From OpenStreetMap Wiki
Jump to navigation Jump to search

Why landuse, and not surface?

Text removed from article: Often, if the area would no longer be actively maintained as grass, it would turn into some other natural feature; most likely into a natural=heath or natural=scrub. Likewise, even though humans take care of the landuse=grass areas, these areas are mostly not used by humans for anything other than growing said grass. Therefore, using the landuse=* is, if not justified, at least not wrong, even if it at first seems out of line with the other natural=* and surface=* uses.

I have copied this text here from the main page given that it is primarily a discussion about the tag and not of great relevance to someone wanting to use it. PeterIto 22:23, 1 June 2011 (BST)

Landuse=grass should be for areas used to PRODUCE grass! Grass is grown here, harvested (with some top soil) moved to where it is required and planted there. Back at the growing area more grass is planted, grown etc. It is a 'farm' for grass. See as an example. Warin61 (talk) 00:19, 9 January 2017 (UTC)

This is gross deviation from established usage pattern of landuse=grass in OSM. You can't redefine it without prior discussion and acceptance from community! Please revert your changes. Rafmar (talk) 08:49, 9 January 2017 (UTC)

What about mulch?

Here in Southern California, due to the recent drought, grass lawns have been replaced with mulch or gravel. I know that front lawns of houses aren't that relevant because they're rarely mapped, but there are large areas that are surfaced with mulch and I can't find a tag for that. These areas are generally not to be walked on, and they are clearly distinguished from their surroundings. For now, I just tag it with grass because they serve similar purposes (asthetics) and it shows that there's something on that area on the map. Should there be a mulch surface tag, or should I continue with what I'm currently doing. For an example of an area like this, see: [1] --AragonChristopherR17z (talk) 01:36, 20 January 2017 (UTC)

landcover=* should be used to describe the cover... cover is not a use. So uselandcover=mulch or landcover=gravel Warin61 (talk) 02:03, 20 January 2017 (UTC)
leisure=garden is often used for residential gardens, including front and back yard lawns. This matches British and German usage of the word "garden", and better describes a private yard in most cases than just "grass". If the mulched/gravel former lawans still have some ornamental bushes and trees and still function as a (private or public) garden, you can use "leisure=garden", and add accesss=*. Also consider if the area has some other main feature, which might also be tagged surface=gravel/mulch if there is a large area which doesn't qualify as a garden. This would depend on the particular feature. --Jeisenbe (talk) 12:37, 10 August 2019 (UTC)

"Controversy on this subject"

The text currently says the use of landuse=grass for any area of grass might be a problem: "consider it as a problem as grass in that case not a use of the land but a cover of it and that one landuse may reside in the other landuse. The landcover=grass is proposed to remove this problem. There is some controversy on this subject." This doesn't seem to represent the de-facto situation: this tag has been used over 3 million times and over 500,000 features were added with this tag in the past 12 months, compared to 20k total uses of landcover=grass. I really don't think the latter tag needs to be mentioned up at the top of the page. The idea that the key "landuse=" refers to "landuse" in a technical, zoning / GIS sense is also incorrect when considering landuse=forest, landuse=reservoir, landuse=basin and so on: several of these represent landcover (grass, trees, water, intermittent water), furthermore, the other values of landuse also represent a type of real "landcover", eg: landuse=residential represents cover by houses / residential buildings and gardens, while landuse=industiral areas are covered in industrial buildings and facilities. These are the visible characteristics that mappers can verify to be real features. --Jeisenbe (talk) 07:28, 10 August 2019 (UTC)

