Aims & Goals
The third meeting of the Consortium for Ocean Sea Ice Modelling in Australia (COSIMA) was held in Canberra on 7-8 May 2018. This annual COSIMA workshop aims to:
- Establish a community around ocean-sea ice modelling in Australia;
- Discuss recent scientific advances in ocean and sea ice research in a forum that is inclusive and model-agnostic, particularly including observational programs;
- Agree on immediate next steps in the COSIMA model development plan; and
- Develop a long-term vision for Australian scientific advances in this area.
Participants
The 2018 workshop is our largest workshop yet, with 30 talks and 49 participants.
Attendees included:
Gary Brassington (Bureau of Meteorology), Matt Chamberlain (CSIRO), Chris Chapman (CSIRO), Fabio Dias (UTAS/CSIRO), Prasanth Divakaran (Bureau of Meteorology), Peter Dobrohotoff (CSIRO), Catia Domingues (UTAS), Matthew England (UNSW), Russ Fiedler (CSIRO), Annie Foppert (CSIRO), Leela Frankcombe (UNSW), Bishakhdatta Gayen (ANU), Angus Gibson (ANU), Stephen Griffies (NOAA/GFDL), Nicholas Hannah (COSIMA), Aidan Heerdegen (ANU/CLEX), Petra Heil (AAD & ACE CRC), Andy Hogg (ANU), Ryan Holmes (UNSW), Shane Keating (UNSW), Andrew Kiss (ANU), Vassili Kitsios (CSIRO), Veronique Lago (UNSW), Clothilde Langlais (CSIRO), Andrew Lenton (CSIRO), Kewei Lyu (CSIRO), Jie Ma (CSIRO), Simon Marsland (CSIRO), Paige Martin (University of Michigan), Josue Martinez Moreno (ANU), Richard Matear (CSIRO), Laurie Menviel (UNSW), Mainak Mondal (ANU), Ruth Moorman (ANU), Adele Morrison (ANU), Terry O’Kane (CSIRO), Peter Oke (CSIRO), Ramkrushnbhai Patel (UTAS), Paul Sandery (CSIRO), Abhishek Savita (UTAS-CSIRO), Kate Snow (NCI), Paul Spence (UNSW), Kial Stewart (ANU/UNSW), Veronica Tamsitt (UNSW/CSIRO), Mirko Velic (Bureau of Meteorology), Marshall Ward (NCI), Luwei Yang (IMAS, UTAS), Rui Yang (NCI), Jan Zika (UNSW)
Status
The workshop was structured to focus on scientific questions on Day 1, particularly in the first two sessions. In these sessions, topics ranged from from Antarctic shelf processes to oceanic convection, from reversibility of the Earth system to frictional drag. The final session on day 1 focussed more on technical issues, including assessment of the optimisation status of existing models. On Day 2, talks focussed more on strategic issues, including an outline of Bluelink, ACCESS, CAFE and coastal programs. These strategic talks transitioned to small-group discussions (see synthesis below). The workshop finished with a tutorial on the COSIMA Cookbook framework for model analysis.
The Australian landscape in ocean-sea ice research involves a number of interleaving programs, each of which was represented at this workshop. The figure below outlines the linkages between these programs:
By way of explanation:
ACCESS-CM2/-ESM1.5 will be Australia’s input to CMIP6, and use MOM5 and CICE at 1°.
CAFE is the decadal prediction system in development, which uses MOM5.
ARCCSS/CLEX, ARC CoE programs, use high-resolution ocean-sea ice models for process studies.
Bluelink/OFAM is the ocean forecasting and reanalysis system which will adopt ACCESS-OM2-01 in future versions.
CSHOR is the Centre for Southern Hemisphere Oceanographic Research; it focuses on observational studies but we hope to establish two-way interactions with this program.
Coastal Modelling includes the Australian coastal oceanography community, as well as Antarctic nearshore programs within AAD and ACE-CRC.
A major theme of the workshop was to review the status of the ACCESS-OM2 model which is the focus of COSIMA. In short, we have had success with model releases at 1° and 0.25° resolution – these models are now actively being used for scientific runs, and are available for download and use by the community. They include a recent upgrade to the file-based atmosphere (YATM) and new JRA55-do forcing datasets. The 0.1° version of the model has progressed significantly in the last year; there are outstanding tasks to evaluate model output and further optimise the model configuration.
