[PROJ] Add trivial support for EPSG:1026 operation method?

[Martin Desruisseaux]

Hello all

PROJ does not seem to support the following operation method at this time:

  * EPSG:1026 — Mercator (Spherical)
    https://epsg.org/coord-operation-method_1026/Mercator-Spherical.html

Would it be possible to add it? (I can contribute a patch if I get some 
guidance about which file to edit). The implementation can be very trivial:

 1. If the ellipsoid is not a sphere, raise an error.
 2. Otherwise treat as synonymous of EPSG:1024 — Popular Visualisation
    Pseudo Mercator.

Step 1 could be replaced by the radius of the conformal sphere as 
suggested by EPSG, but the above trivial implementation would be 
sufficient for now for the purpose described below.

The rational for adding EPSG:1026 support would be for the definition of 
extra-terrestrial CRS. For compatibility with existing software, some 
members of the OGC Planetary working group want to use the 
Pseudo-Mercator projection. However many extra-terrestrial CRS are 
defined on a sphere rather than an ellipsoid. In that particular case, 
Pseudo-Mercator (EPSG:1024) is mathematically equivalent to Spherical 
Mercator (EPSG:1026). The CRS could use former, but I would encourage 
the planetary group to use the latter because it carries a semantic 
difference. The EPSG guidance notes warns the reader about the problems 
of Pseudo-Mercator (non-conformal, etc.). However those issues do not 
apply when the datum uses a sphere. The EPSG guidance notes do not 
mention that fact maybe because Pseudo-Mercator has been introduced for 
use with ellipsoids and is useless on spheres (because the existing 
classical Mercator projection was already doing the exact same thing). 
The use of Pseudo-Mercator on a sphere may give to non-experts the false 
impression that Pseudo-Mercator problems still apply, while actually it 
got the properties of a classical Mercator projection, in particular the 
fact that it become a conformal projection. Using "Mercator (Spherical)" 
operation method instead makes that fact clearer. Given that the 
implementation can be trivial — identical to Pseudo-Mercator with only a 
check for making sure that the datum uses a sphere — it seems to me that 
it would be a reasonable addition to what can be expected from software 
(not only PROJ), and that the gain in semantic value is worthy.

     Martin

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/proj/attachments/20230525/6a1ad836/attachment.htm>