REFLECTANCE OF SNOW AS MEASURED INSITU AND FROM SPACE IN SUB-ARCTIC AREAS IN CANADA AND ALASKA

Quantitative measurement of snow reflectance from space has been possible since the launch of the Landsat Thematic Mapper (TM) in 1982. Efforts to validate the accuracy of Landsat derived reflectances are ongoing to enable mapping global snow reflectance, and eventually snow albedo, using data from future satellite sensors. Visible and near-infrared measurements of clean snow and dirty snow (both actively melting) were acquired near Yellowknife, N.W.T., Canada, in May 19" using a portable spectrometer, and the Landsat TM. Similar measurements were also acquired over snow covered glaciers in Alaska in August 1987. The in situ reflectance at nadir of the dirty snow was approximately 30% lower in the visible part of the spectrum than was the reflectance of the cleaner snow near Yellowknife. The shapes of the reflectance curves are also different, the curves for the dirty snow being flatter than the curves for the clean snow. The shape of the reflectance curves of actively melting snow near Yellowknife versus fresh snow as measured in Alaska, was similar in the visible region of the spectrum, but diverged in the near-infrared region, with the reflectance of the melting snow decreasing moreso than for fresh snow. In Alaska, the off-nadir look angle measurements of snow consistently exceeded 1.0 because of enhanced forward scattering from the snow at off-nadir angles when the sensor is facing the Sun. TM-derived reflectances acquired near Yellowknife are not directly comparable with in situ measurements because a mixture of cover types is included within each 30 x 30 m pixel. Because the variability in solar zenith angle causes major differences in absolute reflectance, the shape of the reflectance curves and the anisotropic reflectance properties are found more useful in differentiating snow type and moisture state than are actual reflectances.