Country Centroids are a known data quality issue within the GBIF network.
This is a companion discussion topic for the original entry at https://data-blog.gbif.org/post/country-centroids/
Country Centroids are a known data quality issue within the GBIF network.
Interesting post. Determining the centre of a country, however, is not always easy. For example, for Australia there are five different (accepted) methods for determining the centre of the country:
Centre of Gravity Method 23° 07’ South, 132° 08’ (-23.1166667, 132.33333) (0 in the ALA records within 1 km)
Lambert Gravitational Centre 25° 36’ 36.4"S, 134° 21’ 17.3"E (-25.610111, 134.3548056) (In the ALA, Lots of records within 1km - also it comes up on the ALA map about 300 m from where Geosciences Australia places it! which sounds like it may be a Datum shift?)
Furthest Point from Coastline 23° 02’S, 132° 10’E (-23.033333, 132.1666667) (0 records in ALA Database within 1 km)
Geodetic Median Point 23° 33’ 09.89"S, 133° 23’ 46.00"E (-23.5527472, 133.396111) (8 records in ALA within 1 km)
Johnston Geodetic Station 25° 56’ 49.3"S, 133° 12’ 34.7"E (-25.9470278, 133.2096389) (lots of records in ALA within 1 km)
Some of those records maybe good records with a low uncertainty. The place where ALA places records that just say “Australia” or “Nova Hollandia” doesn’t coincide with any of the above
ALA -25.2744, 133.7751333 (24 records within 1 km in the ALA Database)
Incidentally - the Getty Thesaurus of Geographic Names (TGN) rounds the location (for Australia) down to -25.0000, 135.0000
Thanks for pointing out these other centroids, @ArthurChapman .
I was using the centroids published with the R package CoordinateCleaner. I am not sure what method they used to calculate their centroids, but obviously more centroids need to be added.
CoordinateCleaner uses these two centroid locations for Austrailia:
"CoordinateCleaner" (-27.000, 133.000)
"CoordinateCleaner" (-24.973, 136.189)
As you point out, these two points are not in your list:
“Centre of Gravity Method”, -23.1166667, 132.33333
“Lambert Gravitational Centre”, -25.9470278, 133.2096389
“Furthest Point from Coastline”, -23.033333, 132.1666667
“Geodetic Median Point”, -23.5527472, 133.396111
“Johnston Geodetic Station”,-25.9470278, 133.2096389 (lots of records in ALA within 1 km)
“Getty Thesaurus of Geographic Names”, -25.0000, 135.0000
This is probably why I do not detect a significant country centroid in Australia even though there is one at “Johnston Geodetic Station”,-25.9470278, 133.2096389
No idea where Coordinate Cleaner got those centroids. Nowhere near where any of the other methods detect it - ?maybe they used the external territories (Norfolk Is, Christmas Island etc.) as extremes and then got the centre. That wouldn’t explain being 2 degrees south of the others. Also, in the first case, rounding to a degree of latitude and longitude doesn’t make sense. I will chase up when I get home in a few weeks time. I am working on another project in Argentina at the moment.
From the R docs: "Coordinates are based on the Central Intelligence Agency World Factbook https://www.cia.gov/library/publications/ the-world-factbook/ and http://thematicmapping.org/downloads/world_borders.php."
So I’d say @ArthurChapman is right on the money, external territories are skewing the centroid. If I remember correctly, the world_borders shapefile includes Macquarie, Heard & McDonald Islands as part of Australia, which could explain the 2° southward skew.
The center I prefer is to draw a circle that passes through an area’s extreme cardinal points (usually 3, not including the northernmost point due to earth’s curvature effect) For Australia’s mainland points the center is at about -24.07,133.55. Including offshore islands, Tasmania the center is about -23.80262, 136.57716.