Using ParameterGroup to annotate ancillary data (such as quality flags) of a variable

We've been considering the use of ParameterGroup to annotate quality flag variables which are directly associated to, for example, observational variables. In the CF conventions this is referred to as ancillary data.

The spec also uses an example with mean values and an associated stddev. So this seems to indicate that the intended use of ParameterGroups can be the same as the ancillary data that the CF conventions describe.

What's missing from the current specification is to indicate which variable of a ParameterGroup is the primary variable and which are the supporting ('ancillary') variables. What are your thoughts on this? Would it help to define this somehow in a ParameterGroup? For example by defining that the first member of a group is the primary variable. Another option would be to do this similarly as the CF conventions and add an attribute ancillaryVariables: list[str] to Parameter.

I've commented on an existing and related issue https://github.com/opengeospatial/ogcapi-environmental-data-retrieval/issues/634#issuecomment-3442567024 in the EDR spec which talks about how ParameterGroup can be used in EDR.

Start by reviewing the current ParameterGroup definition, then compare the CF conventions ancillary data section and the linked OGC API EDR specification example. Read the related EDR issue discussion for context. Done means reaching agreement on how a primary variable and supporting variables should be identified and recording that decision in the CoverageJSON specification.