Re: [CF-metadata] Representing statistics in CF

[Jon Blower]

Hi David,   Many thanks for this, that makes things clearer to me. So do I understand correctly that:   1.       There would be a file containing the ensemble statistics (e.g. mean and variance). The variables in this file would have a cell_methods of “realization: mean” etc. 2.       There would also be a file for each ensemble realization. The variables in this file do not need to have a “realization” dimension. 3.       The standard names of the variables in each file (realizations and statistics) can be the same (e.g. “sea_water_potential_temperature”). No standard name modifier is necessary. 4.       Information about how all these files relate to each other is out of scope for CF.   Best wishes, Jon         From: David Hassell <david.hass...@ncas.ac.uk> Date: Friday, 28 April 2017 16:39 To: Jon Blower <sgs02...@reading.ac.uk> Cc: Jonathan Gregory <j.m.greg...@reading.ac.uk>, "cf-metadata@cgd.ucar.edu" <cf-metadata@cgd.ucar.edu> Subject: Re: [CF-metadata] Representing statistics in CF   Hi Jon, I think that Jonathan is suggesting the mechanism described in 7.3.4 http://cfconventions.org/cf-conventions/v1.6.0/cf-conventions.html#cell-methods-no-coordinates This mechanism would be saying "the data are means over realizations (where 'realization' is well defined by the standard name table), but I'm not going to give you any information about the nature of those realizations". This interpretation assumes that there is no netCDF dimension called "realization" (nor a scalar coordinate variable called that, either) in the file, which is what you I have, I think. No connection can be made to the other individual files that you mention. Hope that helps, All the best, David   On 28 April 2017 at 16:09, Jon Blower <j.d.blo...@reading.ac.uk> wrote: Dear Jonathan, Many thanks for your reply. Looks like my case 2 (statistics over time) is pretty straightforward. But I’m afraid I don’t quite understand your suggestion for case 1 (ensemble statistics): ➢ You can do this with cell_methods in exactly the same way, "e: mean" and "e: standard_deviation" where e is the ensemble dimension. If you don't have an ensemble dimension, you could put "realization: mean" and "realization: standard_deviation", since "realization" is a standard_name for ensemble member designators. In our case we don’t have an ensemble dimension (the individual realizations will be in different files). So I guess we are looking at your second suggestion. Could you help me understand the usage of “realization” please? From the standard name table I can see that “realization” is intended to label a dimension. Given that each realization will be in a different file, does this mean that each file will have a “realization” dimension, whose corresponding coordinate variable contains a single value (the realization label)? In this case, do the variables containing the ensemble mean and spread go in new files? I assume so, but then how do we associate the “realization” in the cell_methods of the statistical variables with the “realization” dimension of the individual files? (And, for that matter, is there a way of expressing that the individual realization files all belong to the same “dataset”?) Best wishes, Jon _______________________________________________ CF-metadata mailing list CF-metadata@cgd.ucar.edu http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata -- David Hassell National Centre for Atmospheric Science Department of Meteorology, University of Reading, Earley Gate, PO Box 243, Reading RG6 6BB Tel: +44 118 378 5613 http://www.met.reading.ac.uk/

_______________________________________________
CF-metadata mailing list
CF-metadata@cgd.ucar.edu
http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata