Revision to leap seconds and calendar conventions (v2)

[JonathanGregory]

This PR is a work in progress and not yet ready for general comments

See issue #XXX for discussion of these changes.

Release checklist

• history.adoc up to date?
• Conformance document up to date?

[pvanlaake]

Suggested change

	Moreover, a given datetime in the **`tai`** calendar represents an instant that is later than the same datetime in the **`utc`** calendar.
	Moreover, a given datetime in the **`tai`** calendar represents an instant that is earlier than the same datetime in the **`utc`** calendar.

[JonathanGregory]

Thanks for noticing this. I think you're right, and therefore the corresponding statement for UTC is also wrong. It should be, "Moreover, a given datetime in the utc calendar represents an instant that is later than the same datetime in the tai calendar." The fact that I got it wrong suggests to me that it's worth spelling it out, provided we do so correctly! I'll change them both. Also it now seems to make more sense to swap the order of two of the sentences, thus (for TAI):

For any given instant, the tai datetime is ahead of the utc datetime, where "ahead" means the same as it does when describing a timezone to the east as being ahead of one to the west. The difference between the two datetimes (tai minus utc) for a given instant of time is 10 seconds plus the net number of leap seconds introduced since 1972-01-01. Moreover, a given datetime in the tai calendar represents an instant that is earlier than the same datetime in the utc calendar.

[ChrisBarker-NOAA]

Honestly, I think we're putting more that necessary description in the CF doc -- if you don't know what TAI and UTC mean, you shouldn't be using them :-(. -- but anyway, I think an example is in order -- examples are always more clear to users. Something like:

For example: 2025-11-13T21:00:43 TAI is the same instant as: 2025-11-13T21:00:10 UTC, as there have been 33 leap seconds introduced as of that date.

[Example taken from: http://www.csgnetwork.com/timetaidispcalc.html, which I presume is correct]

[JonathanGregory]

I think you're commenting on the tai entry. Rather than adding an extra example, we can mirror what is said for utc, as follows (new and changed text in italics here in order to highlight it, but not in the document):

For any given instant, the tai datetime is ahead of the utc datetime, where "ahead" means the same as it does when describing a timezone to the east as being ahead of one to the west. The difference between the two datetimes (tai minus utc) for a given instant of time is 10 seconds plus the net number of leap seconds introduced since the instant with the datetime 1972-01-10 in the tai calendar, when the current definition of UTC came into force. This instant has the datetime 1972-01-01 00:00:00 in the utc calendar. Moreover, a given datetime in the tai calendar represents an instant that is earlier than the same datetime in the utc calendar.

Thanks for the comment. I agree that if you're going to write data in UTC or TAI you should already know what they mean. However, we can't depend on users of the data being familiar with the detailed definitions. Yet they need to know them in order to interpret the data correctly.

[ChrisBarker-NOAA]

This looks OK to me.

Thanks for the comment. I agree that if you're going to write data in UTC or TAI you should already know what they mean. However, we can't depend on users of the data being familiar with the detailed definitions. Yet they need to know them in order to interpret the data correctly.

Or have tools that know :-)

Anyway, I generally think more info is better than less.

[JonathanGregory]

I've just corrected a mistake I made when adding Lars's table to the document. In this version, he had "yes" for "Leap days" in the non-model calendars. In order to give more information, I changed "yes" to "Julian" or "Gregorian", but I forgot that standard is both, so I've now put "Julian/Gregorian" for its entry.