Notes from each meeting of the Atmosphere Working Group are added as a reply to this topic.
Make each reply a wiki post , that way other attendees can edit the notes. This can greatly reduce the burden of note-taking allowing that person to participate fully in discussions without worrying too much about missing important details.
Key aims â establish Australian atmospheric modelling community, currently missing
Why are you / we here?
Community building: atmos modelling community hasnât been as well organised as other communities, e.g. land, ocean. Would be great for atmos community to reach that level of community organisation. Stronger to have a coherent voice, rather than singular voices.
Identify strategic priorities
Several people identified as being new to UM. Atmos WG is way to access training, docs, tools etc? Acknowledged that this is a priority for NRI. Noted that different ACCESS configs require different training
Advanced training would also be valuable for more experienced users â allow them to solve more complex problems
Think outside of box in delivering training / info. Videos?
Noted that responsibility for infrastructure is spread across NRI teams / groups
ToR accepted by those in meeting, can be revised in future if needed
Structure / near-term objectives
Structure / content of meetings
Previous ACCESS evaluation meetings (held a few times a year by Lawrie Rikus at BOM) are missed by community. Resurrect those in some format?
Science presentations, long / short
User problem solving, new and experienced
What are people planning, where are the gaps?
Pick a topic every meeting? Have discussion around that. Where is atmos community placed? e.g. CMIP ambitions. Noted that there is an Earth System WG. How to integrate model in community? Promote use of UM. How to collaborate between WGs, communities
Community experiments
what does community want?
goal for hackathons
establish as community resource
publicise pre-existing simulations that could be valuable
How to reach non UM users, e.g. WRF community
Meeting frequency
other WGs meeting fortnightly
what is community preference?
Great to have at least one in person meeting per year, make it a full day
otherwise ~6 meetings per year? Or monthly? Do we have enough to discuss to meet monthly?
Initially it is important to communicate what we are all doing, higher frequency might help with that
Monthly meetings initially, mix of science / technical topics
how to use forum would be valuable
Several 10-12 min talks, informal discussion style? Or a single longer 20 minute talk? Lightning talks also to cover intros etc?
Meeting length 1 hour
Compute resource
Atmos WG has 1.25 MSU available
Guidelines to be released shortly, used for community experiments
Atmosphere WG Satellite meeting: to be held during ACCESS-NRI annual workshop 5-8 September 2023. More details to follow!
** Note** Please watch the Atmosphere Announce topic on the ACCESS-forum to receive information on upcoming meetings and events! This youtube video explains how to watch a topic on the ACCESS-Hive Forum.
** 22 June 2023 (11:00-12:00)
14 participants online via Teams
From: BoM, CSIRO, UTas, UNSW, NRI, UoM, CLEX
Admin:
Yi Huang volunteered to take todayâs meeting minutes.
Sonya Fiddes volunteered to take the next meetingâs minutes (in July).
Need volunteers for science talks next meeting - ideas welcome!
Science talks:
a) ACCESS-A: Development of the Australian domain convective-scale numerical weather prediction model (Shaun Cooper, Belinda Roux, Mat Lipson BOM).
Questions/discussions: What resolution is used for ensemble runs? What GPM products are used for rainfall evaluation.
b) Atmospheric modelling over the Southern Ocean (Sonya Fiddes, AAPP)
Questions/discussions: Uncertainties of aerosol climatology over the Southern Ocean. Observed vs simulated radar reflectivities.
WG Training needs:
General needs:
What Scott Wales has originally setup (BOM documentation?) was useful for beginners.
Strengthened links with the Met Office to feed changes into model development.
Helplines (e.g. answering specific questions).
ACCESS-NRI Workshop (Sept 5-8) specific needs:
Training on how to make changes in the UM code/ to the trunk (currently need a lot of hand-holding from Met Office).
What do students need:
Documenting model setup and simple instructions to run the model, including an overview of basic aspects like where ancillaries are stored.
Rose-cylc tutorial for any configuration (basic-level users)
Other suggestions for basic-level users:
It would be useful for a new model user to document the difficulties and problems they encountered along the way and how their problems got resolved (for more experienced users this is somewhat foggy).
Do we need a AUS2200 training session at the ACCESS-NRI Workshop?
ACCESS-NRI encourages ECRs to learn and run the model.
Hackathons are useful exercises, but focused more on coding, post-processing, using tools to analyse of model data.
