Proposal talk:Bare rock

From OpenStreetMap Wiki
Jump to navigation Jump to search


I would prefer to render not a filling, but just a dotted line for the vegetation-limit. Or a rock-symbol-filling with alpha, so you can still see the hillshading / elevationcolors. -- Dieterdreist 18:19, 19 July 2009 (UTC)

My is only an idea but a rock-symbol-filling with alpha is better --Cade 06:08, 20 July 2009 (UTC)

Single big rock?

In Finland, some maps have signs for single, noticieable rocks. That's what I thought this tag to be used for. Example:


--Japa-fi 14:55, 10 August 2009 (UTC)

  • I totally agree this feature should be added as node, not area. Big rocks are important navigational-marks while hiking, and can also double as emergency shelter. I don't see much point in marking an area with this tag, as we already got '=scree' and '=cliff'. Antwelm 01:01, 3 September 2009 (UTC)
When i make the proposed of this tag I hadn't considered for a single rock! But we could extend the use alsa at point and not only for area! For 'scree' and 'cliff' their are a particular and specific type of rocks area.--Cade 06:16, 28 September 2009 (UTC)
I have taken a look at IOFmapping#Rock_and_boulders, it is worth looking at, maybe we should have a natural=boulder. By the way, the tag natural=rock is great for what IOF call bare_rock./Johan Jönsson 20:00, 30 September 2009 (UTC)
Natural=cliff doesn't render for areas, and there are plenty of areas defined by a cliff edge which are then a slope of rocks into the sea. Natural=rock would mark those areas (usually between the attractive clifftop path and the attractive beach below!). Rendering cliff for areas is tricky because they have a distinct high edge that's not render-obvious, so this would be useful down here on the southwest England coast too. Andygates
Note, that there already exist Tag:natural=stone to be used for this. --Kslotte 09:49, 28 March 2010 (UTC)

Submerged rocks

Should also apply for submerged rocks, such as some skerries and navigational dangers, together with a depth=* tag, and for single rocks that mark out landscape features in the same way as dominant lone trees natural=tree --Skippern 18:26, 19 July 2009 (UTC)

Rock islets

Some islets are also termed rocks. --seav 18:30, 19 July 2009 (UTC)

It's true, we always extend the tag for all rock and not only for the montain areas. --Cade 06:12, 20 July 2009 (UTC)
Along coastline, single big rock islets are a useful navigation feature too - I definitely support rendering a 'node' rock. Perhaps as a slightly-irregular black dot? --Andygates

What about other features

What would this tag be used for, that is not already covered by natural=scree? There is also a natural=rocks documentet, which could easily be confused with natural=rock. Gustavf 20:32, 19 July 2009 (UTC)

There's several features that have a rock face visible:
  • natural=scree has loose rocks lying around
  • the thing that this proposal addresses, that is reasonably flat bedrock visible as a larger area, usually in mountainous or hilly regions
  • natural=cliff as a (near) vertical rock face, but possibly also used for the previous item
  • boulders and other distinct large rocks (I thought I saw a proposal for these...)
  • (the proposed) natural=rocks as sea navigation hazard; a partially or fully submerged rock formation
  • I guess some would use natural=rock instead of place=islet for very small islands... Alv 06:33, 20 July 2009 (UTC)
The tag natural=scree is only for scree and not for hard rocks, for example like [this].The tag natural=rocks is only a stub and you can't find it in the proposed features, so apologize me but I didn't know it ;); besides {{tag|natural|rocks} page is only for rock on water surface so why not create a tag that will be used with all rock areas and not only for mountains on a water surface?--Cade 06:33, 20 July 2009 (UTC)
Why would someone tag a small island with something described as "A key to tag mountain areas made only by rocks, where there isn't vegetation or it's minimal."? And how do you make a description of this tag and the scree tag, to make sure it is not confusing? I am not saying this tag is a bad idea, but that it needs more work to make sure it is well defined and easy to understand. -- Gustavf 08:07, 20 July 2009 (UTC)
Yes, you right; if we make a general tag the description will be edit--Cade 16:44, 20 July 2009 (UTC)

The tag natural=rocks

The similar tag natural=rocks is Undefined (inactive), it's not in the Proposed features page and never it was vote. Besides is use is very limited and natural=rocks tag will be include in rock tag.

