Planning the virtual ESMValTool workshop, 24–26 November 2026

[rbeucher]

Hi everyone! Following the date announcement, our next virtual ESMValTool workshop will take place on 24–26 November 2026. ACCESS-NRI (@rbeucher and @flicj191) have agreed to coordinate the organisation and programme. @ESMValGroup/esmvaltool-developmentteam,

We would like to try a format that makes it easier for people in Australia, Europe and North America to take part. There is no single convenient live meeting time for all three regions, so here is a starting proposal:

• Hold some key discussions during the best available Europe–Australia overlap (European morning / Australian late afternoon), and continue or repeat discussions at times that work for Europe and North America.
• With presenters’ agreement, record talks and Q&A for replay in another time zone, followed by a live discussion there. Shared notes would carry questions and outcomes between groups.
• Offer coding sessions at useful local times, including Australian working hours before the discussion sessions and European afternoons afterwards. Shared GitHub tasks and short handovers could help work continue across time zones.

This is a proposal, not a fixed timetable. Some topic ideas are already in the date announcement; please build on those here.

We would welcome your comments on:

1. Programme: What talks, discussions or coding tasks would you like to propose? Could you present, lead, chair or help with a session?
2. Timing: Which time zone are you in, and what times could you realistically join live?
3. Format: What would help you participate across time zones? Would the recording, replay and shared discussion approach work for you?

We will use your feedback to shape the programme and post a draft agenda here. Looking forward to planning it together!

DEADLINE FOR PROPOSALS: Friday 6th of November 23:59UTC!

[rbeucher]

I'd be happy to give a presentation on using AI to convert NCL diagnostics to Python, including lessons learned and how to review the converted code. I'd also be glad to discuss the use of AI more generally in ESMValTool development—where it helps, where it falls short, and how to check its output.

[rbeucher]

I'd also like to propose a discussion about the discoverability of ESMValTool recipes. In meetings and workshops, I find it difficult to answer questions about which recipes are available for a particular use case, because it is hard to get a clear overview of what exists. How could we make recipes easier to find and understand?

I've been working on a user interface to help people explore available recipes and design new ones, and I'd be happy to show its current state as a starting point for the discussion.

[k-a-webb]

I would be very interested in this topic, as while I've found this, https://esmvaltool.dkrz.de/shared/esmvaltool/stable_release/, generally helpful, I do think there is room for improvement.

The recipe overview:

• It would be nice if the search bar was more comprehensive; could filter on variables, datasets, "new" or "updated", etc.
• It would be nice to have an alternative layout that minimizes scrolling, perhaps just a table?

Within the summary of a recipe's output:

• Within an output summary the sections are labelled according to the <diagnostic alias>:<diagnostic script alias> which is only helpful if the recipe uses sufficiently descriptive aliases (an example of where this could be improved).
• Only figures are shown. I would be nice to display tables (I'm not a fan of tables being shown as images) or at least indicate they exist..
• There is excellent documentation of the "description of the recipes" in the ESMValTool documentation, however this is separate from both the recipe, and the nice HTML output summary. Would there be a better way to pair recipes with their documentation?

[rbeucher]

It looks like I have not been very successful in gathering suggestions so far, so apologies if I am being a bit pushy! 😄

Here are a few more ideas inspired by the minutes from the last ESMValTool Monthly Meeting.

I think CMIP7 and REF should be high on our agenda, and it might be worth dedicating a full session to them. Some possible topics:

• Updating recipes for CMIP7: Could this be a task for one of the coding sessions? @rswamina?
• ESGF downloads: Testing data discovery and download workflows for CMIP7.
• Recipe compatibility: Adapting existing CMIP5 and CMIP6 recipes to work with CMIP7 data.

@bouweandela @valeriupredoi – Is there anything from the @ESMValGroup/technical-lead-development-team that we should include?

@hb326 @alistairsellar – Would it be useful to have someone from the Steering Committee provide an update on the longer-term strategy and direction of ESMValTool?

It would also be great to hear from the @ESMValGroup/userengagementteam, particularly about recent activities and plans for the IPCC author training.

These are just a few ideas to get the discussion going. Please feel free to suggest other topics, priorities, or activities you would like to see included.

I would really appreciate your input so I can start putting together a draft programme.

