Add rendering for landuse=salt_pond

[matthijsmelissen]

The following landcover types are currently not rendered:

• landuse=harbour, see also https://trac.openstreetmap.org/ticket/3612 rejected in Add rendering for landuse=salt_pond #545 (comment) and Add rendering for landuse=salt_pond #545 (comment)
• landuse=salt_pond, see also https://trac.openstreetmap.org/ticket/4470
• natural=dune, see also https://trac.openstreetmap.org/ticket/3815
• natural=fell rejected in Add rendering for landuse=salt_pond #545 (comment) and Add rendering for landuse=salt_pond #545 (comment) , see also https://trac.openstreetmap.org/ticket/3033
• and natural=scree (done), see also https://trac.openstreetmap.org/ticket/3033
• natural=reef, see also https://trac.openstreetmap.org/ticket/4416 done

It would be nice to have these tags rendered.

[Klumbumbus]

natural=wetland is rendered. See http://www.openstreetmap.org/way/276099816

[matthijsmelissen]

Thanks, didn't realize that one was in the water file rather than the landcover file. Removed.

[Rovastar]

natural=sand is also rendered.

Here is a quick change I just did on a golf course
http://www.openstreetmap.org/#map=19/53.53682/-2.24720

[matthijsmelissen]

Yes I noticed, I removed it already.

[dieterdreist]

2014-05-20 17:21 GMT+02:00 Rovastar [email protected]:

natural=sand is also rendered.

"sand" is IMHO not a good value for natural, as it is not a geographic
feature but a surface / material. Using the landcover key (or surface in
some cases) would be better. E.g. natural=beach & landcover=sand

landuse=quarry seems to be rendered already.

[Rovastar]

I am not sure justification for natural=beach being natural but not natural =sand.
Landcover is another issue entirely. Personally I think landuse and landcover is just semantics and they should be are are used interchangeably sand, grass, etc. But this is the wrong palce for this discussion.
Do we even use landcover in cartocss?!

[imagico]

Add to that natural=bare_rock, see #298.

[daganzdaanda]

Maybe also worth rendering is natural=shoal, even if it's not used a lot yet.

[PauloCarvalhoRJ]

It would be nice to have landuse=salt_pond because there is a salt production region near where I live. It bothers me to see big blank map as if there were nothing on OSM.

[matkoniecz]

I would tag salt ponds as landuse=industrial, with industrial=salt_pond (or maybe man_made=salt_pond)? landuse=salt_pond is extremely specific (though, unfortunately it seems to be frequently used - over 4k instances on taginfo).

[PauloCarvalhoRJ]

@mkoniecz Thanks for the suggestion.

[daganzdaanda]

landuse=plant_nursery is also missing, see #187.

[vyskocil]

natural=fell (or similar) is needed for medium alpine area above the tree line but not already bare_rock

[thor]

Regarding natural=fell, is there anything in particular one can contribute with?
Visually speaking it's a tag that seems appropriate to render with a colour slightly grayer than grass, with perhaps a symbolic/pattern of sorts indicating the uneven-ness of the area itself?

[matthijsmelissen]

Not at the moment, I think. We should first roll out #947, which makes all the landcover labelling consistent. After that, we will need to look at the colours of the existing landcovers, and I think only after that is done it makes sense to start thinking about the colours/rendering of new landcover.

[thor]

@math1985 understandable! With that in mind, is there any particular issues except for #947 that people could help with; with regards to colours of existing landcovers, does any overview/table of the various mappings exist?

Basically, how can a CSS-experienced OSM-addict help, if possible? 👍

[matthijsmelissen]

Sorry, I gave the wrong issue, I meant #941 rather than #947 (although probably both should be rolled out first).

does any overview/table of the various mappings exist?

The code can be found here: https://github.com/gravitystorm/openstreetmap-carto/blob/master/landcover.mss
A picture of the current rendering: https://cloud.githubusercontent.com/assets/5251909/4258446/def475f0-3ad2-11e4-81c9-b3d7c519aac9.png

All colours with a Lch vaue commented have been reviewed recently. All others still need reviewing. Some, like desert, look particularly bad. It would be nice if we could give all or most colours the same lightness, while still being able to distinguish them.

If you're looking for a small project, you could start with the sports colours. See also #919 for that.

Another thing that needs to be done is changing the landcover images so that they are a factor of 2 wide/tall, see #937 - but see also #938.

Probably not everything is documented as clear as it could be, but if you have any questions, let us know and we will clarify.

[matkoniecz]

natural=dune conflicts with more widely used natural=sand, is used on areas less than 800 times, never went through any kind of approval process. I would suggest dropping it from this list.

