[GRASS-dev] Re: [GRASS-user] how to import NetCDF data??

[Hamish]

Michael wrote:
> We had the same errors with a couple
> of files. We sent Tom the simplest and smallest of them.
> Once we figure out what is happening we can let everyone
> know. It would probably be good to begin a WIKI page on this
> once we have some info.


there already is one: (albeit a bit empty)
   http://grass.osgeo.org/wiki/NetCDF

see also:
   http://grass.osgeo.org/wiki/GRASS_and_GMT


and the gdal_translate trick to extract the data layer into a geotiff,
then import the geotiff into GRASS. (probably a direct import is
possible, but the geotiff makes problems easier to understand)
that method is explained here:
   http://grass.osgeo.org/wiki/MODIS#Method_1

also in the above link is instructions for applying georef to the geotiff
using gdal_translate as GDAL wasn't doing so, which is I think the same
problem you'll be seeing.


you get the 'illegal latitude for north' error because it has no georef
(as far as gdal is reporting) and so falls back to 1 row per pixel. If
the image is more than 90 pixels high that exceeds 90 degrees N and is
illegal for a lat/lon location. You can override that in recent versions
of GRASS with the r.in.gdal -l flag, but then Must use r.region directly
after import to set the bounds to what they should be.




Helena wrote:
> > If there are still problems and if we can identify what needs to be
> > developed to make working with netCDF in GRASS more robust perhaps
> > this would be a good project for Google summer code
> > (hopefully OSGeo is participating this tear again)

hopefully :)

anyway I think if there are georef problems they probably need to be
addressed at the GDAL side of things, not in GRASS per se.



Kirk wrote:
>>> I had to import a bunch of SSURGO NetCDF data not too long ago
>>> Michael (I was going back and forth with GRASS and MATLAB).

any reason not to use r.in.mat, r.out.mat for that? i.e. any short-
comings in those tools which should be addressed? 


regards,
Hamish