[PROJ] About geoid_model in proj.db

[Even Rouault]

Le 10/11/2021 à 19:12, Javier Jimenez Shaw a écrit :
> Hi
>
> In proj.db there is a table "geoid_model" with the list of accepted 
> GEOIDMODEL tags, like GEOID18, GEOID12B, OSGM15 or GGM10 (I added that 
> last one because because it is used with NAVD88 height, for consistency)
>
> https://github.com/OSGeo/PROJ/blob/04928472dbee772b6b31c78352dc486fb6543a91/data/sql/customizations.sql#L241 
> <https://github.com/OSGeo/PROJ/blob/04928472dbee772b6b31c78352dc486fb6543a91/data/sql/customizations.sql#L241>
>
> However there are many geoid grid files in PROJ-data that do not have 
> a geoid model name there.
>
> What is the policy to set geoid models there in that table?
That's a good question... I guess when there's a well accepted name for 
the geoid.
> I already have a patch for that table with many of them, like 
> AUSGeoid2020, DNN, EGM08-REDNAP, RAF09, RAF18, NZGeoid2016 or 
> SAGEOID2010. By the way, we use them to "know" that there is a grid 
> file for a vertical CRS using 
> https://proj.org/development/reference/functions.html#c.proj_get_geoid_models_from_database 
> <https://proj.org/development/reference/functions.html#c.proj_get_geoid_models_from_database>
This is a bit fragile. I'm not sure I have a solution with the existing 
API to propose though. You could issue a raw query to the database in 
the grid_transformation table with the source/target CRS being your 
vertical CRS.
>
> Does it make sense to add that to the PROJ repository? If yes, would 
> it make sense that any new geoid grid file added in PROJ-data adds a 
> geoid model there?
Not a big fan of this
> See that this is the tag GEOIDMODEL in WKT2 (if I understood 
> correctly) so it should not change "so often".
> I do not know how sensible it would be to changes in EPSG database.

We're actually using the GEOIDMODEL node of WKT2 differently from what 
EPSG does (I think I realized that after it was already implemented in PROJ)

In EPSG, GEOIDMODEL is just a reference to one or several 
transformations between the vertical CRS and geographic CRS. So the 
names are the ones of the transformations, like "WGS 84 to EGM2008 
height (1)".  The GEOIDMODEL[] node was initially included in the WKT2 
output, but EPSG changed their policy, and it is no longer in the output 
on epsg.org(I've already forgotten the reason). See 
https://epsg.org/crs_3855/EGM2008-height.html that references 2 geoid 
models but https://epsg.org/crs/wkt/id/3855 doesn't include them.


>
> Thanks.
> .___ ._ ..._ .. . ._. .___ .. __ . _. . __..  ... .... ._ .__
> Entre dos pensamientos racionales
> hay infinitos pensamientos irracionales.
>
>
> _______________________________________________
> PROJ mailing list
> PROJ at lists.osgeo.org
> https://lists.osgeo.org/mailman/listinfo/proj

-- 
http://www.spatialys.com
My software is free, but my time generally not.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/proj/attachments/20211110/c90a9fa8/attachment.html>