Simulating Southern Ocean Aerosol and Ice Nucleating Particles in the Community Earth System Model Version 2

Abstract

Southern Ocean (SO) low‐level mixed phase clouds have been a long‐standing challenge for Earth system models to accurately represent. While improvements to the Community Earth System Model version 2 (CESM2) resulted in increased supercooled liquid in SO clouds and improved model radiative biases, simulated SO clouds in CESM2 now contain too little ice. Previous observational studies have indicated that marine particles are major contributor to SO low‐level cloud heterogeneous ice nucleation, a process that initiates a number of cloud processes that govern cloud radiative properties. In this study, we utilize detailed aerosol and ice nucleating particle (INP) measurements from two recent measurement campaigns to assess simulated aerosol abundance, number size distributions, and composition and INP parameterizations for use in CESM2. Our results indicate that CESM2 has a positive bias in simulated surface‐level total aerosol surface area at latitudes north of 58°S. Measured INP populations were dominated by marine INPs and we present evidence of refractory INPs present over the SO assumed here to be mineral dust INPs. Results highlight a critical need to assess simulated mineral dust number and size distributions in CESM2 in order to adequately represent SO INP populations and their response to long‐term changes in atmospheric transport patterns and land use change. We also discuss important cautions and limitations in applying a commonly used mineral dust INP parameterization to remote regions like the pristine SO.

Characterizing Ice Nucleating Particles Over the Southern Ocean Using Simultaneous Aircraft and Ship Observations

“…The lack of blank correction may also explain the higher INP concentrations reported from ACE measurements relative to those from CAPRICORN-1, CAP-2, and SOC, although differences in sampling time and location likely also play a role. Parameterizations for dust INPs based on global INP data and number concentrations of aerosols larger than 500 nm (DeMott et al, 2015) or aerosol surface area concentrations (Ullrich et al, 2017) overpredicted the observations from SOC and CAP-2, corroborating the growing consensus that the majority of boundary layer-INPs in this region are local, and of marine origin (Burrows et al, 2013;McCluskey et al, 2018aMcCluskey et al, , 2019McCluskey et al, , 2023Vergara-Temprado et al, 2017;Welti et al, 2020). The Although there is overall agreement between the campaigns, the SOC filter and CFDC measurements in the MBL are offset toward warmer concentrations relative to those from CAP-2.…”

“…Above cloud and upper tropospheric N n500 were consistent with the DeMott et al ( 2015) parameterization for dust (Figure S6b in Supporting Information S1), but N s values were still 1-2 orders of magnitude lower than the Ullrich et al (2017) dust parameterization (Figure S6c in Supporting Information S1). This discrepancy between the effectiveness of dust aerosol number-and surface area-based parameterizations has been reported previously for this region (McCluskey et al, 2019(McCluskey et al, , 2023. About half of the SOC research flights collected paired filter samples, with one in the MBL and one above cloud.…”

“…About half of the SOC research flights collected paired filter samples, with one in the MBL and one above cloud. Examples from RF05 (1/25-1/26/18) and RF14 (2/21-2/22/18) are shown in Figure 7, along with SOC CFDC data separated into MBL and above-cloud measurements, in-cloud data on the CVI, and simulated CAM6 INP concentrations (Section 2.4, Text S4 in Supporting Information S1) co-located with the SOC observations (McCluskey et al, 2023). When available, CAP-2 CFDC data during G-V overpasses of the RV Investigator are also shown for comparison.…”

“…An earlier version of the SOC IS filter data set was presented in Järvinen et al (2022) as a comparison to in-cloud measurements of ice crystal number concentrations, and the concentrations were used in McCluskey et al (2023) and Zhao et al (2023) to evaluate model INP predictions (Section 2.4, Text S4 in Supporting Information S1). Preliminary CAP-2 filter results were shown in McFarquhar et al (2021), along with previous measurements made in the SO.…”

Radar‐Derived Snowfall Microphysical Properties at Davis, Antarctica

“…This riming frequency is similar to results over mid‐latitude Europe and the Arctic (Kneifel & Moisseev, 2020), although the riming frequency over Davis is slightly higher than the four Northern Hemisphere sites (3 − 5% at the Northern mid‐latitudes at −3°C, and 2% over the Arctic). Cleaner sites, with fewer aerosols and in particular fewer INPs (C. S. McCluskey et al., 2023; McCluskey et al., 2018; Raman et al., 2023; Vignon, Alexander, et al., 2021), have fewer but larger supercooled liquid water droplets, which enhance the possibility of riming. A high riming frequency was found during one summer event of triple‐wavelength radar observations at McMurdo Station, Antarctica (Tridon et al., 2022) which is qualitatively consistent with the Davis summer results presented here.…”