Review how the conventions map to the FAIR Principles

[sadielbartholomew]

Many of us will be familiar with the principles for FAIR data and there's a checklist for criteria for data and metadata to be considered FAIR available on this official page. I was thinking it could be useful to explicitly review and record how the CF Conventions promote, and themselves abide by, the FAIR principles, according to those criteria? Perhaps as some text under a dedicated question on the FAQs page, or in some other location that seems appropriate on the website?

It would show we are striving to abide by the FAIR Principles and also would help us to identify areas where we perhaps aren't as FAIR as we could be (these might be some or all of items where I have put '???' in my notes below to indicate I am not sure what to put there).

To illustrate what I am thinking of in that respect, I have made some brief notes to record my interpretation on this (see below). Note that it is quite obvious how the conventions promote those principles in relation to the netCDF data, namely they primarily concern item R1.3, "data meet domain-relevant community standards", and indeed they are listed as an example resource under the dedicated page for that item, which is really great! So that is simple and easy to record.

But I am also thinking about how FAIR the metadata defined by the conventions are (in practice, this means going through the checklist linked above and reading 'metadata' rather than 'data' for '(meta)data', though there are a few items which aren't applicable in this case, namely F2 and A2, so don't need to be considered in his case).

---

My suggestions in the form of quick notes to illustrate

How CF metadata maps to the FAIR criteria (cross-reference with the items here and the further information provided on the pages each item links to) as far as I can tell or know of, with '???' marked on items I am unsure of that would be good to discuss if it is agreed that my proposal to include something like this somewhere on the website is good:

Findable

• F1. URL https://cfconventions.org/index.html provides a canonical home covering all aspects of the CF Conventions.
• F2. N/A, this item applies to non-meta data only.
• F3. N/A, since with netCDF metadata is included in the same file as the data.
• F4. Conventions document at https://cfconventions.org/conventions.html and standard names table at https://cfconventions.org/standard-names.html register the metadata, and the former is searchable via browser word searching or similar, and the latter has a dedicated search utility.

Accessible

• A1. HTTPS is the main protocol for the website (others may work too).
  • A1.1 Yes, HTTPS is open, free, and universally implementable.
  • A1.2 Yes, HTTPS provides authentication.
• A2. N/A, this item applies to non-meta data only.

Interoperable

• I1. ??? We have a (now official) data model for the metadata as covered in detail in this paper and see item (I2) below regarding "commonly used controlled vocabularies, ontologies, thesauri".
• I2. This question applies only to the standard names for this context and I think we are in the process of defining more concretely the controlled vocabulary and grammar that underlies the current standard names, where there is some notes to summarise in a loose way how the names are constructed and work related to investigating the grammar e.g. under the 'CF Standard Names' FAQs.
• I3. '???' Maybe the fact that standard names are often referential to other standard names, e.g. 'air_temperature_anomaly' and 'difference_between_sea_surface_temperature_and_air_temperature' both refer to another variable that is itself a name, namely 'air_temperature', and that they are all tied to canonical units, would count under this?

Reusable

• R1:
  • R1.1. '???' (It seems like we don't use any license?)
  • R1.2: '???'
    • The proposal and acceptance process is open and persistently accessible on GitHub (now, and Trac previously) and covers such details as the origin story via the opening comment from the user who submits it as a proposal, along with the subsequent discussion.
    • The update history of the conventions overall are covered by the versioning of the document and the standard names table, along with context provided on the official website.
    • Relation to other metadata are described in detail, such as COARDS
      (covered in this dedicated section) and UDUNITS (covered within this section).
  • R1.3 '???' (What community standard do aspects of the CF Conventions abide by, if any, say in terms of formatting of aspects? Is this item even applicable in our context?)

[larsbarring]

I think that in relation to the Find and Accessible components CF is making a big leap forward by having DOIs 👍