Dutch Coordinate System Not Working

[lxnyce]

From a customer :

Just so we know what we are talking about: the Dutch coordinate system is ‘Rijksdriehoeksstelsel’. Luckily that is commonly abbreviated to ‘RD’.

Most of that is just parameters, and we can fill them out basically. But what is fundamentally lacking is the actual projection formula used in that coordinate system. It’s the ‘Double Stereographic’ projection, used worldwide only in NL and on Nova Scotia. It is different from the regular stereographic projection, one that is supported….

Please find enclosed two point shapefiles, depicting the same points. One is in RD, the other one is in geographic coordinates on the WGS84 datum. Please also find enclosed a PDF (published by the Dutch Cadastre, the ‘owners’ of the coordinate system) describing the specs of the coordinate system (chapter 4.2.1 for latlong à X, Y, and 4.2.2 for X, Y à latlong)

[warmerdam]

Lennox,

I don't believe there is much I can do within GDAL till there is a corresponding PROJ.4 projection method. So I think the first step is to file a ticket in the PROJ.4 Trac, and to contact Gerald Evenden (via the PROJ.4 mailing list) asking if he would consider implementing this projection or if there is an existing one that actually matches it.

Once it is in PROJ, we can add it to OGRSpatialReference for use in GDAL and OGR.

PS. It would be helpful if you could append an ESRI .prj file for this coordinate system so I can ensure OGR WKT will match what ESRI does.

[lxnyce]

The ESRI generated prj file

[hogend]

This actually is EPSG:28992, so: +proj=sterea +lat_0=52.15616055555555 +lon_0=5.38763888888889 +k=0.9999079 +x_0=155000 +y_0=463000 +ellps=bessel +units=m +no_defs

[warmerdam]

There was an issue with the Amersfoort datum which is addressed in #1987. However, this ticket also suggests that proj=sterea is not a good implementation of the double stereographic projection. Is that right?

[vincentschut]

Frank, thanks for adding me to the cc. However, I'm by no means a dutch RD proj(ection) expert (just happen to live there), and adding the datum transform values solved the problem for us then. That is, the results looked as we expected and appeared to fit with other datasets.