[Proj] proj and Dutch RD

[Roger Oberholtzer]

On Thu, 2006-01-12 at 00:17 +0200, Martin Vermeer wrote:
> On Wed, Jan 11, 2006 at 12:43:13PM -0800, Eric Miller wrote:
> > I think Martin's comment is a little to brief.
> > 
> > Your source coordinate system is geodetic coordinates using the WGS84 Datum (implies WGS84 ellipsoid)
> > In cs2cs terms:  +proj=latlong +datum=WGS84
> > 
> > Your destination coordinate system is RD (new)
> > In cs2cs terms: +proj=sterea +lat_0=52.15616055555555 +lon_0=5.38763888888889 +k=0.9999079 +x_0=155000 +y_0=463000 +ellps=bessel +units=m +towgs84=593,26,478
> > 
> > I'm not sure about the signs for the towgs84 parameters (dx,dy,dz), maybe they're supposed to be negative.  Anyway, the +towgs84 tells cs2cs how to transform the WGS84 coordinates (in XYZ space) to whatever the output datum for RD is.  The +towgs84 or +datum for a different target coordinate systems will likely be different.  But, your source definition would stay the same.  If you have some reference coordinates, you can probably work out whether the towgs84 numbers are supposed to be positive or negative (provided those are good values).
> > 
> > Clear as mud?
> 
> This is precisely what I meant. Thanks!

But the 'fix' was to change the description of the source lat/longs from

	+proj=latlong +ellps=WGS84
to
	+proj=latlong +ellps=bessel

After the '+to' I always had +ellps=bessel to describe the desired
result. That was not changed.

Confusion here continues...

> 
> So play around with those +towgs84 parms, until you can reproduce your
> reference values. There must be decimetre precision shift values
> somewhere too, as I know the triangulation to be that precise. They must
> know at the cadastral survey in Apeldoorn. Here is their on-line
> transformer:
> 
> https://RDinfo.kadaster.nl/rd/transformator.html

I have looked at that. In summary, the conversion that does the best
(almost identical values) is:

+proj=latlong +ellps=bessel +to
+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

The one that was off by at least 60 meters was:

+proj=latlong +ellps=wgs84 +to
+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

-- 
Roger Oberholtzer
OPQ Systems AB