Estimating Arctic sea-ice deformation using data assimilation

In response to the variations in large-scale atmospheric circulation on interannual and decadal time scales sea-ice motion over the Arctic Ocean exhibits a strong model structure composed of coherent and recursive anomalous patterns such as ones associated with the NAO and the AO. The question arises as to if and how these variations in sea-ice motion would lead to change in sea ice thickness, a critical climatic parameter. It has been argued and assumed in model calculations that aside from thermodynamic influence through atmospheric (e.g. changes in precipitation and a poleward heat flux) and oceanic (e.g. an oceanic heat flux) processes a dynamics process directly linking changes in sea ice thickness and sea ice motion is horizontal ice deformation and its small scale manifestation of ridging. This process is likely controlled by uniaxial contraction especially against the coastline. However this is a rare class of events as on the average the Arctic ice motion is nearly non-divergent. In order to understand and quantify this mechanism it is thus necessary to improve our estimates on sea ice deformation, our first step toward the construction of deformation climatology.

We present preliminary results from our analyses on improved ice motion data by a means of data assimilation which show that this event occurs at a higher frequency during the midwinter in the area near the Canadian Archipelago and Greenland. We interpret this as direct evidence for the importance of the relative frequency of the uniaxial contraction, which also sets daily to synoptic time scales as a relevant time step for studying this process. This knowledge is further used to construct an adjoint sea-ice model that is designed for providing improved estimates on sea-ice deformation. In particular we discuss the implication of this temporal scale in both thermodynamic and dynamic model components.