The prevalence of Arctic multilayer clouds and their observed and modelled characteristics

Abstract. Multilayer clouds (MLCs) are common in the Arctic. With a limited-area setup and 2.5 km horizontal grid spacing, 32 ICON simulations from 22 August to 23 September 2020 were analysed to examine the MLC abundance and characteristics across the Arctic. The model was evaluated against observations from the MOSAiC campaign. An immersion freezing parameterisation was developed to capture the local ice-nucleating particle concentration, increasing the cloud ice number concentration by up to 16 % at temperatures above −12 °C. Overall, the model captured most cloudy events with a dry (moist) bias at lower (higher) altitudes. Simulated water paths were underestimated, roughly 3-fold for liquid water and 100-fold for frozen hydrometeors. A 35 %–65 % MLC occurrence, smoothly distributed across the Arctic region, was simulated. Modelled MOSAiC occurrence frequencies span 42 %–76 %, compared to an observed 32 %–59 %. While large differences in the total MLC occurrence are found, two-layered systems occur with a systematic frequency of about 22 %. The sub-saturated layer between cloud layers is typically < 1 km, indicating a high likelihood of the seeder-feeder mechanism (up to 52 %), consistent with observations.