Wiki edit wars. Both landuse=grass and landcover=grass are tags for the same thing. To edit one so it appears to be better than the other simply due to the amount of use ignores the time each tag has been available. Landuse=forest should be for the production of tree produce, usually lumber. Its misuse may have came from the presence of landuse=grass. And there are instances of landuse=scrub, sand, etc. Is landuse=* to be for anything you see, that is a land cover??? If residences are underground, what you see is not houses yet the land is used for residential purposes. Some cattle ranches look to be native vegetation, landuse=natural_vegetation despite their agricultural use? Land use is not about what you see, but the function that humans gain from it! Grass or flowers in the centre of a roundabout is not what that land is used for. Some roundabouts have concrete centres - the use is still the same, the use is for the road to provide safety and control the use is not about the grass, concrete or flowers. Warin61 (talk) 10:16, 10 August 2019 (UTC)
> "to edit one so it appears to be better than the other simply due to the amount of use ignores the time each tag has been available". The current and historic use of the tags is a fact that can be confirmed. Please check out the graph at Taghistory. lancover=grass has been around since mid 2011 (8 years ago) and has been increasing by slowly but steadily since 2012. But landuse=grass increased from 600k uses at the start of 2011 to over 3 million uses today, and 500k were added just in the past year. The usage pattern shows that there isn't any real controversy about the preferred tag: landuse=grass is 50 to 100 times more popular. This is a quite important fact, and if it influences mappers to prefer a certain tag, there may be a good reason for it, no?
Re: "Land use is not about what you see, but the function that humans gain from it" - this is true in an academic or political sense, but "the function humans gain from land" is not a "real and current" feature which mappers can confirm to exist when they visit a place in person. So instead, what we are mapping is the general appearance. When we map a landuse=retail we are saying "the buildings in this area are full of shops on the first floor", not that the area was zoned by the city for retail goods and services or anything like that. That's why the tag landuse=grass became popular: mappers see that an area has a lawn or some grassy "green space", so that's what they map.
Similarly, landuse=forest became used for all sorts of tree-covered areas (just like natural=wood) in most places, because that's what we can see and map. There's no way to know that an area covered by trees will ever be cut down for wood or timber in the future - we can only guess, so let's map what we can know to be true.
re: > Some cattle ranches look to be native vegetation, landuse=natural_vegetation despite their agricultural use?" I've never seen that tag, but I'd consider natural=grassland or =heath or =scrub depending on the main type of vegetation. We could use a new landuse=rangeland tag to describe land used for low-intensity, seasonal=only grazing - but the problem would be that this tag is hard to verify. The cows or sheep might only be rotated though a certain area of rangeland for a few weeks each year.
re: > "Grass or flowers in the centre of a roundabout is not what that land is used for". Well, is it really used for anything else? What if the neighbors put in a few benches and a little fence around it, or build a memorial statue, or add a great many flowers with small footpaths between? Couldn't it become a small leisure=park or a leisure=garden instead of landuse=grass? In Southern California it could certainly be left unmaintained and turn back into natural=scrub, the local native landcover.
Most people will recognise a 'roundabout' due to it function, not what is covering it. If it is grass covered, covered in flowers or covered on concrete it is still a roundabout that performs a human function distinct from the present land cover. Similar arguments for the other points. Warin61 (talk) 23:55, 10 August 2019 (UTC)
The roundabout is the circular roadway, not the hole in the center, no? Same with a donut: the hole is a defining characteristic, but not part of the feature itself. Many large roundabouts/traffic circles are filled with a monument or park - that's a separate thing from the circular highway. Re "Similar arguments for the other points" - I don't see how the other things I mentioned are very similar to roundabouts. Please explain. --Jeisenbe (talk) 01:28, 11 August 2019 (UTC)
Without the 'hole in the center' the roundabout would not exist, just as a donut requires a 'hole in the center'. The argument about landcover vs landuse is similar for roundabouts and the other things, I had hoped that roundabouts would be easiest to demonstrate the differences. Most roundabouts only exist for the roadway, they have no other function. So for most roundabout centers the human use is for the roadway, while the landcover could be grass, flowers, concrete etc. Some larger roundabout may have other additional functions, some large roundabouts only have the one function, deflecting to these other functions ignores the majority of roundabouts. Warin61 (talk) 05:02, 11 August 2019 (UTC)
To summarize, I believe most landuse= and natural= area values mainly refer to the visible landcover, only secondarily relating to the amount of human use of the land, if at all, and the current use of tags in the database supports this. Therefore, there's no need to push mappers to a new key. --Jeisenbe (talk) 11:51, 10 August 2019 (UTC)

landcover is meant to be orthogonal to landuse, not to replace it. Thus any arguments that compare their numbers fail. --Polarbear w (talk) 17:14, 10 August 2019 (UTC)