[matthijsmelissen]

natural=dune conflicts with more widely used natural=sand

And with natural=scrub, depending on the type of dunes. See here.

[polarbearing]

The problem is that natural formations are mixed with the substances in the tags. Beach can be sandy or stony, dune can be sand or nicely grown over with dune grass.

[dieterdreist]

2015-01-11 1:01 GMT+01:00 polarbearing [email protected]:

The problem is that natural formations are mixed with the substances in
the tags. Beach can be sandy or stony, dune can be sand or nicely grown
over with dune grass.

+1, this is clearly a tagging problem. natural=sand/mud/bare_rock are "bad"
because they are describing properties of an object rather than a class of
object, thereby creating tagging conflicts/incompatibilities. They should
go under another key where they would fit, e.g. "surface" or "landcover".
Unfortunately these tags are rendered currently and there are presets for
them, so that mappers are encouraged to continue using them.

[fkv1]

What's up with natural=fell now? If you can't decide how to render it, you may treat it as a synonym for natural=grassland or landuse=meadow. That's better than leaving the areas white.

[imagico]

The problem of natural=fell is that it is somewhat poorly defined in the wiki and accordingly used for a wide variety of purposes. Areas tagged natural=fell often consist at least in parts of the following:

• natural=grassland
• landuse=meadow, meadow=agricultural (i.e. pasture)
• natural=heath
• natural=wetland + wetland=bog
• natural=bare_rock
• natural=scree
• geological=moraine

As @dieterdreist explained this is partly due to the mixture of tags for landcovers and landforms we have in OSM. For rendering in the map with a color fill/pattern landcover tags describing the local observable properties of the land surface are better suited than landform tags where definition arises from the context and the boundaries are often blurry - in this case fell in a strict sense refers to the upper lying parts of a hill landscape near the tree line. Practical use of the tag is however much more vague and inconsistent.

natural=tundra has probably been set up to overcome this problem since it does not have the landform component but is purely ecologically defined. It is however also very broad and equally overlaps all of the above tags to some extent.

The most prominent use of natural=fell is probably for alpine tundra where coloring it like natural=grassland might seem appropriate. Due to the broad definition of natural=fell mapping however often is also extremely broad which makes it not so suited for rendering in grassland green like here:

http://www.openstreetmap.org/relation/1354524
http://www.openstreetmap.org/way/197815853
http://www.openstreetmap.org/way/27869519
http://www.openstreetmap.org/way/61374902

These comply with the wiki definition of natural=fell (except the last maybe) but IMO are too diverse to be rendered in grassland color.

From the mapping side i think it would be a better approach to tag alpine vegetation above the treeline as natural=grassland/landuse=meadow/natural=heath with a supplementary tag (alpine=yes or similar)

[fkv1]

On 03.05.2015 15:13, imagico wrote:

The problem of natural=fell is that it is somewhat poorly defined in the
wiki and accordingly used for a wide variety of purposes.

It has been part of map features for ages, and I think that the definition
is quite clear. If you find it unclear, discuss it in the wiki or in the
tagging mailing list.

Areas tagged
natural=fell often consist at least in parts of the following:

• natural=grassland
• landuse=meadow, meadow=agricultural (i.e. pasture)
• natural=heath

There may be some overlap with the above, but look, all of them are green.

• natural=wetland + wetland=bog
• natural=bare_rock
• natural=scree

These differ from natural=fell by definition.

• geological=moraine

This is independent from natural=fell, as a moraine may be covered by forest
or a meadow or whatever. And it seems that most geological=moraine were
created by an import. The tag was never proposed.

As @dieterdreist https://github.com/dieterdreist explained this is partly
due to the mixture of tags for landcovers and landforms we have in OSM.

natural=fell is basically a vegetation related tag. That's why it is in the
"vegetation related" section of the natural=* wiki page. That's how it is,
even though some people such as Dieterdreist don't like it.

There's no point in putting approved and widely used tags in question here.
natural=fell is used 9461 times, it is approved, and it has its wiki page,
thus it should be rendered. You may render it using a pattern of red hearts
and purple stars, although a light green or gray tone might be more intuitive.

The most prominent use of natural=fell is probably for alpine tundra where
coloring it like natural=grassland might seem appropriate. Due to the broad
definition of natural=fell mapping however often is also extremely broad
which makes it not so suited for rendering in grassland green like here:

http://www.openstreetmap.org/relation/1354524
http://www.openstreetmap.org/way/197815853
http://www.openstreetmap.org/way/27869519
http://www.openstreetmap.org/way/61374902

These comply with the wiki definition of natural=fell (except the last
maybe) but IMO are too diverse to be rendered in grassland color.

