In the case of 'raw' output from numerical models, it probably makes sense to
use the end-point of the time interval rather than the mid-point. That's the
moment for which the model stores the data, whether they're instantaneous
values (intensive variables) or time-averages over the previous timestep
(extensive variables).
If you used the mid-point of the interval for extensive variables, they
wouldn't have the same time coordinates as the intensive variables, which would
be very confusing. Using the end-point keeps everything aligned.
--Seth
On Thu, 12 Nov 2009 14:41:26 +0000 (UTC)
Thomas Lavergne <thomasl at met.no> wrote:
>Dear Jonathan,
>
>----- "Jonathan Gregory" <j.m.gregory at reading.ac.uk> wrote:
>> Dear Thomas
>>
>> I'm not saying the coordinate *must* be the mid-point. If there's a
>> good reason
>> for it being something else, then you could choose it to be so. I was
>> suggesting that we could recommend it should be the mid-point if there
>> is
>> no strong basis for making another choice. We could also say that it
>> must not
>> be outside the bounds.
>
>I agree with your recommendation.
>
>But I was also trying to gain support on "which axis value should I choose for
>my variable" and your answer does not help :-).
>
>I have rather little basis for making the choice of the end time for
>representing an accumulated quantity but, at least, CF does not forbid it. I
>guess I have to seek agreement inside my scientific community and that it is
>not CF's role to decide upon that.
>
>Are there people interested in taking the discussion further? We seek the
>answer to the question: "In which cases would another choice (other than
>mid-point) be relevant?".
>
>Thomas
>
>
>>
>> You are right, it cannot be missing data. That would break some
>> applications,
>> anyway.
>>
>> Cheers
>>
>> Jonathan
>> _______________________________________________
>> CF-metadata mailing list
>> CF-metadata at cgd.ucar.edu
>> http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata
>_______________________________________________
>CF-metadata mailing list
>CF-metadata at cgd.ucar.edu
>http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata
Received on Thu Nov 12 2009 - 09:21:13 GMT