[k-a-webb]

I think it is important to replace CMIP5 for all recipes we want to keep working and where that is possible. The access to CMIP5 will become more difficult, DKRZ did already remove them from the work storage, which could cause issues for the next releases.

Would it be possible to include a facet in any given dataset (e.g., CMIP5) like skip_if_not_found which permits the recipe to run if the dataset is not available, but will use it if it is available? This could perhaps avoid the loss of backwards compatibility.

[rswamina]

If we have any IPCC chapter scientists, perhaps we can get input on how to prioritize recipes for CMIP7/REF/IPCC. I am not sure how exactly to do this. It will involve at the minimum changing variable names to branded names and testing on CMIP7 data?

[axel-lauer]

Would it be possible to include a facet in any given dataset (e.g., CMIP5) like skip_if_not_found which permits the recipe to run if the dataset is not available, but will use it if it is available? This could perhaps avoid the loss of backwards compatibility.

🚀 I like this idea!

[bouweandela]

Would it be possible to include a facet in any given dataset (e.g., CMIP5) like skip_if_not_found which permits the recipe to run if the dataset is not available, but will use it if it is available?

There is a command-line/configuration option for that, but it may not work in various cases, e.g. when the recipe authors have chosen to regrid all datasets to the grid of the one that is missing. Example usage: esmvaltool run --skip_nonexistent=True recipe_example.yml, from the docs

[bettina-gier]

I think the point is to have it as a dataset facet, so you can add CMIP3/CMIP5 data that's more rare, while still keeping some vital datasets to the analysis as mandatory

[rbeucher]

@bouweandela Will you be happy to discuss the work on the AI policy?

[bouweandela]

Sure!

[rbeucher]

DEADLINE FOR PROPOSALS: Friday 6th of November 23:59UTC!

Thank you!

[axel-lauer]

One more item for the CMIP7 topic could be on updating the scripts for derived variables: ESMValGroup/ESMValCore#2984 (maybe for a coding session).

[k-a-webb]

I would appreciate this discussion, especially a coding session

[rswamina]

The User Engagement Team usually have a one hour meeting and some time to work on tutorials. @LisaBock - what do you think?

[LisaBock]

Thanks, @rswamina ! Sorry for not bringing it up earlier...
Yes, we would need one hour (separate) for the UET meeting. Ideally with an Europe–Australia overlap. If these time slots are already needed for other topics, I am thinking about organising a UET meeting shortly (the days) before the ESMValTool workshop.

While we have coding time, we could definitely work on the tutorial (porting it to the rst-format).

[malininae]

With the giant efforts from literally everyone to submit papers pre March 2027, I was wondering if some sort of triage system/policy should be created for the recipes influx? And is it worth to discuss it all together.

[LisaBock]

If it is of general interest, then I can briefly report on our planned support/webinars for IPCC authors and chapter scientists, together with @flicj191.

[hb326]

It would be great to have a discussion about the different available pipelines we have developed now to run ESMValTool in the production of model simulations. We have developed ICONEval here at DLR, and I heard that there are potentially similar frameworks available from Australia and Canada. Maybe @k-a-webb could present something for Canada, and @schlunma something for ICONEval?

[k-a-webb]

I'm happy to briefly present how we (ECCC, CCCma) have developed a wrapper around ESMValTool to support model development. It's similar to REF in being a framework to pair datasets with relevant recipes/diagnostics.

[ehogan]

Would CMEW fit into this? 🤔

[bettina-gier]

Should we rediscuss the paper that was started since the last workshop? Maybe have a last push to finalize a draft?

[hb326]

Absolutely! The plan is to have a way more mature draft at the time of the workshop so that we hopefully only have to add the final touches to it.

[valeriupredoi]

pls pen me down to give an update on technical matters, including eg https://github.com/NCAS-CMS/cmip7_repack and https://github.com/NCAS-CMS/PyActiveStorage Danke muchly 🍻

[malininae]

Not a topic, but just an FYI: BC is not changing time anymore, which screws up a lot of timezone stuff.

[bouweandela]

I can present the brand new esmvaltool language server that I've been developing, it gives you a lot of information in your editor when writing recipes: https://github.com/ESMValGroup/esmvaltool-language-server