OpenStreetMap supports mapping landuse (a use classification of different areas of land) by drawing in simple areas and using a tag from Key:landuse, or other land classifications such as Key:natural. Renderers tend to support many of these tags by showing areas in different colours as an underlay beneath highways and other features.
In OSM we have some nuances when tagging simple forest.
Short definition: what the area is used for by humans.
- landuse=* - main tag to mark land-use
- some of leisure=* tags might be considered as land-use.
- some of man_made=* tags might be considered as land-use (man_made=wastewater_plant). Please note than many tags here without proposal and placed under man_made are due to how OSM was evolving.
- other tags may refer to human intervention (operator=*, managed=*, craft=*)
Landcover is used to describe the physical material at the surface of the earth. Land covers include grass, asphalt, trees, bare ground, water, etc.
Recommendations on approach to landuse mapping are still to be finalised. See #Open questions below.
Example: Wakefield has a lot of landuse areas mapped out. Is Wakefield a "good" example? You decide:
In this image you can see land classified with tags such as:
Here are some awkward questions to do with landuse:
- Are we aiming to classify every area of land with a landuse? Could be good, but in practical terms, how do we deal with farm land (or in some countries desert/wasteland) which occupy the majority of the landmass outside of cities? While they are half-mapped the map looks weirdly incomplete, with big brown stains over patches where farms are drawn in. Why are the renderers set-up like that?
- How should we map very large landuse areas. These forests in British Colmbia for example have been added by splitting them up in rectangular chunks. This is a pragmatic solution to help renderers, but it results in rendering artefacts. Is that a bug to fix in the style config / software?
- Should adjacent landuse areas share nodes? (overlapping way-segments)? Also should abutting highways along the edge of landuse area share nodes? People have different approaches to this. Some people leave a small gap between landuse areas and surrounding elements. Sharing nodes makes the data more awkward to work with (you have to use clever key-presses to separate the data out if you want to change it).
- Similar question to the above. Should landuse areas ever be "attached" (share a node) with intersecting highways? It is common to do this for highway=footway as they cross into leisure=park, but it means the landuse data is less easily separated out into different layers.
- What granularity do we go for when mapping farms? Is a landuse=farmland area drawn around an individual field, individual farm, or do we draw massive areas covering lots of farmland?
- What granularity do we go for when mapping in cities? and if we do very detailed landuse areas, do we end up mapping a small landuse=retail area around every standalone corner shop? how does this sit with the use of building=*?
In addition to these general questions about landuse mapping, there are also a great deal of questions about tagging. Taking a step back and looking at the array of landuse tags and natural tags we have defined, it can look like a hopeless mess, but remember that OpenStreetMap has never operated with a strict ontology. If you're using the data, it's easy enough to define mappings to fit with how you would like to rationalise classifications.
Persuading the mapping community to stick to more sensible tagging classifications, is a more difficult task. Many tags are used passionately by a small number of contributors in a specific area, but sit uncomfortably with other defined tags. It's not even worth attempting to solve the problems all at once. You are encouraged to cross-link between similar tags on the wiki tag pages, and use the wiki talk pages and other Contact channels to point out inconsistencies and help nudge things in a more rational direction, but don't expect everyone to agree. Have patience.