COSIMA Model Output Collection

An increasingly important aspect of model simulations is to be able to share our data. Over the last few years we have been working on methods to routinely publish our most important simulations. This publication process is designed to allow any users, worldwide, to be able to pick up our model output and test hypotheses against our results. It will also allow journal publications to be able to cite our model output.

Currently we have  5 different datasets within the headline COSIMA Model Output Collection, which can be found here:

For users with NCI access this data is housed under the cj50 project.

We are planning to add new datasets in the coming months.

New Repeat Year Spinup

We are currently undertaking a new spinup simulation using our highest resolution ocean-sea ice model: ACCESS-OM2-01. Our goal is to run at least 50 years of this simulation in the third quarter of 2019, using the  1990-91 Repeat Year Forcing strategy (that we call RYF9091). After the first month of the quarter, we have managed to complete 23 years of the simulation.

Early indications are that temperature and ice biases are reduced in the new simulation (compared with our previous RYF8485 spinup) but that many large scale circulation metrics, such as ACC transport) are largely unchanged.

Globally Averaged temperature from RYF9091 spinup.
ACC Transport from RYF9091 spinup

For those who are interested, this simulation is being stored in /g/data3/hh5/tmp/cosima/access-om2-01/01deg_jra55v13_ryf9091. Timeseries describing the simulation, periodically updated, can be found here, and additional diagnostics are available upon request.

COSIMA IV, ANU, 3-4 Sept 2019

Announcing that our 4th Annual COSIMA workshop will be held at ANU on Tuesday 3rd and Wednesday 4th September.

The workshop will feature our regular mix of talks and discussions, covering a mix of physical oceanography, sea ice, model development and technical advances. Our workshops are not restricted to ACCESS-OM2/MOM users – we welcome contributions from all who consider themselves part of the COSIMA community.

Registration is now closed, but you can join via Zoom – see below.

We will begin with morning tea at about 10am on the first morning (3 Sept), with the first talk at about 10:30.

Click here for the workshop program (updated 1st Sept).

The workshop report provides slides from the presentations and a summary of the discussion. Shweta’s post on the workshop is here.

Joining via Zoom
The workshop will be streamed via Zoom. When joining via zoom, please mute your mic and hide your video. Unfortunately we probably won’t have an ability to take questions from remote participants.

Connection details:

Join from a PC, Mac, iPad, iPhone or Android device:

Join from a H.323/SIP room system:
Dial: +61 2 6222 7588
or H323:8232211676@ (From Cisco)
or H323: (From Huawei, LifeSize, Polycom)
or or (U.S.)

Meeting ID: 8232211676

ACCESS-OM2 at the CLEx Winter School

This week saw the ARC Centre of Excellence for Climate System Science hold its annual Winter School. This year’s edition is on modelling the climate system, hosted by the University of Melbourne. Yesterday was “Oceans Day” at the Winter School, and the highlight was an afternoon lab session based on ACCESS-OM2. Tasks included:

  • To run the ACCESS-OM2 model (about 65 of the 70 students managed to do this); and
  • To analyse some existing ACCESS-OM2 model output using the COSIMA cookbook (most students also completed this, with about a third of students producing great plots to show some new and intriguing results).

Progress was enabled by erstwhile helpers from CLEx’s CMS Team – Aidan, Claire Holger, Paola and Scott!

ACCESS-OM2 Workshop
Feverishly running ACCESS-OM2 at the 2019 CLEx Winter School.

ACCESS-OM2 Evaluation Paper

A community paper evaluating the performance of ACCESS-OM2 has been submitted to Geoscientific Model Development (GMD).  The paper has 30 authors from across the Australian community. GMD has an open review process, so you can track its progress.

This paper outlines the performance of ACCESS-OM2 at three different model resolutions, as indicated in the figure below.  It aims to spell out which versions of the model are suitable for different types of studies, and highlights the performance of the 0.1° resolution configuration (referred to as ACCESS-OM2-01). The paper shows that ACCESS-OM2 does a good job of representing many features of the ocean. Historical sea ice extent trends are well-represented, and the surface properties and transects in each ocean basin compare well with the observational record. The large scale overturning circulation, flow through the Indonesian archipelago and patterns of boundary currents are realistic, supporting the notion that this suite of models is competitive with similar models from other institutions. Areas for improvement include the relatively weak barotropic transport in the midlatitude gyres, particularly in the Pacific Ocean, the weaker than observed Drake Passage transport and the weak AMOC in the 1° configuration. For full details, feel free to browse the paper!