Its also irritating to have two subtley different tags with very similar names. Ruan 08:28, 28 August 2009 (UTC)

More descriptive tag name

Please use a more descriptive tag name. Just looking at natural=rock it could mean: a distinctive single boulder, a partially submerged navigation hazard at sea or an expanse of bare rock surface, etc. --Cartinus 20:37, 19 July 2009 (UTC)

For a rocky seabed I would use seabed=rock, for me rock is submerged or semisubmerged rocks forming a navigational hazard, we can also tag natural=boulder and natural=scree, and obviously I need to reads natural=rocks. I would hesitate to use a landcover tag that have a syntax very similar to other tags, and natural=rock is too close to natural=rocks --Skippern 02:51, 20 July 2009 (UTC)
For this (large visible non vertical rock surfaces) I'd go for bedrock. There's some talk at natural=cliff that it could be used for rock areas, too, but it's a bit ambiguous if used so. Alv 06:33, 20 July 2009 (UTC)
natural=bedrock is probably a good idea for anywhere where bedrock is visible, maybe this proposal can adopt that? such change will also make it significantly different from natural=rocks, reducing the probability of them being mixed. --Skippern 09:35, 20 July 2009 (UTC)

I changed the description of the tag

Since we are likely to cause problems with the existing tag, I changed the description of the tag to make use of the tag more general and not restricted to the mountains. Now the proposal is for a tag that indicates the areas made principally of rock, inside which there are cliffs, scree, coastline, glacier and more.

Note that we will use this tag for the Corine Land Cover import in France. This is called "Bare rocks" in their nomenclature ([1]). -- Pieren 14:37, 23 July 2009 (UTC)
Since glaciers consist of frozen water and not of rock I removed that. --Cartinus 02:01, 24 July 2009 (UTC)
Ok, but usually under the glacier you find rocks, and the rendering will cover the rock area with the glacer--Cade 05:11, 24 July 2009 (UTC)
If claciers are included in rock, than forests can also end in this category. Besides natural=glacier is already tagged. --Skippern 07:54, 24 July 2009 (UTC)
You right, I just edited --Cade 21:51, 24 July 2009 (UTC)

I think this is a very useful tag in west Sweden coastlands. A question on this definition: "areas made principally of rock, inside which there are cliffs, scree and coastline.", Does it mean that inside a area of natural=rock, there could be smaller areas(maybe added later with more precision) of natural=scree (rubble), natural=cliffs (slopes) and that flat areas of solid bare rock simply is tagged natural=rock? /Johan Jönsson 14:55, 27 September 2009 (UTC)

Yes, it does!--Cade 06:08, 28 September 2009 (UTC)


Would it be worth expanding this to include rock material? natural=rock with rock=granite for example.Ruan 08:28, 28 August 2009 (UTC)

I think that this is possible but it also have a secondary importance--Cade 06:09, 28 September 2009 (UTC)

So let's use this

I don't see a reason not to use this tag. If you have an area of exposed rock that is neither cliff (not steep) nor scree (not loose - and well, "scree" is a pretty well defined term), then mark it "natural=rock". Shall we formally vote? Stevage 13:36, 11 December 2009 (UTC)

We shouldn't use it because, like stated numerous times by people above, the meaning of such a tag would be unclear. No wiki definition will ever fix that. --Cartinus 06:54, 12 December 2009 (UTC)
At least the tag natural=rocks should be scrapped before this is to be taken into use, or find another way of tagging this. Two tags with almost identical names for similar things is very bad. --Skippern 10:51, 12 December 2009 (UTC)
I say, lets use it for land cover. But change it to natural=bare_rock to make it somewhat less ambiguous, just "rock" seems to mean too many things. I am not native in english but the dictionaries i have used seem to say that bare rock means visible solid stone ground. This is supported by IOFmapping#Rock_and_boulders/Johan Jönsson 17:38, 26 March 2010 (UTC)
I support the tag natural=bare_rock. IOF has probably done some research about correct naming. --Kslotte 09:51, 28 March 2010 (UTC)
bare_rock sounds good. I've got use for this tag, as soon as it is approved. Indigomc 20:00, 28 March 2010 (UTC)
Don't wait for any approval just tag. I just made the bold move to change the proposal ;) --Cartinus 21:10, 30 March 2010 (UTC)
I am already tagging, rock domes (tepuyes) in the Colombian Jungle, see here on OSM [2] or as photo here: [3].--Federico Explorador 02:27, 4 February 2011 (UTC)

We have plenty of big featureless rocks and rock masses in Australia!

I think the definition is perfectly clear. A simple example that most would understand is Uluru. Have a look at how it currently renders [4], simply a peak. Doesn't do it justice. :)

There are many such rocks and sheets of rock, in Australia. In such a vast continent, this tag as a polygon would be very sweet.

