[gdal-dev] netCDF signed nodata value

[Simon Lyngby Kokkendorff]

Thanks for the tips Kyle and Even and for the quick fix!
Not a big netCDF user myself, but have noticed that it seems to be the
standard format for many of my colleagues who work in climatology. So good
to get this fixed!

Cheers,
Simon


On Tue, Oct 27, 2015 at 6:08 PM, Even Rouault <even.rouault at spatialys.com>
wrote:

> Le mardi 27 octobre 2015 16:54:56, Kyle Shannon a écrit :
> > Simon,
> >
> > On Mon, Oct 26, 2015 at 5:04 PM, Simon Lyngby Kokkendorff
> >
> > <silyko at gmail.com> wrote:
> > > Hello list,
> > >
> > >    Just observed a puzzling behaviour of the netCDF driver. When
> creating
> > >    a
> > >
> > > netCDF dataset of Byte type, the netCDF driver seems to interpret the
> > > nodata value as a signed byte.
> > >
> > >  For example a nodata value assigned with -a_nodata 200, will become
> -56.
> > >  Of
> > >
> > > course one can work around this by simply casting the nodata value (as
> > > e.g. returned by band.GetNoDataValue()) to an unsigned byte.
> >
> > It appears the driver has a sort of data type mis-match with no data
> > and the band type.  A work around, if available, is to set FORMAT=NC4
> > as a creation option.  I am not sure if it is in gdal or the netcdf
> > library at this point.
>
> My understanding is netCDF 3 only supports signed byte and GDAL uses a
> special
> metadata item _Unsigned=True to correct the interpretation. But it failed
> to
> use it to interpret correctly the nodata value. Should be fixed by
> https://trac.osgeo.org/gdal/ticket/6175 (happy to have feedback from
> anyone
> more knowleagable)
>
> --
> Spatialys - Geospatial professional services
> http://www.spatialys.com
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/gdal-dev/attachments/20151027/ccb6c0d6/attachment.html>