[Qgis-developer] Fwd: Re: Datum Transformation - parameters for mainland Portugal

[Marco Hugentobler]

>Since there is now a duplication, I think you can delete the grids entry with coord_op_code 100004 and 100007.

With which other entries are they duplicated? Seems to me there are no two entries with same source_srs / dest_src / gsb file ?


>No, those parameters that have +towgs84=565.04(...) are not associated with any portuguese transformation. These parameters belong to epsg_nr 1571. The only point of contact between this entry and the portuguese entries, is that >both uses 4258 (Geographic CRS ETRS89). epsg_nr 1571 uses it as source_crs_code, and some transformations for Portugal (ntv2 grids transformations) uses it as target_crs_code.


Not sure what you mean.

transformation epsg:1571 says there is a transformation between 4258 and 
4326 with +towgs84=565.04(...). So this transformation can be used for 
all projections based on 4258 to transform to wgs84 ellipsoid (and a 
second transformation is listed to get from wgs84 ellipsoid to the 
destination datum).

Regards,
Marco

On 20.06.2014 13:13, Pedro Venâncio wrote:
> Hi Marco.
>
>>> Added two ntv2 entries and updated two entries, hope it is correct now.
> Great Marco, it is working now with ntv2 grids.
> Since there is now a duplication, I think you can delete the grids entry with coord_op_code 100004 and 100007.
>
>
>
>>> I still feel uncomfortable to delete / update towgs84 entries in the qgis db only. I'm concerned this will get a mess over time if there are lots of updates (and the qgis.db is synchronized during install with the gdal table).
>>> If the updates are very important, I'd suggest the Portuguese user group to provide a modified copy of srs.db for people to download.
> Yes, this is a possibility.
>
>
>
>>>> However I came across something quite odd. In the Datum Transformation window, automatically arise some strange transformations (red in the image http://goo.gl/iIWOuw). These transformations give huge errors, and I do not understand what they represent.
>>> QGIS reads all possible combinations to get from one datum to the other. Seems to me these towgs84 transformations are all present in the transformation table
> No, those parameters that have +towgs84=565.04(...) are not associated with any portuguese transformation. These parameters belong to epsg_nr 1571. The only point of contact between this entry and the portuguese entries, is that both uses 4258 (Geographic CRS ETRS89). epsg_nr 1571 uses it as source_crs_code, and some transformations for Portugal (ntv2 grids transformations) uses it as target_crs_code.
>
>
>
>>> (though I don't know . Is it 4258?)
> The biggest mess in my mind is precisely this. Is that, even the possible incorrect match between source and target codes does not make sense, because the geographical crs related to 30004 is 4326 (+towgs84). We only use 4258 for ntv2 grids.
>
> So, these options in red in the image should not show up. I think this problem only happens with the portuguese transformations, right Giovanni?
>
>
>
>>> Maybe it could be done in the packaging (location is system dependent and there is not proj function call to get the right location).
>>> Budling ntv2 files with the application is a frequent request. We can discus thiss further after the release. Problems I see are:
>>> - too large install package if lots of files are present
>>> - requires to check every file license
>>> Problem 2 can easily be solved if a volountear steps up for maintaining.
> I can check the portuguese grids license.
>
>
> Thank you very much Marco!
>
> Best regards,
> Pedro
> _______________________________________________
> Qgis-developer mailing list
> Qgis-developer at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/qgis-developer


-- 
Dr. Marco Hugentobler
Sourcepole -  Linux & Open Source Solutions
Weberstrasse 5, CH-8004 Zürich, Switzerland
marco.hugentobler at sourcepole.ch http://www.sourcepole.ch
Technical Advisor QGIS Project Steering Committee