ORAC reads netcdf files by first applying a fill value (-999) and then applying the scale + offset specified in the netcdf attributes.
This is problematic in the case that -999 is a valid value in the unscaled netCDF data (such as for short int data in the range -32766 -> +32767).
In this PR I correct the netcdf read code so that this problem doesn't arise any more. I think I also have it set up to cover the case that the input netcdf data doesn't contain a fill value but can't actually check this as I don't have any such netcdf files available.
ORAC reads netcdf files by first applying a fill value (
-999
) and then applying the scale + offset specified in the netcdf attributes. This is problematic in the case that-999
is a valid value in the unscaled netCDF data (such as for short int data in the range-32766
->+32767
).In this PR I correct the netcdf read code so that this problem doesn't arise any more. I think I also have it set up to cover the case that the input netcdf data doesn't contain a fill value but can't actually check this as I don't have any such netcdf files available.