Using QGIS3.10, I have tried Distance Matrix and NNJoin (distance results 0 - 10.32 for both, likely in degrees). I have 93 points of 2 types in one shapefile. I have the same distance issue if the points are in 2 files. UTM projections are problematic in this region because the study area spans at least 2 and often 3 UTM zones (e.g., x: -63 to -78). The CRS is WGS84 (EPSG: 4326). The points (caribou locations) fall into 3 clusters 150-300 km apart, but one point is in a central position between the 3 clusters and it is about 154 km from its nearest neighbor (i.e., QGIS's Measure Line tool readily provides km for WGS84).

How could I resolve this to get nearest neighbor distances in km?


One quick way would be to choose North Pole Azimuthal Equidistant (EPSG:102016), but it may not return satisfying result if your study area is not close enough to the north pole.

Second option: Create custom projection by modifying the above mentioned EPSG:102016. For instance, if your central position is located at (latitude: 86, longitude: 135), then the proj4 string is +proj=aeqd +lat_0=86 +lon_0=135 +x_0=0 +y_0=0 +a=6371000 +b=6371000 +units=m +no_defs.

enter image description here

On a separate note, working on proj4 string on QGIS custom projection has been tricky recently- due to ongoing PROJ improvements. It seems I had luck with QGIS 3.12.3; please carefully test it in your environment.

  • Thanks for this suggestion. It led me to search for polar projections, then to projectionwizard.org, that provides PROJ WKT code. I entered the area extent as N: 68.1, S: 61.75, E: -61.1, W: -78.4. Equal-area option provided very long codes with unrecognizable names. The Conformal option provided: Lambert conformal conic (most realistic landform shapes, but maybe not for NN distances?), and PROJ WKT code: +proj=lcc +lon_0=-69.75 +lat_1=62.8083333 +lat_2=67.0416667 +lat_0=64.925 +datum=WGS84 +units=m +no_defs (to be continued) – Mike Jun 22 '20 at 5:49
  • Continuation ... Equidistant option provided: "Oblique azimuthal equidistant PROJ WKT - distance correct along any line passing through the center of the map (i.e., great circle)", and PROJ WKT code: +proj=aeqd +lon_0=-69.75 +lat_0=64.925 +datum=WGS84 +units=m +no_defs The code looks very similar to your suggestion, with the center corrected. How might it compare to the Conformal option? – Mike Jun 22 '20 at 5:54
  • @Mike It's a vast area, and a little bit far from the north pole. So please forget the suggested aeqd option, which is valid only for distances which involves the center point. I like your Lambert conformal conic option which two standard parallels are close to NS extents of your working area. – Kazuhito Jun 22 '20 at 9:10
  • @Kazukito. Thanks for your confirmation. It agrees with a technical report DOI: 10.13140/RG.2.1.2809.2240 that I just found, which recommended Azimuthal Polar Equidistant and AP Stereographic for the high Arctic (71-90 N), and Lambert Conic Conformal and LC Equidistant for the sub-Arctic (50-75 N). – Mike Jun 23 '20 at 21:15

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.