Evaluating remote sensing retrievals of CER, either spaceborne or airborne, typically is done via comparisons with CER derived from droplet size distributions measured in situ by airborne cloud probes. However, such seemingly straightforward comparisons in practice involve numerous confounding factors that have consequential implications on the interpretation of comparison results. For the remote sensing retrievals, these include imager radiometric calibration, spectral channels having differing vertical sensitivities, spectral above-cloud atmospheric corrections, and forward radiative transfer model assumptions such as plane-parallel atmospheres, assumed droplet size distributions, and the temperature dependence of the liquid water complex index of refraction, each of which can have impacts on the retrieved CER. For the cloud probes, confounding factors include the sampling strategy within the cloud, sensitivities to different portions of the size distribution, and known or unknown sizing uncertainties or errors.
In this presentation, we discuss results of an extensive comparison of airborne imager remote sensing retrievals of liquid cloud CER from the Enhanced MODIS Airborne Simulator (eMAS) against spatially and temporally co-located multi-angle polarimetric retrievals from the Research Scanning Polarimeter (RSP) and in situ cloud probes obtained for marine boundary layer stratocumulus during the NASA ORACLES field campaign. eMAS, a multi-spectral imager having spectral capabilities similar to MODIS and VIIRS, has multiple spectral channels in the CER-sensitive 1.6µm (2 channels) and 2µm (4 channels) SWIR window bands and the 3.7µm (1 channel) MWIR window band that enable a variety of bi-spectral retrieval channel pairings beyond those of any single spaceborne imager. Moreover, during ORACLES, on several occasions the aircraft flight heading was such that eMAS observed the backscatter glory in a portion of its swath; we show CER and droplet size distribution effective variance inferred from this single-scattering reflectance feature using a peak-matching technique similar to the polarized cloud-bow, and compare against both the bi-spectral and polarimetric cloud-bow retrievals. The impacts of numerous confounding factors are explored, including bi-spectral retrieval assumptions, and the broader implications of retrieval differences due to differing fundamental sensitivities of multi-spectral imagery and multi-angle polarimetry are discussed.

