A-priori statistics for AOS cloud retrievals: A case study in cloud drop size dispersion

[13-Dec-2022]
Abstract: Remote sensing of cloud microphysical properties is an inherently under-constrained problem requiring a-priori assumptions regarding the cloud microstructure. The nature of these a-priori assumptions can have a dramatic influence on the a-posteriori estimation of the cloud state. Therefore, a concerted effort is planned by the Atmospheric Observing System (AOS) cloud algorithms group to develop an extensive database of a-priori cloud and precipitation microphysics across a broad range of cloud regimes to improve the realism of the resultant retrieval products. In this presentation we will use the parameterization of the Drop Size Distribution (DSD) of liquid-phase clouds as a case study demonstrating the importance of this planned activity. Specifically, we demonstrate the impact of the assumed droplet spectral width for retrievals of cloud microphysics. Existing remote sensing algorithms tend to assume a fixed value for the droplet spectral width. However, we show that probe data from several field campaigns consistently show a negative correlation between drop number concentration and drop spectral width. We further show that this correlation behavior is largely driven by DSD's associated with precipitation, which is consistent with a suppressed coalescence broadening of the DSD as drop number increases. The sign of this correlation, which is applicable to the context of remote sensing is opposite to widely used parameterizations used in climate models that are instead derived from large-scale averages of the in-situ data, which highlights the importance of defining a-priori databases relevant to the remote sensing problem. To exploit these data, we develop a new parameterization of droplet width that takes into account both drop number and precipitation, which can be applied to remote sensing algorithms. Using A-Train data, we demonstrate that including the precipitation effect on the drop width parameterization increases the derived number concentration from MODIS by up to 20% regionally. We further describe how multi-sensor algorithms for AOS exploiting observations from multi-angle polarimeters, cloud radar, and lidar can exploit this parameterization to provide consistent retrievals of cloud and precipitation microphysics for low-altitude liquid-phase clouds.