Effect of environmental conditions on observed, modeled, and assimilated sea ice motion errors

• Published in: Journal of Geophysical Research. C. Oceans 2003-05, Vol.108 (C5), p.21.1-n/a
• Main Authors: Meier, Walter N., Maslanik, James A.
• Format: Article
• Language: English
• Subjects: data assimilation; Earth, ocean, space; Exact sciences and technology; External geophysics; Geophysics. Techniques, methods, instrumentation and models; Marine; passive microwave remote sensing; Physics of the oceans; Sea ice; sea ice modeling; sea ice motion
• ISSN: 0148-0227; 2169-9275; 2156-2202; 2169-9291
• DOI: 10.1029/2002JC001333

Previous evaluations of ice motion have primarily consisted of comparisons with in situ observations and general sensitivity studies with wind speed and direction. An assimilation framework allows the quality of ice motions to be studied as an explicit function of model and model forcing parameters. Here we investigate the effects of local conditions on remotely sensed, modeled, and assimilated motions. We use local conditions, as given by the ice model, to examine ice motion errors based on (1) proximity to the coast, (2) ice thickness, and (3) wind‐forcing. In comparison with buoy observations, both modeled and remotely sensed ice motions have lowest errors in moderately thick ice, in low winds, and away from coastal regions. Wind speed plays the largest role in the errors, and higher winds speeds lead to substantially higher motion error standard deviation. However, higher wind speeds also increase the correlation of modeled and remotely sensed motions with buoy observations. An optimal interpolation assimilation methodology, which led to substantial improvements in ice motion quality on a regional and seasonal basis, was employed in this study. Assimilation improved the quality of the ice motions (lower error, higher correlation) throughout most ranges of wind speed and ice thickness and both near the coast and in the central ice pack. This study suggests that adjusting assimilation weights based on local model conditions could efficiently yield even higher quality assimilated motion fields.