[MapProxy] Problem with EPSG:31254 (MGI-Austira_West) remians

[Siegel Andreas]

Dear list,

maybe i have spotted a bug in mapproxy.
I'm very reluctant to make such a claim but it is the only conclusion left to me after some intensive testing.
Or maybe i did something stupid in mapproxy.yaml.

The goal is to use mapproxy as a service provider for several WMTS-Serivce.
In our country we use EPSG-31254 as aour Standard-CRS.

Projection properties:
Name: MGI / Austria GK West
Axis Info [cartesian]:
- X[north]: Northing (metre)
- Y[east]: Easting (metre)
Area of Use:
- name: Austria west of 11°50'E of Greenwich (29°30'E of Ferro).
- bounds: (9.53, 46.77, 11.84, 47.61)
Coordinate Operation:
- name: Austria Gauss-Kruger West
- method: Transverse Mercator
Datum: Militar-Geographische Institut
- Ellipsoid: Bessel 1841
- Prime Meridian: Greenwich


Test-Case:
There are 2 services (in 31254) available as sources: one by mapserver an one by geoserver.


Test-configuration:
very minimal. No grids, no chaches, just the 2 services with the following coverage:
    coverage:
      bbox_srs: 'EPSG:3857'
      bbox: [888355.66,5710350.78,1341807.99,6234817.80]


Findings:

EPSGs 3857 and 4236 just work fine.
31254 doesn't.

There is a problem with the BBOX used in the getMap request:
Mapproxy: bbox=46924.27,-42727.44,360519.68,134328.75
This is wrong.
The image is then placed south-east of Reims in France.

Correct would be: bbox=49575.70,-183696.50,404276.30,126502.90

I can not understand how mapproxy computes this wrong bbox-values.


Please note. EPSG-2056 (CH1903+ / LV95 Switzerland) is even worse.


Please have a look at the config used for the testing.
Download-Link: 
https://vogis.cnv.at/inspire-download/mapproxy-31254-problem.yaml



Runtime-Environmnent.
OS: Debian-12
Python: 3.12.2 in a virtual env
Pyproj: 3.7.1

Thanks in advance
Andy


-----Ursprüngliche Nachricht-----
Von: MapProxy <mapproxy-bounces at lists.osgeo.org> Im Auftrag von Thorsten Hildebrand (WhereGroup) via MapProxy
Gesendet: Donnerstag, 13. März 2025 06:25
An: mapproxy at lists.osgeo.org
Betreff: [SPAMVERDACHT] Re: [MapProxy] Problem with EPSG:31254 (MGI-Austira_West) (north, west coordinates)

Hi Andy,

have a look at the settings for axis_order_ne and axis_order_en at: 
https://mapproxy.github.io/mapproxy/latest/configuration.html#id13

Try to explicitly specify the WMS version with your requests.

> If you follow the definition of 31254
> (coordinate system: Cartesian 2D CS. Axes: northing, easting (X,Y). 
> Orientations: north, east.)
> X should be north and Y should be east.

...
The axis ordering defines in which order coordinates are given, i.e. 
lon/lat or lat/lon. The ordering is dependent to the SRS. Most clients and servers did not respect the ordering and everyone used lon/lat ordering. With the WMS 1.3.0 specification the OGC emphasized that the axis ordering of the SRS should be used.

Here you can define the axis ordering of your SRS. This might be required for proper WMS 1.3.0 support if you use any SRS that is not in the default configuration.

By default MapProxy assumes lat/long (north/east) order for all geographic and x/y (east/north) order for all projected SRS.
...

> In our production environment we still use an old Centos-Box (Python 
> 3.6 and mapproxy 1.12) and there everything works fine.

Also note the change in this topic in MapProxy version 1.13.0 ...
Changed in version 1.13.0: MapProxy can now determine the correct axis order for all coordinate systems when using pyproj>=2. The axis_order_ne/axis_order_en are ignored in this case.
...

Greetings from Berlin,
Thorsten

--
Thorsten Hildebrand
WhereGroup GmbH
Bundesallee 23
10717 Berlin
Germany

Tel: +49 (0)30 / 513 02 78 - 75
Fax: +49 (0)30 / 513 02 78 - 11

thorsten.hildebrand at wheregroup.com
www.wheregroup.com

Geschäftsführer:
Olaf Knopp, Peter Stamm
Amtsgericht Bonn, HRB 9885


_______________________________________________
MapProxy mailing list
MapProxy at lists.osgeo.org
https://lists.osgeo.org/mailman/listinfo/mapproxy