[gdal-dev] gdal_translate into netCDF file will lose some important metadata items such as scale_factor and add_offset

[Even Rouault]

Zhan,

adding back the list

There was an issue in the writing code of the netCDF driver. It was supposed supposed to 
copy the offset and scale settings from the source dataset, but there was a logic error. Fixed 
per 
https://github.com/OSGeo/gdal/pull/2906

Even


> Hi Even,
> 
> Thanks a lot for your reply! Yes here in this example, scale_factor and
> add_offset do not matter. I put them here just for demo purposes. In my
> actual data (quite large), I'd like to use 1e-3 as scale factor but could
> not write scale_factor into the metadata of netCDF via gdal_translate. Even
> if I use VRT as an intermediate file and manually add scale_factor into the
> VRT per band through "<MDI key="scale_factor">1e-3</MDI>". It still does
> not work :(
> 
> But it sounds like the implementation of the netCDF driver in GDAL has some
> explicit tweak of metadata items? Like you said, valid_range is explicitly
> discarded. Maybe scale_factor and add_offset and such are also part of
> metadata items that the driver explicitly discards? In that case, I'll
> stick to scale_factor=1 and add_offset=0. Not ideal though. Otherwise, this
> loss of metadata items for netCDF files should count as a bug to be
> reported?
> 
> Zhan.
> 
> On Mon, Aug 31, 2020 at 11:06 AM Even Rouault <even.rouault at spatialys.com>
> 
> wrote:
> > On lundi 31 août 2020 00:11:54 CEST Zhan Li wrote:
> > > Hi, when I use gdal_translate to convert a raster file (e.g., either
> > 
> > GTiff
> > 
> > > or VRT) into netCDF format, it always loses some important metadata
> > > items
> > > 
> > > such as scale_factor, add_offset and etc. What am I doing it wrong? Can
> > > 
> > > anyone help me to solve this problem?
> > 
> > As scale_factor=1 and add_offset=0 in that dataset, they have no effect,
> > and are thus not stored when converted to GeoTIFF.
> > 
> > 
> > 
> > valid_range is explicitly discarded from source metadata when converting
> > into netCDF. I'm not completely sure of the rationale for that.
> > 
> > 
> > 
> > Even
> > 
> > 
> > 
> > --
> > 
> > Spatialys - Geospatial professional services
> > 
> > http://www.spatialys.com


-- 
Spatialys - Geospatial professional services
http://www.spatialys.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/gdal-dev/attachments/20200831/a56ac955/attachment.html>