The effect of ice crystal surface roughness on the retrieval of ice cloud microphysical and optical properties

The effect of the surface roughness of ice crystals is not routinely accounted for in current cloud retrieval algorithms that are based on pre-computed lookup libraries. In this study, we investigate the effect of ice crystal surface roughness on the retrieval of ice cloud effective particle size, optical thickness and cloud-top temperature. Three particle surface conditions, smooth, moderately rough and deeply rough, are considered in the visible and near-infrared channels (0.65 and 3.75 ????m). The discrete ordinates radiative transfer (DISORT) model is used to compute the radiances for a set of optical thicknesses, particle effective sizes, viewing and illumination angles, and cloud temperatures. A parameterization of cloud bi-directional reflectances and effective emittances is then developed from a variety of particle surface conditions. This parameterization is applied in a 3-channel retrieval method for Moderate Resolution Imaging Spectroradiometer (MODIS) data at 0.65, 3.75, and 10.8 ????m. Cloud optical properties are derived iteratively for each pixel that contains ice clouds. The impact of ice crystal surface roughness on the cloud parameter retrievals is examined by comparing the results for particles with smooth surfaces and rough surfaces. Retrieval results from two granules of MODIS data indicate that the retrieved cloud optical thickness is significantly reduced if the parameterization for roughened particles is used, as compared with the case of smooth particles. For the retrieval of cloud effective particle size, the inclusion of the effect of surface roughness tends to decrease the retrieved effective particle size if ice crystals are small. The reversed result is noticed for large ice crystals. It is also found that surface roughness has a very minor effect on the retrieval of cloud-top temperatures.