[netCDFJava #RIZ-451320]: why can't this be read as a grid

Hi Don:

We will have a look to see if we can safely disambiguate the conflicting info. 
Can you contact CDO and let them know they have a bug?

thanks
John

> Dear Lansing, John and Ethan-
> 
> Thanks for your responses.  I had read the CF conventions about the
> standard names and understand that grid_latitude/grid_longitude are the
> standard names for a rotated pole projection.  That's why I indicated to
> Lansing that this was the problem.
> 
> However, as Ethan points out, the units are wrong for a rotated_pole and
> there is no grid_mapping_name variable associated with the tas variable.
> So, tas is not defined by a rotated pole coordinate system.
> 
> The original grid is a gaussian dataset and the original standard names
> were latitude and longitude, so CDO does change the standard names.
> 
> From my perspecitve, though, the latitude and longitude variables meet
> the requirements for being standard latitude and longitude variables in
> CF by the units and axis attributes:
> 
> http://cfconventions.org/1.6.html#latitude-coordinate
> 
> (correct units and axis attributes) much more than for the rotated pole
> case (wrong units, no grid_mapping_name).  So, I would still content
> that this is a bug.
> 
> Don
> 
> On 5/16/14 4:43 PM, Unidata netCDF Java Support wrote:
> > Hi Don,
> >
> > The use of "grid_latitude" and "grid_longitude" standard names indicates a 
> > rotated pole grid mapping (defined in Appendix F: "Grid Mappings" section 
> > [1] of the CF spec). But lat and lon have the wrong units for a rotated 
> > pole grid (they should be "degrees") and a rotated pole grid mapping is not 
> > defined. (Here's the definition for "grid_latitude" [2] from the CF 
> > standard name table [3].)
> >
> > Was the original dataset on a rotated pole grid?
> >
> > Ethan
> >
> > [1] http://cfconventions.org/1.6.html#appendix-grid-mappings
> >
> > [2] "Latitude is positive northward; its units of degree_north (or 
> > equivalent) indicate this explicitly. In a latitude-longitude system 
> > defined with respect to a rotated North Pole, the standard name of 
> > grid_latitude should be used instead of latitude. Grid latitude is positive 
> > in the grid-northward direction, but its units should be plain degree."
> >
> > [3] https://badc.nerc.ac.uk/help/formats/netcdf/cf-standard-name-table.html
> >
> > Don Murray wrote:
> >> It doesn't like the standard_name for latitude and longitude (using ncml
> >> to remove them makes the grid readable).  I don't have the ability to
> >> remove them in the workflow, and this seems like a bug.
> >>
> >> Don
> >>
> >> On 5/16/14 3:35 PM, Unidata netCDF Java Support wrote:
> >>> There's confusion somewhere, ToolsUI is looking for a projection instead
> >>> of recognizing the lat lon coords.
> >>>
> >>> -Lansing
> >>>
> >>> On 5/16/2014 12:55 PM, Don Murray (NOAA Affiliate) wrote:
> >>>> New Ticket: why can't this be read as a grid
> >>>>
> >>>> Hi-
> >>>>
> >>>> I have a netCDF file that I created using the Climate Data Operators
> >>>> (CDO) package by subsetting a larger file.  The original file can be
> >>>> read into the IDV/ToolsUI, but the subset cannot (is not recognized as a
> >>>> grid).  I don't see anything obviously wrong with the file, except that
> >>>> maybe the CDM is getting confused by the use of x and y?
> >>>>
> >>>> Why doesn't the CDM recognize this as a grid?
> >>>>
> >>>> Don
> >
> >
> > Ticket Details
> > ===================
> > Ticket ID: RIZ-451320
> > Department: Support netCDF Java
> > Priority: Normal
> > Status: Closed
> >
> 
> --
> Don Murray
> NOAA/ESRL/PSD and CIRES
> 303-497-3596
> http://www.esrl.noaa.gov/psd/people/don.murray/
> 
> 


Ticket Details
===================
Ticket ID: RIZ-451320
Department: Support netCDF Java
Priority: Normal
Status: Open