Call for hackathons for the 2025 CF workshop

[efisher008]

Hello all,

This is an initial call for hackathons for the 2025 CF meeting. This will be taking place from Monday 22nd - Thursday 25th September 2025, and the format this year will be entirely virtual. Please see the GitHub issue above for details of specific timeslots. The meeting agenda will be published on the CF website in due course.

The hackathons will follow the same format as previous meetings, i.e. parallel breakout groups will each discuss a specific issue with the aim of progressing an enhancement to CF, where "progressing" can mean anything from simply increasing understanding of a topic, to writing a pull request to an already agreed issue.

The first two days of the meeting (22nd - 23rd) will have around 8 hackathons run in parallel, depending on the response from the user community - the idea being that it's then likely that there's something of interest for everyone in each session. Additionally, there will be several roadmap-focused hackathon sessions in the last two days (24th - 25th), which will be reserved for collaborative discussions to make progress on sections of the CF roadmap.

The only pre-requisite for suggesting a hackathon is a commitment to provide (by the time of the meeting) a short paragraph introducing the topic, and to lead (or at least initiate) the discussion during the breakout session. The hackathons are meant to be a mainly group-based activity, so no presentation is necessary, unless you want to show a few slides by way of an introduction.

Any subject can be considered for a hackathon, but we have found in the past that ongoing GitHub issues can provide good topics. Below I have provided a list of open CF enhancement issues (with the original author tagged) in case that provides anyone with inspiration. Where these have resulted in another issue being opened, they are marked with a *.

If you would like to suggest a hackathon topic, please reply to this discussion.

Many thanks,
Ellie, for the workshop organising committee

Discussions

• Composite/array values in string valued global attributes @benjwadams started Jul 17, 2024
• Strengthen "Groups" conventions @pvanlaake started Jul 8, 2024
  *Linked to Conventions for placing CF constructs in a group hierarchy cf-conventions#533
• Allow user-defined types as a data type @pvanlaake started Jun 30, 2024
• Concluding discussions @JonathanGregory started May 10, 2024
• Creating a blacklisting certain characters from variable and attribute names @larsbarring started May 31, 2024
• Can CF offer better support for the representation of climate anomalies? @TomLav started Apr 12, 2024
  *Linked to Conventions for anomalies cf-conventions#582
• CF Coordinate Subsampling (section 8.3) and affine transformations @ethanrd started Apr 10, 2025
• Conformance requirement on actual_range @pvanlaake started Mar 24, 2025
• latitude_longitude grid mapping @taylor13 started Jan 29, 2025
• UTC & leap seconds (again! again!!!) @pvanlaake started Dec 22, 2024
• Rethinking the infamous standard_names @amdasilva started Jan 8, 2025
• Proposal: Modify Cell Methods Guidelines to Accomodate Empirical Sea Ice Datasets @Bruce-T-T-Calvert started Jun 4, 2025

cf-convention/discuss/issues

• Introduce global attributes "license" and "copyright"? opened Mar 11, 2024 by @rjhogan
• cell_methods for covariance opened Jan 2, 2024 by @achho
• Localized metadata in NetCDF files opened Jun 1, 2023 by @turnbullerin
  *Linked to Support localization of natural language attributes and variables cf-conventions#528
• Multiple Feature Types in one CF file opened May 3, 2023 by @gaochen-larc
• Use of "where" in cell_methods opened Aug 16, 2022 by @taylor13
  *Linked to Clarification of weighting in cell_methods cf-conventions#447
• "area: mean where type1 over type2" question opened Mar 29, 2022 by @taylor13
• String/Char labels (section 6) mismatch between Conformance and CF Document opened Jun 1, 2021 by @DocOtak

