A couple months ago, I made some road orientation graphics. I thought it would be interesting to repeat the exercise using the borders between countries.

The method is slightly different for this graphic, so I’ll explain it in full. The polyline shapefile from the data source was separated by continent. Country borders that divide continents were attributed to both continents, but continental borders that divide a single country (e.g., the Asia-Europe border in Russia) were not considered. Oceanic coastal boundaries were not part of the shapefile, and were not of interest – the goal was to see how people define boundaries. Each continent’s polyline was then divided into 1 km segments. This weights the borders by length, and is at a small enough scale that it captures the oddities of the borders. The original shapefile was somewhat generalized, and I was concerned that any smaller length unit would return spurious results. For each continent, the azimuths of the segments were measured in the respective continental Lambert conformal projection; because continents are very large, there will be a small error associated with this measurement. These azimuths were then duplicated using opposing directions (e.g., 45° was repeated as 225°) to ensure rotational symmetry on the rose diagrams, which are plotted in 5° bins.

Please keep in mind that the map in the graphic is a Miller cylindrical projection, which is not conformal. So while a continent’s international borders may not seem to match the rose diagram, there is some distortion in the map projection. In addition, because the measurements are made at the kilometer scale, long borders that have an overall trend in one direction (e.g., Argentina-Chile) may actually have a significant contribution to an oblique or orthogonal direction. You could compare this to walking along a trail with switchbacks: you are generally heading in one direction, but most of your time is spent walking in a perpendicular direction.

The results of the analysis are fascinating – I’d love to hear what Jared Diamond has to say about these. All five graphs have a strong W-E signal; this is especially surprising with South America and Africa, which have significant latitudinal spans compared to their longitudinal extents. I’m not much of a history buff, so I won’t even try to interpret the meaning of the results…

Data source: http://www.naturalearthdata.com/downloads/10m-cultural-vectors/ (land boundary lines)

Posted on 23 May, 2014

600 notes

  1. gisgames reblogged this from hparks2
  2. emelineblack reblogged this from vizual-statistix
  3. my-anime-gaydar reblogged this from vizual-statistix
  4. spooky-dyke reblogged this from trashmitzvah
  5. trashmitzvah reblogged this from ding-us
  6. ding-us reblogged this from andyhin
  7. mespetitesmeditations reblogged this from vizual-statistix
  8. freetogoandgo reblogged this from vizual-statistix
  9. 23700 reblogged this from vizual-statistix
  10. capdocq reblogged this from vizual-statistix
  11. heylizzyhey reblogged this from vizual-statistix
  12. vivipenguin reblogged this from vizual-statistix
  13. therotherman reblogged this from vizual-statistix and added:
    im actually smarter having read this
  14. jryanm reblogged this from galaxyclusters
  15. marcusfields52 reblogged this from vizual-statistix
  16. everythingcurrentlysucks reblogged this from mapsontheweb and added:
    The african borders are really abritrary, made where different euros could control, not on ethnicity. The NA borders are...
  17. euthanasiainthehouseoflords reblogged this from vizual-statistix
  18. crystalmochi reblogged this from vizual-statistix
  19. stefaniajoy reblogged this from mapsontheweb
  20. imaginingthenostalgia reblogged this from we-are-star-stuff
  21. years-strange reblogged this from songofsunset
  22. ununnilium reblogged this from orvillegoggles
  23. spookyzombies reblogged this from songofsunset
  24. nutterbutterbisquit reblogged this from songofsunset
  25. songofsunset reblogged this from mapsontheweb