I am georeferencing a static map in JPEG format. I would prefer to use QGIS for a variety of reasons. I've never georeferenced with QGIS before, so I may not have the process quite right.

I georeferenced the image using the georeferencer plugin and a Bing aerial map from the OpenLayers plugin. The result looks great in QGIS.

I used WGS84/ Pseudo Mercator for both the project and for the output TIFF. The issue arose when I loaded up the georeferenced TIFF in ArcGIS for comparison.

It appeared about 16 miles from where it was supposed to be. It's in the ballpark, but clearly nowhere near good enough.

Could this be some sort of issue with on-the-fly projections?

At every step of the way in both programs I set the coordinate system to WGS84/ Pseudo Mercator.

What could be going wrong?

  • 7
    I think the coordinate system coming from QGIS isn't correct for "web Mercator". Try redefining (not reprojecting) the raster in ArcGIS with the WKID 3857 aka projected coordinate systems, world, WGS 1984 Web Mercator (Auxiliary Sphere). Note: 16 miles sounds classic for interpreting the earth model as an ellipsoid rather than a sphere. – mkennedy Oct 16 '15 at 21:11
  • There seems to be no suggestion that this is an ArcGIS for Desktop problem so I'm removing the tag for that so that the question can be focussed on QGIS. – PolyGeo Oct 21 '15 at 4:41
  • 16 miles off, but in which direction? The usual problems with Mercator seem to result in N-S offsets (often because spherical/ellipsoidal differences). E-W differences of 16 miles would be much harder to explain. – Spacedman Oct 21 '15 at 7:12

The problem indeed arises from arcmap interpretation of the web/psuedo mercator projection.

I've digitized Central Park in QGIS set to EPSG:3857 (WGS 84 / Psuedo Mercator) qgis web mercator

Opening the same layer in ArcMap defaults to WGS_84_Psuedo_Mercator, which as you can see is not an actual crs in ESRI based programs, and is interpreted as being north of the Central Park by some 16 miles, 26 Km.

enter image description here

Using the define projection tool (works for both raster and vector) one can define the crs to WGS_1984_Web_Mercator_Auxiliary_Sphere, and then the layers do line up correctly with aerial imagery

enter image description here

Worth noting that this is only a QGIS to ArcMap issue. Creating a layer in ArcMap with Web Mercator Auxillary Sphere does import correctly as Web Mercator in QGIS

  • 1
    Just to add, this is mainly a Web Mercator issue. Converting the tif to a standard GIS CRS (using transverse mercator, lambert or albert projections) before handing over to other software would avoid those problems as well – AndreJ Mar 14 '16 at 7:18

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.