Stand at 73.68 degrees south, 122.29 degrees west in the middle of a July night and the nearest human being is at the South Pole, 1,815 km away. The second nearest is at Scott Base, 1,864 km. The third is at McMurdo, 1,866 km. The interesting part is not the distance. It is that all three lie in the same direction. Turn your back on the pole and there is nothing at all, all the way to the coast and past it.
So: what is the most remote place in Antarctica? That point, in winter. It is on the West Antarctic ice sheet behind the Amundsen Sea coast, it has no name, and no map of Antarctic research stations shows it, because every such map draws the dots rather than the space between them. We measured the space. The result is a page and the number is a distance.
How it is measured
Two inputs. The first is the COMNAP master list of Antarctic facilities, which carries 114 of them run by 31 nations, with a position, an operator and a season for each. Of those, 103 are in use and 11 are temporarily closed. Only 44 of the 103 stay open through the winter. The second is the ice mask from the ITS_LIVE velocity mosaic, which covers 12.55 million square kilometres of grounded ice and floating shelf.
Every cell of that mask gets a distance to the nearest facility, twice: once against all 103 open places, and once against only the 44 that do not shut. The furthest cell in each field is the answer. Everything else on the page, the medians and the bands, falls out of the same two fields.
In summer it is a different continent
Open every seasonal camp and the furthest you can get drops to 924 km, at 71.75 south, 98.76 west. The nearest facility there is Sky Blu, a British blue ice runway in Ellsworth Land, and the second nearest is the American camp at WAIS Divide, 930 km away. Half the distance, and a completely different kind of neighbour: not a permanent base with a winter crew but a hut and a landing strip that is packed up when the light goes.
The seasonal collapse is bigger than the headcount implies. Fifty-nine of the 103 open facilities shut for the winter, and what is left is not a thinner version of the same map.
| measured on the mapped ice | summer | winter |
|---|---|---|
| median distance to a facility | 300 km | 515 km |
| furthest point from a facility | 924 km | 1,815 km |
| share more than 500 km away | 17.2% | 52.6% |
| share more than 1,000 km away | 0.0% | 12.3% |
Nothing on the mapped ice is more than 1,000 km from a person in January. In July, an eighth of it is. Well over half the human presence on this continent is a summer visit.
Whose ice is it
Turn the same calculation inside out and every wintering station gets a territory: the ice for which that facility, and no other, is the nearest thing to a person. It is a Voronoi cell drawn on the sphere rather than on a flat map. Amundsen-Scott South Pole holds 1.52 million square kilometres of it. Vostok holds 1.37 million, Belgrano II 1.34 million, Syowa 0.93 million and Concordia 0.88 million. Ten facilities are between them the nearest thing to 72.6% of the mapped ice. That number is the whole argument in one line.
Three ways to get this wrong
Distance. Antarctic maps are drawn in polar stereographic projection, which is true to scale at 71 degrees south and runs about 8% long towards the pole. Measure on the flat map and 1,800 km comes out 140 km too big, which is larger than the whole gap between first and third place in that ring of stations. Every cell here is unprojected to its true position on the ellipsoid and the distance is great-circle.
Area. A cell in that projection is not a constant patch of ground either. Every area quoted is weighted by the local scale factor squared, not counted.
Who counts as a person. The 11 temporarily closed facilities are in neither field. Including them does not move the headline, the furthest summer point stays the same distance from the same station, but it moves the median by 23 km and would draw eleven dots where there is nobody.
Two things this number is not
It is not a share of Antarctica. The mask is 12.55 million square kilometres against the continent's roughly 14.0 million, so every percentage above is a share of the mapped ice and is written that way. The headline is a distance for exactly that reason: both furthest points sit deep in the interior where no coastline convention can reach them, so the distance is the one figure no definition can argue with.
And it is not a headcount. COMNAP publishes a peak population for 93 of the 114 facilities and a winter population for none of them, so there is no defensible answer to how many people are on the continent in July. This site does not print one. The count of open buildings is a fact. The count of people inside them is not available, and saying so is more useful than guessing.
Why the emptiness is worth a page
Because it quantifies a caveat this site was already making about itself. Four of the six penguin species mapped on the breeding atlas, whose whole geography we wrote up yesterday, have a median breeding site 13 to 17 km from a research facility. The emperor's median colony is 204 km away, and only 13% are within 50 km. The emperor is also the one species here mapped from orbit rather than visited. Where people can go shapes what gets counted, and the only way to say how much is to measure the map of where people are not.
The station list is the free, keyless COMNAP Antarctic Facilities Master, published by the Polar Geospatial Center. Every facility on it has its own page here, and the 80 that are nearest to any mapped ice now print their own territory on it. The live map is at /antarctica.