[CF-metadata] How to define time coordinate in GPS?

From: Jonathan Gregory <j.m.gregory>
Date: Thu, 21 May 2015 18:13:11 +0100

Dear Tim

You certainly can have a time variable, with plain time units (without "since
whenever"), which is just a count of elapsed time, like any other sort of
coordinate. I'm not sure that's what you mean though, is it?

Best wishes

Jonathan

----- Forwarded message from Timothy Patterson <Timothy.Patterson at eumetsat.int-----

Date: Wed, 20 May 2015 19:47:07 +0200
From: Timothy Patterson <Timothy.Patterson at eumetsat.int>
To: "cf-metadata at cgd.ucar.edu" <cf-metadata at cgd.ucar.edu>
Subject: Re: [CF-metadata] How to define time coordinate in GPS?


>From my (perhaps naive!) viewpoint, it seems the convention is trying to use a single "time" variable to encode two different concepts; the simple "elapsed time" since a given time-zero, monotonically increasing without discontinuities, in line with all other measurement variables like distances, temperatures, etc.); and an encoded "look-up-table" or "date" into which leap second discontinuities are inserted at random junctures and which may not even be defined outside certain boundaries.

I expect that any request to define, say, a temperature variable in this way - sometimes in units of Kelvin and sometimes in a different measurement system which jumped certain temperature values - would be quickly dismissed, but the current proposals for the time variable all seem to be advocating this approach instead of perhaps reserving "time" for storing the actual count of seconds, analogous to all other CF coordinate measurements, and introducing a separate "date" variable for those who want to encode whatever discontinuous, non-linear date system they choose.

I should note that this proposal might not be fully backwards compatible (i.e. it's probably entirely incompatible).

Regards,

Tim
Received on Thu May 21 2015 - 11:13:11 BST