@larsbarring I had started out advocating for constraining ourselves to a single new calendar that would signal that the values in the time variable were metrical. The current direction appears to be that the epoch date and time stamp in the **`units`** attribute would, in each calendar, conform to the calendar. The epoch time stamp for a TAI calendar would be a TAI time stamp. The epoch time stamp for a UTC calendar would be UTC time stamp. The epoch time stamp for a GPS calendar would be a GPS time stamp. The differences between the time stamps for the same event would be as seen at [this link](http://leapsecond.com/java/gpsclock.htm).
If we are going to have new calendar names for each, I have no problem with the names **`tai`**, **`utc`**, and **`gps`**. The naively converted UTC time stamps calendar could be called **`utc_stamp`** or some such. -- 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-437006877
