Sea ice is complex…
When sea water freezes it forms sea ice, a composite of ice and brine. Sea ice exhibits varying structural, thermodynamic and mechanical properties across a range of length- and time-scales. It can be subcategorised into numerous different types of sea ice depending on where is grows and how old it is.
However, climate models do not simulate the evolution of floes (they model floes as cylindrical) or the floe size distribution, which has implications for ice melt rates and exchange of heat with the atmosphere and ocean. Sea ice also hosts algae and small organisms within brine channels in the ice, which can be important for nutrient cycles. This is a developing area of earth system modelling.
How much complexity do global climate models need to sufficiently model the interactions of sea ice with the ocean and atmosphere?
The representation of sea ice in global climate models is actually very simple, with minimal sea ice types and thickness categories. The main important feature of sea ice for global climate models is its albedo, which is much greater than that of open water, making it important for the surface energy balance. So, it is important to get the correct area of sea ice. Global climate models need sea ice:
- to get the correct heat exchange with the atmosphere and ocean
- to get a realistic overturning circulation in the ocean.
- because salt release during sea ice growth is important for the ocean salinity structure, and therefore important to get the correct amount of sea in/near deep water formation sites.
- sea ice is not important for sea level projections.
So, do the complex features of sea ice matter, or are simple parameterisations sufficient?
Schematic showing some dynamic features of sea ice 3.
Which leads to a lot more questions…
- Where does the balance between sufficient complexity and computational cost lie?
- Does adding extra model complexity actually make it harder to understand what the model is doing and therefore to interpret the results?
- Do climate models need any further improvements to sea ice in order to better simulate global climate? There is still large uncertainty surrounding other climate model components, such as clouds and ocean eddies, which are believed to explain a lot of the discrepancy between models and observations, particularly in the Southern Ocean.
A lot of these questions depend on the scientific question that is being asked. And the question is not necessarily always ‘how is global climate going to change in the future’. Sea ice is fascinating because of its complexity, and there are still many interesting questions to investigate, hopefully before it all melts!
Images clockwise from top left: grease ice 4, pancake ice 5, surface melt ponds 6, ice floes 7.
The Future Developments in Climate Sea Ice Modelling Workshop
This blog stems from a one day workshop I attended on ‘Future developments in climate sea ice modelling’ at the Isaac Newton Centre as part of a four month programme on the ‘Mathematics of Sea Ice Phenomena’. The format of the day was that three different strands of sea ice researchers gave 40 min talks giving their strand’s point of view of current sea ice developments and what the focus should be for sea ice modelers, each followed by 40 mins of open discussion with the audience.
The three (very good!) talks were:
- Dirk Notz: What do climate models need sea ice for? A top-down, system level view of what sea ice models should produce from the perspective of a climate modeller.
- Cecilia Bitz: What sea ice physics is missing from models? A bottom-up view of what is missing from current sea ice models from the perspective of a sea ice scientist.
- Elizabeth Hunke: What modelling approaches can be used to address the complexity of sea ice and the needs of climate models?