1

I want to calculate the area of counties (1st admin level) in Kenya in ArcGIS. What is the right projection for that? I know I should use an equal area projection, but when I use them (I tried six different) it varies heavily to the county areas given in wikipedia. I could also use UTM37, but by definition it's not an equal area projection.

PolyGeo
  • 65,136
  • 29
  • 109
  • 338
Enzo Baldini
  • 2,327
  • 1
  • 15
  • 29
  • Did you always get the same answer regardless of which equal-area projection you used? If not, then something erroneous is happening and you will need to tell us more about how you are computing those areas. If so, then everything is working fine and you should challenge the Wikipedia figures. – whuber Nov 28 '15 at 21:04
  • No, the values varied. But I am wondering why they are all different to the "official" area data. – Enzo Baldini Nov 28 '15 at 21:14
  • Since they varied, there must be some aspect of your calculation that is incorrect. That would explain the discrepancies with Wikipedia, too. – whuber Nov 28 '15 at 21:26
  • These are my number for Kenya (admin_0 level): UTM37: 586416 Albers Equal Area:585983 Lambert Equal Area:509155 Eckert2,4,6: 589920 Wikipedia: 581309 – Enzo Baldini Nov 28 '15 at 21:54
  • UTM is effectively equal-area within the one UTM zone, with deviations from equivalence being minor. – Paulo Raposo Nov 28 '15 at 23:12
  • Ok. I did some calculations for Nigeria with the same method in ArcGIS (calculate geometry). albers: 7351, UTM32: 7353, lambert: 6582. And wikipedia says 7315. the nearest to wikipedia is albers (difference of 36 kilometer which is 0,004%, not much really). The reason for the small difference could be the level of generalization of my nigerian states shapefile. So you would recommend to use UTM or Albers, right? In both cases (Kenya, Nigeria) Lambert had the biggest difference (although it says equal area). So in the future I won't use Lambert for field calculations... – Enzo Baldini Nov 28 '15 at 23:25
  • 2
    If you are using the same spherical datum for different equal-area projections, then any differences among your results will arise only from floating-point imprecision (and should be extremely small). – whuber Nov 30 '15 at 15:10

0 Answers0