DETAILED MODELS OF SEA ICE DYNAMICS
Daniel Feltham and Alexander Wilchinsky
Sea ice forms when seawater freezes at the surface of the Earth's polar oceans. Sea ice floats on the ocean, is typically several metres in depth and, at its maximum extent, covers about 10% of the Earth's surface.
Sea ice is important to the climate because it forms a barrier to the transfer of moisture, heat and momentum between the ocean and atmosphere. In particular, it has a higher albedo than the ocean and plays an important part in the well-known albedo feed-back mechanism. Sea ice is also important because of its role in the thermohaline circulation of the Earth's oceans: As sea ice forms, it releases heavy, salty water (brine) that sinks to form bottom water and, when sea ice melts, it freshens the upper ocean.
Sea ice dynamical models model the transfer of momentum between the atmosphere, ocean and sea ice. This momentum transfer is crucial to determining the extent, concentration and thickness of sea ice and thence the energy fluxes between atmosphere and ocean and sea ice melt/freeze rate.
Much of the sea ice cover consists of ice floes with horizontal sizes of 0.1-10 km, interspersed with thinner ice or open ocean with widths of 0.1-2 km and lengths that span many floes. The sea ice cover is constantly deforming, for example through rafting, bending, and breaking to form pressure ridges and diverging, often along shear lines, to form linear regions of open ocean called leads. Sea ice dynamical models have been studied for about 50 years although they have not drastically changed their form since the great improvements made during the joint American-Canadian, Arctic Ice Dynamics Joint EXperiment during the 1970s. Since AIDJEX, sea ice dynamical research has focussed on the most appropriate models of sea ice rheology, the most well-known being Hibler's viscous-plastic model with elliptical yield curve.
(Sanderson, T.J.O., 1988, Ice mechanics - risks to offshore structures. Graham & Trotman, London.)
Despite great successes, there are fundamental problems with sea ice dynamical models that plague their useful application in Global Circulation Models. Specifically, ad hoc tuning of model parameters such as air to ocean drag ratios and ice strength are employed in order to achieve qualified success in reproducing data sets of sea ice extent, concentration and buoy motion (not, as yet, ice thickness - see Representation of sea ice in regional/global climate models). Parameter tuning is required because current sea ice models do not resolve physical processes below grid sizes of 100 km. Such tuning can, at best, provide a good fit to reality over limited periods of time. Accurate simulation over decadal and longer timescales requires models with parameters that can be independently assessed.
The lack of sub-100 km scale model physics is compounded by the fact that, because ice floes are of typical dimension of 0.1-10 km, the continuity assumption of existing models breaks down below 100 km so that observed discontinuities in, e.g., ice velocity, thickness and motion cannot be represented.
Although these problems have been recognised for almost three decades, recent advances in numerical climate prediction make study of the impact of improved sea ice physics especially relevant. For example, HadGEM1, the UK Meteorological Office Hadley Centre GCM under current development, will include one of the most comprehensive, elastic-viscous-plastic models of sea ice rheology. Further improvements to sea ice dynamical models requires fundamental advances in understanding sub-grid scale physics.
The aim of this project to take a fundamental, continuum mechanical approach to the dynamics of sea ice that necessitates a partial return to the insights developed during AIDJEX. However, many of the ideas developed then and subsequently can be substantiated through modern developments in satellite remote sensing of the polar regions. At the heart of our approach is the recognition that sea ice is a granular material. We are developing rheological models of the sea ice cover that link its large-scale response to the interactions between individual floes, air drag and the ocean. We see this as the most promising way to remove undesirable tuning of model parameters and to quantify difficulties associated with treating sea ice as a continuum. Among the techniques we are using are analytical analysis, numerical simulation, `toy' and thought experiments. These techniques help us gain insight into the best ways of modelling the ice cover.
For a detailed review (195 pages) of models of sea ice dynamics, which includes the contributions of, for example, Pritchard, Erlingsson, Tremblay and Mysak, Ukita and Moritz, Gray and Morland, Hibler and Schulson and Coon, contact Daniel Feltham.