this bug's URL: http://intevation.de/rt/webrt?serial_num=4877
-------------------------------------------------------------------------
Subject: d.grid -g, ps.map: incorrect geogrid rendering
Hi,
"d.grid -g" and ps.map's geogrid command are not calculating the correct grid
lines for me. (lines are drawn 90m to the north)
Perhaps it is only doing the projection, and not the datum transform ?!!?
Lon is *exactly* correct though, only lat is just slightly 3" to the north. Weird.
"d.where -w" gets it correct. Also tested with a v.mkproj in LL location +
v.proj into projected location. So I am pretty happy the data is ok just the
grid lines are drawn wrong.
Happens with both NTv2 grid and 7-term transform parms,
proj=nzmg
datum=nzgd49
?
Hamish
-------------------------------------------- Managed by Request Tracker
Hello Hamish
On Wed, 19 Jul 2006, Request Tracker wrote:
this bug's URL: http://intevation.de/rt/webrt?serial_num=4877
-------------------------------------------------------------------------
Subject: d.grid -g, ps.map: incorrect geogrid rendering
Hi,
"d.grid -g" and ps.map's geogrid command are not calculating the correct grid
lines for me. (lines are drawn 90m to the north)
Perhaps it is only doing the projection, and not the datum transform ?!!?
Looking at the code, this seems quite possible. I have applied a patch in CVS - can you test? If it's working now, I expect we should be able to apply a similar patch to ps.map.
Slightly more in-depth explanation:
this behaviour stems from a questionable "feature" of PROJ (my words really although Frank has agreed too---see PROJ bug 368 http://bugzilla.remotesensing.org/show_bug.cgi?id=368 ) where if one co-ordinate system has datum transformation parameters specified and the other does not, a kind of partial transformation is done. IMHO if the datum information is not fully specified it should be ignored completely, but that is not possible the way the datum information is contained as part of the co-ordinate system in PROJ. Thierry Laronde and KerGIS were doing this a different way to avoid this problem as far as I remember...
Paul