Uncertainty of Arctic summer ice drift assessed by high-resolution SAR data

Time-space varying uncertainty maps of monthly mean Arctic summer ice drift are presented. To assess the error statistics of two low-resolution Eulerian ice drift products, we use high-resolution Lagrangian ice motion derived from synthetic aperture radar (SAR) imagery. The Lagrangian trajectories from the SAR data are converted to an Eulerian format to serve as reference for the error assessment of the Eulerian products. The statistical error associated with the conversion is suppressed to an acceptable level by applying a threshold for averaging. By using the SAR ice drift as a reference, we formulate the uncertainty of monthly mean ice drift as an empirical function of drift speed and ice concentration. The empirical functions are applied to derive uncertainty maps of Arctic ice drift fields. The estimated uncertainty maps reasonably capture an increase of uncertainty with the progress of summer melting season. The uncertainties range from 1.0 to 2.0 cm s(-1), which indicates that the low-resolution Eulerian products for summer seasons are of practical use for climate studies, model validation, and data assimilation, if their uncertainties are appropriately taken into account.