Intercomparison of observed and simulated ice motion for a one year time series

Heat transfer and ice production in the ice-covered seas is controlled to a large extent by ice motion, which is in turn a function of winds and atmospheric boundary layer conditions. While sea-ice models appear to capture the general elements of this interaction, relatively little data have been available to verify model performance or to assess the spatial and temporal variability of ice motion. Several satellite remote sensing instruments provide the capability of detecting ice displacements. In combination with drifting buoy data, it is now possible to create detailed motion fields suitable for studying mesoscale responses of the ice pack to wind forcing. AVHRR, SAR, and drifting buoys have been merged using optimal interpolation techniques to generate daily gridded ice velocity fields for the Beaufort Sea from June 1992 thru July 1993. These motion fields are compared to simulations using a dynamic-thermodynamic ice model with different ice rheologies and heat transfer processes. The motion fields are discussed in terms of responses to various atmospheric synoptic regimes, and sources of differences between observed and simulated motions are considered. The remotely-sensed ice motion fields show seasonal and temporal variability not apparent in the relatively widely-spaced buoy network. Ice velocities and divergence rates agree in general with the simulations. The cavitating fluid and viscous-plastic ice rheologies yield similar drift directions, but can differ substantially in drift speed under strong winds. The overall utility of AVHRR-based ice motion fields is discussed. Possibilities of assimilating the daily motion fields directly into the ice model to assist in refining the thermodynamic portion of the model is described