This is a work in progress.
In OSM, the anarchic approach to tagging has led to proliferation of tags which compete for the same space on OSM. The problem tags are land use, natural, land cover, amenity, surface and vegetation. This competition complicates rendering, confuses new volunteers and ultimately diminishes the quality of the product.
This page examines each of the tags, identifies problem definitions and tagging practices, suggests good tagging practice and an approach to rendering.
- 1 Land use
- 2 Natural
- 3 Land cover
- 4 Amenity
- 5 Surface
- 6 Vegetation
Land use is the human use of land. Land use involves the human management and modification of the natural environment into a human-made environment of urban and rural settlements, agricultural areas of all types, leisure spaces, transport utility spaces and military spaces. Land use does not usually apply to pristine wilderness or to oceans, though there are exceptions.
Structuring a decision
The first question when tagging land use to ask is whether the feature is managed or modified by humans. If so, use 'landuse' but if not, another key like Natural or Surface may be appropriate. The second question is to determine nature of the primary human activity in the area. The most common values are landuse=forest, landuse=residential, landuse=grass and landuse=meadow.
Land use tagging problems
There most frequently used tags exhibit some of the ambiguities that create problems for mapping.
Forest, as an English word, means an area covered by trees. There is no implied use of the trees, so it is not directly useful as a land use value. Forestry and more accurately silviculture do refer to a use, being the cultivation of trees by humans. To avoid major changes OSM retains the value 'forest' but OSM defines 'forest' as a synonym for silviculture. An unmanaged 'wild' area covered with trees should not be coded with a landuse=forest tag. See 'forest' in the main land use article.
Residential is a good land use value.
Grass is really a land cover or a vegetation type, not a land use. The human use of 'grass' is not clear. Some possibilities are landuse=garden, landuse=park, landuse=farmland.
Meadow is similar to grass. The human use of landuse=meadow is not clear. OSM defines meadow as longer grassland, not regularly mowed. There is no statement of the actual human use of the meadow. Most likely it should be coded as landuse=farmland or landuse=conservation or if its a meadow not actively managed by humans, then it has no landuse value at all.
Watery land use
Some oceanic and inland water areas do have a landuse= tag. Examples are marine conservation reserves, military practice areas, salmon farms and oil production platforms. Offshore wind generation towers are usually smaller than 5m x 5m so should be coded as a point feature not as landuse=.
As some areas are unused by humans, or almost so, there will be gaps in land use. Main article , Proposals
Some text here...
Natural tagging problems
Some text here
Land cover is different to land use in that it makes no distinction as to the cause of the feature. Land cover describes the physical material at the surface of the earth. Land covers include desert, semi-desert, grass, scrub, trees, asphalt, concrete, rock, ice, sand, water, etc.
Structuring a decision
The deciding question in determining values for landcover is to ask what is seen in an aerial image. Do not ask what its purpose or its use is, but define only what is covering the surface. Landcover was first used to describe the images of earth developed by the 1972 Landsat satellite. Using multispectral scanning, surface water, geology and vegetation could be identified by their spectral signature, but the satellite system could not deduce the human use of the features, thus the images were land cover, not land use.
Differences between land use and land cover
The land use and the land cover are related but not similar. On a GIS they would be stored as different coverages or layers. OSM does not use layered data, so the uncomfortable combination of land cover and land use on one layer causes confusion.
Some practical examples are provided to assist in illustrating the difference between land use and land cover:
A military bombing range (land use) may be covered in patches of grass, scrub, water bodies and trees (land covers). A regional sports facility (land use) may have land cover of grass, asphalt, building, concrete and water. A shopping centre (land use) will have land cover of asphalt, building and sometimes a token patch of garden.
A school (land use) in Busia, Kenya may have land cover of buildings and earth while a school (land use) in St. Albans, U.K. may have cover of building, grass, asphalt and trees. In 14th Street, Manhattan, NY a school (land use) has only the land cover of building. The land use is identical in each case, but three different land covers apply.
A contrary example is that an airfield (landuse) will be land cover of concrete, asphalt, buildings and grass. An abandoned airfield (land use) will have precisely the same land covers. The grass may be longer, the concrete cracked, the buildings derelict, but the land cover is unchanged. Two different land uses apply.
An Amenity is a useful and important facility for visitors and residents. The tag covers an assortment of community facilities including toilets, telephones, banks, pharmacies and schools.
The surface key is used to provide additional information about the physical surface of roads/footpaths and sports facilities, particularly regarding material composition and/or structure. It is not approved as a general key for areas other than roads and sports facilities.
There has been talk about it taking on more general use but the current tagging does not show widespread use outside the transport and sport tags.
Vegetation could be an independent layer of information, but is not widely used as a primary tag. From taginfo, the KV pair vegetation=grassland has been used 1353 times, and vegetation=wood has been used 967 times, followed by vegetation=none with 117 uses. The wiki does not support vegetation as a key. Most tags featuring vegetation have vegetation forms as values in natural=, landuse= and leisure= KV pairs.
On a GIS, vegetation can be separated out as a layer. On OSM this is not possible. One possible work-around is for vegetation rendering to use only symbolic fills that illustrate structural class. In this way vegatation can overlap land use and land cover without conflict, behaving as its own layer.
Less developed areas
In less developed areas, vegetation classification by stuctural class is a simple, rapid means of storing mappable information. An area would be trees, shrubs, herb, grass, sedge, ferns etc. More details are provided below.
In urban areas, the fragmented pattern of landscaped parks, gardens, traffic islands denies the use of vegetation structural classes. Features instead are described by their form. A tree, a row of trees. A shrub, a shrubbery, a hedge. Flower beds. Grass lawn, grass meadow. Crops in urban areas should not be classified as vegetation, but coded as landuse= and landcover=
In farmlands, crops are not normally classified as vegetation, but isolated trees, hedges and rows of trees can be classified by their form as is done in urban areas. Larger areas of non-crop trees, shrubs or grasses can be classified by structural class.
Vegetation growth form
Biological scientists will classify vegetation in many different and complex ways. For OSM, the form of the vegetation, which is a feature that can be accurately observed on an aerial photo or recorded on a field sheet is an adequate classification for mapping purposes.
Ecologist Ray Specht devised a structural classification system which uses only height, growth form and dispersion. His work is globally applicable and it is simple to use. Specht specifies that Height is specified as one of tall, low or dwarf. Growth form is one of tree, shrub, grass, sedge, herb, fern and dispersion is either closed or open. If tussock grasses and hummock grasses are merged into "grass", there are 43 distinct classes which provide complete cover of every vegetation form.
Vegetation structural classes
Examples are Tall Closed Forest, Low Open Shrubland, Fernland, Short Open Forest, Grassland.
Most classes have a variety of local common names like Tundra, Miombe forest, Mangrove forest, Prarie, Pampas, Peat marsh, which are worth recording as a secondary type= value. In this manner: vegetation=short_closed_forest;type=miombe;
In OSM much use has been made of the 'natural' key to describe vegetation, with classification including natural=wood, natural=scrub, natural=grassland, natural=heath, natural=fell. Vegetation under the 'landuse' key includes ...
Text goes here
Text goes here
A note on tagging land cover when it is vegetation
Simplified vegetation classes for use as land cover values...