[netCDF #GGI-666903]: question about standard grids in CF convention

John,

> Yes, it explains the situation.  Do you know of anyone using alternative 
> approaches - what do modelers do?  I have a time series global grid 6km, 
> about 10% of the grid with values. So every time step (i.e., netCDF file) 
> repeats every lat-lon value for each grid point.  We are using custom clients 
> that ignore the lat-lons and use grid/projection standard parameterization 
> for proj-4 transformations.

Yes, the modelers for AR5 under the PCMDI project at LLNL are using GRIDSPEC, 
which has been
proposed as a CF Conventions extension.  The approach taken there is to store 
shared grid
specifications in a separate file.  The proposed enhancement is here:

  https://cf-pcmdi.llnl.gov/trac/ticket/63

There's also the approach of the THREDDS Data Server (TDS), which permits 
datasets with well-specified
standard grids that are not CF compliant, as John Caron writes:

   The CDM/TDS uses netcdf/CF as a way to transfer binary information
   in WCS and other web services. Adding two-dim lat/lon fields can
   triple (worst case) the size of the file. For that reason CDM/TDS
   allows the user to specify if they want 2D lat/lon fields or
   not. This makes the files not strictly CF compliant, but we leave it
   to the client to decide what tradeoff they want.
 
   The point is that web service binary encoding is a use case likely
   not originally envisioned by the CF committee.

--Russ

> -----Original Message-----
> From: Unidata netCDF Support [mailto:address@hidden]
> Sent: Tuesday, April 05, 2011 11:36 AM
> To: Moses, John F. (GSFC-5860)
> Cc: address@hidden
> Subject: [netCDF #GGI-666903]: question about standard grids in CF convention
> 
> Hi John,
> 
> > We've been developing a netCDF3 product and trying to comply with CF
> > conventions. We want the product on one of the standard grids, e.g.,
> > global equal area cylindrical. We would like to use a netCDF3 standard
> > grid so that we don't need to include the latitude/longitude pair for each
> > gridpoint - but were told that the CF convention requires a lat-lon value
> > for each gridpoint regardless of the grid projection selected/used - i.e.,
> > that the grid and projection parameterization in the CF convention is not
> > complete enough for clients (e.g., IDV) to compute a gridpoint lat-lon.
> >
> > Just want to confirm - doesn't seem to make sense to me - why have
> > standard grids if you need to include lat-lon values for every gridpoint?
> >
> > I'd welcome any clarification you can offer.
> 
> There's a discussion of exactly this issue on the CF conventions
> mailing list that clarifies why lat/lon pairs are still required for
> CF compliance.  The mailng list thread of 12 postings starts here:
> 
> http://mailman.cgd.ucar.edu/pipermail/cf-metadata/2009/006550.html
> 
> Steve Hankin's response is especially comprehensive and persuasive.
> For more discussion of the issue, you can see the CF Trac tickets #9
> and #18
> 
> I hope that helps.
> 
> --Russ
> 
> Russ Rew                                         UCAR Unidata Program
> address@hidden                      http://www.unidata.ucar.edu
> 
> 
> 
> Ticket Details
> ===================
> Ticket ID: GGI-666903
> Department: Support netCDF
> Priority: Normal
> Status: Closed
> 
> 

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



Ticket Details
===================
Ticket ID: GGI-666903
Department: Support netCDF
Priority: Normal
Status: Closed