The COSIMA Cookbook tutorial was attended by about a third of participants, and some progress was made. The aim of this tutorial was to entrain more active users to the system and encourage input from those users. The Cookbook is similar in style to the analysis system being developed for CAFE and it may be possible to merge elements of each framework at some stage in the future.
Program
Where available, talk files are linked from the presenter’s name.
Monday 7 May | |
10:00 | Arrival & Morning tea |
10:30 | Session 1 (Chair – Andy Hogg) |
Stephen M Griffies (NOAA/GFDL): Understanding and projecting global and regional sea level: More reasons to include refined ocean resolution in global climate models | |
Andrew Kiss (ANU): Overview of the ACCESS-OM2 model suite | |
Andrew Lenton (CSIRO): Ocean Reversibility in ACCESS-ESM | |
Catia Domingues (UTAS): Global and spatial temporal changes in upper-ocean thermometric sea level | |
Fabio Dias (UTAS/CSIRO): Mean and seasonal states of the ocean heat and salt budgets in ACCESS-OM2 | |
Adele Morrison (ANU): Circumpolar Deep Water transport towards Antarctica driven by dense water export | |
Jan Zika (UNSW): Getting an ocean model to obey: Prescribing and perturbing exact fluxes of heat and fresh water | |
12:30 | Lunch |
13:30 | Session 2 (Chair – Clothilde Langlais) |
Petra Heil (AAD & ACE CRC): ACCESS-OM2-01 sea ice | |
Paul Sandery (CSIRO): Sea-ice data assimilation and forecasting using an Ensemble Transform Kalman Filter | |
Paul Spence (UNSW): Does the Southern Ocean have sleep apnea? | |
Veronique Lago (UNSW): Impact of projected amplification of Antarctic meltwater on Antarctic Bottom Water formation | |
Ryan Holmes (UNSW): Numerical Mixing in the COSIMA Models | |
Luwei Yang (IMAS, UTAS): The impacts of bottom frictional drag on the sensitivity of the Southern Ocean circulation to changing wind | |
Vassili Kitsios (CSIRO): Stochastic subgrid turbulence parameterisation of eddy-eddy, eddy-topographic, eddy-meanfield and meanfield-meanfield interactions | |
Matt Chamberlain (CSIRO): Using transport matrices to probe circulation in ocean models | |
15:30 | Afternoon tea |
16:00 | Session 3 (Chair – Petra Heil) |
Nicholas Hannah (COSIMA): ACCESS-OM2 Software Development | |
Marshall Ward (NCI): ACCESS-OM2 performance analysis | |
Rui Yang (NCI): Parallel IO in MOM5 | |
Angus Gibson (ANU): Towards an adaptive vertical coordinate in MOM6 | |
Jie Ma (CSIRO ): Investigating interannual-decadal variability of Indian Ocean temperature transport in an eddy-resolving model | |
Paige Martin (University of Michigan): Frequency-domain analysis of energy transfer in an idealized ocean-atmosphere model | |
17:30 | Close |
19:00 | Workshop dinner (Debacle – 24 Lonsdale St Braddon) |
Tuesday 8 May | |
9:00 | Session 4 (Chair – Andrew Kiss) |
Andy Hogg (ANU): Are we Redi for 0.25° ocean-climate models? | |
Kial Stewart (ANU): The Repeat Year Forcing for JRA55-do | |
Terry O’Kane (CSIRO): Coupled data assimilation and ensemble initialization with application to multi-year ENSO prediction | |
Gary Brassington (Bureau of Meteorology): Ocean forecasting status and outlook | |
Peter Oke (CSIRO): Bluelink activities and plans | |
Matthew England (UNSW): A proposal for future projection simulations using COSIMA ocean-ice models | |
Richard Matear (CSIRO): CSIRO Decadal Climate Forecasting, update of the project’s progress | |
Simon Marsland (CSIRO): Preparing ACCESS for CMIP6 | |
Clothilde Langlais (CSIRO): Downscaling towards the coast – a perspective on where the coastal modelling group would like to go | |
11:00 | Morning tea |
11:30 | Discussion: COSIMA planning and strategy |
13:00 | Lunch |
14:00 | Strategy and planning summary |
14:30 | COSIMA Cookbook tutorial |
16:00 | Close |
Synthesis of Discussion
Tuesday afternoon included discussions of present and future needs and directions of the COSIMA community, via breakout sessions on the topics Sea Ice, Coastal / Forecasting, Coupled Modelling, Process Modelling, Biogeochemistry, and Technical. The overall threads of these discussions are summarised here.
