[gdal-dev] gdal_translate differences between versions 3.4.3 and 3.5.1?

Even Rouault even.rouault at spatialys.com
Thu Jul 28 11:29:00 PDT 2022


Matt,

The output of gdalinfo that matters here was the one on the subdataset 
you transform, so

gdalinfo NETCDF:"./test.nc":analysed_sst

I've a very poor Internet connection for the next 10 days so can't 
download the full .nc file yet, but from the .ncdump file you provided, 
I believe I managed to reproduce your issue

I've queued a fix in https://github.com/OSGeo/gdal/pull/6129

There's no much workaround I can think of, apart removing the 
'coordinates' attribute on the sst_anomaly variable.

Even

Le 27/07/2022 à 21:19, Matthew Savoie a écrit :
>
> Hi Even,
>
>     > I have run this command with gdal 3.4.3 and gdal 3.5.1 and get
>     different output files.
>
>     Can you compare the output of the two versions of gdalinfo ?
>     This should determine whether the difference is when reading or
>     writing.
>
>     Good idea, I did that and there is no difference in the output of
>     gdalinfo between 3.4.3 and 3.5.1 on the test.nc file.
>
> Are you 100% sure about that ? It is highly unlikely that the 
> difference would be on the side of the output driver (GeoTIFF), and 
> much more likely on the reading side (netCDF). There have been 
> adjustments recently in the netCDF driver to be a bit more restrictive 
> regarding exposing geotransforms (the driver could expose non relevant 
> geotransforms), that have had side effects on some files where they 
> could be legitimate.
>
> You can likely workaround the issue by definining the configuration 
> option / environment variable GDAL_NETCDF_IGNORE_XY_AXIS_NAME_CHECKS 
> to YES (cf https://gdal.org/drivers/raster/netcdf.html#georeference).
>
> Yes, I rechecked the output of gdalinfo on the file.  Here is a link 
> to my shared drive.
>
> gdal-test 
> <https://o365coloradoedu-my.sharepoint.com/:f:/g/personal/savoie_colorado_edu/EjihTcOSuZdJiZiiJmXkEiABHpbXHPFab6MbnjSfyfDLfw?e=qhQYmF>
>
> The files are the test.nc file (which is in fact a copy of a dataset 
> file 20210429090000-JPL-L4_GHRSST-SSTfnd-MUR-GLOB-v02.0-fv04.1.nc)
>
> test.nc.dump which is the output of ncdump -h on the file
>
> gdal.3.4.3.out and gdal3.5.1.out which are the result of two commands 
> gdalinfo –version and gdalinfo test.nc. You can see the only 
> difference  is the version.
>
> I will look at the config option to see how to get around this.  Thanks.
>
> Matt
>
-- 
http://www.spatialys.com
My software is free, but my time generally not.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/gdal-dev/attachments/20220728/44c43370/attachment.htm>


More information about the gdal-dev mailing list