[CF-metadata] Best practice to include spatial resolution for projected data

From: Seth McGinnis <mcginnis>
Date: Tue, 16 Aug 2016 11:06:35 -0600

You are correct that the scale_factor_* attributes are not the right
place for this information. The grid_mapping variable is a container
for the parameters that describe the projected coordinate system, but
they don't actually depend on the resolution of the grid, just the
Cartesian coordinates attached to the grid.

I am not aware of any established best practices in the world of
regional climate modeling; the spatial resolution generally gets lumped
in with the definition of the domain.

Given that lacuna, I think the best practice would be to extend the
closest existing recommendation according to the principle of least
surprise. In this case, I would follow the NCEI's suggestion of using
global attributes and simply drop the _lat_ component. I.e.:

// global attributes:
    :geospatial_resolution = "12.5 km" ;

(Or, if they're different, change it to geospatial_x_resolution &
geospatial_y_resolution.)

In other words, what Jim said. He just got to the "send" button before
me. :)

Cheers,

--Seth

----
Seth McGinnis
Associate Scientist IV
NARCCAP Data Manager
IMAGe / CISL / NCAR
-----
On 8/16/16 10:33 AM, Mary Jo Brodzik wrote:
> 
> I am producing a gridded data set using CF-1.6 conventions.  In a given
> file, the data are projected onto one of three possible projections:
> Northern or Southern Lambert Azimuthal Equal Area (LAEA), or Cylindrical
> Equal-Area (CEA).  With each data variable in the file, I do include the
> grid_mapping attribute and I am populating my grid_mapping variable with
> the required projection information as listed here:
> 
> http://cfconventions.org/cf-conventions/v1.6.0/cf-conventions.html
> 
> Also in a given file, the spatial resolution will be one of 25, 12.5,
> 6.25 or 3.125 km, and although this is obvious from the x and y
> dimension variable values, I am looking for a place to include the
> spatial resolution for a human who is reading the metadata.
> 
> I assume an attribute in the grid_mapping variable would be the place
> for this.  At first I thought it was one of the "scale_factor*"
> attributes, e.g. "scale_factor_at_projection_origin", but for CEA, this
> would be a misnomer, since the value is the scale at true latitude, not
> the projection origin at all.  Also, the convention says to include
> "either standard_parallel or scale_factor_at_projection_origin", so now
> I'm beginning to think these "scale_factor_*" attributes are not the
> correct place at all for this information.
> 
> I see in the NCEI template file for gridded data
> 
> https://www.nodc.noaa.gov/data/formats/netcdf/v2.0/grid.cdl
> 
> that they recommend global attributes like "geospatial_lat_resolution"
> and "geospatial_lon_resolution", with a string value of "0.1 degree" or
> "100 meters"  which is almost what I am looking for, but of course since
> I have projected data it would not be "lat" or "lon".
> 
> Can anyone recommend a best practice for my projected data case that I
> can follow?
> 
> Thank you,
> Mary Jo Brodzik
> 
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> Mary Jo Brodzik, Senior Associate Scientist, 303-492-8263
> NSIDC/CIRES, Univ. of Colo. at Boulder, 449 UCB, Boulder, CO 80309-0449
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> _______________________________________________
> CF-metadata mailing list
> CF-metadata at cgd.ucar.edu
> http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata

Received on Tue Aug 16 2016 - 11:06:35 BST