[Esip-documentation] cdm_data_type defintion

[John Graybeal]

Hi Ed,

I'm going to give you the off-the-cuff summary. I can look up the 3 or 4 detailed discussions if you need something more referenced, but it won't add much. 

In short, this was not resolved in a deterministic way.

Background:

As originally defined in 1.1, ACDD pointed to a particular implementation list (the one you cite). But Unidata and its libraries moved on from that specific code list; as I recall, several different lists were proposed as replacements, and there has not been an established and community-agreed URL which is guaranteed to provide authoritative information. (Plus, some participants did not want to reference an external URL as an authority for this purpose, due to concern the content could change and cause havoc.)

Meanwhile, CF created its own discrete sampling geometries formulation (http://cfconventions.org/Data/cf-conventions/cf-conventions-1.6/build/cf-conventions.html#discrete-sampling-geometries) using the featureType attribute. This formulation was fairly robust -- but could only be used if the variables were in fact organized as one of the defined DSGs. (In other words, featureType is not a descriptive attribute, but a forcing one.). This attribute seems to be a more commonly used attribute in new data sets (though likely not always as part of a DSG), to the point that most of the committee was not advocating for the use of cdm_data_type any more (note that it is only Suggested now).

I think if you look at general practice for use of cdm_data_type, you will find some users strictly followed the 1.1 specification reference, others used lower case, and others used lists from other sources, either within Unidata or outside of it. So that doesn't help much. In the 1.3 specification, if you still want to use cdm_data_type, the specification is pointedly vague about exactly what list to use. 

My personal recommendation (others may suggest otherwise): 

If I search on "Unidata's Common Data Model Scientific Data types" (the phrase in the current standard), the first hit is a Unidata page (http://www.unidata.ucar.edu/software/thredds/current/netcdf-java/CDM/) that says "Feature Type definitions, APIU, and encodings are still being developed, so applications using these must be able to evolve along with the APIs." This page points to a .doc that has a summary list that includes 
Point
Station
Profile
Station Profile
Trajectory
Section
Grid
Moving Grid
Swath
RadialSweep
Though no controlled vocabulary is specified in that document, I would (naively) probably use whichever of those are consistent with the THREDDS implementation I was dealing with. Others might point to the latest Java or other library code; I'll let them make that argument.

John

---------------
John Graybeal
Project Lead
Marine Metadata Interoperability Project: http://marinemetadata.org


On Feb 19, 2015, at 09:51, Armstrong, Edward M (398G) via Esip-documentation <esip-documentation at lists.esipfed.org> wrote:

> 
> Hi folks 
> 
> I’ve come across a request for the proper usage of the ACDD attribute cdm_data_type.  This has been retained in  ver 1.3.   But what are the examples of its implementation ? According to:
> 
> http://www.unidata.ucar.edu/software/thredds/current/tds/catalog/InvCatalogSpec.html#dataType
> 
> attribute examples could “Grid” ,“Swath”, “Point" etc.  But some popular implementation embedded in existing examples use a lowercase syntax such as “swath”
> 
> Which is the proper implementation, upper vs lowercase ?  Or perhaps this is still nebulous.  I did find this note from Bob Simmons in our recent ACCD v1.3 discussion on a google doc spreadsheet (but I don’t remember the consensus  on it):
> 
> cdm_data_type should not be tied to
> 
> http://www.unidata.ucar.edu/software/thredds/current/tds/catalog/InvCatalogSpec.html#dataType
> which is out-of-date and obsolete
> 
> Any insight appreciated.
> 
> -ed
> 
> Ed Armstrong
> JPL Physical Oceanography DAAC
> 818 519-7607
> 
> 
> 
> _______________________________________________
> Esip-documentation mailing list
> Esip-documentation at lists.esipfed.org
> http://lists.deltaforce.net/mailman/listinfo/esip-documentation

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.deltaforce.net/pipermail/esip-documentation/attachments/20150219/0d1674df/attachment-0001.html>