What causes some values to become `nan` when one saves to_netcdf without using `encoding=`?

edit

Values can become nan in a netCDF file when using to_netcdf without specifying the encoding because the default encoding used by xarray (the library that provides the to_netcdf function) is not always suitable for all types of data.

In particular, if the data contains missing or undefined values (e.g., inf, -inf, None), these values can be converted to nan if the encoding is not specified correctly. This is because the default encoding for netCDF files uses a floating point data type with a specific precision and range, and values outside this range are converted to nan.

To avoid this issue, it is therefore recommended to specify the encoding when saving to netCDF, using the encoding argument in the to_netcdf function. This allows you to specify the data type and fill value for each variable, and ensure that missing values are properly preserved in the output file.