Stability of the winter-time Arctic Ocean boundary layer in CMIP6 climate models evaluated against Soviet drifting stations, SHEBA and MOSAiC observations

The atmospheric boundary layer in the Arctic winter is characterised by strong and long-lived low level stability which arises from long-wave radiative cooling of the surface during the polar night. This atmospheric temperature inversion is a necessary condition for the positive lapse rate feedback, which is a major contributor to Arctic Amplification. In this study, we assess the low-level stability of the winter-time Arctic boundary layer using ground-based and radiosonde observations collected during the MOSAiC (2019-2020) and SHEBA (1997-1998) expeditions, and from Soviet drifting stations (1955-1991). We compare these observations with the representation of Arctic boundary layer stability in models participating in the latest phase of the Coupled Model Intercomparison Project (CMIP6). The observations show a bimodal distribution of clear and cloudy states which has been reported previously. In the clear state, longwave radiative cooling from the surface leads to strong inversions and a stably stratified boundary layer. Whereas, in the cloudy state, inversions are weaker and not confined to the surface. Previous work has shown that many CMIP5-era climate models fail to realistically represent the cloudy state and often overestimate low-level stability. Here, we assess the extent to which the CMIP6 models also show such biases and examine the representation of surface net longwave radiation and turbulent heat fluxes as potential sources of the biases. Finally, we show that across CMIP6 models, low level stability over sea-ice is correlated with inter-model variation in Arctic amplification.