Atmosphere WG Meeting 19.03.2025 minutes
19 March 2025
27 participants on Zoom
Matt Woodhouse convening.
Updates from WG Chairs.
Heidi:
- ACCESS-NRI are about to release a configuration of the Regional Nesting Suite ACCESS-rAM3 (beta release).
Charmaine:
- New starter at BoM. Ian White. Working on ACCESS-A/AE km scale systems.
- Working with met office on RAL3/Lfric. Lfric is the replacement for the UM. We will be running that RAL3 for three domains for high impact weather case studies.
Yi:
- AUS2200 project research is ongoing using the model datasets. Black summer case study has been published in JGR atmosphere. The first major publication from the AUS2200 data. Other papers are in draft. Thanks for the community support.
- In 21stCenturyWeather we are discussing running larger simulations. A ‘flagship’ simulation using NRI and rAM3. Discussions are ongoing.
- Trying to get AM3 running with Cable at ~25 km spacing (N512). Cable ancillaries are currently only available at 1 degrees.
- We have some new ECRs.
- Ian W. Originally from the UK, now at BoM w/Charmaine.
- Qingqand Gao. From Cambridge PhD, now post-doc with Yi at Unimelb. Looking at low clouds in hi-res, convective permitting models. Looking at BARRA-R2 and C2. Working with Chun-Hsu at BoM.
- Alanah Chapman. Working with Yi, Matt, Robyn and Sonya on AM3. Looking at southern ocean clouds, specifically sources of dimethyl sulphide and sea ice melt.
- Nasim Aln. Starting post-doc with Steven Zim and Monash. Wanting to run ACCESS-UM examining convection schemes and transition from shallow to open convection.
- ANU students Emma, et al. working with Sarah Perkins-Kirkpatrick.
Topics for ACCESS workshop in September 2025:
Jasmeen:
- User training team at ACCESS-NRI has created a training topics poll
- Please fill out - should take only 2-3 minutes. Feedback is greatly appreciated!
Comment From Paul Leopardi :
- Looking for extra technical requests/ material for training days for breakout rooms. UKMO is upgrading suites to cylc8 and Github. Fab etc. The Momentum User workshop runs before ours, these will be included in our community workshop.
Christian Stassen science talk: Added value (AV) of regional models over Australia
- RCMs improve many outputs from GCMs.
- But they have their own set of problems, i.e. the remaining parameterisations.
- RCMs over Australia are still not km-scale convection resolving.
- Comparing BARPA-R (~17 km) and CCAM12 (~12.5 km), both with model tops at ~40 km, against ERA5 + 8 X CMIP6 models downscaled
- Comparing GCM and RCM historical runs against observations from 1985-2014, looking at MSE, bias and correlation.
- Looking at metrics at two different length scales to avoid double-penalty problem (e.g. penalised with spatial error sin extreme precipitation events at higher resolution, which GCMs can’t even simulate)
- We upscale RCM to GCM, and interpolate GCM to RCM.
- Looking at Hot, cold, wet and dry extremes.
- RCM improves tasmax over most of Australia, but struggles with Tmin and precip over SE Australia
- Generally BARPA and CCAM add value at GCM scale. ‘Drizzle problem’ in GCM is improved, so RCMs give AV in number of drought days.
- BARPA and CCAM add vallue at RCM length scale.
Q : Martin. Can you say whether RCMs make bad GCMs less bad, and good GCMs better?
A: Not much AV for ERA5 and Euro-Earth as these are already at fairly hi-res. We do improve MPI models but some biases (e.g. cold bias) can’t be removed by CCAM. It is metric and model dependent.
Q : Matt. A preprint yet?
A: Don’t know. I’ve been away for the past week! Will let you know.
Matt - a preprint in the minutes would be great.
Paul Gregory : Showing 21st Century Weather Wiki on introductory rose/cylc training for new ACCESS-UM users.
See Introduction to Rose/Cylc — 21st Century Weather Software Wiki
Q : Chermelle. Be careful with out suite ids. Please remove Scott’s suite and waite for rAM3.
A : Sure, no probs.
Q : Davide. Overlapping with NRI and Hive docs.
A : Will link to existing documentation where ever possible so there is only one, supported and maintained, source of truth within the community.
Q : Heidi. Reach out to me and Davide as we maintain the Hive docs.
Q : Paul L. Keep in mind migration , e.g. cycl 8 and github, rosie migration.
Martin Dix : AM3 updates
- Uses updated UM and CABLE and will eventually become ESM3
- Using GM GA9 with CABLE instead of JULES.
- Running at N96. Issues with CABLE ancilliary generation at higher resolution.
- Unifying CABLE science code to a common library.
- Some CABLE science improvements which also apply to ESM1.6
- Hoping to have alpha release soon.
- CABLE development is on github.
- AM3 development on private github repo. Better than MOSRS and svn.
- Can’t wait for UKMO to transition to github. So much better than MOSRS and svn.
- Will we backport to MOSRS UM, JULES and roses repos? Feedback sought.
- Next step - Evaluation of AM3 and extendend simulation with CM3 (with JULES, MOM6 and CICE).
- Adopt new NRI built and test framework (already used for OM3)
- Ideally we want CABLE in a version or rAM3. First with UM, then with Lfric.
Q: From Matt W. Switching on chemistry?
A : Need to check surface deposition and UKCA stuff. Can be messy w/different vegetation types. Should be a single switch in the same way.
Q : Matthew Lipson. Land working group is moving to workflows for UM ancillary generation. Are you working with Claire to align with us?
A : Existing methods will work for high-res global, but we will need to move to other sources for high-res km-scale. We are looking at CCI now. We are using ANTS suite.
Heidi - ACCESS-rAM3 (beta-release) and WG compute resources
- Watch Hive Docs for documentation on new beta-release of ACCESS-rAM3.**
- Please apply for WG resources. We didn’t use all our quarterly kSUs. If you have a good case for suite testing, please put on experiment category of atmosphere for resources you want. See here for how to apply for Atmosphere WG Resources.
Next WG meeting: April dates? Let us know about any community conference/workshops to avoid clashes