I quote from Landuse "Land use is distinct from 'Landcover' - land cover describes the physical thing/surface covering the land. The two concepts are complementary and can be used together. The land use might be 'military' but the land cover might be 'heathland' or the land use might be 'leisure' but the land cover might be 'grass'. " I recommend you read both Landuse and Landcover, there are distinct differences between them. Warin61 (talk) 23:50, 10 August 2019 (UTC)
I understand that you would like to create a new key using the term "landcover=*", but I've tried to show above that the tag landuse=grass is generally used to define an area covered in grass. Either you would like landcover=grass to replace landuse=grass, or not. Which one is it? Would you like landuse=grass to be double-tagged with landcover=grass? --Jeisenbe (talk) 01:28, 11 August 2019 (UTC)
I did not create the key nor tag. I use it, and usually double tagged once for being correct the other for renders. What is 'generally used' I view as a bad practice. Firstly it is not a Landuse. Secondly it encourages more misuse e.g. landuse=sand, landuse=scree, landuse=heath. Third it makes teaching new mappers who have good English more difficult. It appears that you desire that landuse be replaces with landcover? If so what does military land cover look like? Warin61 (talk) 05:19, 11 August 2019 (UTC)
"usually double tagged" - why not just use landuse=meadow/grass and natural=grassland, natural=wood? There is no debate that landuse=meadow, =grass and natural=grassland are always covered by grass (+similar herbaceous plants), and that natural=wood is always covered in trees. You can also use surface=grass instead of landuse=grass for roundabouts and highway verges, if landuse=grass is anathema. See One feature, One OSM Element.
Re: "you desire that landuse be replaced with landcover?" No, I'm not sure what gave this impression? I don't think the key "landcover=" is needed, because we already have tags that describe every significant type of landcover. Maybe we still need one for gravel that isn't scree or shingle, and maybe one for clay soil and mixed bare soil (eg in desert areas), but these can go under natural=gravel and natural=clay or natural=bare_earth - like most vegetation and mineral landcovers. Re: landuse=sand/scree/heath, there will always be "mistakes" (like natural=grass, natural=trees) and a long tail of synonymous tags - adding another key will only increase the number of duplicates. --Jeisenbe (talk) 04:04, 13 August 2019 (UTC)
landcover, as described in original landcover proposal, is a failed attempt to deprecate several tags like landuse=grass or surface=* on areas Mateusz Konieczny (talk) 23:52, 11 August 2019 (UTC)
Failed? Or evolving? Various mappers support and use the key landcover. Warin61 (talk) 00:33, 12 August 2019 (UTC)
It is not the primary goal of 'landcover' to deprecate other keys. The goal is to provide a solution of the problem that 'landuse' is inappropriately used in many situations. The patch of grass between two parking lanes is not a landuse. The use is a car park, and it constitutes of some areas with parking spaces, some ways to drive to them, and some green areas that make it more appealing. --Polarbear w (talk) 16:30, 12 August 2019 (UTC)
The link above is to the section "tags and deprecated tags" in the original proposal, which suggests deprecating 3 or 5 tags, including landuse=grass, plus the entire key "surface=":
"It would deprecate the following tags: natural=mud, natural=sand, landuse=grass, surface=*, [and possibly]... natural=scrub and natural=water."
It's clear that the proposal authors saw all of these tags as types of landcover, which would be replaced by landcover=*. Without deprecating existing tags, the key landcover= contributes to tagging fragmentation. --Jeisenbe (talk) 04:04, 13 August 2019 (UTC)
Also see "Deprecating landuse=forest" and "Table with all the values" from the talk page, which both recommend deprecating landuse=forest (but not natural=wood?), suggesting that many "landcover=" supporters are confused about the difference between landuse= and landcover= too. :-) --Jeisenbe (talk) 04:29, 13 August 2019 (UTC)
The people there have wrongly used landuse=forest and don't like natural=wood and would prefer landcover=trees which indicates the desirability of the landcover=*. Their confusing probably come from the presence of the oxymoron landuse=grass. However the problem with forestry is there is no, as yet, suitable value for those areas used to provide lumber that are presently correctly tagged landuse=forest. Grass, on the other hand, is simply a presence of a land cover and can be migrated without harm. Is this discussion about grass or trees? Warin61 (talk) 05:26, 13 August 2019 (UTC)
landuse=* is about the primary use of land by humans, it is not about what is seen but about the use to which it is put. landuse=grass fails to be a 'land use' and deserves to be abandoned and depreciated. A replacement is available with landcover=grass. Migration to the better tag is provided by dual tagging. Dual tagging is a common method of providing compatibility and testing. -- Warin61 (talk) 04:20 13 August 2019 (UTC)
> "landuse=* is about the primary use of land by humans" Mainly. But landuse=* is just a key, a string in the database. And for landuse=grass it implies "an area of mown and managed grass". Whether or not you or I think this is a "rea" landuse or not isn't very important; the definition is clear, and when you analyse OSM data you get to assign keys to whatever columns you want, in your routing/rendering/geocoding/GIS database. You can easily import "landuse=grass" as "landcover=grass" and "natural=wood" as "landcover=trees" on your own server, when you want to do this. There is no need to change a well- tag, when it's not causing problems. --Jeisenbe (talk) 04:43, 13 August 2019 (UTC)
I repeat. It is causing problems. Not in the data base. Not in the renders. But causes problems with mappers, particularly new mappers. Firstly it is not a land use, it does not match the easy common definition of a land use and this makes it hard to learn, remember. Secondly it encourages more misuse e.g. landuse=sand, landuse=scree, landuse=heath, particularly with new mappers who having learnt that 'landuse' can be used for 'land cover' from the example of 'landuse=grass' continue this with other land covers. Warin61 (talk) 23:43, 13 August 2019 (UTC)

