Open anissa111 opened 6 months ago
I went down the rabbit hole a bit and I'm guessing it's because this (i.e. the interpretation of calendars for days_in_month) seems to all be custom in NCL: https://github.com/NCAR/ncl/blob/edd46f5426286146313b31bd6826e3d3c360c0dc/ni/src/lib/nfp/datesW.c#L760
And defined in ways that don't align with the CF conventions or UDUNITS.
This is also relevant: https://github.com/NCAR/ncl/blob/edd46f5426286146313b31bd6826e3d3c360c0dc/ni/src/lib/nfp/datesW.c#L1015
So effectively in NCL for this function anyway (and likely some others, but maybe not all) the "none", "standard", and "gregorian" calendars are all "proleptic_gregorian" calendars 🙃 .
44 discovered some differences in the NCL and
cftime
gregorian and standard calendars.We should add more receipts testing to fully understand the differences.
Depends on #44