OpenStreetMap wants to map the whole world, but there is a whole continent that has been neglected a bit: Antarctica. Of course there isn’t much there besides rocks, ice, and penguins and not too many people live there. But still, it is a huge area and it should appear properly on our maps.
I have looked into the current situation and what could be done and want to write that up here as a first step towards improving the situation.
The Mercator projection generally used in online maps only covers the area between about 85.0511 degrees South and 85.0511 degrees North. While this is not a problem in most cases, it obviously is a problem for Antarctica. JOSM can edit data South of -85.0511 degrees when you change the projection from Mercator to “WGS84 Geographic” (EPSG:4326), but Potlatch can’t. And the usual aerial images aren’t available, because they also use the same Mercator projection.
If the source tags are right, the coastline seems to have been mostly imported from GADM. Apart of the licensing issues, this data seems to be old and not very accurate. The real coastline dips below 85.0511 degrees South in the Ross sea, but in OSM it is cut off there, probably because not many people are actually creating maps not using the Mercator projection. There are newer and better sources for the Antarctica coastline around and it should not be too difficult to throw the current one out and replace it by a better one.
There are huge ocean areas around the land mass of the continent that are covered in ice shelves. They should be tagged and appear on the map because they are large features and they affect the way we perceive Antarctica.
There is no tag for ice shelves yet, there is natural=glacier, but that is not the same. I propose adding a tag natural=ice_shelf for this. There are sources for the borders of the ice-shelf, so it wouldn’t be a big problem to add them.
Bing and Mapbox seem to be using older and not very detailed Landsat images. There are newer Landsat images with better resolution available, see below. Landsat images don’t go below 82.5 degrees South because of the orbit of the Landsat satellites. The images from MapQuest Open are better, they are from Landsat and mostly from NASA’s BlueMarble NG.
If you want to have a look around, maybe try out some of the data yourself, here are some pointers.
I have asked Frederik and he has added Antarctica to the list of OSM extracts that are generated daily and published on download.geofabrik.de. It is not much data and you can load all of it into JOSM, though it will be a bit slow then.
The current OSM coastline is available from OpenStreetMapData.com, it is usually updated daily.
There are several sources of reasonably current data that seem to be quite accurate and detailed.
The Mosaic of Antarctica (MOA) contains good satellite images for Antarctica and, derived from it, shapefiles. You can look at them and download the shape files. It seems that Natural Earth Data derives their Antarctica coastlines from these. The shape files can’t be used directly for import, though, because the vectorization is bad and the data contains lots of ziggzaggy lines.
(Thanks to Christoph Hormann for these links and insights into the data.)
There are also several more links on the wiki on the Antarctica page. I haven’t looked at all of them yet.
Of course we have to look into the details, but licensing should not be a big issue. Almost all of this data is ultimately derived from NASA satellite images or other satellite data which is in the Public Domain. Either NASA themselves or other US government or research agencies have been processing this data to create the data that is available at the places mentioned above.
I can display all of this data in EPSG:4326 in Quantum GIS (QGIS). The Antarctic Polar Stereographic Projection EPSG:3031 is much better, though. Unfortunately there are some rendering problems with it in QGIS. I haven’t tried Mapnik or other renderers yet. This is something I must explore some more.
There is one other problem with OSM data in Antarctica: Just like the spot in the Atlantic with 0 degrees latitude and 0 degrees longitude, Antarctica seems to attract some bogus data. Most probably from software bugs, but it is also a likely spot to add bogus things for the fun of it without anybody noticing. Last time I looked, I found (and removed), among other strange things, a restaurant, a restroom, and a pub called “Very Cold” somewhere in Antarctica and a parking lot and a tree from failed imports at the South Pole.
I think we can probably just throw away the existing coastline and import the coastline and ice-shelf data from NaturalEarth. We don’t actually need that much detail in Antarctica. If and when we need it we can always add the more detailed data from MOA either in the whole of Antarctica or just in selected areas. This is much less work than cleaning up the better MOA data and making it ready for OSM and we’d still have a large improvement over the current situation. If it turns out that there are areas where the data currently existing in OSM is better than the NaturalEarth data, we should, of course, keep that, or check with the better MOA data.
So far these are just some notes from my research. There is more work to be done. I hope this article will start a discussion. Please tell me if you have any other knowledge, ideas, insights. Before actually doing any kind of import, I will, of course, write up in some more detail what is planned and discuss that according to the Import Guidelines.