New models of sea ice.    It has been proposed that GCMs should incorporate a description of sea ice that captures the discrete nature of the ice cover. Indeed rather than a continuous medium sea ice is formed of a network of fractures (leads) that delineate polygonal ice blocks of various sizes (floes) reminiscent of granular media (figure 4). At the Center for Polar Observation and Modeling (CPOM) we have developed new theoretical models of sea ice. First we have used a discrete element model that embodies the granularity of the ice pack by explicitly considering individual multiyear floes in a matrix of first-year ice floes, and showed that under realistic wind forcing sea ice becomes progressively anisotropic with diamond shaped floe aggregates. Based on this observed sub-continuum scale anisotropy of the ice cover we have derived and implemented a new anisotropic rheology into a GCM and tested its impact on the Arctic sea ice [4].
Figure 6: Long range redistribution of stress [1].
Finally an exciting new direction of research would consist in apply- ing the novel theoretical results de- veloped in the framework of the me- chanical properties of glassy materi- als to the geophysical sciences and particularly to the rheology of sea ice. In these models [1] flow occurs through a succession of global elas- tic deformations and localized plas- tic rearrangements associated with a microscopic yield stress (figure 6). These localized events induce long-
range elastic modifications of the stress over the system, thereby creating long-lived fragile zones where flow occurs. Flow in these systems is thus highly cooperative and spatially heterogeneous: a dynamically active region will induce stress fluctuations of its neighbours and thus a locally higher rate of plastic rearrangements. Correlations between plastic events are accordingly expected to exhibit a complex spatiotemporal pattern. Early studies with a similar elasto-plastic rheology ap- plied to the Arctic sea ice have produced promising results. In particular the inclusion of the long range elastic redistribution of internal stress in the ice seems to confer a better representation of the degree of localization of the deformation and to reproduce more realistically the shear pattern seen on figure 3.