Abstract
Radiative transfer calculations for a one-dimensional column model of the atmosphere containing a plane-parallel, homogeneous cloud are used to show that the common procedure of assuming that fields of view for high resolution satellite imagers are either overcast or cloud-free will lead to biases in the infrared albedo and surface insulation for typical (250 km)2 weekly to monthly mean climate scales. The biases arise because cloud fields on the (1–8 km)2 scale typical of satellite imager resolution are often broken rather than overcast and because the anisotropy of the radiance held for overcast regions differs from that for cloud-flee regions. Furthermore, the anisotropy for overcast regions is a nonlinear function of cloud optical depth; consequently, there is no equivalent overcast system that will reproduce the anisotropy of the radiance field reflected by broken clouds. The results of the calculations indicate that the biases are a function of the sun-earth-satellite viewing geometry. For example, for a solar zenith angle of 30° and for typical satellite viewing geometries the biases are estimated to reach 5% in the planetary albedo and −5% in surface insulation. The biases increase with increasing solar zenith angle, but on average the percentage bias is fairly insensitive to cloud optical depth. The constancy of the percentage bias should allow it to be largely removed from climatological datasets.