  • Is more consistent with tagging of other features such as reservoirs, which are tagged natural=water, along with their land use.
  • This approach is recommended for use in Russia as a result of discussion and further voting in local OSM community.
• Approach 2
  • Mapping forests does not require the tagger to make a distinction between managed and virgin woodland
  • Information about forest management is stored in a tag separate from natural=wood and landuse=forest
• Approach 3
  • More commonly used after a bot was used to retag existing woodland this way.
• Approach 4
  • As for Approach 1
  • Uses only existing tags.
  • Existing tagging largely retains its meaning.
  • Further landuses might be specified to distinguish between forest managed for decorative/leisure use, lumber/pulp production, or ecological improvement.
• Approach 5
  • Simple. No need to identify use nor origin of the trees
  • Can be used in addition to landuse=forest, landuse=orchard, and natural=wood
• Approach 6
  • Mapping forests does not require the tagger to make a distinction between managed and virgin woodland

Disadvantages of each approach

• Approach 1
  • Conflicts with approach 2, 3, 4.
  • Requires the mapper to make a distinction that can be very difficult. People mapping forests are likely to be unable or not interested in making this distinction.
  • Conflicts with widespread interpretation of landuse=forest as marking of forested areas. Potential switch to this tagging scheme would require updates on side of data consumers.
  • Any forest mapped using approach 6 requires resurvey.
• Approach 2
  • Conflicts with approach 1, 3 and 4
• Approach 3
  • Requires the mapper to make a distinction that can be very difficult. People mapping forests are likely to be unable or not interested in making this distinction.
  • Conflicts with approach 1, 2, 4.
  • Any forest mapped using approach 6 requires resurvey.
  • Does not distinguish between landuse=forest, natural=wood tagged using this approach and other approaches.
• Approach 4
  • Uses the wood=* key which has gone out of favour for other purposes.
  • Conflicts with approach 1, 2, 3
  • Any forest mapped using approach 6 requires resurvey.
  • Forest tagged by mapper unable/uninterested in making distinction between managed and unmanaged forest would be ignored by typical data consumers. As result people are unlikely to start using this scheme.
  • Is fairly uncommon.
• Approach 5
  • landcover=trees is not common (taginfo oct 2019: 100000) and rarely supported by data consumers or editors (JOSM], iD both explicitly rejected support for landcover=*). As result either original mapper or somebody editing later in a given region will use other approach - instead or in addition to.
  • 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.
• Approach 6
  • Does not distinguish between landuse=forest, natural=wood. As result it is disliked by people that would prefer these tags to have a different meaning.

General problem with this tagging scheme

When mapping from aerial or satellite images is usually very difficult to determine if a woodland area is used for forestry. This can be difficult even for the observer on the ground. As result, most mappers use landuse=forest and natural=wood fairly arbitrarily.

As result tagging schemes that rely on a particular difference between natural=wood landuse=forest are problematic. For a given area it is complicated to guess which tagging scheme used by the original mapper. It is nearly impossible to do it automatically and there are no known tools for doing this. As result other mappers and data consumers are unable to use data that was intended to be added by selecting natural=wood or landuse=forest.

Historical development

The tag natural=wood was first used in England in the early days of OpenStreetMap, before mapping had spread to other countries. There it was used to map a wood or a forest in the countryside. (At the time, very small patches of trees in parks or urban areas were not yet mapped). The tag landuse=forest was added to areas that were clearly used for managed forestry - usually plantations of conifer trees which were planted to produce timber and lumber. These are distinct from the semi-natural broadleaf woodlands in England.

However, when mappers in other countries wanted to map a named "forest", they often used landuse=forest. Other mappers considered that woods are not really "natural" when they are managed by humans, so used landuse=forest for all kinds of woodland.

In other regions, especially in the sub-arctic and the tropics, there is no clear distinction between woodland areas which are actively managed for timber and lumber production, and areas of woodland which are lightly used by humans. In these countries it is usually not possible for a visiting mapper to determine the level of human management of a particular area of trees.

Over time it became popular to map small areas of trees within residential areas, university campuses, urban parks, and other features. Both landuse=forest and natural=wood were sometimes adopted for this purpose by different mappers, which expanded their meaning from "a forest or wood" to "an area covered by trees (which is not an orchard)". Some use landuse=forest because urban trees are "managed", others use natural=wood because these areas not used for forestry.

