Loading model data with 360_day calendar using datetime.datetime objects

[Adrian Matthews]

Hello,

I am trying to read in model data with a 360_day calendar for the first time in iris. I  have always used datetime.datetime objects to set a time constraint (using data with a Gregorian calendar) but this now fails with the error

TypeError: Cannot determine the order of cftime.datetime objects

Can I use datetime.datetime objects to load in data with a 360_day calendar? (Note I am using sensible times that do exist in the data, not eg 31 Dec). If not, what is the best way to set a time constraint to load in data on a 360_day calendar eg between 1986-01-01 and 1986-12-30?

Thanks

Adrian

[Adrian Matthews]

Many thanks to Andrew Dawson for providing the following solution. The key is to use cftime.Datetime360day objects instead of datetime.datetime objects, and to set the iris time Constraint on the numerical values (i.e. the point attributes) on the time coordinate, rather than the default datetime like values. Sample code below for anyone else with this problem.

start = cftime.Datetime360Day(1985, 1, 1)

end = cftime.Datetime360Day(1985, 12, 30)

time_selection = iris.Constraint(time=lambda c: start <= c.point <= end)

iris.load_cube('my_filename.nc', time_selection)

Happy Christmas

Adrian