Kiss, A. E., Hogg, A. McC., Hannah, N., Boeira Dias, F., Brassington, G. B., Chamberlain, M. A., Chapman, C., Dobrohotoff, P., Domingues, C. M., Duran, E. R., England, M. H., Fiedler, R., Griffies, S. M., Heerdegen, A., Heil, P., Holmes, R. M., Klocker, A., Marsland, S. J., Morrison, A. K., Munroe, J., Oke, P. R., Nikurashin, M., Pilo, G. S., Richet, O., Savita, A., Spence, P., Stewart, K. D., Ward, M. L., Wu, F., and Zhang, X.: ACCESS-OM2: A Global Ocean-Sea Ice Model at Three Resolutions, Geosci. Model Dev. Discuss.,, in review, 2019.


Over the last day we have completed initial spinup runs of the ACCESS-OM2 model.
The model has been run at 3 different resolutions, as listed below:

ACCESS-OM2 = [MOM5.1 + CICE5.1 + OASIS3-MCT + YATM + JRA55v13-do]


  • 1° resolution, 50 levels
  • 252 cores
  • 48 yrs/day – 160 SU/yr
  • 5 InterAnnual Forcing (IAF) cycles complete
  • Numerous Repeat Year Forcing (RYF) cases


  • 0.25° resolution, 50 (KDS) levels
  • 1824 cores
  • 16 yrs/day – 2800 SU/yr
  • 5 IAF cycles complete
  • Additional IAF and RYF cases run


  • 0.1° resolution, 75 (KDS) levels
  • 5744 cores
  • 2.2 yrs/day – 63 kSU/yr (provided dt=600 sec)
  • Minimal config with 2064 cores, ~1 yr/day
  • 40-year RYF spinup with variable parameters, tweaks, date fixes during spinup.
  • Single IAF run from 1985 to 2017

For a full recount of today’s COSIMA meeting presentation, see the COSIMA Update slides.

COSIMA 2018 Workshop Details

The 2018 COSIMA workshop will be held on 7 & 8 May at the Australian Centre for China in the World on ANU Campus.

You can download the latest draft of the COSIMA Workshop Program (updated 3rd May). The program includes instructions for uploading your talk, guidelines on how to contribute to our discussion and some preparatory homework for anyone attending the COSIMA Cookbook Tutorial.

Please contact Andy Hogg or Andrew Kiss if you have any queries.


Over the last few months, COSIMA folks have been working hard on releasing our ACCESS-OM2 suite of models. The current status is that we have now completed a 500 year spinup for 3 different cases using the JRA55-do (Tsujino et al., 2017, personal communication) forcing dataset. Some preliminary results can be seen in the figures below. We are also spinning up a CORE-NYF comparison case. For a more complete analysis have a look in the COSIMA Cookbook.

Plans in the coming weeks are to finalise spinups of our 0.25° case (ACCESS-OM2-025), and to begin running our flagship 0.1° simulation, ACCESS-OM2-01.


COSIMA Linkage Project

The COSIMA consortium is partly supported by an ARC Linkage Project. This project is now active, following sign-off from all partners in December 2016. The management committee for the project formally met for the first time on Friday 9th December. Following are abbreviated minutes from the meeting which outline our plans for 2017:

  1. To advertise for a Postdoctoral/Research Fellow to coordinate COSIMA projects.
  2. To enhance the vertical resolution of existing MOM5-SIS 0.1° model, with no change in horizontal resolution;
  3. To enhance model bathymetry in coastal regions to take advantage of reduced minimum depth stemming from vertical resolution changes;
  4. To couple MOM5 with CICE at 0.1° using OASIS3-MCT (we have a contract with Nic Hannah from Double Precision to perform this work);
  5. To evaluate, refine, optimise and document the new model configuration.

Furthermore, we have confirmed that the 2017 COSIMA meeting will be in Sydney on May 25-26.

COSIMA I: Workshop Report

The first meeting of the Consortium for Ocean Sea Ice Modelling in Australia (COSIMA) was held in Hobart on the 26-27 May 2016. There we 38 attendees (20 of whom gave presentations), representing 10 different institutions. A full participant list is included at the end of this report.


The goal of the workshop was to formalise a consortium across universities, BoM, AAD and CSIRO to build global model configurations that will form the basis for high-resolution forecasting, reanalysis, process modelling and ultimately coupled climate modelling. The consortium recently received 4 years of funding from an ARC Linkage Project to build a model configuration which underpins a variety of applications.

