Partitioning ice water content from retrievals and its application in model comparison

Retrieved bulk microphysics from remote sensing observations is a composite of ice, snow, and graupel in the three-species ice-phase bulk microphysics parameterization. In this study, density thresholds are used to partition the retrieved ice particle size distribution (PSD) into small, median, and large particle size modes from millimeter cloud radar (MMCR) observations in the tropics and global CloudSat and CALIPSO ice cloud property product (2C-ICE) observations. It shows that the small mode can contribute to more than 60% of the total ice water content (IWC) above 12 km (colder than 220 K). Below that, dominant small mode transitions to dominant median mode. The large mode contributes to less than 10%-20% at all height levels. The PSD assumption in retrieval may cause about 10% error in the IWC partition ratio. The lidar-only region in 2C-ICE is dominated by the small mode, while the median mode dominates the radar-only region. For the three-species ice-phase bulk microphysics parameterizations, the cloud ice mass mainly consists of the small mode. But snow and graupel in the models are not equivalent to the median and large modes in the observations, respectively. Therefore, they need to be repartitioned with rebuilt PSDs from the model assumptions using the same partition technique as the observations. The repartitioned IWCs in each mode from different ice species need to be added together and then compared with the corresponding mode from observations.