Hello,

In the original proposal for an ensemble like axis we used the standard_name 
'realization' because it included things that weren't obviously model ensembles 
- they may be forecasts or projections generated by statistical methods - but 
that could be handled in a way similar to model ensembles.  The paper:  
http://www.climateprediction.net/wp-content/publications/nature_allen_051000.pdf
 is an example of this sort of technique.

The use of 'realization' has been questioned before on this list - so I'd be 
very happy to review its use.  If 'ensemble' is more familiar and useful then 
that would be fine.  (I think this can be handled with aliases - is that right?)

I think there is a complication with ensemble based statistics (sorry I haven't 
followed this latest thread close enough to know whether it has been talked 
about).  You often weight each ensemble memember e.g. based on some estimate of 
that ensemble member's likelihood - before calculating the ensemble statistic. 
I think there is a thread somewhere like 
http://mailman.cgd.ucar.edu/pipermail/cf-metadata/2007/046103.html which 
discussed this previously.

Now apologies: in my usual useless style on this list - I probably won't manage 
to follow this up in any detail - but since Jonathan mentioned me I thought I 
needed to try to fill in some background.

Jamie


> -----Original Message-----
> From: Jonathan Gregory [mailto:[email protected]] 
> Sent: 15 November 2013 17:09
> To: [email protected]
> Cc: [email protected]; Steve Hankin; Kettleborough, 
> Jamie; Hedley, Mark
> Subject: [CF-metadata] Standardizing how ensemble 
> (realization) axes are encoded
> 
> Dear all
> 
> This is partly a reply to various off-list emails. 
> 
> I don't believe that the current CF standard says anything 
> specifically about
> ensemble axes. We did discuss introducing axis="E" but that 
> wasn't adopted. The
> axis attribute is generally redundant (the same came be 
> deduced from mandatory
> units, positive att or standard_name). I think it would be 
> better to use the
> standard_name to indicate an ensemble axis, whether collapsed or not.
> 
> Jamies Kettleborough has reminded us that we introduced the 
> standard_name of
> realization for ensemble members. (I did have a memory of 
> that, but I searched
> the table for a spelling with S rather than Z!) So I withdraw 
> my suggestion of
> new standard names for this. I think that the existing 
> convention permits a
> cell_methods entry of "realization: METHOD", when there is no 
> dimension named
> "realization", to indicate that an ensemble axis has been 
> collapsed using
> the METHOD stated. In the case where a cell_methods entry 
> gives a name that is
> not a dimension name, it's interpreted as a standard name, 
> and it means "over
> all available values".
> 
> Cheers
> 
> Jonathan
> 
_______________________________________________
CF-metadata mailing list
[email protected]
http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata

Reply via email to