And does it matter what system your data layer begins with if you choose the right system for your project? Do some applications use more precise projection systems than others?

  • 2
    Perhaps you mean "accuracy" instead of "precision"? See gis.stackexchange.com/questions/8650/… for the distinction.
    – whuber
    Feb 20, 2013 at 22:48
  • Accuracy is how often a target gets hit in a certain spot (reliability). The nice thing about math, is it's highly reliable. No, I mean which system comes the closest to projecting coordinates with the precision I'd find walking through the city with a tape measure. Or is this just a ridiculous comparison. I'm willing to look ridiculous, gotta pay my dues.
    – Fred
    Feb 20, 2013 at 23:23
  • I suppose part of my distorted paradigm is distortion itself. If we talk about a group of data points all having varying degrees of distortion due to there distance from a conceptual projection curve then I agree there can be a difference in Accuracy of points in a set relative to different projections. I'd like to know if using one coordinate system will get a given group of points closer to the real than another (higher precision) and conversely if some should be avoided as archaic. But thanks for the ref. I'll look deeper.
    – Fred
    Feb 20, 2013 at 23:36
  • 1
    You appear to be confused about the distinction between accuracy and precision. Almost all GISes will compute projections to sub-micron precision. It sounds like you are interested in questions that compare the metric distortion of projections.
    – whuber
    Feb 20, 2013 at 23:46
  • Yes, Yes, it's not the precision of the calculation I care about if it doesn't hit the mark with precision. But the distance between the calculated stone on the ground and the stone on the ground is it's precision. How many times it gets the same result is it's accuracy.
    – Fred
    Feb 20, 2013 at 23:49

1 Answer 1


From my admittedly incomplete understanding of projections, they should all be equally precise, at least within reasonable limits. Its just a matter of using more decimal places. The main difference between projections are what aspects are "true to life" and which aren't. For example, some projections preserve N-S distance, others E-W, others preserve area at the expense of distance, still others preserve constant bearings or true bearings, and some just try to minimize a variety of errors. If you're working in a small area, I'd recommend using the appropriate UTM zone for your area, as the distances are "true" N-S and the zones are set up to minimize E-W distortion within a zone. If you're working on a larger project, use UTM if it is oriented N-S, or a Lambert Conformal Conic if it is oriented E-W. For a fuller discussion of what you need to know, read the answers to this question.

  • 1
    You're correct about the precision issue (+1). Some notes: (1) The distortion in the gnomic projections is so great that although the line segment on the map does accurately trace a geodesic, its scaled length on the map will usually be far from the actual geodesic length. (2) UTM has a 0.04% error built in as a compromise to make it (merely) adequate across a large east-west swath. It can indeed work well in small areas, like cities, and be highly accurate, provided that its built-bias is adjusted. For instance, no property surveyor would use UTM; it's just not accurate enough.
    – whuber
    Feb 20, 2013 at 23:50
  • Good point on the gnomonic projections. Just because it preserves the LOCATION of the shortest route doesn't mean it preserves the LENGTH. I hadn't thought about it that way. As for the UTM issue, I had been under the impression that surveyors would use UTM with grid/ground corrections, or create a custom local coordinate system and attach it to a UTM coordinate. But then again, I'm a geographer and not a surveyor :). Feb 21, 2013 at 0:01
  • Exactly right, Jay: the custom local coordinate system is one example of the adjustments needed to increase the accuracy.
    – whuber
    Feb 21, 2013 at 0:01
  • Wow, I'm glad I asked. I thought I was being tedious. But I never would have figured this out on my own. I am getting data as WGS 84 (EPSG:4326), had chosen NAD83 (EPSG:2272) because it had Pennsylvania in the name, but I think I'll use PSAD56 (UTM zone 18N / EPSD:24818) for now cause I'm working pretty much on Philadelphia. Thanks so much for your help.
    – Fred
    Feb 21, 2013 at 0:05
  • PSAD = Provisional South American Datum! How about using one of the NAD83-based UTM zones like EPSG:26918 or EPSG:3748 (NAD83 HARN) or EPSG:3725 (NAD83 NSRS2007).
    – mkennedy
    Feb 26, 2013 at 21:01

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.