2

I need 5 km buffer of a basin boundary of Brazil using ArcGIS 10.1. The basin boundary is in GCS. I have tried it giving buffer distance 5 km and 0.0454545 degree decimal (assuming 1degree=110km). I have got two different results for two distances. Which one is more accurate?

underdark
  • 84,148
  • 21
  • 231
  • 413
user19120
  • 121
  • 2
  • Try it with a degree of latitude being 111.2km, that's based on one nautical mile being one minute. – Pete Jun 14 '13 at 14:25
  • Will it give same output like 5km? if not which output is more accurate?????? – user19120 Jun 14 '13 at 17:15
  • Accuracy depends on where in Brazil you are. Because it is Equatorial, equating degrees with meters is actually fairly accurate: at the northernmost tip of the country, the error is at most 0.4%. Because the southern part of the country is furthest from the Equator, errors are larger there, reaching almost 21%. Incidentally, you should be using approximately 111.1 Km per degree: see http://gis.stackexchange.com/questions/8650/how-to-measure-the-accuracy-of-latitude-and-longitude/8674#8674. – whuber Jul 24 '13 at 13:51

2 Answers2

2

As of ArcGIS 10.1, the Buffer Tool now creates geodesic-based buffers for points, lines, or polygons, if the data's coordinate reference system is geographic (latitude-longitude). You should use the 50 km value for the buffer distance.

Here's a link to the What's New online help topic.

Note: I work for Esri.

mkennedy
  • 18,916
  • 2
  • 38
  • 60
1

You need to reproject your basin boundary into a meter projection.
I don't think just changing the projection of the document (right click on layer {the word layer which is usually the default name for the data frame} in TOC) to a meter (local) projection will cause it to use meter.
You should know that gcs is not an equal distant crs. So lat is not the same length as Lon. If your basin is of any significant size the only way of getting accurate area is to use meter.

Brad Nesom
  • 17,412
  • 2
  • 42
  • 68
  • My question is whether the distortion from projecting and then using a straight line, ignoring the curve of the elipsoid, will result in less error than the change in distance of degrees in the converging lines of longitude. Normally I would completely agree that projecting and using a buffer with km would be best, but if the basin is almost all equitorial, then I'm not so sure. – Pete Jun 14 '13 at 14:44
  • 1
    You are concerned for distortion of projecting which is designed to minimize distortion, but consider accepting a known distortion that is accentuated more with every km from equator. It comes down to acceptable inaccuracy. – Brad Nesom Jun 14 '13 at 14:55