[Proj] ERROR 6: No translation for Double_Stereographic to PROJ.4 format is known.

[Mikael Rittri]

Frank Warmerdam wrote (31 March 2009): 

> OvV_HN wrote:
> > I don't get it. I always thought the existing PROJ (and libproj) 
> > function sterea was a full implementation of the double
stereographic 
> > projection, suited for Dutch RD, New Brunswick / Prince Edward
Island 
> > stereo and Romanian Stereo 70.
> 
> Oscar,
> 
> Well, that is possible.  Is someone willing to do some leg work 
> confirming if that is the case, and documenting the correct
formulation 
> of the coordinate systems described in
http://trac.osgeo.org/gdal/ticket/2487
> for PROJ.4?  Some sample test points would also be helpful.  Given
that I 
> would be happy to update GDAL to support double stereographic and
update 
> some other resources.

I think Oscar is right (at least for the Netherlands,
very likely for the Canadian projections, and I don't 
know anything about Romania.)

I found some older posts in the Proj Archives, which 
I try to summarize here. (Apologies if there are 
errors in my summary.)

Gerald Evenden wrote a post, mainly about the 
terminology, but also citing Rueben Schulz, who 
wrote that 
> The EPSG does not have any plans to include 
> the USGS form in their database since they 
> do not know of any coordinate reference systems 
> that use it.
http://lists.maptools.org/pipermail/proj/2004-May/001216.html

Jan Hartmann agreed that the Dutch Stereographic 
should be implemented by proj=sterea, not proj=stere, 
and attached two images showing 
(a) that the United States variant, proj=stere, would
give errors up to 8 meters in the Netherlands, and 
(b) that the correct proj=sterea gives errors up 
to 0.2 meters, anyway. (?!)
http://lists.maptools.org/pipermail/proj/2006-January/002021.html

Oscar van Vlijmen replied that proj=sterea should 
give errors at most 0.2 millimeters in the Netherlands 
(and that's because the Dutch authorities use an 
approximative algorithm claimed to have that accuracy, 
while proj=sterea is exact (or more exact)). 
The up to 0.2 meter error, found by Jan Hartmann, 
can be attributed to a grid correction used by 
the official transformation tool RDNAPTRANS, writes Oscar.
If I understand him correctly, the official tool conflates 
the map projection with a datum shift via grid file.
There are some test points in Oscar's post, but he wrote    

> I got with my version of sterea: 

so I guess that's his own implementation of sterea, 
rather than the implementation in Proj.4. 
http://lists.maptools.org/pipermail/proj/2006-January/002022.html

Best regards,

--
Mikael Rittri
Carmenta AB
SWEDEN
www.carmenta.com