[Proj] Proj-5.0.0 cct path to gridfiles

[Kristian Evers]

> On 20 Feb 2018, at 21:55, stevejking <stevejking at talktalk.net> wrote:
> 
> Kristian,
> Definitely the ellipsoid the grids use is important. You won’t be able to
> replicate the transformations otherwise. The official transformation tool is
> here
> 
> https://www.ordnancesurvey.co.uk/gps/transformation/
> 

A few example transformations with cs2cs:

$ echo -3.5 55 0 |cs2cs -f %.10f +proj=latlong +nadgrids=null +to +proj=latlong +nadgrids=OSTN15_NTv2_OSGBtoETRS.gsb +geoidgrids=OSGM15_GTX_OSGBtoGRS80.gtx +ellps=GRS80
-3.4986556560	54.9999433768 -53.2179254054
$ echo -3.5 55 0 |cs2cs -f %.10f +proj=latlong +nadgrids=null +to +proj=latlong +nadgrids=OSTN15_NTv2_OSGBtoETRS.gsb +geoidgrids=OSGM15_GTX_OSGBtoGRS80.gtx +ellps=bessel
-3.4986556560	54.9999433768 -53.2179254054
$ echo -3.5 55 0 |cs2cs -f %.10f +proj=latlong +nadgrids=null +to +proj=latlong +nadgrids=OSTN15_NTv2_OSGBtoETRS.gsb +geoidgrids=OSGM15_GTX_OSGBtoGRS80.gtx +ellps=airy
-3.4986556560	54.9999433768 -53.2179254054

Do you believe me now when I say that grid shifts doesn’t take note of the ellipsoid you specify?
In your original case the ellipsoid specification is applied when the geodetic coordinates are projected with tmerc.
The projection is the last step in the transformation, which is why I only added the ellipsoid definition for the last
step in my cct pipeline demo.


> Cs2cs gets the correct result using these grids, which are latitude
> longitude grids of shifts realised on the Airy ellipsoid. The GTX comprises
> shifts in metres to the GRS80 ellipsoid. Whilst the GSB file comprises sets
> of 4 values for each grid point. These are latitude shift, longitude shift
> both in arc seconds, and two values indicating error.
> The previous versions
> of proj applied these shifts after first transforming the projection to
> latitude longitude on its own ellipsoid, and the resulting latitude and
> longitudes were then by definition wgs84 (which etrs is standing in for in
> this case).

This is still the case with cs2cs.

> 
> Steve
> 
> 
> 
> 
> 
> --
> Sent from: http://osgeo-org.1560.x6.nabble.com/PROJ-4-f3840930.html
> _______________________________________________
> Proj mailing list
> Proj at lists.maptools.org
> http://lists.maptools.org/mailman/listinfo/proj