Sea-ice elevation profiles and thickness measurements have been collected during summer 2011 in the Central Arctic. These two different data sets have been combined in order to obtain surface and bottom topography of the sea-ice. From the bottom profile, the keels of ridges are detected. Then, a parameterization of oceanic drag coefficients that accounts for the keels depth and density is applied. The calculated oceanic drag coefficients are highly variable (between about 2 × 10-3 and about 8 × 10-3) within the range of observed values. In order to estimate the contribution of variable drag coefficients on the Ekman pumping, the calculated drag coefficients are used in an idealized model experiment, where sea ice is drifting at constant velocity on an ocean at rest. The resulting variations of the Ekman vertical velocity are in the same order of magnitude as for variable ice velocity at the surface. In most state-of-the-art general circulation models, the variations of drag coefficients are not taken into account. The simple experiment carried out in the present study suggests that neglecting this contribution can lead to an incorrect representation of the momentum exchange between ice and ocean and to an underestimation of the Ekman pumping, with consequences for the large scale ocean circulation.