cf-convention/cf-conventions/issues

• Conventions for placing CF constructs in a group hierarchy opened Jul 29, 2024 by @pvanlaake
• Support localization of natural language attributes and variables opened Jul 8, 2024 by @turnbullerin
• Clarification of weighting in cell_methods opened Aug 22, 2023 by @JonathanGregory
• Convention for HEALPix grid parameters opened Mar 18, 2023 by @uweschulzweida
• Clarification of cell_methods opened Nov 23, 2022 by @taylor13
• Color blindness accessibility of document figures & tables opened Sep 27, 2022 by @sadielbartholomew
• Location / Identification for grid specs outside of a file opened Mar 10, 2022 by @ChrisBarker-NOAA
• Provision of netCDF files for the netCDF header examples shown in the Conventions? opened Jan 10, 2022 by @atmodatcode
  *Linked to https://github.com/orgs/cf-convention/discussions/355
• Support Swath Data in CF opened May 14, 2020 by @ajelenak
• Proposal: Example for Independent Latitude, Longitude, Non-spatiotemporal Variable, and Time Axes opened Mar 2, 2020 by @GeyerB
• Merge CONTRIBUTING and rules opened Jul 26, 2022 by @erget
  *Linked to Add "GDAL" to list of CF-aware software cf-convention.github.io#545
• Conventions for anomalies opened Dec 28, 2024 by @JonathanGregory
• Use BCP 14 or inspiration of that in Conventions document opened Sep 20, 2024 by @erget
• Where to put figure source files? opened Oct 23, 2024 by @DocOtak
• Provide a better reference for what units are supported opened Nov 5, 2024 by @ChrisBarker-NOAA
• Feedback Request: Improving Document Versioning and Colophon Details opened Dec 12, 2024 by @cofinoa
• Attributes in separate TSV file opened Dec 17, 2024 by @lhmarsden
• Centralize author metadata for consistency across document and zenodo.json opened Dec 17, 2024 by @cofinoa
• Proposal to Revise Section 4.4 to Support String-Based Datetime Representations Using ISO 8601 Templates opened Dec 22, 2024 by @cofinoa
• units_metadata contradiction between temperature and time variables opened May 8, 2025 by @benjwadams

[sadielbartholomew]

Hi Ellie, all,

Sorry for a somewhat late response but I am happy to lead a hackathon and perhaps the issue you suggest that I opened, namely 'Color blindness accessibility of document figures & tables' (cf-convention/cf-conventions#404), would be a good one indeed - it is certainly something I think we could make real headway with it in only an hour or two.

My tentative plan is that we could take the images for figures that seem particularly inaccessible for those with colour blindness and update the colours to a form that passes the test that it appears distinguishable after colourblind filters are applied, which could be as simple as applying some colour filtering using (free) photoshop-like software, but if we can get access to the original source for the image (e.g. source for UML for the the CF Data Model diagrams) we could update them at source as a preferred/reproducible way (I anticipate accessing the source for the schematics will not always be possible). Then switch those in for the old version using a PR.

Please let me know if you think this would be a topic/plan that could work and that there would be interest in.

[sebvi]

Hi all,
I would like to propose organizing a hackathon session focused on "Mapping CF Metadata and WMO GRIB2 Metadata" to support and streamline the conversion between these two formats. We plan to initiate this effort internally in September, beginning with a comprehensive mapping of all variables in our upcoming ERA6 reanalysis dataset. Ultimately, the goal is to extend this work to encompass any parameters generated by our models. To reinforce this initiative, we are also in the process of hiring a new full-time team member dedicated to this task (starting September 1st).

[JonathanGregory]

I'm happy to hear about this initiative. I'm sure it'll be useful, and it seems likely you'll have new proposals to make for standard names and perhaps CF conventions. It's really good that you'll have a member of staff working on it.

[atmodatcode]

Sounds like a great initiative — I intend to join this hackathon!

[feggleton]

Hi Sebastian, I'd be really interested in this topic. Would be happy to share what we've been doing at the Met Office in this area. We've mapped quite a few of our models. I know we've spoken in WMO meetings about this problem before in relation to identification of parameters across different centres and elements used from the different format properties. I believe @japamment was also interested in this topic. We've discussed before about universally agreeing on a standard set of properties/information from each format that's sufficient to describe all parameters, especially considering probabilities and percentiles with the addition of thresholds and vicinities.

