A new parameterization of single scattering solar radiative properties for tropical anvils using observed ice crystal size and shape distributions

Parameterizations of single scattering properties currently used in cloud resolving and general circulation models are somewhat limited in that they typically assume the presence of single particle habits, do not adequately account for the numbers of ice crystals with diameters smaller than 100 μm, and contain no information about the variance of parameterization coefficients. Here, new parameterizations of mean single scattering properties (e.g., single scatter albedo, asymmetry parameter, and extinction efficiency) for distributions of ice crystals in tropical anvils are developed. Using information about the size and shape of ice crystals acquired by a two-dimensional cloud probe during the Central Equatorial Pacific Experiment (CEPEX), a self-organized neural network defines shape based on simulations of how the particle maximum dimension and area ratio (ratio of projected area to that of circumscribed circle with maximum dimension) vary for random orientations of different idealized shapes (i.e., columns, bullet rosettes, rough aggregates, and particles represented by Chebyshev polynomials). The size distributions for ice crystals smaller than 100 μm are based on parameterizations developed using representative samples of 11 633 crystals imaged by a video ice particle sampler (VIPS). The mean-scattering properties for distribution of ice crystals are then determined by weighting the single scattering properties of individual ice crystals, determined using an improved geometric ray-tracing method, according to number concentration and scattering cross section. The featureless nature of the calculated phase function, averaged over all observed sizes and shapes of ice crystals, is similar to that obtained using other schemes designed to account for variations in sizes and shapes of ice crystals. The new parameterizations of single scatter albedo, asymmetry parameter, and extinction efficiency are then determined by functional fits in terms of cloud particle effective radius; there was no statistically significant dependence on either ice water content or temperature. Uncertainty estimates incorporated into the parameterization coefficients are based upon a Monte Carlo approach. Comparisons with previously used parameterizations and with parameterizations developed using single crystal habits are made to show that the determination of representative crystal habits is still a major unknown in the development of parameterization schemes.