Estimating turbulent exchange over snow-covered surfaces--in energy budget studies or in atmospheric models, for example--almost always invokes Monin-Obukhov similarity theory. Through this theory, we can predict the fluxes of interest--momentum, sensible heat, and latent heat--from measured or modeled values of routine quantities such as wind speed, temperature, and humidity. The catch, though, is that, to use Monin-Obukhov similarity theory, we must also be able to predict the aerodynamic properties of the surface: namely, its roughness lengths for wind speed (z₀), temperature (z_T), and humidity (z_Q).

Here we report on parameterizations for z₀, z_T, and z_Q that we have developed from eddy-correlation and other supporting measurements over snow-covered sea ice during two long polar experiments: one in the Antarctic (Ice Station Weddell), and one in the Arctic (SHEBA, the Surface Heat Budget of the Arctic Ocean). These two experiments provided nearly ideal conditions for studying turbulent exchange over snow: The sites were expansive, flat areas with continuous snow cover, and both were at least 300 km from any topography that might have complicated the atmospheric flow.

On the basis of our analyses, we parameterize z₀ as a continuous function of the friction velocity u_*. z₀ displays three regimes. In low winds, the flow is aerodynamically smooth, and z₀ scales as 1/u_*. In high winds, the saltation layer sets the roughness length, so z₀ scales with u_*². In between these two extremes, the macroscale or "permanent" roughness of the surface determines z₀.

Our z_T data tend to corroborate Andreas's (1987, Bound.-Layer Meteor., 38, 159-184) theoretical model for this quantity. Our z_Q data provide the first meaningful test for Andreas's prediction of this roughness length. Although the z_Q data do not fit the model as well as the z_T data do, they are still generally within an order of magnitude of its predictions--a result we consider encouraging since z_Q is so difficult to measure at low temperatures.