Identifying Cloud Phase States from Multiple ARM Remote Sensors

Cloud phase has a strong impact on cloud radiative and microphysical processes, yet remains difficult to measure and simulate correctly due to the possibility of finding super-cooled liquid at temperatures between -40 and 0 degrees C. In order to improve our understanding of cloud phase and the processes that impact it, we need quality measurements of phase and an assessment of its accuracy. In addition, identifying cloud phase is a necessary step in retrieving additional cloud microphysical properties like water contents, size distributions, and number densities. Quantifying and reducing uncertainties in cloud microphysical retrievals from remote sensors remains a significant challenge due to insufficient independent information to fully describe the cloud state, and the complexity of finding a standard “truth” with which to validate the retrievals. There are cases, however, when a great deal of confidence exists in expert interpretation of cloud phase state from lidar depolarization ratio and cloud radar Doppler spectra, even in mixed-phase conditions. Here we present work to assess and improve the accuracy of automated cloud phase retrievals in comparison with manually classified cloud phase. These results will be used to test hypotheses about the statistical nature of cloud phase in different geographical regions and atmospheric conditions.