[PROJ] EPSG v10 update status

[Even Rouault]

Nyall,

For the record, I'm running now QGIS master against PROJ's epsg10_part2 branch and things 
seem to go fine (but I didn't stress QGIS much)

QGIS doesn't use proj_crs_get_datum() (which will return NULL in PROJ 8 / epsg10_part2 
branch for a CRS based on WGS 84), so at first sight, I don't think any code change is required.

> Am I correct in my understanding that a ensemble is (at this stage) a
> "metadata only" concept, which indicates that the datum is only
> suitable for approximate spatial referencing? Or does a datum being
> flagged as an ensemble also influence the operations determined by
> PROJ and their order of precedence?

This should have no consequence on the order of precedence of operations returned by 
PROJ. At least for now since in PROJ, for convenience of code, in the createOperations() 
code paths, everytime a datum is needed, if the CRS intead uses a datumEnsemble it is 
internally converted to a datum object. The sorting logic of operations doesn't specifically 
take into account the presence of a datum ensemble (other than the fact that 
transformations from/to WGS 84 should normally have at least a >= 1m accuracy)

> How would you imagine a downstream project like QGIS should respond to
> this concept? Should we add a user-visible flag on any CRS definitions
> associated with a datum ensemble to say "Warning: ensemble datum, not
> for use in accurate spatial referencing"? What other user-facing
> changes do you think should be introduced?

That's a tough question. Depends if we want to stress users or not :-)

For example, when you open a dataset referenced to WGS84, I don't think we should put a 
warning. The user isn't responsible for the choice of the data producer.

With PROJ 8 / epsg10_part2, the WKT:2019 output of a CRS using WGS84 will show the 
ENSEMBLE[] and the 2m ensemble accuracy, so that's already a form of warning (but 
probably most people not aware of the issue won't realize what this means)

When a user saves a layer from a CRS that doesn't use a datumensemble to another one that 
uses one, perhaps a warning could be issued. But I'm not sure if we should do that. That also 
depends on the accuracy of the data itself. Reprojecting something with a 100m accuracy to 
WGS84 isn't really a problem. There are so many ways in which users can shoot themselves in 
the foot. Almost anything involving datum transformation introduces extra inaccuracy.

There's also the fact that there are none projected CRS available in EPSG based on any of the 
realizations of WGS84 or ETRS89... So users have no easy choice of a better alternative, if 
they want to stay in the 'family' of the datum ensemble.

And not all datum ensembles are the same. (well, there are just 2 geodetic datums in EPSG 
for now :-)). But the ensemble accuracy of ETRS89 is 0.1 m. Not the same story as the 2 m of 
WGS84. So for European users ETRS89 is quite a reasonable choice for a lot of use cases. I 
don't believe the use of a given realization of it is very frequent, at least in the GIS field. 
People would actually rather use a national datum when available (in France RGF93 which in 
its latest version is a realization of ETRF2000 at epoch 2009.0, similar story for CHTRF95 in 
Switzerland, etc.), mostly for legal reasons, than a generic pan-european ETRF realization.

That's just my current thoughts. I've no definitive opinion. Looks to me datum ensemble are 
more a band aid to reflect that the current situation is not ideal, but the ideal has not been 
drawn (yet?)

Even

-- 
Spatialys - Geospatial professional services
http://www.spatialys.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/proj/attachments/20201011/c3c05273/attachment-0001.html>