In Eastern Melbourne we have several suburbs that are basically houses and roads put into forests. How should we map and tag the landuse there, where landuse=residential overlaps with landuse=forest?
Areas with relatively dense housing without any landuse
Areas with relatively dense housing with landuse=residential
Areas with relatively dense housing with landuse=forest (NE corner)
Areas where landuse=residential and landuse=forest overlap (although I read that this approach is not very welcomed, “One landuse only!”; I observed that the default rendering puts the forest symbols above the residential area, which I personally think is an interesting approach, but we’re not mapping for the renderer) (SW corner)
Where would you draw the line between landuse=forest and landuse=residential, and are there any established criteria?
Thanks for your help, I hope for a productive discussion.
for start, overlapping areas if landuse=residential and landuse=forest (or natural=wood) are 100% fine if that matches situation on the ground.
There are many cases where overlapping landuse=* are valid. Especially with forest, to the point that special supports was added for better display of small areas within large forested areas.
This is further complicated by many landuse= not mapping land use (landuse=forest is used for all tree-covered areas) and many land use info tagged with different tags - for example amenity=school or amenity=prison
This is why many push for landuse=forest to be limited to areas with logging or other “managed woodland” activities and use natural=wood to describe places “with trees on them”.
Other groups push for other distinctions between landuse=forest and natural=wood (common one is trying to make distinction between woodland more or less influenced by humans that is contradictory to this one)
Personally I see no hope to creating some systematic distinction between this two tags, if I would be some Tagging Imperator then I would deprecate landuse=forest.
I mostly don’t overlap landuse=residential with natural=wood/landuse=forest. In the areas that I map, humans generally cut down most of the trees where they build houses. Even with most of the trees cut down, the remaining ones can make a residential area appear as if it is forested from above (especially with leaf-on aerial imagery). However, on the ground, the character of these areas is not that of a forest. They are full of lawns, flowerbeds, gardens, driveways, patios, and all sorts of other things characteristic of a residential area. So I map these human landscaped areas as landuse=residential and then where the human modified landscape ends and the true forest starts I map as natural=wood. Here’s an example of a rural residential area I mapped showing this style.
I would overlap landuse=residential with natural=wood/landuse=forest in place where a bunch of small houses or tree houses exist in a forest and are permanent residences. There would have to be very little tree clearing or landscaping so the character of the area was that of a true forest. I’ve never mapped a place like this, but I’m sure there must be some places like it in the world. Overlapping would make sense in this case because the character of the place is very much both forest and residential.
I would not support that. The definition of landuse according to the wiki is
Mainly used to describe the primary use of land by humans.
There can only be one “primary” use. Either it is residential or forest. If it is residential, forest can only be secondary use and there is no key landuse:secondary=*. The same applies to overlapping landuse and natural. If any area is in use by humans, it cannot be natural any longer (or vice versa).
Contrary to that I do not see any problem in overlapping a landuse with an amenity. If any amenity is located in a residential landuse area the primary use is still residential, no conflict here.
note " Note: Two values of landuse=* may be viewed as not strictly land use. These are landuse=grass and landuse=forest. Please refer to the pages of these for more information."
In general, attempted definition of keys such as landusehistoricleisureamenity are not very useful and just try to find common features which often it are not really present.
I have noted “mainly” and “may be viewed as not strictly landuse” before and this is OSM. Some mappers may see it this way, others don’t. I prefer a simple structure according to the KISS principle.
A certain area is either left to nature (natural) ore used by humans (landuse). If landuse the primary landuse is it - if a secondary landuse should be added we should implement landuse:secondary. This is a simple structure and easy to follow. If we start overlaying one landuse with another one the next step could be to put a third one on top and eventually add some natural tag. Does that make sense … I doubt.
The distincton between “forest” and “wood” is another issue. Grass is no landuse at all (although widespread in use) but simply a landcover - which is another issue as well.
nevertheless natural=* are also used for man-made features and landuse=* tags are in use also for areas with minimal human intervention
it may: flowerbed in forested area, in factory in a military base may have overlap of:
landuse=flowerbednatural=woodlanduse=industriallanduse=military (I do not have real world example of such extreme case, but valid cases for lesser overlaps can be found)
Mainly used to describe the primary use of land by humans.