[feggleton]

I've chatted to Alison and Ellie on the workshop planning committee about the potential for a presentation or lightning talk related to the hackathon about about metadata mappings and thought I'd capture it here. I've been working with a colleague at the Met Office, Adam Jacobs, who has been prototyping a tool to help with mappings between NetCDF and GRIB using the Fluid language. I think this could be really relevant to present, even though it's just an idea at the moment and will fit well with the other conversations we're having around mapping the two formats.

[efisher008]

Thank you for highlighting this, Fran @feggleton. This sounds potentially very relevant and a great fit with the proposed hackathon on CF and GRIB2 above.

[feggleton]

@efisher008 Hi Ellie, thanks for getting that into the agenda, would it be possible to change the title to 'Tools for mapping between NetCDF and GRIB' just to remove the mention of fluid specifically.

[ChrisBarker-NOAA]

Hackathon Proposal: SGRID conventions.

CF has a set of standards for defining "cells", which work fairly well for data on single coordinate axis or simple rectangular grids.

And thanks to the addition of the UGRID standard, unstructured grids.

But the methods for defining cells get unwieldy for larger dimensional rectangular grids, and even more unwieldy for curvilinear grids.

Even more unwieldy for the results from oceanographic (and met?) models that use various schemes where results are "on" different parts of the grid cells, i.e. "Staggered Grids" (e.g Arakawa C grid).

Fundamentally, CF is still lacking (IMHO) in describing a grid (or mesh, choose your words) per se as a single entity -- you have to define all the cells independently -- which is quite unwieldy for data producers, and even more so for data consumers.

In practice, there are a LOT of otherwise CF-compliant datasets out there that do not define the cell structure explicitly, and expect users to imply the meaning from other variables, which usually requires knowledge of the particular model that produced the results.

To address this, some years back the SGRID standard was proposed:

https://sgrid.github.io/sgrid/

It is in use by at least some folks (including me :-) )

It would be great to integrate it with CF, similarly to what we did with UGRID.

If we do this hackathon, it would be really great if we could get @hrajagers to join.

Also, someone from the NEMO community.

[JonathanGregory]

NOTE: this comment has been addressed with an update to the orignal proposed idea. (-CHB)

Dear Chris

If I understand correctly, SGRID is not about curvilinear grids specifically; it's about describing the relationship between different grids, such as the T, u and v points of an Arakawa C-grid, for instance. Is that right? CF already has ways to describe curvilinear structured grids, such as map projections and the NEMO tripolar grid. However, CF doesn't have ways to describe the relationship between grids, even when the different grids are defined by a single UGRID unstructured mesh.

Best wishes

Jonathan

[ChrisBarker-NOAA]

Yes, that is correct, I wrote that poorly -- SGRID builds on the existing CF concepts for curvilinear grids.

However, CF is still lacking (IMHO) in describing a grid (or mesh, choose your words) per se as a single entity -- you have to define all the cells independently -- which gets quite painful for curvilinear grids, and even more so for "staggered" grids.

Which is what SGRID (I think) is all about.

The "I think" is what I'd love to hash this all out with a few folks at a hackathon ....

[japamment]

Hi all,

At recent CF workshops we've had "housekeeping" hackathon sessions that have resulted in good progress on maintaining/improving the website and reviewing issues that are close to resolution and just need one final push to get them agreed, or potentially closing some older issues as "dormant". I propose that we do the same at this year's workshop.

Best wishes,

Alison

[lhmarsden]

Hackathon proposal

I would like to propose a hackathon on working with 'example datasets' for CF. This has been discussed here:
cf-convention/cf-conventions#348

In the hackathon we could discuss how to incorporate example datasets into the CF repository and perhaps into the CF conventions document. Depending on the group and progress, we could work on providing a simple example as a proof of concept.