PS we've plenty of salt lakes too, what about a tag for that? :)

PPS I agree. Scrap 'rocks'.

Agreed. There are large areas of rock shelf on the coastline of Australia that could/should be tagged as natural=bare_rock.

bare_rock proposal

natural=bare_rock is supposed to map the landscape type or the surface characteristics in areas where the bedrock is mostly uncovered leaving large areas of bare rock in big slabs or as a featureless area. It could be smooth or rugged with cracks and crevasses and don´t imply any special class of passability. It could be plain or leaning, when it gets to steep consider natural=cliff instead.

The original natural=rock proposal where supposed to map all areas mainly uncovered by vegetation, including stony ground, scree and the like. natural=bare_rock is for solid rock.

The area should have little or no vegetation. In some areas there is a natural development with liches->moss->grass->shrubbery, one rule could be that the area form sparsely vegetated areas where 90 % of the land surface is covered by rocks.

The above almost follow the definition [5] of "European Topic Centre on Land Use and Spatial Information", with the exception that they include mountain top debris, a.k.a. scree.

Here is a table (please feel free to edit)

Value for iof= key notation Element Comment Rendering example (not to scale) natural=bare_rock used? corresponding osm tags

node way

Passable or impassable cliffs.


NO natural=cliff barrier=retaining_wall

optionaly: foot=passable



Rock pillars or gigantic boulders

IOFrock pillars.png

MAYBE natural=cliff area=yes


Boulders of different sizes


NO natural=stone or maybe a new natural=boulder.

node area

Stony ground

IOFstony ground.png

NO natural=scree although the name is rather specific for mountain slopes.


Bare rock

IOFbare rock.png


Steep cliff areas

There seem to be a problem with rendering natural=cliff as areas acc. to Andygates above. If the extended steep area isnt a scree it probably would be suitable with natural=bare_rock, unless there is some other tag for steep areas.

Rock pillars

There are some natural features that rise high straight up, both on land and in water.natural=cliff should be the first choice, but when they get bigger, the whole area could be tagged with appropriate surface cover. If its bare rock then use that.

Australian "rock"s: Ayer´s rock and their like. I´ll let User:Wahroonga farm comment on this.


These big stones should be mapped with natural=stone. See discussion above

Submerged rocks and other marine stuff

Skerries, navigational dangers should be bale to use natural=stone, or is it not consistent with marine language? Small islets that is mapped as an area could be tagged with appropriate surface cover. If its bare rock then use that. See discussion above

For a rocky seabed Skippern recommends seabed=rock or more precisely seabed_surface=rocky

Rocky coasts

I guess there are some proposals on different coastal types, if there isn´t use bare_rock. I think natural=beach is only for flat areas, mostly sand and gravel but maybe also for the bare_rock limestone pavement beaches?


There is also the tag landuse=quarry, but if we want to follow the rule "don't tag for the renderer", so we need the tag natural=bare_rock for areas of rock, in order to tag correctly.

What is the point you are trying to make? There is nothing natural about a quarry, so landuse=quarry is obviously more correct than any natural=* tag we can dream up. --Cartinus 13:01, 6 February 2011 (UTC)

Stone deserts

hamadas and reg(see wikipedia), largely barren, hard, rocky plateaus, with very little sand. Could be tagged natural=stone natural=bare_rock. /Johan Jönsson

natural=stone describes single (or a couple of) stones according to it's wiki page. So that doesn't sound like a good tag for a stone desert. --Cartinus 11:43, 2 April 2010 (UTC)

Sand and salt deserts

erg and other sandy deserts should have another tag. Also Salt lakes.

natural=rock was used for similar purpose

Just to add a note, we used natural=rock in the Corine Land Cover french import for something that is really close to this proposal sletuffe 18:03, 27 January 2011 (UTC)

Summary of RFC januar 2011

I sent the natural=bare_rock to the mailing list for comments. Here is a summary: tagging maillist. In short, there is a need for a tag for this use but I found two things to develop more:

  • landcover=bare_rock instead of natural=bare_rock, see Proposed features/landcover
  • should this tag only be for solid rock or should it also cover splintered blocks and loose rock?

I am now reverting this proposal to a draft./Johan Jönsson 13:16, 6 February 2011 (UTC)


There is a Proposed features/landcover that I think wants to:

  • use natural= for distinct natural features like springs, waterfalls, scree. things that stand out.
  • use landcover= for general description of the terrain/landform.

