Satellite discrimination of snow/cloud surfaces

Differentiation between cloud cover and snow surfaces using remotely sensed data is complicated by the similarity of their radiative temperatures, and also by their similar reflectanccs at visible wavelengths. A method of cloud analysis over snow-covered regions is presented, using 1.51 – 1.63 μm data from an experimental sensor on board a U.S. Air Force Defense Meteorological Satellite Program platform. At these wavelengths, snow appears relatively ‘black’ while clouds are highly reflective. The spatial structure of the 1.51-1.63 μm reflectivity fields over a continuous snow surface are examined. Plots of mean reflectance against coefficients of variation for 4×4 pixel areas reveals a cluster of points have low reflectivity and low variability, corresponding to snow-covered (cloud free) areas, and a similar cluster with high reflectances corresponding to 100 per cent cloud cover. For the case of a single layered cloud, the radiances associated with partially filled fields of view are also inferred.