Atmosphere WG meeting 29 July 2026
12 participants on Zoom
Community updates:
Bethan has rotated on as the third WG co-chair, joining Sonya and Andrew. Yi has now rotated off – thank you for all your work, Yi!
Sonya and Yi are seeking roadmap input in advance of the ACCESS-NRI Scientific Advisory Council meeting in September. See the Hive post here ACCESS NRI - community input for major science priorities over next 10 years – either reply to the post, or message Sonya/Yi privately if you don’t want it to be public.
Science talk 1 – Leena Khadke and Sam Dahl, ACCESS-rAM3 runs for SHEP project.
rAM3 being used for case study, ensemble, and PGW approaches to sub-hourly extreme precipitation events.
Case study approach: 4 simulations x 9 cities
Ensemble approach: 10 ensembles, 5-6 cases each 50-60 simulations in total
PGW setup: 4 simulations x 9 cities
Process-based not forecast-based (don’t need exact timing and location compared to observed case).
Development of a benchmarking framework – doesn’t exist in the literature for these kinds of short-duration events.
Science talk 2: Pao Corrales, n512 AM3 evaluation and GitHub repo walk-through
AM3 configured to run at n512 (~25km), set of test runs done and asking for community evaluation help, goal of checking everything looks ok!
To get involved, nee;
- access to gb02 NCI project
- a GitHub account
- access to the access-am3-evaluation repo
- and possibly ssh keys set up on GitHub.
Pao walked us through how to clone the evaluation repo to your local gadi directory, and how to move analysis files (e.g. notebooks) into the relevant dir in the repo, and how to add it to the local repo and push back to git.
Pao also showed us how to upload files to the git repo through the browser, and highlighted that this is a simpler interface method but also highlighted the limitations of this compared to working with the repo in a terminal, especially when working with complicated files.
Pao told us how to upload figures to the web page for discussion, by opening an issue, uploading a figure and starting a discussion.
Pao’s presentation slides are here:
We need your help! AM3 Hi-Res evaluation! - Google Slides
There will be an AM3-evaluation meeting 10:30 on 4 Aug to discuss AM3 N512 vs N96 analysis, including a tutorial of how to contribute scripts on the AM3-evaluation GitHub repository – come along!
Upcoming workshops:
Don’t forget the Atmosphere WG meeting at the ML workshop - only a 45 min meeting, but come along for discussion about how to integrate machine learning techniques into our working practices.
Then the following Land-Coupled modelling workshop 31 Aug – 2 Sep will have some AM3 training on cylc8, as well as a hackathon.
Mat: The Land and Coupled Modelling workshop themes are quite broad:
Coupled modelling
• Uses of Earth System Models versus Climate Models
• Response of the Earth system to externally forced changes
• Ocean-atmosphere-ice interactions
• Tropical climate variability and change (e.g. double ITCZ, El Niño Southern Oscillation, tropical Pacific SST trends)
• Representation of convection and deep ocean circulation at high latitudes
Pathway towards high resolution coupled modelling with ACCESS.
Heidi mentioned that NCI will be removing support for cylc7 in early 2027 (!).
Do we need to flag this more widely for the community now? What about cylc7 suites that are outside of ACCESS-NRI’s development umbrella (e.g. the n512-AM3 suite) – is there any support for transitioning these suites to cylc8?
WG co-chairs introducing a new “ask the community” section to working group meetings.
Started off this week with Sonya asking whether anyone knows if CASIM can be run in a convection-parameterised (eg. GA8/GAL9) configuration instead of PC2. (Nobody knew).
Heidi: compute and storage:
Apply for Q3 compute if you have a project:
Can also apply for storage – note that it is temporary. Up to about 2TB is generally ok.
Heidi: next meeting:
In person at ML workshop on 20 August.
Next online Atmosphere WG meeting in September TBA.