Hi Rich,
As the following may convey, my familiarity with CF featureTypes is
all of ~10 minutes old :) From what I gather, featureTypes give a
sense of how to treat collections of related variables. One could
imagine a logical extension to featureType that allowed specifying an
"ensemble" as a collection of groups to be treated in a certain way.
The case for some sort of "ensemble" metadata is to provide hints.
Aggregating raw CMIP5 files into groups in a single file is all well
and good for a single user like me, because I know my own conventions.
What I envision a CF convention providing though is more ambitious.
It would tell data providers how to indicate to a non-specialist user
how/if group in hierarchical group file are inter-related, e.g.,
different realizations of a single-model, or simulations of different
models with the same initial and time-varying boundary conditions, ...
The ensemble type then determines how analysis should proceed.
Best,
cz
Le 16/09/2013 10:52, Signell, Richard a ?crit :
> Charlie & Co,
>
> Also, regardless of whether these hierarchical structures are stored
> in NetCDF4 or flattened NetCDF3, we get a big boost in
> interoperability when we write datasets with known featureTypes
> (profile, time series collection, swath, etc), because then workflows
> that have performed a
> catalog search and returned dataset endpoints knows what to do. If
> the dataset endpoints contain ad hoc heirarchies it will be a lot more
> difficult.
>
> So if we create hierarchical datasets, I hope we create known
> featureTypes to accompany them (like gridEnsembleStructure).
>
> Thanks,
> Rich
--
Charlie Zender, Earth System Sci. & Computer Sci.
University of California, Irvine 949-891-2429 )'(
Received on Mon Sep 16 2013 - 23:18:13 BST
