2

EDITED

I started a project with CRS WGS 84 /Pseudo Mercator and I created a shapefile including point features in different locations around the globe. I have been already working on that for a while and all the operations done so far to measure distances and areas have worked fine enabling "on-the-fly" mode (eg. drawing poly-lines connecting these points, measuring the length of those poly-lines, density of the points, etc).

Now I need to export XY coordinates of the points to use them as an input for a tool external to QGIS (csv, xlsx, or similar formats are fine), but I noticed that the values of XY coordinates do not reflect the actual ground distance between points at each location. For example, for 2 points located in Iceland being 600m away from each other on the x-axis, the difference of X coordinates is 1400m because of my initial choice of CRS. What I would need is XY coordinates relative to the location of each subset of points (eg, with origin on one of the points itself) reflecting the actual ground distance. Note that the geographical position of the points is not important anymore.

Is there a way to extract those coordinates at this point, even with a poorly chosen initial CRS, or to reproject my point shapefile to a more suitable CRS, in order to be able to work with those coordinates?

  • 1
    This is all very vague, and seems more like hand-waving to me. If the first projection was valid, there's no reason you can't just use it, as-is. If it wasn't valid, there may have been data loss even before deprojection/reprojection, making further steps pointless. Please Edit the question to specify the exact projections and the envelope of the data involved. – Vince Apr 18 '18 at 10:07
  • Are you sure that you want "to work with the geographical distances between the points"? It seems you would have enough with the geemotrical distance you have in your projection. – Marco Apr 18 '18 at 10:51
  • By exporting, in wich format is enough? plain csv files? – Marco Apr 18 '18 at 10:52
  • @Marco csv, xslx or anything similar would be enough. I do not actually need to work with the distances, just with XY coordinates, but the XY coordinates I get now seem to be calculated on the Equator and Greenwhich Meridian respectively, while I need them to reflect the actual local position with respect to each other. The issue is that for example 2 points being 600m away from each other result having a difference of 1400m in the x coordinates because X coordinates are calculated on the equator with respect to the intersection with the Greenwhich Meridian. Sorry if I'm not being clear. – M Taglia Apr 18 '18 at 12:06
  • 1
    Please Edit the question to provide critical information, like the fact that you're usng a Mercator to measure distances. This completely changes the question and invalidates the current answer. – Vince Apr 18 '18 at 13:56
2

Pseudo Mercator is not a reliable CRS when it comes to measuring distances, areas, or anything.

For small areas, you can use a local transverse mercator based on a central point. On a global scale, you would need an azimuthal equidistant projection on a central point.

In any case you need to reproject your coordinates, not just change the CRS.

See http://www.progonos.com/furuti/MapProj/Dither/CartProp/DistPres/distPres.html for further explanantions.

  • The link with the explanations was quite useful, thanks! Given that I have around 40 subset of points in different locations, some of them really far away from each other, but I need to analyse each subset separately and each subset is limited to relatively small areas (max 500m radius), would you suggest reprojecting each subset separately? – M Taglia Apr 19 '18 at 7:42
  • 1
    Yes, 40 local CRS would be the best way. – AndreJ Apr 19 '18 at 12:42
  • Thanks, I started reprojecting them subset by subset and it seems to work fine. – M Taglia Apr 19 '18 at 14:29

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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