Open and accessible code, configurations, output and analysis
Transparency, accessibility and reproducibility of model code development, run configurations and output data were named as priorities by many groups. Nic Hannah’s proposed REDB (Reproducible Experiment Database, http://redb.io) was widely supported as a means to tie together and curate the source code, configurations, output and analysis of model experiments. Using consistent shared codebases was also a priority. Containerisation was suggested as a method to make experiments self-contained. Extension of the database to include idealised experiments was also suggested.
Model evaluation
There is a need for more model evaluation against observations. Several groups highlighted the importance of better integration of observations for model validation and a desire for this functionality to be better supported in the COSIMA Cookbook. Comparison of CICE to SIS-1 at 1 and 1/4 deg was also suggested.
Technical validation is also needed – e.g. BGC, bit reproducibility, broadened test suite, regression testing. Model performance and stability priorities include: resolve crashes, balance load, MPI benchmarks and stress testing.
Usability
Suggestions included a glossary for beginners, an online portal for control runs, and to minimise difficulty of running new model configurations. Standardised output files and naming conventions would facilitate analysis. Improved functionality and versatility of the COSIMA Cookbook was also suggested.
Documentation was a priority for many, in particular an ACCESS-OM2 documentation paper, but also open/evolving documentation as the models develop.
Parameter selection was also a concern for many – how to choose appropriate parameters (e.g. for ice or BGC), how to assess model sensitivity to parameters, how to document why parameters were chosen or altered. Data assimilation was suggested a way to improve ice parameter selection, including assimilation of under-ice observations (e.g. temperature). BGC was suggested as a way to constrain the dynamics.
It was pointed out that the payu run management software underpins model runs, yet formal funding for its continued development is presently lacking.
Model enhancements
Suggestions for enhanced modelling capability included: interannual forcing, WOMBAT BGC, coupling to an atmosphere model, 1-way nesting, coupling to wavewatch, explicit tides, wet/dry cells.
Community coordination, synergies and strategy
Suggestions included a streamlined process for providing community feedback and deciding on priorities, and for community involvement in developing the BGC component. It was also suggested to foster engagement with atmosphere and sea ice specialists, and have a more formalized ice group. The technical team is also seeking more input from scientists, especially regarding sea ice.
Regarding modelling strategy, it was suggested to have intelligent model diversity (not too many versions), a consensus on standard perturbation experiments, and to decide on resources to commit to MOM5 vs. MOM6.
Summary of Priority Tasks
The following list of tasks was identified as a priority for the near term. Volunteers to lead or assist with tasks much appreciated.
- IAF Runs: With the addition of YATM, we now have the facility to run Interannual Forcing (IAF) runs from the JRA55-do forcing dataset in ACCESS-OM2. Once YATM has been tested, we will conduct IAF runs at all resolutions, starting with 1°.
- Model Documentation: Production of a model documentation paper is a high priority for the coming months. This will be achieved by:
- Writing a larger technical documentation report (https://github.com/OceansAus/ACCESS-OM2-1-025-010deg-report) that will be stripped down to feed into a paper; and
- Inviting community evaluation of existing model output.
- Model evaluation and analysis: We propose the COSIMA Cookbook as a framework for users to contribute model analyses. In particular, we encourage observational comparisons with existing model output, and also encourage users to submit bug reports and feature requests via https://github.com/OceansAus/cosima-cookbook/issues
- WOMBAT: In the coming months we will look to implement the WOMBAT biogeochemistry model (already running in MOM5) into the ACCESS-OM2 framework.
- Capability gaps: The COSIMA community has been able to leverage expertise from a number of different programs. However, our community as a whole remains subcritical in several areas, including sea ice modelling and atmospheric dynamics.
- REDB: Nic Hannah proposed a new system for tracking simulations and the output data. This system was identified by many discussion groups as a potential solution to some of our collaboration roadblocks. We will investigate the viability of such a system.
- MOM6: Plan is to begin transition to MOM6, building up experience in the latter half of 2018.
Recommendations for COSIMA 2019 workshop
- Institute a James Munroe award for contributions to COSIMA
- Extend to a 2.5-day workshop to allow more time for discussion (not extra talks)