Ice and Supercooled Liquid Water Distributions over the Southern Ocean Based on In-Situ Observations and Three Climate Model Simulations

Abstract:

Clouds have a significant impact on Earth’s radiation budget. Identifying cloud phases in remote regions such as the Southern Ocean has been one of the key challenges in global climate model simulations. In this work, an evaluation of the NCAR Community Atmosphere Model version 5 and 6 (CAM5 and CAM6) and the U.S. Department of Energy (DOE) Energy Exascale Earth System Model (E3SM) is conducted by comparing with 1-s and 200-s spatially averaged in-situ airborne observation from the NSF Southern Ocean Clouds, Radiation, Aerosol Transport Experimental Study (SOCRATES) campaign. The cloud characteristics are examined in terms of thermodynamic, dynamic conditions, and aerosol indirect effects at temperatures from -40°C to 0°C where supercooled liquid droplets and ice crystals coexist. For cloud phase occurrence frequency, CAM6 model shows the most similar results to observations, allowing supercooled liquid water to exist from 0°C to -35°C, while CAM5 model only allows it to exist above -10°C. The E3SM model shows the occurrence frequency of mixed phase at 0.4, which is much higher than the observed frequency at 0.1 – 0.2 between -35°C and 0°C. Both CAM6 and E3SM simulate similar magnitudes of average liquid water content (LWC) compared with 1-s observation, but significantly underestimate the average ice water content (IWC) by 2 – 3 (1 – 1.5) orders of magnitude compared with 1-s (200-s) observations. Such a result is likely due to the insufficient growth of ice particles in sedimentation, which is consistent with narrower ranges of in-cloud RH in the simulations compared with observations. For aerosol indirect effects, LWC, IWC, number concentrations of liquid and ice (Nliq and Nice), glaciation ratio, and cloud fraction are found to have positive correlations with aerosol number concentrations in the observations. However, CAM6 and E3SM simulations only show a small positive correlation for Nliq but no obvious correlations for other variables. This study provides valuable information for future model development on cloud microphysics parameterizations in CAM and E3SM models.