Some detailed mapping (splitting the areas up in fell/bare_rock/etc.) is to
be done in these examples. So far nobody cared, because natural=fell is
"rendered" white, and white looks like rock and glacier.

I've seen lots of areas which should be natural=fell but are actually tagged
landuse=grass just to make them look green.

This is a chicken-egg problem. You will not get people to use the proper
tags as long as you don't render them properly.

[gravitystorm]

it is approved, and it has its wiki page, thus it should be rendered

This is absolutely not how we decide what things to render.

[Nic787]

The problem about looking if people use x tag or not to decide to render is a bit weird.

From what I remember that wasn't why they decided not to render natural=fall. @imagico was pretty clear it's because the definition on the Wiki wasn't clear and because fall's can involve a lot of other natural features. Which I totally agree with. According to the Wiki falls are "habitats" and that's a pretty vague, ambiguous term that could mean literally anything.

In other words, it had nothing to do with a "roll of the dice" and more the inherent problems of the tag/Wiki article that no one (mainly pointing the finger at @fkv1 here) appears to have bothered to address or deal with in the six years they had an opportunity to (six years!).

Alright.

From my understanding after reading a bit of informations about Fell vs tundra.

Tundra = composed of small tree, shrubs, grasses, mosses, and lichens. Mostly cold and windy area. (North of Canada, Iceland, North of Scandinavia, North of Russia, etc.)

Fell = slope where it has a tundra or alpine vegetation, mixed with scree. Could be used in alpine, mountainous area where there isn't a lot of vegetation.

[fkv1]

@imagico was pretty clear it's because the definition on the Wiki wasn't clear

I already explained in #545 (comment) why this is nonsense and that he is supposed to ask on https://wiki.openstreetmap.org/wiki/Talk:Tag:natural%3Dfell if he does not understand what this tag means. This tag has been used 20000 times by mappers all over the world who obviously know what it means, and I can assure you that at least here in Austria it is used consistently according to its short description.

I suspect that those who don't get the meaning of this tag live themselves in areas where no fells exist. I suggest you look beyond your borders. If you don't want to travel, browse the web to see what the world looks like.

the inherent problems of the tag/Wiki article that no one (mainly pointing the finger at @fkv1 here) appears to have bothered to address or deal with in the six years they had an opportunity to (six years!).

My above-mentioned reply was on 2015-05-03, less then 3 hours after Imagico's post. He never answered, nor did he raise his concern on the talk page in the wiki as I suggested.

[Nic787]

My suggestion :

I don't know if there area already something using those colors.

[imagico]

For info for @Nic787 who seems to be new here - this issue has been closed and the and the request to render natural=fell has been rejected by the maintainers for the reasons provided above in the discussion. Without new arguments being provided (in particular on the key points of the discussion that led to closing this) it is unlikely that we will reach consensus to re-open this.

Like @matkoniecz i will not engage in an off topic discussion or a re-iteration of discussions we already had on this matter. Discussion on the supposed meaning of tags or the meaning of the English language terms used for tag values should go to the wiki or the tagging mailing list.

[fkv1]

@Nic787:

My suggestion :

The darker green may be ok for tundra, but not for fell which in the real word is lighter green and often intermixed with white rock fragments (depending on rock type of course). I have suggested (and still do) to render natural=fell like landuse=meadow and landuse=grass. I don't know how natural=grassland is currently rendered, but it should also be the same.

@imagico: Why do you keep posting here when you admit that you are not interesting in engaging in the discussion?

Discussion on the supposed meaning of tags or the meaning of the English language terms used for tag values should go to the wiki or the tagging mailing list.

Just to put it clear: It's you, not me, who does not understand what natural=fell means. So it's up to you to start that discussion in the wiki, as I've already suggested here twice.

[matkoniecz]

So it's up to you to start that discussion in the wiki, as I've already suggested here twice.

People are not obligated to rescue/discuss/improve all tagging schemes. More specifically, if some tag turned out to be low quality, poorly defined, mismatching mix of various things, and such issues were raised since 2010 ( https://wiki.openstreetmap.org/wiki/Talk:Tag:natural%3Dfell ) other people are not obligated to start discussions about it.

I have long queue of things that I want to do in OSM ( partial TODO list on https://wiki.openstreetmap.org/wiki/User:Mateusz_Konieczny ) and inventing a good tagging scheme for alpine landcover is not one of them.

Presumably @imagico also has list of things to do in OSM that is quite long.

