Hi all,
We had a great meeting discussing ACDD’s future with all of the participants of The Future of ACDD: Charting the Course Ahead session as part of the January 2026 Earth Science Information Partners meeting.
A significant number of attendees supported updating the ACDD. However, attendees noted that any update needs to be done carefully, following an explicit process with identified decisionmakers and participants. We need to be clear about our motivations for making updates and what benefits data providers and users would experience from updates to the ACDD.
As a result of these conversations, at least two questions should be addressed:
Who wants to be involved and in what capacity? (I created a separate issue to address this question, so please feel free to post there.)
What is our governance framework for this work on the ACDD? (I created a separate issue to address this question, so please feel free to post there.)
Third, what important questions did I miss?
Session Summary
Finally, I’ve summarized the session, particularly for those who were unable to attend the session. Please feel free to comment, ask questions, or provide your own insights from the session.
The session included a few presentations and then time for discussion and a breakout session, facilitated by @MathewBiddle.
@akrelling and I, who work with Integrated Ocean Observing System (IOOS) data at the National Centers for Environmental Information at NOAA, discussed the following new global attribute recommendations as a solution for dealing with spelling variations of institution and creator/publisher names:
- for individuals: creator_id, publisher_id
- for institutions: creator_institution_id, publisher_institution_id
(See discussion at ioos/ioos-metadata#52. Thanks @relphj for identifying these issues, and @MathewBiddle and @mwengren for solutions.)
One participant suggested the inclusion of additional attributes like creator_id_vocabulary to record the type of PID used, but noted that as more requirements are added, the chance of compliance may decline. Alternatively, the ACDD could stipulate the use of a specific vocabulary, such as ORCID for individuals. Another participant proposed an attribute such as creator_uri instead of creator_id, where the value is a URL, such as the ORCID URL, which would then identify the vocabulary via the URL.
@ethanrd with NSF Unidata / UCAR provided an overview of the early history of ACDD, which he built as an extension of CF for NetCDF files, targeting harvesting into THREDDS catalogs and digital libraries. He originally proposed its inclusion in CF, but ultimately published it as a separate convention. ACDD 1.0 was published in 2005 by Unidata.
@ethanrd then presented slides about the last ACDD update, on @graybeal's behalf. Seventeen people collaborated for 23 months to publish the ACDD 1.3 update. @graybeal provided some recommendations based on his experience, including
- Decide up front what you will allow (to change), e.g. attributes, definitions;
- Determine a process by which a group can approve changes;
- Have good, agreed-upon, persistent tools for documentation and discussion;
- Expect to spend time on logistics, especially for the chairs and support team; and
- Know when to stop accepting changes so that a new version can be released.
@ethanrd also noted that the CF community has strong standards and a clear process for approving changes because it is hard to get a lot of people to agree. Limiting choices is hard to do, but important for making measurable progress. Finally, the handling of logistics will be easier if more than one person is contributing to the logistics effort.
Participants then broke into small groups to facilitate discussion. Participants identified a variety of benefits to and reasons for updating the ACDD, including
- Improving data quality for discovery, archiving, and storage;
- Support for cloud and AI-ready data;
- Improved documentation for PIDs; and
- Improved vocabulary support.
Missing from the current version of ACDD are also elements that map easily to catalog-level standards (i.e., ISO 19115). Another participant noted that ACDD may not be granular enough in its conventions for some users, and this might be a reason for making updates. Updates could be made to include guidelines on how to extend and customize the convention for one’s own purposes.
Concerns about updating included
- Knowing when to stop updating,
- The addition of metadata recommendations and requirements that might overly complicate the convention,
- Potential negative impacts to existing data (such as through deprecation),
- Potential overlap with the CF convention, and
- The resources and coordination required to support the development and maintenance of the standard.
Hi all,
We had a great meeting discussing ACDD’s future with all of the participants of The Future of ACDD: Charting the Course Ahead session as part of the January 2026 Earth Science Information Partners meeting.
A significant number of attendees supported updating the ACDD. However, attendees noted that any update needs to be done carefully, following an explicit process with identified decisionmakers and participants. We need to be clear about our motivations for making updates and what benefits data providers and users would experience from updates to the ACDD.
As a result of these conversations, at least two questions should be addressed:
Who wants to be involved and in what capacity? (I created a separate issue to address this question, so please feel free to post there.)
What is our governance framework for this work on the ACDD? (I created a separate issue to address this question, so please feel free to post there.)
Third, what important questions did I miss?
Session Summary
Finally, I’ve summarized the session, particularly for those who were unable to attend the session. Please feel free to comment, ask questions, or provide your own insights from the session.
The session included a few presentations and then time for discussion and a breakout session, facilitated by @MathewBiddle.
@akrelling and I, who work with Integrated Ocean Observing System (IOOS) data at the National Centers for Environmental Information at NOAA, discussed the following new global attribute recommendations as a solution for dealing with spelling variations of institution and creator/publisher names:
(See discussion at ioos/ioos-metadata#52. Thanks @relphj for identifying these issues, and @MathewBiddle and @mwengren for solutions.)
One participant suggested the inclusion of additional attributes like creator_id_vocabulary to record the type of PID used, but noted that as more requirements are added, the chance of compliance may decline. Alternatively, the ACDD could stipulate the use of a specific vocabulary, such as ORCID for individuals. Another participant proposed an attribute such as creator_uri instead of creator_id, where the value is a URL, such as the ORCID URL, which would then identify the vocabulary via the URL.
@ethanrd with NSF Unidata / UCAR provided an overview of the early history of ACDD, which he built as an extension of CF for NetCDF files, targeting harvesting into THREDDS catalogs and digital libraries. He originally proposed its inclusion in CF, but ultimately published it as a separate convention. ACDD 1.0 was published in 2005 by Unidata.
@ethanrd then presented slides about the last ACDD update, on @graybeal's behalf. Seventeen people collaborated for 23 months to publish the ACDD 1.3 update. @graybeal provided some recommendations based on his experience, including
@ethanrd also noted that the CF community has strong standards and a clear process for approving changes because it is hard to get a lot of people to agree. Limiting choices is hard to do, but important for making measurable progress. Finally, the handling of logistics will be easier if more than one person is contributing to the logistics effort.
Participants then broke into small groups to facilitate discussion. Participants identified a variety of benefits to and reasons for updating the ACDD, including
Missing from the current version of ACDD are also elements that map easily to catalog-level standards (i.e., ISO 19115). Another participant noted that ACDD may not be granular enough in its conventions for some users, and this might be a reason for making updates. Updates could be made to include guidelines on how to extend and customize the convention for one’s own purposes.
Concerns about updating included