The only problem I see with landuse=grass is that some people use it instead of landuse=meadow for pastures, but adding landcover=grass would be much worse in this regards. Do you want mappers to stop tagging landuse=meadow and start using landcover=grass instead? The current situation is better. --Jeisenbe (talk) 04:43, 13 August 2019 (UTC)
How would it be worse? I would prefer landuse=grass to be replaced by landcover=grass. Again I repeat my thinking of what they did - mappers correctly tagged what they saw, or had from their import data, the presence of grass, that is what it was seen and tagged correctly using the present practice of tagging landuse=grass, they did not intend indicating a land use as such. They had mapped land cover not land use, so landuse=meadow did not enter their thinking. If you want to tag land use go right ahead. There should be no problem tagging the presence of grass to the exclusion of land use. Similarly there should be no problem tagging the presence of a land use to the exclusion of the land cover. Warin61 (talk) 23:43, 13 August 2019 (UTC)

Why is it better? The landuse=meadow should not say what the land cover is but the use of the land - to provide fodder/hay. Those who use landuse=grass instead of landuse=meadow are tagging what they see - a land cover of grass, they do not presume to know the use of the land. It is not a mistake but a lack of knowledge. Removing landuse=grass and replacing it with landcover=grass would better indicate a lack of knowledge of the use of that area. Warin61 (talk) 05:26, 13 August 2019 (UTC)
I don't think we are talking about the same situations. I'm mainly thinking of what happened in the central Netherlands, where huge areas of pasture and meadow were imported as landuse=grass from the national landcover database. I don't see how this would be avoided by adding a 4th tag for grass landcover with landcover=grass; this would just increase confusion. And I don't see how replacing landuse=grass with landcover=grass would prevent sloppy mapping or sloppy imports, rather it would blur the distinction between GIS / satellite raster "landcover" concepts and the real feaures we map in OSM.--Jeisenbe (talk) 06:58, 13 August 2019 (UTC)
Seems to me the import was done correctly. The imagery shows grass, not the use of the land? And so they tagged grass as best they could using a tag that implies land use but is actually a land cover (hence the confusion). Land uses of meadow, agricultural, sport pitches should not be taken as implying a land cover such as grass. If all of the import had landuse=medow then it would be wrong in some places, but it is correct to say there is grass here from this imagery. There is no prevention of sloppiness but an attempt to reduce confusion. See trees? Tag natural=wood and/or landcover=trees. See grass? Tag landcover=grass (at the moment you have to tag landuse=grass for some/most renders). If the land use is known then tag that too, but seeing trees, grass, gravel does not imply some particular land use.

A luxuriant lawn vs. grass patch

All I know is I used this, and the part of the park with the splended lawn turned an ugly grey on the map. Like I paved the park with cement.

All I also know is people spend a lot of money to maintain lawns instead of some "patch of tall grass".

So must be a cultural difference. Where some people come from they don't have lawns, and grass is just grass. Jidanni (talk) 13:40, 27 September 2022 (UTC)