1

This question already has an answer here:

I received a dataset containing some "XY" coordinates that are supposed to correspond to some locations in natural reserves/national parks in Costa Rica. I'm trying to use the proj4 package in R to try to convert these coordinates to lat/long (using the project(..., inverse = TRUE) command), but I cannot figure out which projection was used to get the XY coordinates, so I cannot invert the projection to get the lat/long.

How can I guess which projection/coordinate system was used?

A few points:

  X,       Y
193087, 1056379
193587, 1056917
211524, 1032512
213201, 1032612

marked as duplicate by Andre Silva, Community May 30 '18 at 23:31

This question has been asked before and already has an answer. If those answers do not fully address your question, please ask a new question.

  • 5
    Guessing is a really bad idea, go back to the person/agency you obtained the data from and ask what CRS the data is in.. did you get metadata also? the answer could be there. It does look like undelimited DMS or decimal seconds (is 19.3,105.63 in Costa Rica?) but to what datum? Also it looks like the X and Y are swapped. It's much better to ask and know for sure than guess and find out you were wrong invalidating your whole process. – Michael Stimson May 30 '18 at 23:00
4

Thanks to http://projfinder.com/ , I think I managed to pin it down as EPSG:32617, which has a proj4 string of +proj=utm +zone=17 +ellps=WGS84 +datum=WGS84 +units=m +no_defs.

That website is actually very helpful, basically you input your mystery XY coordinate and point on a map where you think that point is supposed to correspond, and the website spits out a list of projections that seem to match well.

Of course, the best course of action is to ask the person who created the data for the correct coordinate system. But if that proves impractical for one reason or another, this is a pretty good backup option.

  • 2
    Only when you've exhausted all avenues of knowing for sure the CRS of a dataset should you consider an educated guess of the kind offered by this site... if you are forced to guess or assume a CRS be sure to specify in any report you make and if possible include in the files' metadata that the CRS is assumed. – Michael Stimson May 30 '18 at 23:08
  • 2
    That's a pretty old system, anything modern will probably be based on the WGS84 ellipsoid as used by GPS, and that is EPSG:32617 which is about 50 metres away from EPSG:26717. – Spacedman May 30 '18 at 23:10
  • @Spacedman good call. I switched to that one and it is giving similarly reasonable results – twieg May 30 '18 at 23:28
  • @MichaelStimson that's definitely the best practice. I've reached out to the person who gave me the data, but I wanted to try to work with the data at least a little bit, while I wait for a response. – twieg May 30 '18 at 23:30
  • 1
    That puts a totally different light on the question, if you're just after something reasonable to experiment with while awaiting the correct information I suggest that should have been stated in your question... to avoid multiple downvotes. – Michael Stimson May 30 '18 at 23:54

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