Performance assessment of a five‐channel estimation‐based ice cloud retrieval scheme for use over the global oceans

This work determines the performance of a five‐channel ice cloud retrieval scheme in context of numerical synthetic experiments and real‐world data and examines the implications of these results on the global retrieval of ice cloud microphysical properties over the global oceans. This estimation‐based scheme, designed from information content principles, uses a rigorous, state‐dependent error analysis to combine measurements from the visible, near‐infrared, and infrared spectral regions. In the synthetic experiments, the five‐channel scheme performed as well or better in terms of retrieval bias and random error than the traditional split‐window and Nakajima and King bispectral retrieval techniques for all states of the atmosphere. Although the five‐channel scheme performed favorably compared to the other methods, the inherently large uncertainties associated with ice cloud physics dictate typical retrieval uncertainties in both IWP and effective radius of 30–40%. These relatively large uncertainties suggest caution in the strict interpretation of small temporal or spatial trends found in existing cloud products. In MODIS and CRYSTAL‐FACE applications, the five‐channel scheme exploited the strengths of each of the bispectral approaches to smoothly transition from a split‐window type approach for thin clouds to a Nakajima and King type approach for thick clouds. Uniform application of such a retrieval scheme across different satellite and field measurement campaigns would provide a set of consistent cloud products to the user community, theoretically allowing the direct comparison of cloud properties for the climate processes studies found throughout the literature.