Deriving Sea-Ice Freeboard Height Distributions and Estimates of Ice Thickness From ICESat/GLAS Laser Altimetry

The state of a sea ice pack is well described by distributions of sea thicknesses obtained over spatial scales such as 50 km by 50 km. Information that can be derived from thickness distributions includes the mean sea-ice thickness and the ratio of thicker multi-year ice to thinner ice types. Thickness distributions as a function of time also provide information on the time variability of these parameters and ice growth and decay rates. However, large-scale spatial and temporal measurement of ice thickness distributions has been an elusive goal. The total ice thickness consists of the draft, which is the ice portion below sea level, and the freeboard, which is the ice portion above sea level (defined here to include the snow cover). The total ice thickness can be derived from measurements of the freeboard height using estimates of the snow cover thickness and estimates of the snow and ice densities. This paper describes work to derive freeboard height distributions from measurements of surface elevations over sea ice by the laser altimeter on ICESat. ICESat measures the mean surface elevation over footprints 70 m in diameter spaced at 170 m along track. The along track accuracy of the elevations over surfaces with low slope is in the range of 5 to 10 cm after editing of data with pulse saturation effects and strong atmospheric forward scattering. Estimates of the freeboard height distribution are derived by calculating 50 km running means of the measured elevations, plotting the distributions of the deviations of the elevations from the centered mean for 50 km segments, fitting the distributions with a Gaussian curve, and taking the 10% point of the Gaussian fit as the open water/sea level. The mean freeboard height is then calculated from the distributions with respect to the estimate of sea level. The mean sea ice thickness is calculated from mean freeboard using snow cover climatology. The methodology assumes the existence of some open water or very thin sea ice in each segment and other approximations about the shape of the thickness distributions. Several variations of the method such as changing the averaging length and multi-track averaging in grid cells are described. Preliminary maps of sea ice thickness show some reasonable agreement with known sea ice thicknesses.