Antarctica is quite a bit different from other continents due to the climate and the resulting abundance of ice. This page describes some of these differences and possible ways to represent them in OSM data. Most of these concepts have been discussed at the Karlsruhe_Hack_Weekend_February_2013 - everyone who has additional ideas is invited to share them on the talk page here.
Most of Antarctica is covered by ice up to the coast, therefore ice free areas are the exception rather than the rule. The ice free areas would none the less need to be mapped. Rather than having a huge multipolygon for the ice it might be a good idea to consider ice to be the default land cover for Antarctica and map the ice free areas individually as polygons (mostly natural=bare_rock, natural=scree). This would require support from the renderers. Individual glacial features could still be mapped (for example to name glaciers) by drawing them like elsewhere above the default ice.
It would probably be best to apply this default to all land areas south of 60 degree south - including islands (this line being the 'political border' of Antarctica and also a reasonable logical division).
It could be considered to do the same for Greenland and some Arctic islands but the situation there is different since the Greenland ice only extents to the Ocean in a small part of the coastline. Therefore the advantage in changing the default would be much less than for Antarctica.
While this looks like a very special case, it is actually not that special. Looking around the world we don't have a huge "desert" multipolygon for the Sahara or one huge "forest" multipolygon in Canada. There is just no way we can handle multipolygons that large with our current technology. The worst thing that happens if a renderer does not know about this default, is that the usual "land" color is used for Sahara, Canada, or Antarctica.
The ice shelves around Antarctica are thick plates of ice (thickness of 100m and more) which exist at 44 percent of the antarctic coast. More detailes can be found on . How exactly this looks like in principle and where the various lines talked about in the following are can be seen in the image on the right.
In the current OSM data the ice shelves are not mapped at all, the current OSM coastline is drawn at the inner edge of the shelf ice, the grounding line. This obviously needs to be changed and the current consensus is that the main coastline natural=coastline should be drawn at the outer edge of the shelf ice, the calving line. This is primarily motivated by the following:
- this is the logical coastline with regards to human activities. Typical ocean use (i.e. ships) can only take place outside this line and typical land use (like buildings, airfields etc.) is fairly unrestricted on the ice shelf. Ice shelves are not simply thin floating ice, they are thick and stable over decades and more. Also see Sea ice below.
- the grounding line is often far below the surface, in parts as far as 2000m making its exact position both rather meaningless and difficult to determine.
The ice shelves would be represented as mutipolygons and be - corresponding to Proposed_features/Glaciers_tags - tagged glacier:type=shelf. Map rendering should best show them in a distinct way, different from normal glaciers.
Since some of the ice shelves are quite large, some of them more than 500km across and it would be useful to render the outer edge (the calving line) and the inner edge (the grounding line) in different colors additional tags for the corresponding ways are planned:
Sea ice is relatively thin, much thinner as the ice shelf and it is seasonal, so it changes with the time of year. Around the North Pole you have sea ice, but not ice shelf. Sea ice is not mapped at all in OSM because it changes all the time.
Antarctica islands are called islands no matter if they are in the free ocean or enclosed in ice shelf. Their perimeter (ie the way) should be tagged as place=island in both cases (in case of islands enclosed in the ice shelf the edge would be the grounding line). If and only if they are in open water they are also tagged with natural=coastline. Like with all islands a node with place=island and a name can be put in the middle of the island.
Long straight lines are problematic in OSM everywhere. The earth is round and depending on the projection used for a map those straight lines should sometimes be rendered as curves. But the computer doesn't know that and will still just render a straight line between the two end points. This problem is not specific to Antarctica, but it is seen there more often than in other places, because map distortions are often greater and because features are often drawn with less detail than in other places in the world.
To mitigate this problem you should split up long straight lines every 100km or so with extra nodes.
There are certainly further features which are rare or non-existent on other continents. The above are only the large scale phenomena that will be important for the import. Feel free to add other things here to help develop uniform standards for mapping.