The deformation of sea ice is an important element of the Arctic climate system because of its influence on the ice-thickness distribution and on the rates of ice production and melt. New data obtained from the Radarsat Geophysical Processor System (RGPS) using satellite synthetic aperture radar images of the ice offers an opportunity to compare observations of the ice deformation to estimates obtained from models. The RGPS tracks tens of thousands of points, spaced roughly at 10-km intervals, for an entire season in a Lagrangian fashion. The deformation is computed from cells formed by the tracked points
IntroductionSea-ice deformation is a fascinating and unique component of the Arctic geophysical environment. The deformation rate of pack ice, determined from the spatial gradients in the velocity, is a key parameter in determining the formation of leads and open water, as well as the formation of rafted and ridged ice. The amount of open water and the thickness of the ice are key parameters in the climate system because of the strong effect ice thickness has on albedo, heat exchange, and ice growth rates. In sea-ice models, the ice velocity is established through a balance of forces that depends on the winds and currents (forcing), the model state (mean thickness), the model physics (drag and constitutive laws), as well as the model resolution. Accurate modelling of the ice velocity and deformation rate is essential if ice is to be properly represented in climate models.It is possible to test the mean motion of ice calculated from models by comparing it to the trajectories of buoys and drifting ice stations (e.g., Thomas, 1999; Zhang et al., 2003;Meier et al., 2000). But it has not been possible to test ice deformation computed in the models adequately for lack of an appropriate comparison dataset. The buoys routinely deployed in the Arctic are too few and too widely separated to measure ice deformation accurately at small scales. Thomas (1999) compared buoy-derived and model-derived deformation estimates at large scales, 400 to 600 km. The model he used is similar to the one used here. He investigated different model wind drag formulations as well as a best-fit linear model based only on the geostrophic wind. He found modest correlations for vorticity and shear but the correlations of divergence were statistically insignificant. The best correlations were found for the simple linear model. There was insufficient data to determine the spatial or temporal variability of the correlations. However we now have a new dataset, based on the tracking of ice motion in satellite radar backscatter images, that offers an excellent opportunity to compare modelled and measured deformation rates over a wide area of the Arctic Ocean, for all seasons of the year, and over a wide range of scales.This new dataset is from the Radarsat Geophysical Processor System (RGPS) (Kwok, 1998; Kwok et al., 1999). The RGPS uses synthetic aperture radar images from the Canadian Radarsat satellite to track thousands of points during a...