Workshop Highlights

Presentations included an update on the current status and near-term plans of the major modelling groups, recent scientific advances using ocean and sea ice models and highlights of technical advances in model development. Some edited highlights are listed below:

  • Uptake of the MOM5 model for global applications has progressed well in the last 5 years (Bi). It is currently used broadly across climate and ocean-only configurations at both 1° (O’Kane) and 0.25° resolution (Spence, Holmes). Developments at 0.1° are proceeding (Zhang, Langlais, Matear, Chamberlain, Hogg).
  • Other models with a strong user base in Australia include ROMS (Galton Fenzi) for near-coastal and near-Antarctic applications, MITgcm for regional GFD-style simulations (Nikurashin) and NEMO (Alves) for seasonal prediction.
  • There is a strong need to consider vertical resolution in future high resolution model developments (Stewart).
  • The CICE sea ice model, while no longer under active development at LANL, continues to be the ice model of choice (due primarily to its superior ice physics over SIS 1; Heil, Reid, Bennetts). It is likely that it will remain the model of choice with applications using MOM5.
  • Our MOM-CICE implementations using OASIS3-MCT coupling, but it is not clear that this solution will scale to 0.1° and beyond (Hannah)
  • MOM6 is rapidly developing, and is beginning to gather users within Australia. It looks viable to use MOM6 for global models in the next year or two (Griffies, Gibson).
  • MOM6, being a C-grid model, presents some challenges when working with a B-grid ice model such as CICE. For the time being, it seems that the best ice model to use with MOM6 will be SIS2, which is actively being developed at GFDL to incorporate the vertical physics (Griffies).
  • Computational performance of these models shows that MOM5 is marginally faster than MOM6 (once accounting for vertical resolution differences) and that both models scale well on Raijin. NEMO is faster at small core counts but scales poorly (Ward).
  • Forced ocean-sea ice models should be transferring to the JRA-55 forcing set when possible (Marsland).

The COSIMA Community

There was significant support for the formation of a community of ocean-sea ice modellers around the COSIMA banner:

  • Need to create a website to outline COSIMA activities and developments. [We are in the process of acquiring the domain name, and ANU will fund hosting and a web development team to put together a skeleton site.]
  • We will formalise a code of ethics for COSIMA users to abide by, based on the DRAKKAR agreement. [Spence]
  • We aim to register members on the website, and create a mailing list
  • We will need to formalise the use of technologies to share code configurations, analysis tools and data. It may be possible to have a data project code on NCI to help with this.
  • We will release flagship configurations that are broadly supported by the community. The goal is to make the naming conventions consistent with the ACCESS community where possible, and to overlap with ACCESS developments where possible.
  • COSIMA will hold an annual meeting in the last week of May, for two days. Venue will rotate around the partner institutions. The focus of the meeting will be on science applications of ocean models, but will also include a technical component.
  • We will aim to have more regular communications, including newsletters and video meetings.
  • We will establish working groups within the community, along with a working group chair. Proposed groups include
    • Sea Ice Modelling [Heil]
    • Technical [Ward]
    • Linkage Project [Hogg]
  • The major gap in the community was identified to be sea ice modelling and forecasting. We will all look for opportunities to attract visitors and expertise in this space.

Linkage Project

One of the workshop goals was to receive advice from the community regarding the ARC Linkage Project designed to support COSIMA development activities. Major items of discussion were:

  • There was general agreement that the Linkage Project should fund both the development and evaluation of new model configurations. This point implies that we should equally fund the technical and postdoctoral position, despite the partial funding of the program.
  • In the first year we will look to upgrade current MOM5 implementations, focussing on the vertical grid and the incorporation of CICE.
  • In subsequent years we will look to adopt a MOM6 configuration.
  • As configurations develop and have been properly evaluated, they will be distributed to the community.
  • Suggestions on evaluation include using ESMVal.

Attendee List

(Where available, talk files are linked from the presenter’s name.)

Andy Hogg (ANU)
Gary Brassington (BoM)
Ben Galton-Fenzi (AAD)
Nic Hannah (Breakaway Labs)
Dave Bi (CSIRO)
Oscar Alves (BoM)
Max Nikurashin (UTas)
Simon Marsland  (CSIRO)
Kial Stewart (ANU)
Xuebin Zhang (CSIRO)
Terry O’Kane (CSIRO)
Paul Spence (UNSW)
Richard Matear (CSIRO)
Ryan Holmes (UNSW)
Clothilde Langlais (CSIRO)
Stephen Griffies (GFDL)
Angus Gibson (ANU)
Russ Fiedler (CSIRO)
Marshall Ward (NCI)
Justin Freeman (BoM)
Phil Reid (BoM)
Luke Bennetts (Adelaide)
Matt Chamberlain (CSIRO)
Scott Carpentier (BoM)
Fabio Dias (UTas)
Stephanie Downes (UTas)
David Gwyther (UTas)
Aidan Heerdegen (ANU)
Petra Heil (AAD & ACE CRC)
Andreas Klocker (UTas)
Andrew Lenton (CSIRO)
Jan Lieser (UTas)
Sebastien Moreau (CSIRO)
Siobhan O’Farrell (CSIRO)
Paul Barker (UNSW)
Mirko Velic (BoM)
Xiaobing Zhou (BoM)
Stefan Zieger (BoM)