S. G. Warren
This review will cover new measurements and theoretical understanding of the optical properties of water-snow on Earth since my 1982 review, and of the optical constants of water-ice since my 1984 review.
New measurements are available in the near-UV, where none were previously available; in the near-IR, where previous measurements had large uncertainties; and in the visible, where absorption is now found to be a factor of 4 weaker than earlier reported. However, no new measurements have been reported in the microwave; there is still a factor-of-ten uncertainty in the centimeter-wave region.
Radiative transfer modeling has proven successful in explaining the most accurate observations of spectral albedo. The near-IR albedo is sensitive to grain size and zenith angle; the visible albedo is sensitive to snow depth and impurities. The models use "equivalent spheres" to represent nonspherical snow grains. It is argued that the model snowpack should contain both the same total mass of ice and the same total surface area as the real snowpack, which requires a greater number of grains than in the real snowpack; the surface-to-volume ratio is conserved.
To calculate accurately the vertical distribution of absorbed solar radiation in snow, it is important to divide the solar spectrum into several narrow wavelength bands. The visible light penetrates deeply into the snow but is not absorbed and thus eventually re-emerges upward from the surface if the snow is sufficiently thick. Essentially all of the absorbed radiation is therefore in the near-IR, and this radiation is absorbed in the top few millimeters.
The results of spectral albedo models have been parameterized for the two broad wavelength bands used by general circulation climate models, as functions of grain size, solar zenith angle, atmospheric transmittance, snow depth, and impurities.
Only a few millimeters of snow can dramatically raise the albedo of thick sea ice.
The visible albedo is sensitive to small amounts of absorptive impurities. Per unit of mass, soot is 50 times as effective as clay-mineral soil dust, and 200 times as effective as volcanic ash, at lowering the albedo of snow. There is often sufficient soot in Arctic and midlatitude snow, but not in Antarctic snow, to reduce the spectrally-averaged albedo by 1-2%.
Snow emissivity is in the range 0.96-1.00 for all grain sizes and viewing angles. Radiative transfer models predict slightly lower emissivity for the finest-grained snow, but a recent laboratory experiment found the opposite result.
Snow is the most isotropic reflector of all natural surfaces on Earth, but still exhibits substantial anisotropy. The anisotropy is greater in the near-IR than in the visible, because as absorption increases the bidirectional reflectance pattern becomes dominated by the single-scattering phase function. The normal forward-scattering pattern of a flat snow surface can be altered considerably by natural forms of surface roughness.
Introducing clouds over snow causes an increase of planetary albedo in the near-IR, because cloud particles are smaller than surface snow grains. The visible albedo is unchanged, but clouds do often alter the observed visible brightness in particular directions either positively or negatively, by hiding the snow surface roughness and by scattering from their own irregular top surfaces.
Contact Information:
Stephen G. Warren
Department of Atmospheric Sciences
University of Washington, Box 351640
Seattle, Washington 98195 USA
sgw@atmos.washington.edu