Modelling the impacts of secondary ice production in Arctic mixed-phase clouds

Several observational studies show that ice crystal number concentrations in relatively warm (temperatures above -10 °C) Arctic mixed-phase clouds can exceed 1 L^-1 while concentrations of ice-nucleating particles (INPs), which produce primary ice by initiating cloud droplet freezing, are several orders of magnitude lower. The difference is often explained by secondary ice production (SIP). The three most common SIP mechanics are Hallet-Mossop process also called as rime splintering (RS), ice-ice collisional breakup (IIBR), and droplet shattering during freezing (DS). Our large-eddy simulation (LES) model called UCLALES-SALSA accounts for these processes in addition to aerosol-cloud interactions and different primary freezing processes. In our recent study (Raatikainen et al., EGUsphere [preprint], https://doi.org/10.5194/egusphere-2025-4470, 2025) we used observations from the ACLOUD (Arctic CLoud Observations Using airborne measurements during polar Day) campaign to derive setups for LES simulations which aim at reproducing the observed ice crystal number concentrations exceeding 1 L^-1 at about -5 °C cloud top temperatures. At this temperature, INP concentrations are about three orders of magnitude lower than the observed ice concentration, so secondary ice production is likely occurring. The first simulations showed that rime splintering is the most effective SIP process while IIBR and DS have negligible impact. However, RS cannot produce enough secondary ice to match the observations. The observed ice concentrations can be reached by artificially increasing the efficiency of RS SIP. When ice concentration becomes high enough, SIP starts to maintain itself so that primary cloud droplet freezing is not needed at all. Additional sensitivity tests showed that the same result can be obtained by using different model parameterizations (e.g., mass-dimension-fall velocity or the temperature-dependent efficiency of the RS parameterization) or slightly cooler cloud temperatures. Overall, these results show that rime splintering can explain the observed high ice concentrations in such relatively warm and shallow mixed-phase clouds, but the process is also sensitive to model parameterizations and cloud temperatures.