[CF-metadata] date and time

From: John Graybeal <graybeal>
Date: Thu, 23 Oct 2008 12:34:06 -0700

Hi all. I am trying to understand (or establish) best practices for
date and time using CF standard names. Please note this is distinct
from the time _coordinate_ in the CF conventions; with observational
date, there may be multiple date and time readings from sensors, and
it seems to me these should be describable with standard names.

Right now there is a single standard name called 'time', with no
documentation, and no standard name called 'date'. (In the Coordinates
section, the units discussion for time refers the user to the UDUNITS
library. UDUNITS seems to have a less-than-optimal set of units for
representing time, as there is no way to incorporate 'date in a
Gregorian calendar'.)

In a situation where we have the following measurements being taken,
would it be appropriate to use standard names? And how can we
represent units adequately in this context? Can these data values be
represented without transformation in the netCDF file, while retaining
at least a useful standard name?

Header Description:
    DateTime (ISO8601), Elapsed Mission Time (seconds), Sensor Time
(ISO8601), Sensor Date (ISO8601), Day-of-Year
Data Record:
    2008-08-18T18:57:55.79, 8403.33, 12:12:11,2008-08-18,262

The first field is what might get converted to the time coordinate in
the netCDF file.

The second field could be represented as a regular variable with
standard_name = time and the units seconds_since_(start of mission).

The third field is a time, but the units are difficult to specify.
This is an acceptable ISO8601 format, so it seems to me there should
be a way to say units = 'ISO8601' or something like that; I know of no
way to do this.

The fourth field is a date, not a time. Does CF also refer to this as
a time, as far as standard names are concerned? And the units problem
occurs here also.

The fifth field is at least a 'days since' format, though the
reference point will have discontinuities every new year. I'm a little
at sea as to whether this can be called 'time' or not.

Would it be reasonable to use standard names like date (new!) and
time, and allow ISO8601 as the units? Or what is a better way?

John
Received on Thu Oct 23 2008 - 13:34:06 BST