Sea ice, motion tracking, SAR, kinematics, pattern matching, earth observation. - For the last thirty years, sea ice kinematics have been studied continuously over large areas, but with coarse spatial resolutions. In the framework of this dissertation, the author implemented and modified an algorithm for the observation of high resolution sea ice motion from satellite imagery. Employing the algorithm he shows that sea ice can change its drift patterns completely within less than a day if wind conditions are highly variable. Furthermore, the author compares the ice drift derived from satellite imagery with model simulations. Another part of this thesis focuses on the automated evaluation of the retrieved drift data, since sea ice regions are often difficult to access, in-situ measurements to validate the calculated ice drift are sparse. Here, a new reliability estimate is introduced, combining the correlation coefficient with measures of selected texture parameters and the results of a reversed drift calculation.