This study investigates the representation of stratocumulus (Sc) clouds, cloud variability, and precipitation statistics over the Southern Ocean (SO) to understand the dominant ice processes within the Icosahedral Nonhydrostatic (ICON) model at the kilometer scale using real case simulations. The simulations are evaluated using the shipborne observations as open-cell stratocumuli were continuously observed during two days (26th -27th of March 2016), south of Tasmania. The radar retrievals are used to effectively analyze the forward-
simulated radar signatures from Passive and Active Microwave TRAnsfer (PAMTRA). We contrast cloud-precipitation statistics, and microphysical process rates between simulations performed with one-moment (1M) and two-moment (2M) microphysics schemes. We further analyze their sensitivity to primary and secondary ice-phase processes (Hallett–Mossop and collisional breakup). Both processes have previously been shown to improve the ice properties of simulated shallow mixed-phase clouds over the SO in other models. We find that only simulations with continuous formation, growth, and subsequent melting of graupel, and the effective riming of in-cloud rain by graupel, capture the observed cloud-precipitation vertical structure. In particular, the 2M microphysics scheme requires additional tuning for graupel processes in SO stratocumuli. Lowering the assumed graupel density and terminal velocity, in combination with secondary ice processes, enhances graupel formation in 2M microphysics ICON simulations. Overall, all simulations capture the observed intermittency of precipitation irrespective of the microphysics scheme used, and most of them sparsely distribute intense precipitation (>1mm h-1 ) events. Furthermore, the simulated clouds are too reflective as they are optically thick and/or have high cloud cover.