[GRASS-dev] [GRASS GIS] #2456: read CSV from GDAL data directory
paul-grass at stjohnspoint.co.uk
Mon Apr 25 14:15:53 PDT 2016
On 04/25/2016 09:47 PM, Markus Neteler wrote:
> Paul, still struggling with "SIRGAS2000"
> (g.proj versus testepsg output). Needs a different trick?
It looks to me like the canonical and equivalent names are the wrong way
round. I will attempt a brief summary.
datum.table is the oldest file, going back to the earliest days of
datums in GRASS. It should contain:
1. Short code for datum, used only in GRASS
2. Full EPSG name of datum, in quotation marks
3. ellipsoid used with this datum (GRASS ellipsoid code, from ellipse.table)
4. A single 3-parameter datum transformation that should cover the whole
area the datum is used in
datumtransform.table is newer and allows multiple different sets of
datum transformation parameters to be supplied for any datum. These can
be in any format accepted by PROJ.4, so towgs84= 3 parameters or 7
parameters, or nadgrids=
A datum does not have to have an entry in datumtransform.table, and
there can be multiple entries for each datum if necessary.
Some datums have different names (i.e. different from the EPSG name)
that are used in WKT descriptions. These should be added as a pair of
names in the papszDatumEquiv array lib/proj/convert.c. The equivalent
name comes first in the pair and the EPSG name (as used in the GRASS
datum.table file) comes second.
So for the SIRGAS2000 datum, according to the testepsg program
"Sistema_de_Referencia_Geocentrico_para_las_AmericaS_2000" is the EPSG
name, so this is what should go in datum.table. I don't think you even
need the equivalent
convert.c, unless there have been problems with this?
I would suggest the attached patch to solve both problems (taking into
account Helena's comments that NAD83 HARN should be separate).
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 1365 bytes
Desc: not available
More information about the grass-dev