Evaluation of biases in mid-to-high-latitude surface snowfall and cloud phase in ERA5 and CMIP6 using satellite observations

Abstract. Supercooled liquid-containing clouds (sLCCs) play a significant role in Earth's radiative budget and the hydrological cycle, especially through surface snowfall production. Evaluating state-of-the-art climate models with respect to their ability to simulate the frequency of occurrence of sLCCs and the frequency with which they produce snow is, therefore, critically important. Here, we compare these quantities as derived from satellite observations, reanalysis datasets, and Earth system models from Phase 6 of the Coupled Model Intercomparison Project (CMIP6) and find significant discrepancies between the datasets for mid- and high latitudes in both hemispheres. Specifically, we find that the ERA5 reanalysis and 10 CMIP6 models consistently overestimate the frequency of sLCCs and snowfall frequencies from sLCCs compared to CloudSat–CALIPSO satellite observations. The biases are very similar for ERA5 and the CMIP6 models, which indicates that the discrepancies in cloud phase and snowfall stem from differences in the representation of cloud microphysics rather than the representation of meteorological conditions. This, in turn, highlights the need for refinements in the models’ parameterizations of cloud microphysics in order for them to represent cloud phase and snowfall accurately. The thermodynamic phase of clouds and precipitation has a strong influence on simulated climate feedbacks and, thus, projections of future climate. Understanding the origin(s) of the biases identified here is, therefore, crucial for improving the overall reliability of climate models.