Airports — how this layer is made
What you are looking at
Each point is a place aircraft land — an airport, an airfield, an air base or a heliport, as tagged in OpenStreetMap. 46,787 of them worldwide, coloured by which kind it is and sized by the area of the site.
Click any of them to see what OpenStreetMap records — the IATA and ICAO codes, and who runs it, where a mapper noted that. Where it has a Wikipedia article, a short summary is fetched live and shown in your own language; and every airfield links to the OpenStreetMap object it came from.
Mapped, not modelled
Each airfield is an object a mapper drew, from survey, aerial imagery or an open aeronautical source. Nothing here is modelled or estimated, and no traffic figures, passenger numbers or movement counts are involved — this map does not know which airports are busy, only which ones have a commercial code and how much ground they cover.
What counts, and what does not
Three tags decide what is on the map at all. The colours then split the first of them in two, which the next section explains.
- `aeroway=aerodrome` — a civil airfield of any size, from Heathrow to a farm strip in a field. 45,627 objects, and nineteen features in twenty here;
- `military=airfield` — an air base, whether or not it also carries a civil tag. 2,314;
- `aeroway=heliport` — a landing site for helicopters alone: a hospital roof, an oil terminal, a mountain rescue pad. 782.
`aeroway=airport` is not selected, and the reason is worth recording: it matches exactly zero objects worldwide. It reads like the obvious tag and nobody uses it.
Runways, taxiways, terminals and aprons are not separate points. Where an airport is drawn as an outline, its runways lie inside that outline and are part of the shape rather than features of their own.
Airport or airfield — the line is an IATA code
Two of the four colours divide the civil aerodromes in two, and the dividing line is whether OpenStreetMap records an IATA code for the place — the three-letter code on a boarding pass, LHR or SIN.
- Airport — carries a code. 8,351 features, 17.8% of the layer.
- Airfield — does not. 35,409, three quarters of the layer.
A code is issued to a place with commercial air service, so its presence separates an airport from an airstrip where no tag naming that distinction does. And it separates the map too: the median coded aerodrome covers 1,075,261 m² against 123,187 m² uncoded, and only about one uncoded field in ten is larger than the median coded one.
The colour is doing work that size cannot. 1,275 coded airports have no outline at all, so they are drawn at a nominal size — pixel-identical to a farm strip. Nothing about the dot could tell you otherwise. London City is nearly that case: at 0.49 km² it is twenty times smaller than Heathrow on the ground, and without the colour it reads as one of the gliding clubs around it.
99.9% of the airport class carries a name, the highest share of any class on this site. That is a second, independent sign the code is picking out the right places.
What the colours are not, and three things that would have been better
They do not mean "big", and they do not mean "open". A code is a commercial registration. Manston in Kent keeps MSE and has been closed since 2014. 1,694 uncoded fields cover more ground than the median coded airport. Read the colour as "this place has an airport code", because that is exactly what it says.
They do not mean international. OpenStreetMap has a tag for that — `aerodrome=` — and `aerodrome=international` is on 578 objects worldwide, while its opposite, `aerodrome=domestic`, is on twenty-five. Even trusting the 578, there is no way to draw the other half. `aerodrome:type=` reaches about 10% and mixes ownership, traffic and mode in one key, so `public`, `airfield`, `military/public` and `civil` all appear among its commonest values.
They are not passenger numbers, and the reason is worth your attention because it is the same reason most of this site carries a warning about empty regions. Wikidata publishes yearly passenger figures, and they cover 9.7% of this layer. Tested against 43 airports chosen to be unambiguously international but small, spread over every continent, only 26 had a figure. The missing seventeen included every Pacific airport tested — Port Moresby, Honiara, Port Vila, Apia, Nukuʻalofa, Funafuti, Tarawa — along with Male, Paro, Vientiane, Banjul, Freetown and Maseru, while Exeter, Newquay and Norwich all had one.
Male handles several million passengers a year. Newquay handles 244,674. A map coloured by that would have been a map of which Wikipedia communities write airport articles.
The IATA code passed the same test: 43 of 43, and 35 of 36 major world airports. A code is a short, stable fact a mapper can type; a passenger statistic needs a source and a year, and only some places get one.
Air bases are on two maps, on purpose
The 2,314 air bases here are also on the military sites map, drawn as its airfield class in the same orange. That is not double counting — it is the same object answering two different questions, the way a university is on both the education map and the universities map.
1,839 of them carry a civil aerodrome tag as well. Where a feature is both, this map draws it as a base: a shared civil and military field is more usefully described to a reader as a base, and resolving it the other way would hide every one of them inside a class nineteen times larger. Click one and the panel shows what the feature itself claims.
Size
Each symbol is sized by the footprint of the airfield, in square metres measured on the globe. Symbol area is proportional to that (radius to its square root), so an airport four times the area draws twice as wide.
These are the largest features on HueMaps by a wide margin. The median traced airfield covers 214,053 m² — 21 hectares — where the golf courses map, until now the site's largest ordinary footprints, runs an order of magnitude below that.
About half of all airfields (48.6%) are recorded as a single point with no shape, and those are drawn at a nominal 200,000 m² rather than at the minimum. Being mapped as a point says something about the mapper's time, not about the size of the airfield. The untraced half is almost entirely civil aerodromes: heliports are 82% traced and air bases 85%, against 49% for aerodromes, which is why the nominal is set from a typical traced aerodrome rather than from the layer as a whole.
94.6% carry a name. An airfield without a name is unusual, and most of those are private strips.
The largest dots are not the biggest airports
At the top of the size range are things whose polygon covers far more than an airfield: Baikonur Cosmodrome at 2,730 km² is the leased area, and several United States Air Force entries cover a whole range rather than the base inside it. The scale is capped below them, so they draw at the maximum and do not flatten everything else.
Below that the order is honest. Denver International at 49 km² and Singapore Changi at 22 km² really are among the largest airports on earth by land area — though land area is not passenger traffic, and this map does not know the difference between a vast quiet field and a busy one on a small site.
Where the map is empty — and where it honestly is not
Most layers on HueMaps carry a warning that a sparse region means the record is missing rather than the thing. Here that warning needs qualifying, because a real part of the difference is real.
Counted inside each country's boundary, per million people: Australia 88.23, the United States 47.01, Brazil 15.40, France 12.41, the United Kingdom 11.87, Germany 8.77, Indonesia 2.46, Japan 1.74, China 0.55, Nigeria 0.30, India 0.29.
Australia has three hundred times India's rate per head. No mapping effort explains that on its own — a country of 27 million people with those distances genuinely has thousands of outback strips, and India does not. What mapping effort does explain is the tail: small private and agricultural strips are recorded far more completely in North America, Australia and western Europe than in west Africa or central Asia, so the low end of this map is thinner than the ground is.
A picture of the present
Today's map, not a history. OpenStreetMap is edited continuously; this copy is refreshed periodically. There is no year control, and a closed airfield may still be mapped until a mapper updates it — disused wartime fields are a common case.
Source and licence
© OpenStreetMap contributors, under the Open Database Licence (ODbL) 1.0.