Assessing Simulated Primary Ice Nucleation in Southern Ocean Mixed Phase Clouds

Description

Simulated Southern Ocean (SO) low-level clouds are sensitive to the representation of the primary ice nucleation process and subsequent freezing processes. We utilize observations from the DOE-ARM Measurement of Aerosols, Radiation and CloUds over the Southern oceans (MARCUS) campaign and the Macquarie Island Cloud and Radiation Experiment (MICRE) to understand and simulate the freezing processes occurring in SO mixed phase clouds. Observations are compared with Community Atmosphere Model version 6 (CAM6) simulations. CAM6 utilizes a nudged meteorology approach, which specifies the model dynamics using reanalysis data for the study periods.

Simulated SO low-level mixed phase clouds in CAM6 contain significantly larger amounts of supercooled liquid water compared to the previous version (CAM5). Sensitivity simulations indicate that supercooled liquid water is maintained in CAM6 due to a reduced primary ice nucleation process over the SO, which largely resulted from a change to the primary ice nucleation scheme in CAM6. Specifically, the temperature dependent ice nucleation scheme used in CAM5 was replaced with a dust and temperature dependent ice nucleation rate in CAM6. Preliminary comparisons between simulated and cloud phase observations from MICRE, MARCUS, and SOCRATES (Southern Ocean Clouds, Radiation, Aerosol, Transport Experimental Study) suggest that CAM6 simulated low-level clouds contain more supercooled liquid compared to observations. 

As part of this evaluation, we are assessing current numerical methods for predicting ice nucleating particles (INPs), or particles responsible for forming the first ice crystals in the mixed phase cloud regime, rather than dust ice nucleation rates currently estimated in CAM6.  To predict INP number concentrations (n_INPs), INP parameterizations for sea spray aerosol and dust INP sources have been applied to simulated aerosol from CAM6. Surface level n_INPs observations from MICRE and MARCUS are used to assess predicted n_INPs, revealing sensitivities to simulated dust concentrations. Observed aerosol number and surface area from MICRE and MARCUS have also been compared to the modeled aerosol. Finally, INP parameterizations for dust and sea spray aerosol have been implemented into the single column atmosphere model (SCAM) and CAM6 to investigate simulated cloud properties resulting from the INP-prediction approach for primary ice nucleation.