[Proj] Misplaced SHP obtained with PROJ 4.8.0 called from gvSIG CE

[Hermann Peifer]

On 2013-08-26 18:43, Jorge Arevalo wrote:
>
> Ok, so, if I didn't misunderstand:
>
> - These proj strings should be valid for epsg:23030 to epsg:25830
> transformation using mentioned grid file:
>
> source SRS: +proj=tmerc +lat_0=0.0 +lon_0=-3.0 +k=0.9996 +x_0=500000.0
> +y_0=0.0 +ellps=intl +units=m  +nadgrids=@sped2et.gsb +no_defs

I never heard of the "@gridshiftfile.gsb" option. I only use 
+nadgrids=gridshiftfile.gsb (if the file is where PROJ expects it to be, 
on my MacBook at /opt/local/share/proj/) or 
+nadgrids=/path/to/gridshiftfile.gsb

>
> destination SRS: +proj=tmerc +lat_0=0.0 +lon_0=-3.0 +k=0.9996
> +x_0=500000.0 +y_0=0.0 +ellps=GRS80 +units=m  +nadgrids=@null +no_defs
>
> (not sure about the need of +no_defs part, but included just in case)

+no_defs is not needed in the above case.

> - These proj strings should be valid for the same transformation, but
> without using any grid file:
>
> source SRS: +proj=tmerc +lat_0=0.0 +lon_0=-3.0 +k=0.9996 +x_0=500000.0
> +y_0=0.0 +ellps=intl +units=m  +nadgrids=@ null +no_defs
>
> destination SRS: +proj=tmerc +lat_0=0.0 +lon_0=-3.0 +k=0.9996
> +x_0=500000.0 +y_0=0.0 +ellps=GRS80 +units=m  +nadgrids=@null +no_defs

If you have +nadgrids=@null on both sides (source and target SRS), then 
no actual transformation will be made.

> And the important thing about the change PROJ.4 4.5.0 to PROJ.4 >=
> 4.6.0 is that PROJ.4 version >= 4.6.0 needs a datum specification for
> *both* source SRS and target SRS in order to "make the datum shift".
> So, in the case I don't have a datum specification by default, I
> should *always* add +nadgrids=@null or +toWGS84=0,0,0.
>

Hmm. Blindly adding +nadgrids=@null or +toWGS84=0,0,0 would actually 
bring you back to PROJ 4.5.0 behaviour, and I doubt if you want this. At 
the time, the PROJ developers had good reasons for changing the 
behaviour of PROJ version >= 4.6.0.

However, in your example case the target datum is ETRS89 which is 
practically identical with WGS84, so using +nadgrids=@null or 
+toWGS84=0,0,0 is appropriate.

Hermann