A new evaluation method for the thermodynamic phases of clouds in cloud system-resolving models is presented using CALIPSO observations and a satellite simulator. This method determines the thermodynamic phases using the depolarization ratio and a cloud extinction proxy. For the evaluation, we introduced empirical parameterization of the depolarization ratio of ice and water clouds using temperatures of a reanalysis data and total attenuated backscatters of CALIPSO.
We evaluated the mixed-phase clouds simulated in a cloud system-resolving model over the Southern Ocean using single-moment and double-moment bulk cloud microphysics schemes, referred to as NSW6 and NDW6, respectively. The NDW6 simulations reproduce supercooled water clouds near the boundary layer that are consistent with the observations. Conversely, the NSW6 simulations failed to reproduce such supercooled water clouds.
Consistencies between the cloud classes diagnosed by the evaluation method and the simulated hydrometeor categories were examined. NDW6 shows diagnosed water and ice classes that are consistent with the simulated categories, whereas the ice category simulated with NSW6 is diagnosed as liquid water by the present method due to the large extinction from the ice cloud layers. Additional analyses indicated that ice clouds with a small effective radius and large ice water content in NSW6 lead to erroneous values for the fraction of the diagnosed liquid water. It is shown that the uncertainty in the cloud classification method depends on the details of the cloud microphysics schemes. It is important to understand the causes of inconsistencies in order to properly understand the cloud classification applied to model evaluations as well as retrievals.