This paper uses the Winter Weddell Sea Project 1986, winter Antarctic data set to (1) describe the nature of observed sea ice drift and momentum exchange and (2) determine relevant drag coefficients (linear and quadratic) and parameter values for three formulations of the momentum balance. The large-scale mean divergence of the ice justifies, with some penalty, use of the steady free drift equation in which the air-ice stress is balanced by ice-ocean drag and the Coriolis force. Three forms of the free drift equation are considered: (1) stresses are parameterized with a quadratic drag law, (2) stresses are parameterized with a linear drag law (useful because of its analytically manageable form), and (3) the Coriolis force is ignored (owing to the thin, 0.6-m ice), so ice speed is proportional to wind speed at a specified angle. All three formulations simulate the observed ice drift with the same degree of accuracy. The linear drag law is an excellent approximation to the quadratic law over a broad range of forcing only when the air-ice and ice-water stresses are both parameterized using the linear law (otherwise the ice-water drag coefficient is a nonconstant function of wind speed). The linear drag coefficient and constant of proportionality relating ice speed to wind speed can both be computed directly from knowledge of the quadratic values. These calculations result in estimates within ≲2% of the optimum fitted values. Because of the ~95% ice coverage, ice interaction is frequently significant. During such periods, the ice-water drag coefficient represents an ''effective'' drag, artificially inflated to include the forces arising from this interaction.

We break the ice drift data into 6-hour nonoverlapping windows to allow isolation of periods of true free drift. Both true drag coefficient values and effective values are then estimated. The effective values show a strong correlation to the 4-day average large-scale ice divergence. They also show that the ice-water stress is typically ~1/3 the air-ice stress, indicating a significant role of ice interaction (free drift still provides an excellent parameterization of ice drift but at the expense of neglecting the details of these physics). The optimum quadratic drag coefficient is 1.62¿10-3 with turning angle 15.2¿; the effective value is 3.22¿10-3 with turning angle 18.1¿; the linear drag coefficient is 0.80¿10-3 with turning angle 18.1¿; the effective value is 1.48¿10-3 with turning angle 19.6¿, and in general, the ice drifts at ~3% of the wind speed, ~23¿ to its left. ¿ American Geophysical Union 1990