Sea-ice motion and ocean currents have been studied on the basis of
drifter trajectories in the Baltic Sea, in the northern North
Atlantic, and in the Weddell Sea, Antarctic. Additionally,
meteorological observations, output from meteorological models, and
ice thickness and ice concentration data were utilised. The
relationship between the wind and the drift was particularly
investigated. Although advanced data analysis methods were applied in
the study, the parameters obtained from the simple linear relationship
between the wind and the drift velocities was found to be very useful
as well. The data analysis methods consisted of the optimum
interpolation of the drifter data, buoy array deformation estimation
at varying timescales, solving linear models using the orthogonal
distance regression and calculation of time-frequency diagrams with
wavelets.

The drift was often primarily driven by the wind, but in coastal
conditions of strong currents, tidal and inertial motions, and in the
presence of internal sea-ice force the wind dependence was reduced.
In the case of open water drift, the occurrences of drogue losses were
determined from modifications in the wind-drift
relationship. Otherwise, changes in the wind-drift relationship were
assumed to follow from changes in the drift characteristics and thus
revealing information about the momentum balance of the motion. A
relationship between the air-ice stress and the sea-ice velocity was
applied to estimate the ice-water drag coefficient, which is often a
poorly known variable. The sea-ice momentum balance in the Weddell Sea
in winter was found to be often between two forces, either between
air-ice stress and ice-water stress or between air-ice stress and
internal ice stress gradients, explaining the observed near-linear
relationship between the wind and the drift.

Sea-ice motion was simulated with various wind data and air-ice
stresses. The modelled drift signified the importance of accurate air
stress values particularly in non-coastal environments. The general
agreement between the simulations based on the observed surface wind
and the simulations based on the geostrophic wind, derived from air
pressure analyses of numerical models was good. When the drift was
simulated with wind stress parameterised in various methods, the best
result was achieved by the air drag parameterization depending on the
atmospheric surface layer stratification and ice conditions. This
parameterization produced significantly better simulated ice drift
outside the coastal zone than values with the same wind data, but with
a constant air drag coefficient.