[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[netCDF #NEL-892254]: METAFOR, CIM



Hi Tim,

> I'm writing a 'Chapter 9. Earth General Circulation Models' for
> a Univ. Arizona Press planetary-series book, 'Comparative Climatology
> of Terrestrial Planets.' Here's the context:
> http://www.lpi.usra.edu/books/climatology2012/outline/index.shtml

Looks like a great book!

> I have a section on metadata, and I need some clarification between the
> differences of netCDF and CIM (the climate metadata by METAFOR in Europe
> for CMIP5).
> 
> How much functional redundancy is there between netCDF and CIM? I use
> netCDF, but haven't touched CIM yet.

I think the CF (Climate and Forecast) metadata conventions might be
more comparable to CIM than netCDF.  NetCDF is intended to be a
general-purpose data model and access library for scientific data from
various disciplines.  It is used primarily in climate and ocean
sciences, but has also been used successfully in other disciplines
(fusion energy experiment data, neural imagery, molecular dynamics, 
etc.).

The CF metadata conventions were built on top of netCDF to provide
agreements on more model-specific and climate-relevant representations
in a layer above netCDF (or other suitable data access model).

I haven't used CIM and don't know much about it except what I've heard
in presentations about its intended use for CMIP5, to describe models,
experiments, simulations, data, and grids.  But it is for metadata,
not data.  It appears to be more specific to climate models than other
kinds of forecast models, and aimed specifically at interoperability
in model metadata for the climate community.  NetCDF is primarily for
data, but it also includes mechanisms for including metadata in the
same files with data.  However netCDF doesn't try to specify standards
specific to particular disciplines.
 
> Will netCDF and CIM merge at some point? Will it be painless to the users,
> under the hood?

There are no plans for netCDF and CIM to merge.  I'm fairly confident
that there are no plans for CIM and CF to merge either, and that they
occupy different niches in the infrastructure ecosystem for the
climate community.

> Or, does a CMIP5 researcher need both?

A CMIP5 researcher needs CF, especially to provide model output for
CMIP5, which must use CF and netCDF to be included in the PCMDI
archives of CMIP5 experiment results.  I'm not aware of requirements
regarding CIM, except that every climate modeling group participating
in CMIP5 has been requested to provide answers to the questions in the
CMIP5 Metadata Questionnaire.  CIM looks very useful for
interoperability, but I don't know how widely it's used or where it's
required.

> This is just an executive-summary, so I don't need much, but I'm not
> sure about the basic questions as posed above, and I don't want to pass
> on my confusion into this review chapter.

A better source of information about CIM would be Bryan Lawrence,
Director of the BADC, member of the CF Governance Committee, and head
of STFC Center for Environmental Data Archival.

--Russ

Russ Rew                                         UCAR Unidata Program
address@hidden                      http://www.unidata.ucar.edu



Ticket Details
===================
Ticket ID: NEL-892254
Department: Support netCDF
Priority: Normal
Status: Closed