From OpenStreetMap Wiki
Jump to: navigation, search

two meanings for "cliff"?

a) "cliff" = material
b) "cliff" = up/down

Sometimes a "cliff" is a wide flat rock in a flat landscape. Or a small steap coast line. Or a climbing rock. Or a rocky landscape. Or the Swiss Alps. Or a mix of all...

How do we distinguish this all? --Markus 09:45, 31 May 2009 (UTC)

(a) It seems to me that natural=cliff indicates both "material"=stone/rock and "up/down", this is a useful tag. Additionally I propose a tag for a landscape with bare bedrock, as in mountain areas and other eroded places, influenced by IOFmapping i propose natural=bare_rock. /Johan Jönsson 14:15, 27 September 2009 (UTC)

how to draw it anyway

Some thoughts about this tag:

there are cliffs in both directions, to the left as well as to the right. The tag is not made for this case
there are cliffs surrounding a flat space or a cliffy area. When using the tag for an area, what does that mean?
use as a point could be nice to mark climbing gardens. It does not supply a geographic description, as there is no right or left. Mark a mountain's peak as the topmost point if you want.

I'd suggest to use it as a way if it is a crête (french) or Grat (german). So the slopes are on both sides. We could tag it as oneway=yes, meaning the cliff is on the right side. When used as an area, I interpret it as marking an unpassable area, the cliffy sides on the outer edge. I don't know how to use it if it does not mark a closed area. The area use also does not tell me whether there is more cliffy ground on top of the area, i.e. when passing the outer cliff, what comes after? And what about a cliffy area which surrounds a hole or valley? An abandoned quarry, for example.

So, use as an area does not really fit to mark the rock's edge, but marks a cliffy area instead. Marking the cliff in itself needs a nonclosed way w/cliff=yes. --Thomas 07:35, 09 June 2009 (UTC)

Cliff areas

How should Mapnik render cliff polygons? Where a cliff is at a moderate slope, for example along Devon's Jurassic Coast where it's crumbly sandstone, the cliff occupies a section of land. Commonly this is a thick variable area between a clifftop way and a beach or coastline at the bottom.

Marking a cliff way at the top is OK for marking the edge of a vertical (Dover-style) cliff but it would be good to render these cliff polygons in some way. Rendering the enclosing way is incorrect: that suggests a crater lip. Andygates

There are some proposals, that it would be better to separate steep areas and solid rock faces, but none has gained any further support: natural=bedrock (or similar) for the visible rock face and natural=cliff separately along the highest edge of the steep area. Even then it's a bit incomplete description, but better than a crater or table outline. Alv 11:48, 25 October 2009 (UTC)
That's pretty good: it lets the cliff way define the dangerous lip edge, and fills in the otherwise unmapped area. Andygates


Hi. I've been trying to tag the edge of a Motte (specifically Bedford Castle), and came to the conclusion that this tag was the nearest thing we have, since it describes a steep slope.

What do people think about using something like slope = 45 degrees in conjunction with the cliff tag for stuff which isn't vertical? Ojw 17:38, 2 August 2009 (UTC)

-- I've been using man_made=embankment for these, when mapping hill fort bank-and-ditch structures. They're definitely not natural. Andygates

Crater Rim

I have put a cliff around the "Meteor Crater" in Arizona here [1]. Maybe a new area key called "crater" would be useful. It could use the same rendering as cliff in a large scale. Tagging could be like: natural=crater, crater=meteor/volcanic or even caldera=major/minor with the main caldera rendered in more scales than all only showing up in close viewing. On a second thought maybe crater should only include meteor/volcanic(including maar?) with slopes on the inside while caldera should go to volcano with slopes on both sides. --T.woelk 09:14, 26 September 2009 (UTC)

I found a set of curious craters in Morocco ( and thought it was worthwile mapping them. so I went to look for other volcanic areas I know of and check how they were represented. to my surprise nothing has been agreed upon up to now and some of the most notable examples of volcanic areas are not mapped at all (I checked Vesuvius, Campi Flegrei, Hawaii, Stromboli, Vulcano). so I queried openstreetmap for the term "crater" and I found this one which looks very recognizable ( this is how I came across this natural=cliff. I think it would be a good thing if we could agree on a "best practice" for mapping craters. --User:Mariotomo 2012-09-26 10:58UTC

Classification of drop-offs like cliffs and other near-vertical natural features

Consider the linked picture There appear to be at least three classes of near-vertical features:

  1. a huge near-vertical drop-off (like the mountainside in the picture, casting a large shadow); height probably > 100 m
  2. a small near-vertical drop-off (like the ridge near the foot of the mountain, immediately to the right of the conspicuous white building in the left half of the picture); height maybe 8–10 m
  3. other near-vertical features that you cannot ascend or descend without using your hands, but that are not readily identified as such from a distance due to their small size; height perhaps 2–3 m

(I don't have a picture of type (3) at hand, but gives you an impression of the scale that I'm talking about here. In this kind of terrain, even a very small outcrop of just a metre or two can form a near-vertical barrier.)

