contributor_name and contributor_role

[Italo Gabriele Nicolaus Borrelli]

We've noticed that we can't correctly include multiple contributor_name and contributor_role so that it shows up correctly in metadata. We currently implement it so that the datasets.xml has multiple contributor_name and  contributor_role elements, eg:

<contributor_name>Ocean Networks Canada Society</contributor_name>

<contributor_role>custodian</contributor_role>

<contributor_name>Ocean Networks Canada Society</contributor_name>

<contributor_role>owner</contributor_role>

but only one ends up showing up on the page of the dataset and in the ISO 19115.

Another organization suggested to us using comma separated values:

<contributor_name>Ocean Networks Canada Society,Ocean Networks Canada Society</contributor_name>

<contributor_role>custodian,owner</contributor_role>

but testing that that doesn't do the job. In fact in both cases the role is never used, instead favouring contributor.

We do generate our own ISO 19115 xml internally and may consider switching to using the <iso19115File>. I understand it's difficult to create ISO 19115 files for ERDDAP because of the amount of possible metadata that could be included.

Thanks!

Italo

[bobsimons2.00]

contributor_name and contributor_role are attributes from ACDD 1.3

https://wiki.esipfed.org/Attribute_Convention_for_Data_Discovery_1-3#Suggested

For ACDD, netcdf-files, and ERDDAP, there can't be more than 1 instance of a given attribute (it's a key-value set of attributes).

Per ACDD documentation, when there are multiple values, you need to express the values as a comma separated list.

https://wiki.esipfed.org/Attribute_Convention_for_Data_Discovery_1-3#Comma-Separated_Lists

In this case, the multiple values in contributor_name and contributor_role should parallel each other (i.e., role1 is for name1, role2 is for name2, ...).

You say that when using comma separated values, "but testing that that doesn't do the job. In fact in both cases the role is never used, instead favouring contributor." If contributor_role is defined in the dataset's chunk of xml in datasets.xml, it should show up in ERDDAP (e.g., in the dataset description at the bottom of the Data Access Form). I'll be shocked if that is't the case. If you think that isn't working right, please provide the datasets.xml chunk for that dataset. (Please make sure the dataset was reloaded after you last made changes to datasets.xml.)

Once contributor_role is properly showing up in ERDDAP's web pages for the dataset, then again test how it appears in the ISO 19115 generated by ERDDAP. It is possible that it isn't properly handled. (The ISO 19115 world is complicated and NOAA had their own recommendations/requirements/schema which have changed periodically. I don't remember the details of how contributor_role is handled by the ERDDAP code that generates the ISO 19115 document.) If the role information doesn't appear like you think it should in the ISO 19115 that ERDDAP generates, then let us know as it would be a bug for Chris John to investigate and fix.

Best wishes.

[Italo Gabriele Nicolaus Borrelli]

Thank you! Sorry if I wasn't quite clear. It does show up, but I can see now how the ERDDAP works with the ACDD standard.

Regarding "but testing that that doesn't do the job. In fact in both cases the role is never used, instead favouring contributor." it does show up on ERDDAP, but the role in the ISO 19115 XML is always contributor, even when a role is provided that is on the CI_RoleCode code list: https://wiki.esipfed.org/ISO_19115-3_Codelists#CI_RoleCode. But I can completely understand why it is difficult to map those to the ISO 19115 XML because I can see now that the ACDD allows for roles outside of the ISO 19115 roles.

I don't have any further questions about that, I just wanted to clarify. I don't think that it's a bug. I can see it's just a place where the standards do not overlap. It's just one of those things where it's difficult to convert one standard to another without adding ancillary information. I've been contending with that for many different projects. I can also came across the https://wiki.esipfed.org/Attribute_Convention_for_Data_Discovery_Mappings which provides a lot of interesting information and context for the ACDD -> ISO 19115. ISO 19115 is our most thorough metadata format for most things so it's possible this might help us bridge the gap where we have metadata for one and not the other.

Thanks for your help!