Solar Radiative Transfer Through Clouds Possessing Isotropic Variable Extinction Coefficient

Abstract Solar radiative fluxes were computed for heterogeneous clouds using an extension of the Monte Carlo method of photon transport which assumes that clouds possess isotropic variability. Hence, computation of fluxes for three‐dimensional (3‐D) clouds can be achieved with only an extended, characteristic one‐dimensional (1‐D) transect of extinction coefficient, β. These are easily obtained by aircraft measurements. In order to validate the new scheme, fluxes for 3‐D, stochastic multiplicative cascade clouds were computed by the conventional approach to Monte Carlo simulation. 1‐D transects through these clouds were then strung together and used in the new scheme. Not only do both methods of calculation yield statistically identical flux estimates, but they also produce virtually identical distributions of photon optical pathlengths and number of scattering events. Furthermore, the new technique may require an order of magnitude less computation time, depending on the desired level of accuracy. Cloud microphysical data obtained by aircraft were used to represent characteristic transects of β, and corresponding fluxes were computed with the new scheme. Results suggest that internal inhomogeneity reduces cloud albedo below homogeneous values by about 5‐10% for overcast and isolated cubic clouds. Also, it is predicted that for overcast clouds of optical depth τ < (>) ≈ 40, inhomogeneous clouds absorb less (more) radiation relative to their homogeneous counterparts. Three individual and trivial modifications of a plane‐parallel, homogeneous two‐stream solution of the radiative‐transfer equation appear to be capable of capturing the essential effects of inhomogeneity. This is promising for incorporation of inhomogeneous cloud effects into existing climate‐model solar‐radiation routines. Little evidence, however, was found to suggest that the effects of inhomogeneity alone can explain the spectral aspects of the cloud absorption/albedo anomaly problem.