@JimBiardCics Thanks for the detailed answers to my three questions. 

On the first question, just I just want to dig a bit further (I am not trying 
to be difficult here, just to make things clear, at least for myself). What is 
the calendar used for the reference time stamp of a `gregorian_tai` time? Is it 
`gregorian_tai` (I think it must), or could it be a `gregorian_utc`? The use 
case (possibly only imagined by myself) where this is important is two 
satellites (totally different types, or one replacing the other in the same 
series) where one want to have a common precise reference time scale. 

On the second,  OK let's leave the `year` unit out of the current issue.

On the third question (on calendar names), I first wonder whether the 
`gregorian_` part actually is needed for the `tai` and `utc` calendars? Is it 
to indicate that the conversion to a format for human consumption is the 
gregorian style year, month, day, etc.? But is that not implicit in how these 
time systems are used in general? By introducing the `gregorian_` part an 
unwanted link to the fuzziness of the `gregorian` calendar is created. 
Secondly, I agree that the `gregorian` calendar is well described as a "mixed 
julian/gregorian" calendar. But in this case I think that further improving the 
clarity by introducing the `mixed_julian_gregorian` and deprecating `gregorian` 
is warranted. It makes the distinction to the `gregorian_proleptic` clearer. 
For modellers the `gregorian_proleptic` is relevant, and for long historic 
observational datasets the `mixed_julian_gregorian` would be the relevant one.  

If CF ends up with several `gregorian_` + `specification` calendars, the 
`gregorian_` part essentially is something else than what the `gregorian` 
calendar currently stands for. And this may lead to unnecessary confusion.  

Kind regards,
Lars        

-- 
You are receiving this because you commented.
Reply to this email directly or view it on GitHub:
https://github.com/cf-convention/cf-conventions/issues/148#issuecomment-436892572

Reply via email to