There can only be one “primary” use. Either it is residential or forest.
the definition like this is not complete, because some “landuse” is actually meant to be landcover (the mapper has applied it with the intention to show landcover), for forest this is particularly observable.
I would also not believe there can only be one primary use of land, rather the term is meant to distinguish from secondary landuse, having two equally important, primary uses on the same land, e.g. vertically stacked, seems a possible situation, and what can be considered primary and secondary would depend on the use case/point of view. We don’t have established tagging for these situations (3 floors of retail and 5 commercial in the same building?) on the landuse level, but the pois and possibly detailed buildings and polygon pois offer a way to describe the situation if you have a lot of time and dedication.
I believe everyone is aware that there are very very few places in the world which are definitely not used by humans at all. Every natural=wood is used for recreational purpose, scientific studies and the like, even the romotest places like Antarctica are used in one or the other way.
That is why I like the “primary” idea. If an area is primarily nature, I use “natural”. If it is primarily utilized by humans, I use “landuse”. The same applies for the type of landuse.
This is not what we are talking about here imho. The issue here is overlapping areas with different landuses so that the shared part of these areas has two valid landuse tags.
A flowerbed within a forest (not very common indeed - a better sample would be a meadow within a forest which is quite usual) can easily be separated from each other by mapping a simple MP, so that both areas (the flowerbed as well as the surrounding forest) will still have a single landuse tag.
it may: flowerbed in forested area, in factory in a military base may have overlap of:
landuse=flowerbed is one of those tags I would deprecate, despite being very reluctant to deprecation in general.
man_made=flowerbed
and btw., landuse=flowerbed is just 0,1% of landuse objects: https://taginfo.openstreetmap.org/tags/landuse=flowerbed
A factory in a forest where both can apply is something that could only be possible if either the factory was elevated like an oil platform or underground (what seems more likely for your example in the military context), otherwise you would expect a hole in the forest where the factory is, or example required. For underground installations I believe we do not use landuse, but unsure about it.
Sorry but I do not get your point. If there is a well hidden (secret?) factory within military premises how does one know about it? And is it desirable to map this in detail? Anyhow, if so this area would be landuse=military and if it is known some of the buildings are productuion sites for military goods these could get the tag man_made=works and if further specification ist wanted another tag produce=* could be added. I cannot see any need or additional value for a second landuse=industrial tag in this case.
I believe everyone is aware that there are very very few places in the world which are definitely not used by humans at all.
this is not the question though, the question is whether humans determine what they are, like a planted and controlled forest vs. one that is there naturally and without significant human intervention. With very low density, the people are not determining the environment, rather living within, not landuse as they “use” only a fraction
Every natural=wood is used for recreational purpose, scientific studies and the like, even the romotest places like Antarctica are used in one or the other way.
not really, the fact that there are a handful of research stations in a vast area does not mean the land is used for research, not beyond the station and the immediate areas around.
You know that I agree with you about landcover to be the better key compared to landuse in many cases but landuse is by far the more common and widespread key even for landcovers like grass, flowers or the like which do not represent any “use” at all.
Anything is amginable in one or the other way but I have not seen anything like that in real life. A common installation is a PV plant put up on framework on former agrigultural land with a landcover of grass used for grazing some sheep or goats. There is no doubt about the primary landuse in such case and I would not think about adding a landuse=meadow here.
Again we do have large scale commercial salt mining close to here. The underground mines cover an area of the size of a large city, spread in 400 m depth beneath a rural setting with farmland and villages and represent an important economical factor for the whole area. Nevertheless I cannot see any reason to add a second landuse here unless we find a way for 3D-mapping of different land layers and tag every layer with it’s own landuse …
Thanks for your samples. In most of these cases I personally would not use overlapping landuses but treat them as described earlier. The overlapping of military and forest is an exceptional case as military exercise areas often include large wooded areas which can be under forest management at the same time. Although the primary use of these areas (at least in Germany) is military with forestry being secondary it is common practise to overlap the 2 landuses here.
The situation OTG would be much better represented by using landuse=military + landcover=wood but unfortunately landcover is not an established key. I accept there is no established alternative here but I still dislike the inconsequence of two primary uses for a single area.