Parameterization of Arctic sea-ice surface roughness for application in ice type classification

Abstract

Statistics of Arctic sea-ice surface roughness have been investigated in order to improve classification of ice-thickness regimes. The data consist of surface roughness and thickness profiles, acquired simultaneously by helicopter-borne laser altimetry and electromagnetic induction sounding. Five thickness classes were identified using the modal thickness as a criterion. For each class, the statistical properties of the surface roughness profiles were analyzed. A classification algorithm was designed, which assigns profiles to the thickness classes on the basis of a set of selected statistical roughness parameters. The algorithm was applied to profiles of different lengths. Best results were obtained for 2 km long profiles, for which it was possible to discriminate well between thick first-year and multi-year ice, and to distinguish these classes from thinner ice. The classification rule was tested on data obtained under winter and summer conditions. The results suggest that statistical surface roughness properties are different for thinner and thicker ice classes. However, individual thin-ice classes cannot be discriminated on the basis of the selected roughness parameters.