In this picture, the 100 m drop-off (1) is immediately recogniseable as a cliff, even when viewed from a large distance. There probably is no doubt that natural=cliff is correct for this one. Whether or not the second kind (8–10 m drop-off) can reasonably considered to be a “cliff” largely depends on the map scale and the size of the surrounding drop-offs, if any. In this particular case, the height seems to disqualify the feature for the term “cliff”, but its length and the fact that it sort of merges into an even steeper area of the mountainside may indeed justify tagging it using natural=cliff. In the end, it remains a decision of cartographic judgement. Finally, minor drop-offs like (3) above might be of great importance to hikers trying to find a path down a mountain or something, but can hardly be called a real cliff; it feels more like a barrier of some kind. barrier=retaining_wall would work pragmatically (those hikers would be able to interpret it correctly), but is wrong semantically because it implies a man-made feature.

The three classes above can perhaps also be roughly expressed in an alternative way:

  1. drops that most certainly kill you
  2. drops that will at least seriously hurt you
  3. drops that you stand a good chance of walking away from more or less unscathed

Using height=*, we are already able to tag the exact size of the feature and leave the problem up for the renderer to solve it. Most (all?) current renderers are unable to achieve the kind of context-sensitive classification (generalisation) that would be needed in this situation. Therefore it seems like a good idea to hint at a possible classification the same way tracktype=* does (in addition to the height, if possible).

Existing tags that see some use are:

For minor drop-offs of type (3), I like the barrier=cliff concept a lot: It emphasises that the feature is a natural one and that it is more of an impediment than a place of interest or a fall hazard.

[User:Arne Johannessen|Arne Johannessen]] 17:14, 1 August 2012 (BST)

To differentiate I guess one could use another tag for the really big mountain side cliffs, maybe mountain_face or precipe. There are some mountain terms in the wiki, natural=ridge, natural=arete. Maybe the climbing community have some nice words for different kinds of cliffs, maybe even a scale (as the mountain biking-people have at mtb:scale). See alsohiking.
I didn´t know that small cliffs wasn´t called cliffs, I wonder what the english-speaking people use instead.
Landmark=* have been suggested instead of conspicious. /Johan Jönsson 18:18, 1 August 2012 (BST)

When natural=cliff is part of natural=coastline

How is this solved? Draw a cliff parallel to the coast? Relation? Skippern 19:00, 1 July 2011 (BST)

In this case we shouldn't use values like "natural=coastline;cliff" because rendering could produce an "overflowing" see. It is better to produce a new way.If the cliff is absolutely vertical, this cliff-way can use the nods of the coastline, but normaly cliffs are more or less inland.--Hurdygurdyman 08:51, 14 October 2011 (BST)
+1 for this. As the page notes the way of the cliff should be drawn on the _top_ side of the cliff. It is very rarely _exactly_ on top of the coastline (where the water meets land). --JaakkoH 11:34, 2 January 2013 (UTC)


Sometimes a cliff is drawn in the wrong orientation and rendered therefore the wrong way. To fix it, now it is necessary to drop it and redraw it (hopefully) correctly. What about having a "reverse=yes", to inform the renderer to draw in the opposite orientation quickly fixing such errors?

You can reverse the way, no need to delete and redraw. Keyboard shortcut 'r' in JOSM, 'v' in ID (the "<<" icon in the round popup tools), and also in Potlatch 2 it's 'v' (or the arrow button right next to the trash can button). Alv (talk) 14:54, 27 October 2013 (UTC)

Combine bare_rock + cliff areas?

Currently it is not very well defined how to specify vertical and horizontal (other than the length) extension of cliffs. One possibility that I see is an area natural=bare_rock with an overlapping way natural=cliff on its top. RicoZ (talk) 09:33, 3 December 2013 (UTC)

Artificial cliffs, e.g. in quarries

How are artificial cliffs, e.g. in quaries, tagged?

In an active quarry this is nearly impossible and also quite nonsensical, because they change. But in an abandoned quarry there are rock faces that are definately not natural, and are not embarkments. So how do we tag those? natural=cliff will render correctly, but we do not tag for the renderer. So, do we need a new tag for that? man_made=cliff? --Gormo (talk) 14:05, 9 February 2015 (UTC)

I tag them natural=cliff anyway, because a cliff is a cliff. I see no reason why we should distiguish natural/man_made cliffs. When a quarry is tens or even hundreds of years old, it's almost impossible to tell whether a cliff is natural. It's the same as with trees. We use natural=tree for all trees, no matter whether they are planted by humans. The only difference lies in history. In OSM, we use to map the current state of things. After all, the rock that makes up a cliff is natural, not man_made.
I use natural=cliff for active quarries as well, for the aforementioned reason. We map the present state. Quarries usually expand to one side only. The other side remains valid in OSM. The expanding side needs to be adjusted from time to time. It's better to have the cliff in an outdated position than not having it at all.
--Fkv (talk) 14:55, 9 February 2015 (UTC)

Landform examples

The wiki-page landform examples looks like it could be useful if it could be updated and linked to all relevant pages. RicoZ (talk) 11:08, 10 February 2015 (UTC)