clarify the dimensions of ancillary variables

[jonathan]

The purpose of ancillary variables (Section 3.4 of the convention) is to provide metadata about individual values in data variables. Given this purpose, I believe that an ancillary data variable should not have any dimensions which its data variable does not have. To clarify this, I propose that the following should be appended to the first paragraph of Section 3.4: "The dimensions of an ancillary variable should be the same as or a subset of the dimensions of the data variable with which it is associated by the ancillary_variables attribute, but the order of the dimensions is not restricted." This text corresponds to a similar sentence in Section 7.2 about cell measures variables.

I'm proposing this as a defect ticket because this is my understanding of what the convention intends. Please comment if you don't think so. Thanks.

Jonathan

[taylor13]

Hi Jonathan,

I support this clarification of the use of ancillary_variables. I think this limitation on dimensions should be imposed.

Thanks for proposing this.

Karl