Feel free to do it if you care about such tagging. Maybe things changes since 2015 ( #545 (comment) )? But comments not adding anything new are not useful and may result in moderation tools being used, such as disabling ability to comment on that issue.

The darker green may be ok for tundra, but not for fell which in the real word is lighter green and often intermixed with white rock fragments (depending on rock type of course).

Note that this project makes map, not vector equivalent of aerial imagery. industrial areas rarely are purple, typical motorway is not red.

I don't know how natural=grassland is currently rendered, but it should also be the same.

Please, check such things before commenting.

[fkv1]

People are not obligated to rescue/discuss/improve all tagging schemes. More specifically, if some tag turned out to be low quality, poorly defined, mismatching mix of various things, and such issues were raised since 2010 ( https://wiki.openstreetmap.org/wiki/Talk:Tag:natural%3Dfell ) other people are not obligated to start discussions about it.

All the issues that were raised back than have been resolved meanwhile. The issue with tundra has been resolved by the addition of a separate natural=tundra tag (see Rudolf's comment from 2 April 2014). The issue with bare rock has been resolved by the approval of a natural=bare_rock tag in 2012. The issue with Wikipedia has been resolved by keeping the old definition (again, see Rudolf's comment from 2 April 2014). Nobody felt the need to re-open any of those issues in the last 7 years, or to raise new issues. You'll hardly find any other tag with so many instances and so little controversy.

I have long queue of things that I want to do in OSM ( partial TODO list on https://wiki.openstreetmap.org/wiki/User:Mateusz_Konieczny ) and inventing a good tagging scheme for alpine landcover is not one of them.

You don't need to invent a new tagging scheme. The tag is already there. Just render it light green. That's it.

Presumably @imagico also has list of things to do in OSM that is quite long.

Don't you believe that I have nothing else to do than to explain you a tag that is perfectly defined in the wiki? It's offensive (to say the least) to tell me that your time is more precious than mine, and it really has nothing to do with the question how to render natural=fell.

But comments not adding anything new are not useful

What news does your own comment add?

and may result in moderation tools being used, such as disabling ability to comment on that issue.

Wow, the great dictator. Silencing everyone who doesn't share your views.

The darker green may be ok for tundra, but not for fell which in the real word is lighter green and often intermixed with white rock fragments (depending on rock type of course).

Note that this project makes map, not vector equivalent of aerial imagery. industrial areas rarely are purple, typical motorway is not red.

You could also render water red and forests blue. Have you ever wondered why water is blue in all maps?

A map should be intuitive. There may be no intuitive way to render industrial areas distinctively, but there is an intuitive way to render areas vegetated by green plants.

Same as for point symbols. A guidepost is rendered with a symbol that looks like a guidepost, because that's intuitive.

I don't know how natural=grassland is currently rendered, but it should also be the same.

Please, check such things before commenting.

I'm here to give you my expertise as a mapper and moutaineer with lots of experience with natural=fell, not as a coder.

[Adamant36]

I suspect that those who don't get the meaning of this tag live themselves in areas where no fells exist. I suggest you look beyond your borders.

Sure, because where you live is the only place in the world with tree lines, "sloped mountainous area where there isn't a lot of vegetation", those things are rare in the world, or like you know where any of the people who think the tag is problematic live in the first place 😆

This tag has been used 20000 times by mappers all over the world who obviously know what it means

I'll give you the benefit of the doubt that you have asked everyone who has used it that they know what the tags means. That aside though, it seems to be mainly used in Europe and the last time I checked Europe isn't "all over the world." If you think it is, then likely you need to look beyond your what I assume are European borders 👍

Just to put it clear: It's you, not me, who does not understand what natural=fell means. So it's up to you to start that discussion in the wiki, as I've already suggested here twice.

Usually it's on the person who is requesting the tag be rendered that has to and should put in the foot work to make that happen. Which includes addressing any issues the Collaborator's have. Someone can't just request any random tag be rendered and expect them to bend over backwards to make it happen just because they asked. That's not how this works.

All the issues that were raised back than have been resolved meanwhile.

No they haven't. A quick read over the Wiki page makes it clear that the tag is still vague and article's tone is almost recommending people use other tags. Also, where's a link to a mailing list discussion about it?

Don't you believe that I have nothing else to do than to explain you a tag that is perfectly defined in the wiki?

It's far from "perfectly defined" in the Wiki. The whole first of the article is nothing but back peddling on the definition and hedging the tag against other ones. That's not well (or perfectly) defined by any means.

Wow, the great dictator. Silencing everyone who doesn't share your views.

