The 40 attributes that CF defines to describe data can be categorized:

• Data provenance: title, institution, source, history, references, comment, Conventions
• Description of data: units, standard_name, long_name, ancillary_variables, missing_value, valid_range, valid_min, valid_max, flag_values, flag_meanings, flag_masks, _FillValue, standard_error_multiplier
• Coordinate systems: coordinates, axis, bounds, grid_mapping, formula_terms, calendar, leap_month, leap_year, month_lengths, positive
• Data packing: add_offset, scale_factor, compress,
• Data cell properties and methods: cell_measures, cell_methods, climatology,
• Discrete sampling geometries (new): cf_role, featureType, instance_dimension, sample_dimension

Most of these CF attributes are optional.

2012 Unidata NetCDF Workshop > CF Conventions