The level of training similar to what was provided at the UM training workshop would be helpful (people could learn to run a model for real).
Documentation:
Addressed in the above discussion on training needs, but further discussion needed.
Navigating around ACCESS-Hive remains a bit challenging. These training videos are helpful.
UM research activities:
Regarding this table, some people cannot see the âEditâ button on the bottom RHS of this wiki post. Heidi followed up on this issue and apparently you need to have a Trust Level 1 to edit a wiki post on the forum, which can be achieved simply by reading only 10 minutes of forum posts.
If you still have difficulties filling out this table, then you can reply directly to the post and we will add it in the table.
ACCESS-NRI Workshop (5-8 Sept ):
Yi Huang gave an update from program committee meeting:
Program has not yet been finalised, but a general training plan was developed.
Confirmed that we only need one day Atmosphere WG meeting.
People encouraged to think about what specific training activities we can help organise.
Table includes contact emails so can be in touch if there is overlap.
To get edit permissions you need to have spent 10 minutes looking at forum posts.
4. Combined working group meeting with forecast and prediction WG (Charmaine Franklin)
Forecast and prediction WG is interested in the single column model
Single-column models are useful for understanding parameterisations, good for idealised studies. Separates dynamics from physics. Assumes independence of columns so less relevant for high-res model.
Meeting will probably be organised by Martin Dix. Martin adds:
Single-column models also useful for debugging.
Working in latest version of the UM.
Could drive it with ERA5 similar to AUS2200.
Are people interested in learning more about the single-column model at NCI? Yes, there is interest.
Combined WG meeting will be after workshop in September.
5. ACCESS-NRI workshop update
Program (Yi Huang):
Two day workshop, first day will include invited keynote speakers; updates for the community.
Second day to focus more on science, including invited talks from individual working groups; breakout sessions.
Breakout sessions to be interdisciplinary.
For our group we need to:
Nominate someone to give an update on Day 1.
Think about who we want to give science talks on Day 2.
In a warming climate, there is a general tendency of reduced hail frequency, but increased hail size.
Mainly running WRF, some AUS400. Also keen to run ACCESS models for future studies.
Questions/discussion:
Whatâs WRF HAILCAST; what is needed to run it (Charmaine Franklin)?
Only profile information needed, two versions (including offline one) are available (Tim Raupach).
Particle size distribution used in microphysics schemes is developed for extratropics â not suitable for tropical conditions. Suggested that Tim test different particle size distribution using ACCESS models (Charmaine Franklin).
Observations may be available in more populated regions, e.g., SE Asia (Martin Dix).
AUS2220 training will use the 2022 Canberra hailstorm as a training case (Yi Huang).
3. ACCESS-NRI workshop update (Heidi Nettelbeck)
ACCESS-NRI Workshop / Training Day / WG Day registration reminder
Registration nearing capacity, closes 20.08.2023
4. Atmosphere WG day 07.09.2023 planning update (Charmaine Franklin)
3 science talks on topics that havenât been discussed much in previous WG meetings;
Speakers: Martin Singh, Matt Woodhouse, Christian Jakob.
Other suggestions: lighting presentations (participation open to all to present current or future work relevant to the WG)
Discussion session on WG Day (45min): call for ideas on discussion topics â post ideas on the forum!
Action item: Heidi to email Atmosphere WG day attendees regarding their participation in lighting presentations
Suggestions on experiments to be run as a community case (Atmosphere WG has resources to support this). Post your ideas on the forum! (Charmaine Franklin)
Next meeting is 7th September (9.30-5pm) at the ACCESS-NRI Workshop.
Charmaine Franklin - Regional atmospheric modelling activities at BoM
Heidi Nettelbeck - ACESS-NRI atmosphere liaison, Atmospheric model team lead
Belinda Roux - BoM Earth System Modelling Group - Atmosphere. Development of ACCESS A/AERegional Model Evaluation and Development (RMED)
Sonya Fiddes - Australian Antarctic Program Partnership, IMAS, UTAS. Using ACESS-AM2 to look at southern ocean radiation biases.
Kim Reid - ACCESS-S2 to enhance sub-seasonal scale precipitation. Also looking at AUS2200 to look at how temperatures in the coral sea influenced the 2022 extreme rainfall.
Sramana Neogi - ACCESS AMIP configuration to understand how much we can learn about climatology from short-duration runs.
