Is coastline tag required for rocks in the sea?

Dear fellow mappers,

As a hobby project, last week I started mapping the surface details of South Georgia, the island in the South Atlantic. Years ago the rough contours of the island were mapped from Bing imagery, but now there is detailed imagery from DigitalGlobe Premium available for this area.

As is usual with rocky islands, there are a lot of loose rocks in sea that have no connection to the mainland. When detailing the coastline of South Georgia these rocks are no longer part of the coastline.

Now I am wondering what to do with them. Currently I have mapped them as natural=bare_rock, without a natural=coastline tag, see this example. However, at other places I see that the rocks have been given a natural=coastline tag.

I searched on the OSM wiki for answers, but so far have not found anything conclusive.
Coastline states that it should be placed at the locations of Mean high water spring, does not state what to do with small masses.
DE:Coastline states that is as well relevant for big landmasses as well as islands.
Tag:place=islet states that an islet (tiny island) also should get a coastline.
Seamarks/Rocks only give info on rendering, not on how to tag the area of the actual rock.

On the Wikipedia for Islet it is stated that according to international low

However, to me it is not clear if a rock which often washes over by waves is considered an islet as well. I’ve searched for that and have not really found a minimum size to be considered a islet. Maybe it is due to not being a native English speaker.

Even rocks that are always slightly below the surface are observable if you hit them with your 2m deep keel, so relevant to map from OSM perpective.

I can imagine the following scheme:

  1. Rocks that are above Mean high water spring receive a natural=coastline tag.

  2. Rocks that are below Mean high water spring are drawn, but do receive a natural=bare_rock tag and tagging from the Seamarks/Rocks scheme.

The challenge with 1 is to determine the waterlevel at the time the imagery was made. Does anybody now if the date and time of Digital Globe imagery can be determined?

What do you think?


As an example, I mapped Rockall both as natural=coastline and natural=bare_rock . Rockall can certainly claim to have a coastline (just), whether your rocks near South Georgia do is something you’ll need to decide.

Thanks for sharing your thoughts. Did you draw two ways that have merged nodes? It is very hard to distinguish the two ways in JOSM (or it’s my lack of skill). Wouldn’t it be less ambiguous to draw one way and than use a multipolygon relation for the other feature?

Generally relations should be avoided unless really necessary. They are expensive to render and difficult for beginning editors to understand.

With JOSM, you can use filters to suppress the way that you don’t want, you can also use one of the other mouse buttons to cycle through possible selections.

Yes, it’s two contiguous ways. A relation wouldn’t do what you want here, since I’d expect that the tags on the relation would lead to it being treated as a “new-style” relation and the other natural tag on the outer way would be ignored.

Oh, it works nicely. You can see for yourselves here:
The way is natural=cliff, which is then used in the relation natural=grassland as outer, and in the relation natural=bare_rock as inner.