Thanks Adele, Pearse, Andrew, Will and others—I think this is a really great discussion.
For what it is worth, I wanted to strongly support Andrew’s point about enabling separate CICE instances, or at least hemisphere-specific CICE parameter control, for global ACCESS-OM3. From a Southern Hemisphere sea-ice perspective, I think this is more than a convenience issue. Arctic and Antarctic sea ice have different dynamical regimes, different snow/thermodynamic regimes, different coastal geometry, and different fast-ice formation mechanisms. A single global sea-ice parameter set is therefore a fairly blunt instrument if we are trying to improve both hemispheres simultaneously.
A concrete example is fast ice. The standard CICE seabed or basal-stress approach is physically useful where grounded pressure ridges interact with shallow bathymetry, but that is a very rare (if at all occurrence) in Antarctica. Around Antarctica, fast ice is often maintained by grounded icebergs, ice shelves, and coastline geometry (similar to the Arctic), however, this all occurs in water depths where a shallow-water grounding scheme is not the right first-order mechanism. In my own work with Antarctic CICE configurations, simply having seabed stress enabled is not necessarily benign at 1/4-degree: it can introduce awkward behaviour in a small number of grid cells, depending on the bathymetry/mask representation. This may or may not remain an issue at 1/10-degree, but I do not think we should assume that without testing.
For that reason, I think a 10-year plan should include the capability to tune, perturb, and evaluate CICE independently in each hemisphere. Separate CICE instances would be a clean way to do this, particularly if it also helps isolate NH and SH choices around rheology, tensile strength, landfast-ice parameterisations, seabed stress, floe-size/wave-ice interactions, and snow/thermodynamic parameters. If separate instances are technically too large a step, then a useful minimum would be a hemisphere-aware CICE configuration pathway, where selected namelist parameters and diagnostics can be controlled independently north and south of the equator.
This also connects to the broader question of what COSIMA wants ACCESS-OM3 to support. As a seperate thought than the one just give. Climate-scale projection is obviously central, but I think we might also name deterministic and ensemble sea-ice modelling as explicit scientific capabilities. Many sea-ice questions are not only about producing one best long climate run; they require controlled perturbation experiments, parameter sensitivity studies, ensemble attribution, process isolation, and forecast-like or event-scale experiments. This is particularly true for Antarctic sea ice, fast ice, wave-ice interaction, polynya/ice production processes, and the coupling pathways to BGC, ecosystems and ice shelves.
So my suggested addition would be something like:
-
support hemisphere-specific sea-ice model configuration and tuning in global ACCESS-OM3, including the option of separate NH and SH CICE instances;
-
prioritise Antarctic sea-ice process fidelity, with waves treated as a key enabling process rather than necessarily as the headline science question;
-
include deterministic and ensemble ocean-sea-ice modelling capability alongside climate-scale simulation, so that ACCESS-OM3 can support process attribution, uncertainty quantification, perturbation experiments and forecast-like sea-ice science.
I think this would sit naturally alongside the existing priorities around Antarctic sea ice, ice shelves, waves, improved parameterisations/tuning, and simplified atmospheric coupling.
… thanks for listening to my Tuesday night thoughts …
