Thickness and roughness variations of Arctic multi-year sea ice

Three lines on a multi-year ice floe in the Beaufort Sea, 200 m, 110 m and 76 m long, were profiled by level to obtain ice surface elevation (freeboard) and snow depths and drilled to obtain ice thickness at two meter intervals. Three models were then constructed to identify the relationship between surface elevation and ice thickness so that top roughness, bottom roughness and thickness could be obtained from a measure of the surface elevation alone: 1. At each profile point isostatic balance was assumed and the thickness calculated from the freeboard by using the average observed ice density of 910 kg/m³. 2. A linear relationship was assumed between freeboard and density with the thicker ice having a lower density (more brine drainage) than thinner ice. Isostatic balance was then combined with the adjusted density to obtain the thickness at each profile point. 3. Finally it was assumed that although the ice was, on the average, isostatically compensated, this compensation does not occur on a point to point basis. Instead, the ice "below" each point was distributed over some lateral distance and was not concentrated directly below the profile point. This is essentially the same as the Wittmann-Markarov model for a pressure ridge and is equivalent to operating on the freeboard data with a linear filter to obtain the bottom profile. The first assumption overestimated the thicker ice and underestimated the thinner ice with errors in the estimate of thickness exceeding one meter. The last two models gave roughly similar results, with prediction errors of0.40m. The prediction accuracy is limited by the fact that the high-frequency roughness of the top ice surface (wave lengths less than 10 m) accounts for considerable variance which does not correlate with the bottom surface. An estimate is also made of the error in predicted ice thicknesses based only on the elevation of the upper ice surface as obtained by airborne laser. The error in predicted thicknesses - - only increased by about 10% over the error obtained from surface observations. Therefore, surface profiles obtained by an airborne laser may be a useful source of ice thickness information.