Hello Jonathan

This topic is not related to CF#108 (which I need to get back to and consider 
in detail).

This is an explicit piece of metadata which is provided by some systems and 
models and is applied to the ensemble collection, rather than statistics 
derived from that collection.

In most cases there will be no CellMethod but there will be a realization 
dimension and coordinate.

This piece of metadata could be considered to be:
  'The expected length of the realization coordinate and dimension, if the 
whole ensemble collection is present.'

This is preserved during subset operations, so that I may be looking one 
ensemble member, with a realization coordinate value of 4, but I know that this 
is from an ensemble of 17 members.

This is useful metadata which is available from some models and we would like 
to preserve it in a standardised way in our CF NetCDF data sets.

I hope this clarifies the question

all the best
mark

________________________________________
From: CF-metadata [[email protected]] on behalf of Jonathan 
Gregory [[email protected]]
Sent: 06 October 2014 18:15
To: [email protected]
Subject: [CF-metadata]  realization | x of n

Dear Mark

Is this related to ticket 108? In that ticket, you and I have very similar
proposals, while Karl has made a more general proposal. In our simple
proposal, the number of realizations would the size of the collapsed axis,
and you could have a cell_methods entry for the collapsed axis such as
  realization: mean (dimension: members)
where "realization" is allowed because it's a standard_name, and "members"
is the name of a dimension which contains the original size of the ensemble
before collapse. This would work in my version of the proposal because it
does not require any coordinate vars to exist with this dimension (though they
could if you want to supply them too).

Best wishes

Jonathan

----- Forwarded message from "Hedley, Mark" <[email protected]> -----

> From: "Hedley, Mark" <[email protected]>
> To: "[email protected]" <[email protected]>
> Date: Fri, 3 Oct 2014 10:52:00 +0000
> Subject: [CF-metadata] realization | x of n
>
> Hello CF
>
> The standard name
>   realization
> is available for use:
>   Realization is used to label a dimension that can be thought of asa 
> statistical sample, e.g., labelling members of a model ensemble.
>
> It is common practice in forecasting and some forecasting formats to 
> explicitly state the number of members an ensemble contained.
>
> This would be generally be expected to be a single value across an ensemble 
> collection.  As such I am not sure that a standard name and a coordinate 
> would be the most useful way to encode this piece of information.
>
> Thus, we could look to provide a new data variable attribute, something like
>   number_of_realizations
> to store this metadata.
>
> However it is not impossible for this to vary across some coherent data set 
> so perhaps a standard_name of
>   number_of_realizations
> would be a sensible approach, even if in most cases this would be a scalar 
> coordinate
>
> Please may I ask for some opinions on this matter?
>
> thank you
> mark

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

Reply via email to