Your being pretty defensive about this. Usually moderation is a perfectly legitimate way to do deal with a discussion where one side is going off on rants about how the other side needs to get out of their basements. It's nothing to do with silencing anyone or being dictatorial, but keeping things on topic and rhetorically toned down. Which I say as someone who has had a few issues I've participated in blocked from further discussion because of my defensiveness.

I'm here to give you my expertise as a mapper and moutaineer with lots of experience with natural=fell, not as a coder.

it doesn't take any coding to look at the map on the OSM website to see what color natural=grassland is rendered as. It's even shown in the natural=grassland Wiki article. Clearly you'd know that if you looked beyond your borders yada yada Etc. Etc. though 😆

[pnorman]

Temporarily locking as this has become too heated and unproductive.

[Gruebeln]

Hi, I have created a proposal in the OSM wiki. Hopefully it can solve the problem with natural=fell and natural=tundra rendering:

https://wiki.openstreetmap.org/wiki/Proposed_features/stone%3D*

[hungerburg]

In the area of my local knowledge fell is almost exclusively used for what can be seen on the photos here https://community.openstreetmap.org/t/identity-crisis-am-i-a-tundra/114432

I created a mock-up of how OSM-Carto could display those mappings around Rumer Spitze:

For comparison, the current rendering:

The colour used purely came out of shifting controls at random :) Pale by design, these are not fertile grounds. Must be distinct from scrubs, grasslands and heaths, those are mapped separately.

On the ground it is a mix of mostly grasses and rocks, a few flowers, herbs, erica and thistles perhaps. This being a mix certainly a problem: Patches of components usually just too tiny to map separately! Nature is not much forthcoming such clean delineations as some wish for.

PS: There is no meadow south of Thaurer Jochspitze, while around Pfeis hut there is more like grassland.

PPS: Here matte do to own colour experiments:

PPPS: One colour experiment - Retag fell into grassland:

The look of it makes me immediately understand why blank area preferred by the mapping community. I guess I will have to retag the meadow south of Thaurer Jochspitze into fell. It clearly is what CORINE calls "sparse vegetation". (Actually it is our local geographers - https://tirolatlas.uibk.ac.at/wsgi/data/corine/text?id=8120010&lang=en - The grasslands they call "Alpine Mats" - https://tirolatlas.uibk.ac.at/wsgi/data/corine/text?id=8120007&lang=en )

Final Update: Here how the Pros classify -- wms:https://image.discomap.eea.europa.eu/arcgis/services/Corine/CLC2018_WM/MapServer/WMSServer?FORMAT=image/png&TRANSPARENT=TRUE&VERSION=1.3.0&SERVICE=WMS&REQUEST=GetMap&LAYERS=13&STYLES=&CRS={proj}&WIDTH={width}&HEIGHT={height}&BBOX={bbox} -- There is a lot of sparsely vegetated area, yet Pfeis considered grassland :)

Of course, they have strict sounding guidelines. Now, after all this research, a bit clueless … Therefore put that up to debate in the forum.

Umpteenth addendum: Both map-tiler and mapbox render fell same as grassland, at least map-tiler also renders refs on paths:

[imagico]

@hungerburg - this is more of a mapping/tagging discussion than about rendering. None the less a few pointers on the matter of border cases/mixed properties in landcover/landuse mapping:

• it is generally accepted that mappers are free to map things in any spatial level of detail they see fit. In the area you discuss this level of detail is somewhere between several hundred meters and a few meters.
• it is also widely accepted that at the chosen level of detail the classification should be based on what is considered the dominant element present. That classification is not invalidated by the presence of another minority class (iconic example: a landuse=residential area remains residential even in the presence of a shop within that area).
• ways to indicate additional landcover/landuse classifications subordinate to the primary one in general have been discussed in the past (like here or more narrowly here) but no such concept has gained significant traction so far. The only case i am aware of where such an idea has been truly established is for the specific case of Streuobstwiesen (https://wiki.openstreetmap.org/wiki/Tag:orchard%3Dmeadow_orchard).
• primary landcover/landuse tags explicitly indicating a mixture of two or more individually distinct classifications are almost completely absent in practical use in OSM.

[hungerburg]

The voting section of https://wiki.openstreetmap.org/wiki/Proposal:Multiple_landuse features a strong party for stacking overlapping polygons -- another contender to solve the fell issue?

While this way freedom to mix whatever is provided, and OSM-Carto already makes some use of that by applying heuristics in such stacks with other mappings, the fell ambiguity cannot be solved that way and this scheme also cannot map sparsely vegetated areas distinctly at all, because there is no way to tell, which is the dominating natural.

It is a pity, that fell was defined so badly.