Dear Lars

I think you have to put the operations in the order they are carried out:
mean within years (to get monthly means), mean over years (to get climato-
logical monthly means), range (to get climatological annual range). However,
you can describe e.g. 30-year climatologies, as you would like, because
the final time dimension can be bigger than one. The climatological time
bounds for it will indicate the 30-year periods concerned.

This is a bit more complicated than the ticket as so far discussed because
of combining multiple time processing with climatological time processing.
The ticket went dormant because no-one had the impetus to conclude it. If
you comment on it now with your new needs, the discussion could resume.

Best wishes

Jonathan

----- Forwarded message from Bärring Lars <[email protected]> -----

> Date: Mon, 16 May 2016 08:39:58 +0000
> From: Bärring Lars <[email protected]>
> To: Jonathan Gregory <[email protected]>, "[email protected]"
>       <[email protected]>
> Subject: RE: [CF-metadata] climatological statistics --- climate indices
> 
> Dear Jonathan,
> 
> Indeed, this looks like good solution. And to follow this line of thoughts to 
> its end; by shifting the order of the operations to
>       time: mean within years (period: 1 month) time: range time: mean over 
> years
> it should be possible to have one file containing a series of successive 30 
> year (say) averages of annual spans between monthly mean temperatures
>  
> I note that ticket 82 has gone dormant since almost a year, there was not 
> much discussion and the comments were largely positive. Any chance that it 
> will be resurrected?
> 
> 
> Many thank you for you constructive input.
> 
> Lars
> 
> 
> -----Original Message-----
> From: CF-metadata [mailto:[email protected]] On Behalf Of 
> Jonathan Gregory
> Sent: den 13 maj 2016 17:23
> To: [email protected]
> Subject: Re: [CF-metadata] climatological statistics --- climate indices
> 
> Dear Lars
> 
> When you compute the third operation (the range over climatological months) 
> you collapse the time axis to size 1, so it no longer indicates that it was 
> months. It might previously have been days, 5-day periods, or anything else.
> I think this case could be dealt with by the proposed (but not agreed) con- 
> vention for multiple processing of axes in cf-trac.llnl.gov/trac/ticket/82.
> Something like
> 
> time: mean within years (period: 1 month) time: mean over years time: range
> 
> might work.
> 
> Best wishes
> 
> Jonathan
> 
> ----- Forwarded message from Bärring Lars <[email protected]> -----
> 
> > Date: Fri, 13 May 2016 14:09:25 +0000
> > From: Bärring Lars <[email protected]>
> > To: Jonathan Gregory <[email protected]>, "[email protected]"
> >     <[email protected]>
> > Subject: RE: [CF-metadata] climatological statistics --- climate 
> > indices
> > 
> > Dear Jonathan,
> > 
> > > > But how about common continentality indices based on the annual range 
> > > > of monthly mean temperatures? 
> > > > Is there a "within months" such that the cell method for the basic 
> > > > element of the continentality index would be "time: mean within months 
> > > > time: range within years" ? Or is there another solution?
> > 
> > > Climatological stats can be described for the annual cycle. The 
> > > climatological monthly mean temperature would be "time: mean within 
> > > years
> > > time: mean over years", and you would use the climatological time bounds 
> > > to indicate the start and end of months (or other portions of the annual 
> > > cycle).
> > > This would produce twelve climatological monthly values (for each 
> > > location).
> > > Do you mean you then apply a third operation to compute the range of 
> > > these twelve values?
> > 
> > Yes, exactly. 
> > 
> > And, just to complete the picture, possibly even a to apply fourth 
> > operation to calculate a climatological average over e.g. 30 years. In 
> > practice, this is of course much simplified if one uses monthly mean 
> > temperature data as input because the operations that is actually performed 
> >  would be described by the cell method "time: range within years time: mean 
> > over years". But this would be an incomplete description of the data.
> > 
> > Kind regards,
> > 
> > Lars
> > 
> > 
> > 
> 
> ----- End forwarded message -----
> _______________________________________________
> CF-metadata mailing list
> [email protected]
> http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata

----- End forwarded message -----
_______________________________________________
CF-metadata mailing list
[email protected]
http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata

Reply via email to