1

I have an old map that I'm trying to georeference. The map says that it's a Bonne projection, with the origin at 34ºN, 130º,30'E.

I downloaded a DEM (in WGS84), and reprojected it using a custom Bonne projection +proj=bonne +lat_1=34.00 +lon_0=130.5 +ellps=bessel +units=m. I then georeferenced the map in QGIS (Transformation type: Projective, Resampling method: Nearest neighbor, Target SRS: the custom Bonne projection).

The georeferenced map is not horrible, however not perfect. I tried reprojecting the original DEM into EPSG: 30162, a projection that's a good fit for my target area, and it seems to line up better (although not perfectly) with my georeferenced map. The project CRS is set to the custom Bonne projection, so I understand that the EPSG:30162 DEM is being reprojected into the Bonne projection on the fly, but why does it work better than the DEM that is in the Bonne projection natively?

Tutorials I've read about georeferencing have told me to define a project CRS that is the same as the map, import a DEM in that projection, and then georeference in that CRS for best results. Can anyone shed some light on why that process doesn't seem to be delivering the best results in my case?

Did I make a mistake georeferencing? Did I make a mistake defining the Bonne projection? Or is there some other step or option that I failed to take into account?

Here is a section of the the DEM with the Bonne projection: enter image description here

Here is a section of the DEM in EPSG:30162 (project CRS still set to the Bonne projection): enter image description here

Here are my GCP points in the georeferencer: enter image description here

I'm still new to georeferencing.

1
  • 1
    Complicated. Apart from how QGIS has treated the CRS by the on-the-fly projection, there would be multiple issues related to the map itself. Geological Survey of Japan has used Bonne projection to map surface geology prior to 1952. Hence the LatLong GCPs you used to georeference were on ellipsoid Bessel 1841. On the other hand your DEM was on WGS84. On and around Japan, they usually show 400m+ displacement between them. EPSG 30162 was built on older Japanese mapping system (and used Bessel 1841). Current CRS for that area (Kyushu) would be EPSG: 3097 (on GRS80).
    – Kazuhito
    Commented Feb 13, 2021 at 13:01

0

Your Answer

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