An intercomparison of Arctic ice drift products to deduce uncertainty estimates

An intercomparison of four low-resolution remotely sensed ice-drift products in the Arctic Ocean is presented. The purpose of the study is to examine the uncertainty in space and time of these different drift products. The comparison is based on monthly mean ice drifts from October 2002 to December 2006. The ice drifts were also compared with available buoy data. The result shows that the differences of the drift vectors are not spatially uniform, but are covariant with ice concentration and thickness. In high (low) ice-concentration areas, the differences are small (large), and in thick (thin) ice-thickness areas, the differences are small (large). A comparison with the drift deduced from buoys reveals that the error of the drift speed depends on the magnitude of the drift speed: larger drift speeds have larger errors. Based on the intercomparison of the products and comparison with buoy data, uncertainties of the monthly mean drift are estimated. The estimated uncertainty maps reasonably reflect the difference between the products in relation to ice concentration and the bias from the buoy drift in relation to drift speed. Examinations of distinctive features of Arctic sea ice motion demonstrate that the transpolar drift speed differs among the products by 13% (0.32 cm s(-1)) on average, and ice drift curl in the Amerasian Basin differs by up to 24% (3.3 x 10(4) m(2) s(-1)). These uncertainties should be taken into account if these products are used, particularly for model validation and data assimilation within the Arctic.