Dear Jonathan,
Thank you for rapid answer. I agree that the mid-point is the most obvious choice, BUT:
Shouldn't we start by identifying what kind of variables we are referring to? Forgive me if my semantic is not accurate:
1) intensive variables : for those there is no problem, since bounds are not needed, nor relevant.
2) extensive variables which are average values : mid-point is the obvious choice (like giving the center point of the grid cell in spatial coordinates).
3) extensive variables which are accumulated quantities: that is where I am slightly uncertain. Of course, mid-point can be justified. But (gut feeling) an accumulated value is not truely achieved before the end of the interval.
Take again the case of sea ice motion as derived from satellite images (sorry, that is what I know best...). If my variable was the "average x and y velocities of the drift *during* the interval between start and stop time", I would use the mid-point, no problem. But again, my ice particle is not arrived at its destination before the time interval is finished, so why not using stop time? IMO, mid-time is worse than stop-time. Even start-time would be better than mid-point.
Are accumulated variables different from average variables?
And what are we trying to achieve? Wouldn't the most correct choice be to have a fillvalue in the axis to force user's/software to use the cell bounds? But we are not allowed to do this because of Unidata/COARDS, correct?
I am almost ready to give up and say mid-point is best for all practical purposes but wanted to hear if others in the CF community gave a thought to these things. It all started with someone proposing/metioning the end-time to be used for "accumulated precipitation".
Cheers,
Thomas
----- "Jonathan Gregory" <j.m.gregory at reading.ac.uk> wrote:
> Dear Thomas
>
> The reason for the asymmetry is that point coordinates are required by
> Unidata
> and COARDS conventions, whereas cell bounds were introduced by CF.
> Most of the
> new features introduced by CF are not mandatory, only recommended.
>
> I agree that it would be helpful if we made a definite recommendation
> for the
> values of the point coordinates, for the case where the data-writer
> does not
> know what to choose. It is important that sensible coord values should
> be
> provided, since generic applications may not consult the bounds. We
> could
> put a recommendation at the end of the preamble to Sect 4.
>
> I'd say the mid-point is the obvious choice. What do you think?
>
> Cheers
>
> Jonathan
> _______________________________________________
> 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 - 07:18:00 GMT
