Interplay between Primary and Secondary Ice Production (SIP) in Arctic Mixed-Phase Clouds (MPCs) as simulated for the M-PACE campaign in ICON

Recent experiments and modelling studies suggest that secondary ice production (SIP) may close the gap between observed Arctic ice nucleating particle (INP) concentrations and ice crystal number concentrations (N_i). Here we explore model sensitivities with respect to the complexity of different INP parameterisations in numerical simulations under the premiss that N_i is governed by SIP. Idealised, cloud-resolving simulations are performed for the marine cold air outbreak cloud deck sampled during M-PACE (cloud-top temperature of -17°C) with the ICOsahedral Nonhydrostatic (ICON) model.

Droplet shattering (DS) of rain drops according to Phillips et al. (2018), and collisional breakup (CB) (Phillips et al. 2017) were implemented and tested in addition to the existing Hallet-Mossop (HM) rime splintering implemented in ICON’s state-of-the-art two-moment bulk microphysics scheme. Furthermore, a fully prognostic temperature-dependent budget representation of INP (Solomon et al. 2015) was implemented and contrasted to a less sophisticated time-relaxation formulation of atmospheric INP concentrations.

Overall, 16 different model experiments (24h runs) were performed and analysed. Despite the considerable amount of uncertainty remaining with regard to ice production mechanisms and their process representation in numerical models we conclude from these experiments that: (i) N_i-enhancement through SIP can close the gap between measured and simulated N_i concentrations during M-PACE in ICON consistent with previous studies (e.g. Sotiropoulou et al. 2020; Zhao et al. 2021), (ii) only simulations where DS dominates the SIP signal (potentially amplified by CB) capture the vertical N_i in-cloud profile correctly, (iii) INP recycling remains necessary for MPC maintenance during M-PACE even if N_i is dominated by SIP, and (iv) experiments using a computationally more efficient relaxation-based prognostic parameterisation of primary nucleation are statistically invariant from simulations considering a prognostic INP budget.