Key points are not available for this paper at this time.
We examine the sensitivity of cirrus cloud albedo and bidirectional reflection function to particle shape using the phase functions of liquid water spheres, regular hexagonal ice crystals, and random‐fractal ice particles calculated at a nonabsorbing visible wavelength of 0.63 μm. Accurate multiple‐scattering calculations for plane‐parallel clouds show that hexagonal ice crystal clouds have systematically larger planetary and global albedos than liquid water clouds of the same optical thickness. There is accumulating evidence that the idealized phase function of regular hexagonal crystals, which causes pronounced halos, is not necessarily the best representation of the range of reflectance characteristics of the majority of ice clouds. A more typical representation of the scattering phase function for ice clouds that are composed of a complex set of crystal shapes and sizes may be obtained using a model of randomly shaped irregular particles. Even larger cloud albedos are obtained for the random‐fractal particle model because of its smaller asymmetry parameter. Our computations also show that a larger planetary albedo does not always imply a larger reflectance and that the relative brightness of ice versus liquid water clouds is highly scattering‐geometry dependent. Use of the wrong particle shape model (crystal instead of water droplet and vice versa) in retrieving cloud optical thickness from bidirectional reflectance measurements can result in an underestimation or overestimation of the true optical thickness by a factor that can exceed 3. At some scattering geometries, use of the wrong model can give an unrealistically large optical thickness or no solution at all. Overall, bidirectional reflectance differences between random‐fractal and regular hexagonal particle shapes are significantly smaller than those between either ice crystal and liquid water spheres, except at the back scattering direction.
Mishchenko et al. (Mon,) studied this question.