[Qgis-user] QGIS/GDAL GML flipped coordinate order with URN/URI srsName

Even Rouault even.rouault at spatialys.com
Tue Oct 11 02:59:34 PDT 2016


Le mardi 11 octobre 2016 10:47:09, Ilkka Rinne a écrit :
> Hi,
> 
> We are experiencing an unwanted flipping of coordinate order interpretation
> when adding GML 3.2 files as new vector layers in QGIS 2.2 Valmiera.
> 
> The GML data is has been encoded as a wfs:featureCollection with EU INSPIRE
> ProtectedSite 4.0 features as collection members (XML Schema:
> http://inspire.ec.europa.eu/schemas/ps/4.0/ProtectedSites.xsd). When using
> the URI-type srsName "http://www.opengis.net/def/crs/EPSG/0/3035" the
> coordinates seem to be interpreted (by GDAL?) in order (easting, northing),
> while the official axis order (according to the EPSG-registry) is
> (northing, easting). If we change the srsName to the URN-type format
> "urn:ogc:def:crs:EPSG::3035" in the GML file, the import seems to use the
> correct coordinate order.
> 
> To comply with the INSPIRE requirements, we have to use the URI format
> srsNames, but this is likely to cause a lot of confusion for the QGIS &
> GDAL users of the data if the coordinates are interpreted in the wrong
> order.
> 
> I'm fully aware of the massively confusing lat/lon (x/y) coordinate order
> interpretation misalignment between the geographers and the (GIS) software
> implementors, but was not aware that this issues also shows it's ugly head
> for other CRSes than CRS 84/EPSG:4326.

It's a mess indeed, reinforced by constant changes in srsName syntaxes.

> 
> Question 1: Is this an intentional feature or a bug?

This was a bug. I've just fixed it per https://trac.osgeo.org/gdal/ticket/6678 
(in time for the upcoming GDAL 2.1.2)

> 
> Question 2: Is it possible to force GDAL from the QGIS UI to interpret the
> coordinate order of these GML files in the official order?

No workaround I can think of, apart the manual replacing as you did.

Even

-- 
Spatialys - Geospatial professional services
http://www.spatialys.com



More information about the Qgis-user mailing list