I am not quite clear over the subtle differences but I have found several lists of land cover classification. I find the United Nations Food and Agriculture Organization, FAO, classification pleasing. See FAO land cover definition. In the appendix they use a way to classify bare rock surfaces and on the lowest level of classification they differentiate between the unfragmented rock called bare rock and the fragmented rock called Gravel/Stones/Boulders.

If the landcover-key is getting in use then a tagging like this would be good: For bare areas (no or sparse vegetation) of rock:

fragmented=yes/no or maybe bare_rock=bedrock/boulder/gravel

If the feature have a specialized name, i.e. stone_run or hammam, then add natural=stone_run or natural=desert + desert=hammam. I guess that the area of the natural-tagging do not have to follow the area of landcover-tagging. /Johan Jönsson 16:01, 6 February 2011 (UTC)

I don't agree, because the discussion on "landcover" is much more general and affects many other tags; so at the end this proposal gets postponed indefinitely.--Federico Explorador 17:52, 3 June 2012 (BST)
I do agree. There is not necessarily a need to have a general discussion. Simply start applying this logic to new tags. As the landcover key is not needed otherwise you can also add this to existing geometry with tags like land use or natural and others. --Dieterdreist 23:40, 22 July 2012 (BST)
There are common cases were "landcover=bare_rock" is used instead of "natural=bare_rock", because "natural=*" is already used for linear features (notably "natural=coastline"); we cannot write "natural=coastline;bare_rock", simply because it breaks the generation of contiguous coastlines, and causes severe havoc in the file generated for later use by Mapnik, with holes arbitrarily filled by strait segments (correcting these bugs in the Mapnik rendering takes months before we realize that this is the coastline file that is incorrect and that the tool generating it does NOT recognize anything else than "natural=coastline" without any extension.
Ideally, linear-only features (like coastlines, or cliffs) and surfacic features (including boundaries which have simultaneous roles) should not use the same natural=* tags.
The only alternative for now would be to duplicate the géometry, or to create a multipolygon containing the other tag necessary for "natural=bare-rock", but this would create a multipolygon with only 1 member for tagging rocks that are also islands with the coastline exactly at the same geometry as the rock itself...
We could solve the warning about the 1-member relation by splitting the rocky island into two separate "natural=coastline" paths for closing its border, and creating the multipolygon tagged with "natural=bare_rock". But various tools or users will detect two successive ways that are members of the same set of relations, and have the same tags, they will merge these paths and then will see again that the relation has only one member, and will attempt to correct this: which value (bare_rock or coastline) will they indicate for "natural=*" ????
Is it too late to change Mapnik's coastline generation tool so that it will recognize another tag for its linear features ? e.g. "natural:line=coastline".Or the alternative "natural:area=bare_rock" (less problematic for the transition and compatibility).
Why not introducing the suffixes ":line" and ":area" to existing tags when necessary for creating a distinction in any tags (that are applicable to closed ways or closed relations, including recursive relations using member multilinestrings)?
Note that this is different from the tag "area=yes/no" (which cannot create any distinction between other distinct tags but applies by default to every tag in the object). The tag "area=yes/no" will be significant only for tags that are not distinguished explicitly by a ":line" or ":area" (may be also by ":node") suffix in their name.
Verdy_p (talk) 08:43, 16 July 2013 (UTC)
The idea behind "warnings" created by error checking tools, is that once you see the warning you then turn on your brain and make a decision whether it needs correcting or not. The goal is not to remove all warnings. So either create a duplicate geometry or a multipolygon and the problem is solved. --Cartinus (talk) 02:33, 18 July 2013 (UTC)
I did not' speak about removing warnings but about correct interpretation of ambiguous tags. Here the main problem is that natural=coastline conflicts with natural=bare_rock on the same tag but two distinct values (Ok this only happens on small islets, but still we will need to duplicate the geometry; the other solution beiwing to create a relation for the land area containing the way for the same coastline; if the coastline of the islet is already splitted into two ways, we already need a relation anyway for the surface; the coastline normally remains a set of connected ways and not a relation). I won't vote for duplicating the geometry when it just complicate things for small items. But the problem is more general than that: the main cause being that naural=* is too much overloaded for things that are completely unrelated (coast lines by themselves are not geographic objects, they are estimated from some rough average, and never visible in reality with the same precision (coastlines are varying each day, sometimes on more than several kilometers in some bays or during the same lunar month). — Verdy_p (talk) 03:36, 28 July 2013 (UTC)

tag natural=bare_rock according to taginfo

On June 3, 2012, this tag has been used 2.243, so let's go forward! See