Davide - Research Software Engineer at ACCESS-NRI. Use UM in geo-engineering. Contributed to development of Hive. Working on workflow for editing ancillary files.
Thi Lan Dao - PhD student UoM. Understand heavy rainfall signal under different combinations of large-scale variability. Using AUS2200 to look at heavy rainfall over Coastal Queensland.
Jasmeen Kaur - Working on Hive.
Claire Vincent - Tropical Variability, using AUS2200 for tropical rainfall experiments. Working on ACCESS-S2 and tropical modes rainfall predictabilty.
Scott Wales - ACCESS Ensemble modelling , Couple modelling
Christian Jakob - Director of W21C. Scientific Interest - convection in weather and climate
Yi Huang - UoM - Key Interests Cloud and Precipitation. Cloud, climate interaction, physical & dynamical processes. Uses modelling and field work. Cloud-atmosphere interactions over GBR & Antarctic sea-ice.
Mark Mallet - Australian Antarctic Partnership Program. Work with observations & ML. Supervising PhD student running regional UM over Southern Ocean. Observational campaigns.
Marty Singh - Monash. Mostly idealised modelling.
Matt Woodhouse - CSIRO - Resolving atmospheric composition in ACCESS
Xinhui Wang - Interested prospective student
Dale Roberts - part of CLEX CMS. AUS2200 production. Analysis of large datasets from a software level.
Kial Steward - ANU, primarily running laboratory experiments. Interested in idealised setups. Polar vortices, jets, stratospheric variability etc.
1. Marty Singh: Process understanding using model hierachies - examples from convection permitting modelling
Experiments / wish list of ACCESS configurations that could support research:
Doubly periodic RCE
Tropical Channel
Simple Physics
Aqua Planet
Mechanism Denial (COOKIE etc)
Discussion
Christian - profile of low level cooling and upper level heating - looks like stratiform profile.
Marty - this is part of the story, but do you get statiform profiles due to the heating profile, or the other way around?
Kial - Question about Double Periodic RCE
Marty - Convective heating balances radiative heating. Use RCE to examine problems around moist convection. Easy to simulate in both models. Also used to study convective aggregation. Experiments into why this happens and what the processes are - led to ways of looking at the real atmosphere.
Charmaine - UM sub-km scale simulates convection too early and too shallow. Could idealised model help understand what is going on.
Marty - Idealised modelling allows you to isolate processes
Scott - UM does support some idealised modelling configurations. There is some documentation, not sure how up-to-date it is. Aquaplanets are not that hard.
Christian - Strategy - we need a plan to incentivise people to develop these configs. Idealised model versions in the UK?
Scott - A lot of material exists, but people may not know about it.
Christian - nudged runs are useful semi-idealised config that would be useful for people to use.
Sonya - nudged runs need full ERA5 model levels - a lot of data is required. A full set of ERA5 model levels would be really useful.
Christian - nudging can maybe be done usig model levels too?
Kial - ERA5 project on NCI - pressure levels.
Marty - Is there an imbalance between storage versus compute at NCI? At the moment, a lot more compute hours than storage.
Christian - A lot of storage is from LEIF grants.
Kial - Can we design better ways of using /scratch - eg download and delete.
Sonya - ERA5 model levels 12.5GB per day
2. Christian Jacob ARC COE The Weather of the 21st Century
â21st Century Weatherâ what they want to be called
We know the Earth will get warmer, extremes will get more extreme
But location specific information on extremes is less certain
People want localised information to make decisions - need to know about future weather rather than global mean states
Weather change is a grand challenge of climate science
The climate is made up of weather, so we need to understand the weather in order to understand the climate. Weather is blobby and complex
Climate change will make us more weather dependent (e.g. renewable energy)
What happens to clouds or winds in the future will affect our future energy generation and climate change adaptation
Centre will have 3 major areas: 1) Circulation Change (Weather Systems, Climate Variability, Climate Change). 2) Weather change (Weather Systems, Weather Resources, High-impact Events). 3) Modelling (hi-res coupled modelling)
Goal by 2026 global 25km coupled model, regional Aust. domain 1km
By 2028 large domain 3km eg Indian and Pacific Ocean and Australia
Km-scale global models by 2030
Strong collaboration with the NRI especially through the CMS team
Often, climate models break at high resolution. Re-tuning necessary.
