Multifrequency passive microwave observations are reported from surface based field experiments carried out over multiyear sea ice in the eastern Arctic Basin, the Beaufort Sea, the Canadian archipelago and the northern Greenland Sea. Dual polarized brightness temperatures at 6.7, 10, 18.7, 37, and 90 GHz were obtained at selected multiyear ice sites covering a wide range of surface properties. Considerable variation was found in the observed spectra over short distances on individual floes under winter conditions where values of <ev(37 GHz)--ev(18 GHz)>/ <ev(37 GHz)+ev(18 GHz)> ranged from nearly zero to approximately -0.08. Concurrent surface characterization observations consisting of temperature, density, salinity, and grain size profiles in the snow and the upper meter of the ice were carried out frequently in conjunction with Cold Regions Research and Engineering Laboratory personnel. The present analysis shows that the magnitude of the spectral gradient of emissivity is directly related to the existence and thickness of a decomposed surface ice layer with very high porosity. Spectra for melt ponds with a frozen surface layer closely resembled those of lake ice and show a positive spectral gradient. As soon as the ponds were completely frozen, their spectra became nearly the same as those for first-year ice. Fine-grained winter snow produced a mild decrease in the emissivities at 90 GHz with increasing snow thickness from 0 to 0.2 m at which point the snow appears to be optically thick. During the melt season the ice emissivities ev depend most strongly on whether the surface layers are experiencing temporary freezing or not.

Undel melting conditions, ev depends on the liquid water content only over a very small range (e.g., 0--5% by mass at 90 GHz). If the liquid water in the surface layers is freezing, which can occur at intervals throughout the melt season, the spectral gradient is negative with a slope which depends on the thickness of the refrozen layer. The variance among emissivity spectra for multiyear ice appears to be influenced primarily by the distribution of melt ponds and by the presence of significant amounts of scattering inhomogeneities greater than 1 mm in diameter in the snow and in the upper 20--30 cm of the ice. We expect the signatures of multiyear ice to be regionally dependent, perhaps yielding environmental information about the Arctic ice pack. ¿ American Geophysical Union 1992