Calendar attribute in outputs should be proleptic_gregorian

[aidanheerdegen]

The gregorian calendar option for MOM5 implements a proleptic_gregorian calendar

MOM5/src/shared/time_manager/time_manager.F90

Lines 2799 to 2807 in baaf7ed

	function leap_year_gregorian_int(year)
	logical :: leap_year_gregorian_int
	integer, intent(in) :: year
	leap_year_gregorian_int = mod(year,4) == 0
	leap_year_gregorian_int = leap_year_gregorian_int .and. .not.mod(year,100) == 0
	leap_year_gregorian_int = leap_year_gregorian_int .or. mod(year,400) == 0
	end function leap_year_gregorian_int

whereas the CF conventions state that a gregorian calendar attribute is mixed Julian/Gregorian, and a proleptic Gregorian must have the attribute proleptic_gregorian.

https://cfconventions.org/Data/cf-conventions/cf-conventions-1.7/build/ch04s04.html

Original issue is here: COSIMA/access-om2#117

A simple fix would be to change this line

https://github.com/mom-ocean/FMS/blob/3bd24800ed3cd8b0eb26643afdecf88e10641beb/time_manager/time_manager.F90#L1564

to

valid_calendar_types = 'PROLEPTIC_GREGORIAN'

as this is the function that is called when the time attributes are written:

https://github.com/ACCESS-NRI/MOM5/blob/baaf7ed1baf4543bc821d97402a94792b4044ac4/src/shared/diag_manager/diag_output.F90#L353-L355

Which results in a netCDF variable definition like so:

        double time(time) ;
                time:long_name = "time" ;
                time:units = "days since 0001-01-01 00:00:00" ;
                time:cartesian_axis = "T" ;
                time:calendar_type = "GREGORIAN" ;
                time:calendar = "GREGORIAN" ;
                time:bounds = "time_bounds" ;

Note that this line appears in the MOM5 source here due to FMS being included as a subtree:

MOM5/src/shared/time_manager/time_manager.F90

Line 3256 in baaf7ed

valid_calendar_types = 'GREGORIAN '