OpenStreetMap
I know https://wiki.openstreetmap.org/wiki/How_to_map_landuse exists, but I’ve got a separate question…
Basically, is it intended for landuse to eventually cover the whole entire globe, so that every available spot of land is marked as either residential, commercial, parkland, or water, etc? Or should landuse only cover a few areas, while the vast majority of the map is left as “standard” empty grey?
The European countries seem to have a lot more landuse cover than the US, so should I use them as an example?
The “residential” landuse is what’s confusing me here, in particular, because I could mark vast swaths of the map (think apartment blocks, suburban neighborhoods) as “residential”, or I could leave them unmarked. Same goes for things like “forest” and “farmland”… should all woods and fields be marked with landuse, or just left empty? Not sure what the usual approach is.
Discussion
Comment from imagico on 22 February 2018 at 13:24
A few things to consider:
So specifically: If you have an area where grain or other non-permanent crops are grown you can map it as landuse=farmland. If you have an area with residential buildings you can map it landuse=residential. If you do so diligently (i.e. making sure there are no larger areas that are not farmland or residential within the area you map) the areas you can uniformly map this way are usually fairly small. If they are larger it makes sense to map them in smaller parts (split at sensible lines) for easy maintainance.
Comment from Warin61 on 23 February 2018 at 01:57
Careful! Do not assume that an area of trees means the land is used for forestry. For an area of trees you can map landcover=trees (and use ‘natural=wood’ for the renders … though it may not necessarily be ‘natural’ in some way).
What about a dessert area? The land may not be ‘used’ at all .. landuse=vacant? Or it may be ‘used’ as a a reserve of some description. Unless you know then it is best left blank as far as landuse goes. It is easy enough to use satellite imagery to add residential areas and some farmland, commercial and industrial areas to the map for landuse. It needs local information to map landuse=forest.
Comment from SomeoneElse on 23 February 2018 at 10:28
I also wouldn’t assume that all comments on diary entries** are anything other than the contributor’s personal view :)
With regard to trees / woodland / forest et al there’s been a lot of discussion about the various tags in use and what they “actually mean”. This page describes 6 approaches. For trees, the main tags used are landuse=forest with 3 million examples and natural=wood with 4.5 million. By contrast, landcover=trees has 19,000. If it expresses what you want to get across by all means use it as a tag but do be aware that many data consumers will be confused by it.
More generally, I wouldn’t worry too much about landuse (at all), especially not about “covering the entire globe”. It’s far more useful to know that roads and paths are present and correct, and that shops and offices (that people actually visit) are up to date.
** including this one