Radiative effects of sub-mean free path liquid water variability observed in stratiform clouds

A unique 4-cm-resolution data set of stratocumulus cloud liquid water content (LWC) made possible, for the first time, the generation of credible models of cloud LWC down to centimeter-scales, well below the photon mean free path (mfp) which is typically tens of meters. These models are exploited to study the errors in the usual assumption of LWC homogeneity within three-dimensional (3D) radiative transfer grid boxes, by doing Monte Carlo computations for 30-m cubes resolved down to IO-cm scales. Cloud models with and without vertical variability are compared. We found that vertically homogeneous but horizontally variable cloud structure produces cloud radiation biases of 3-5% relative to the assumption of complete homogeneity. The addition of vertical variability similar to the horizontal variability reduces these biases well below 1%. We therefore conclude that 3D cloud models resolved down to the mfp scale (20-30 m for marine stratocumulus clouds), with an incorrect assumption of homogeneity below that scale are sufficient for modeling radiative properties averaged over mfp and larger scales in both visible and near-infrared spectral regions. This conclusion is restricted to only the overcast cases.