Re: [netcdfgroup] [Esip-documentation] information on related files

Thanks Ed, Dave, and Aleksandar.
These are all possibilities, although I was hoping there would be an attribute 
in some convention, somewhere, that was more specific. Both 'references',
and 'metadata_link' seem to be intended to provide information about the NetCDF file in which they're found, and 'external_variables' is meant to list 'variables which 
are named by attributes in the file'.
What we were hoping to do was point to external data that was collected at the same place and time as the data in the file, but which is not IN the file for various reasons. This might be because it's not time series data, or because it's from another domain 
where NetCDF is unknown.
Making this information machine-readable would require it to be pretty specific, and might even require more than one attribute - one giving  the location, one describing 
the remote content, and maybe one describing how the data sets are related.
Maybe the terminology that THREDDS uses would be a good place to start. I think I've got some THREDDS documentation ... somewhere around here. Also, NcML has the concept of Aggregation (to create "virtual" NetCDF datasets, according to the NcML page 
at unidata), but I'm not sure how far that's gone since I last looked at it.

Thanks again -
Nan


On 9/19/18 11:48 AM, Aleksandar Jelenak via Esip-documentation wrote:

Hello!
There's also this: http://cfconventions.org/Data/cf-conventions/cf-conventions-1.7/cf-conventions.html#external-variables
Apparently this approach is chosen for the CMIP6 data and is lax on the specifics by design. The context where to find files with those referenced external netCDF variables is to be provided in some unspecified manner.
It is possible to include links to related files on THREDDS dataset pages but this is outside of file. The metadata_link keeps the connection in the file but requires that the application code handle ISO XML. 

-Aleksandar
On Sep 19, 2018, at 11:32 AM, Armstrong, Edward M (398G) via Esip-documentation <esip-documentation@xxxxxxxxxxxxxxxxx <mailto:esip-documentation@xxxxxxxxxxxxxxxxx>> wrote: 

Hi:
I would recommend the ACDD attribute “references”:
references
Published or web-based references that describe the data or methods used to produce it. Recommend URIs (such as a URL or DOI) for papers or other references. This attribute is defined in the CF conventions.
In CF global attributes there is also a “comment” attribute where such info could be placed in a generic sense. I have often thought of this use case too and its an area where ACDD or CF could be improved. The metadata_link could point to an ISO 19115 record where multiple citations could point to other datasets cleanly. 
*From:*Esip-documentation on behalf of David Neufeld
*Reply-To:*David Neufeld <david.neufeld@xxxxxxxx <mailto:david.neufeld@xxxxxxxx>> 
*Date:*Wednesday, September 19, 2018 at 8:18 AM
*To:*Nan Galbraith <ngalbraith@xxxxxxxx <mailto:ngalbraith@xxxxxxxx>>
*Subject:*Re: [Esip-documentation] [thredds] information on related files
Hi Nan,
I'd recommend using a 'metadata_link' attribute.
http://wiki.esipfed.org/index.php/Attribute_Convention_for_Data_Discovery_1-3#Additional_Metadata:_metadata_link_attribute
-Dave
On Wed, Sep 19, 2018 at 8:56 AM, Nan Galbraith <ngalbraith@xxxxxxxx <mailto:ngalbraith@xxxxxxxx>> wrote: 
Hi ESIP, NetCDF, and THREDDS folks -

First, I'm sorry for the cross posting. I have a hard time figuring
out which NetCDF list does what; there are 'best practices' threads
on several lists, interspersed with install and development threads.

The OceanSITES project would like to find the most useful way to
provide a link within a NetCDF file to an external dataset.

> Another piece I am missing ... is an optional field, that link to a
> data DOI. Why? Here's an example: We operate an OceanSITES site off
> the Cape Verdes and the sediment trap data is ... not included in an
> netcdf file but processed and stored elsewhere
>https://doi.pangaea.de/10.1594/PANGAEA.863969

Does THREDDS have a standard way to provide links to related or
co-located data? Is this something we discussed as part of ESIP's
ACDD recommendations? I've been searching through my store
of old messages, and haven't come up with anything.

Thanks very much for any advice on this!

- Nan Galbraith






--
*******************************************************
* Nan Galbraith        Information Systems Specialist *
* Upper Ocean Processes Group            Mail Stop 29 *
* Woods Hole Oceanographic Institution                *
* Woods Hole, MA 02543                 (508) 289-2444 *
*******************************************************