Opened 8 years ago

Last modified 8 years ago

#98 new defect

Clarify constraints on ancillary variable use

Reported by: markh Owned by: cf-conventions@…
Priority: medium Milestone:
Component: cf-conventions Version:
Keywords: Cc:

Description

1. Title

Clarify constraints on ancillary variable use

2. Moderator

required

3. Requirement

An ancillary variable should not be able to be referenced by a data variable in the case where the ancillary variable is defined with respect to NetCDF file dimensions which the data variable with the referencing ancillary_variables attribute is not defined with respect to.

For example, if

    data(x,y,z)

    anc(x,y,t)

then data may not have an attribute of

    data:ancillary_variables = "anc"

The dimensions anc is defined with respect to are not the set or a subset of the dimensions data is defined with respect to so it cannot be an ancillary variable of data.

4. Initial Statement of Technical Proposal

Ancillary Variables

additional text:

For a variable in a NetCDF file to be referenced as an ancillary_variable by a data variable, this ancillary variable must be defined with respect to the set, or a subset of the dimensions defining the referencing data variable.

5. Benefits

The limitations on use of Ancillary Variables will be made clear and explicit and may be checked for compliance by software.

6. Status Quo

None Available. Currently compliant files may be created but not be understood by data users.

Change History (2)

comment:1 Changed 8 years ago by jonathan

Dear Mark

No-one has commented on this here, but was some discussion about it on the email list, as you know. I'm concerned that this is not really a defect. I would say it that you are making a proposal to add a restriction which isn't currently present in the convention. I'm not necessarily arguing against it, but I feel that it needs support to be expressed positively in favour of it, for instance by some of those who engaged in the email discussion earlier.

Cheers

Jonathan

comment:2 Changed 8 years ago by markh

Hello Jonathan

I agree. There were some useful comments on the mailing list but a number of these were not in support of this proposal. It appears more flexibility is desired from ancillary data than I am advocating here

I think that, unless there are further comments on this ticket in the near future, it should be closed and no further action taken.

mark

Note: See TracTickets for help on using tickets.