Difference in future ENSO projections with different resolutions
Decline in global atmospheric modelling in Australia but lots of work in regional modelling
Transition to new version of the UM will occur during lifetime of the centre (unless delayed - highly likely)
Discussion
Sonya Fiddes - global ACCESS model is old and is missing updates around microphysics. Hard to include global ACCESS model changes in the new UM versions, which is why they use the regional model which is more up to date.
Rachel Isphording - High res models needs high res obs for evaluation. Are there discussions with the BoM on this? BoM is a partner.
Yi Huang - What is the COE going to do differently from other international centres doing high resolution modelling? Going to collab and use resources with them. But focus is on understanding Australian weather. Goal is underlying science not predictions or projections.
AUS2200 is the starting point for the new centre and high resolution modelling.
Claire Vincent - Nocturnal jets important for wind energy generation and models currently do a poor job of simulating these. Need for easily relocatable (domain) high resolution model in research sector not just the BoM.
Heidi - how is that different from AUS2200? Relocatable aspect is the key difference
Charmaine Franklin - NRI needs to move away from âconfigurationâ thinking
Overview of ACCESS composition-climate modelling capabilities
Modelling, fieldwork, lab measurements - the three-legged stool of atmospheric chemistry
Land emits lots of gas that interacts with the atmosphere. [Chemical] composition connects different parts of the earth system.
Lots of stuff (that emits gases) floating around the Southern Ocean and Antarctica. Connected coupled system.
Capability starts with people - small group of atmospheric composition modelling in Australia. Mostly at CSIRO + UTAS
Composition in ACCESS. UKCA is the framework for chemistry composition. Advection, emission, deposition. Handles chemistry schemes
Lots of advected tracers makes it very expensive
GLOMAP-mode is the aerosol scheme (included in ACCESS-CM2, ocean optional, can nudge to the ERAs)
Coupled to cloud and radiation schemes
GLOMAP is a big improvement but also why ACCESS became much more expensive. GLOMAP explicitly represents aerosol size distribution. Necessary for cloud-aerosol feedbacks/decent radiative forcing
ACCESS-CM2-Chem includes full chemistry (tropospheric and stratospheric). Simulates ozone hole.
NRI has new visualisation specialist - Owen Kaluza. Valuable outreach tool.
Work on stratospheric ozone. Ozone hole better simulated in later ACCESS versions (with interactive chemistry). Ozone affects SAM which is important for Australian weather.
CM2-Chem 1.9x more expensive than CM2
UM nesting suite over Great Barrier Reef - reef restoration and adaptation program partnership
Assessing how the GBR is responding to coral bleaching and potential interventions
Southern Ocean simulations planned (Sonya Fiddes)
Past success: missing source of marine aerosol, addressing bias in ozone deposition over the ocean, Lightning NOx parameterisation, biases in stratospheric ozone
Make sure improvements are fed back to model development and included in UKESM/2
Future applications: transition to hydrogen economy and consequences (future CMIP input), Southern Ocean warm bias, Ozone hole recovery, stratospheric impacts from volcanoes (next Pinatubo, Hunga Tonga)
Big Questions: To chem or not to chem? Does Australia want/need and Earth System Model? Do we have compute/storage?
Discussion
Charmaine - focus is on marine aerosols, what about land? This is due to focus on Southern Ocean. Limited by number of people in Australia working on this to cover land too.
Can get rid of some things in GLOMAP to make it cheaper but need to evaluate new configurations. âGLOMAP-liteâ
Matt W - Lots of things we may want but how much capacity does the NRI have?
Heidi N - NRI has 4 people in the atmosphere team
The NRI needs input from the community on what is needed including around documentation and tools. Hive is useful for documentation and training- a starting platform.
What the NRI supports needs to pass through the Scientific Advisory Committee and Board.
Matt W- could write instructions of nudging. Need for processing ERA5 data for nudging/input data.
Heidi - use the forum for raising issues
Claire-challenge with prioritising tasks. Procedure necessary. AUS2200 hackathon led to Python scripts in various states on accessing and plotting data etc. Could be developed for future AUS2200 users as tutorials. Guides for accessing AUS2200 data in an efficient way would be helpful.
Yi - propose wishlist on the forum. Need community input to prioritise resources.
Atmosphere WG meeting has monthly meetings - open to feedback on these including around scheduling.