2022: approval of boundary=forest and boundary=forest_compartment

Before this approval, there was no approved way to map delimited forests and their compartments. Such forest routinely include non-wooded areas, such as screes, glades, scrublands, heath… which, although non-wooded, are still considered part of the surrounding forest. For compartments, there was two concurrent tags: boundary=forest_compartment and boundary=forestry_compartment. This led to many problematic uses, such as:

• using natural=wood for non-wooded areas such as ponds, because these areas where part of the surrounding forest;
• overlapping landuse=forest polygons, one for the compartment and one for the forest to which it belonged;
• overfragmentation of natural=wood entities in order to have one polygon per compartment/forest…

Following approval of a new forest-related boundary scheme, the approved tagging scheme for such situations is the following:

• for the whole, mainly wooded area, which is considered a unique forest, use a boundary=forest entity;
• if compartments are present in this forest, use boundary=forest_compartment entity (the tag name is no longer about forestry, because some compartments may not be forested at all);
• map the tree stand, glades, ponds, screes… with usual natural and landuse entities.

Additional tags

• name=* - name of the forest (you should consider boundary=forest in such cases)
• leaf_type=broadleaved/needleleaved/mixed - describes the type of leaves.
• leaf_cycle=deciduous/evergreen/mixed - describes the phenology of leaves.
• managed=yes/no - whether a woodland area is actively managed
• produce=* - to describe the produce of a woodland/forestry area (e.g. produce=wood, produce=timber, produce=cork). Also see crop=* and trees=*, used for landuse=orchard.

Plantation to unmodified forest

Note that even in case where mappers agree that natural=wood should be used for forest more natural than one tagged as landuse=forest there is an open issue of which kind of forest belongs to a given category.

Following are just examples of various cases ranging from nearly unmodified natural forest to a monoculture forest plantation.

• Unmodified by a direct human activity - except global effects such as pollution and climate change
• Affected only by a small scale human activity such as foraging
• Limited logging, done to maintain natural environment (removal of invasive species, removal of trees infected by invasive species etc)
• Commercial logging, officially described as done to protect environment (large scale removal of trees under pretext of pest infestation)
• Limited commercial logging of a natural forest
• Limited commercial logging of a forest in state seeming close to natural, but undergoing active forestry (forest was planted and maintained by humans for centuries)
• Forest is nowadays protected, but within last decades/centuries was deeply affected by human activity (heavy logging 60 years ago)
• Forest growing on area where former monoculture forest plantation was destroyed by outbreak of tree-eating insects
• Forest succession on a former pastures, forest in the future will be logged
• Forest succession on a former pastures, forest in the future will be protected
• Forest with intensive forestry activity intended to protect some specific species
• Forest plantation with multiple tree species, supposed to be become closer to a natural state
• Monoculture forest plantation, transforming nowadays into a natural forest
• Monoculture forest plantation

All this cases may change over time - single forest may have visible traces of following happening over time: natural forest, managed forestry focused on wood production, uncontrolled tree removal, managed forestry focused on wood production and becoming protected natural reserve.

Rendering

For a given area it is impossible to guess what kind of tagging scheme was used by the original mapper, so in practice both natural=wood and landuse=forest are typically interpreted as "area covered by trees".

On maps forests are typically a green area. When leaf_type=* is set one may show little broad leaved or coniferous (or both) icons.

Possible rendering of woodland

Tag	Rendering	Comment	Pictures
leaf_type=broadleaved		Broadleaved woodland.
leaf_type=needleleaved		Needleleaved woodland.
leaf_type=mixed		Mixed woodland.

Rendering of woodland in Alt-colors map style

Differentiated by leaf_cycle and leaf_type. See discussion at http://blog.imagico.de/differentiated-rendering-of-woodland-in-maps/

Possible tagging mistakes

natural=forest

		$1%
		$1%
		$1%
		$1%

More details at taginfo

• natural=forest

If you know places with this tag, verify if it could be tagged with another tag.
Automated edits are strongly discouraged unless you really know what you are doing!

landuse=wood

		$1%
		$1%
		$1%
		$1%

More details at taginfo

• landuse=wood

If you know places with this tag, verify if it could be tagged with another tag.
Automated edits are strongly discouraged unless you really know what you are doing!

See also

• landuse=orchard
• leaf_type=* and leaf_cycle=*
• When should we use landuse=forest rather than natural=wood? on help.openstreetmap.org
• natural=tree - an individual tree
• natural=scrub - area covered with shrubs
• Habitat
• plant_community=*
• species=* - for species name
• taxon=* - for general scientific name
• Vegetation - summary of tags for various types of trees and plants
• boundary=forest and boundary=forest_compartment

Related OSM projects

Wiki for all environment and natural tags and projects Environmental OSM

Related terms: ‹ forest › ‹ wood › ‹ woods › ‹ woodland › ‹ forestry › ‹ park › ‹ jungle ›