The fillvalue on line 88 of dataset_util.py is currently set to False.
This could result in non-CF-compliant fillvalues.
However, when using other (default) fillvalues, in many places (e.g. correlation matrices), zeros are being replaced by fill values (and thus NaNs.
The fillvalue on line 88 of dataset_util.py is currently set to False. This could result in non-CF-compliant fillvalues. However, when using other (default) fillvalues, in many places (e.g. correlation matrices), zeros are being replaced by fill values (and thus NaNs.