Observations and Simulations of Southern Ocean Marine Boundary Layer Cloud Condensation Nuclei during the Austral Spring and Summer

Abstract

During 2017 November - 2018 March Measurements of Aerosols, Radiation, and Clouds over the Southern Ocean (MARCUS) campaign the Atmospheric Radiation Measurement Program Mobile Facility-2 (AMF2) was installed on an Australian icebreaker conducting resupply voyages over the Southern Ocean (SO) (50°S–68°S, 63°E–150°E). The AMF2 captured the seasonal and latitudinal variation of Marine Boundary Layer (MBL) Cloud Condensation Nuclei (CCN) and clouds which play a crucial role in regulating radiative balance in the atmosphere. During the Austral spring and summer, two regions of cloud and aerosol patterns are seen over the north SO (north of 60°S, NSO) and the south SO (SSO) south of 62°S. During MARCUS, the NSO witnessed surface wind speed with a mode of 12-17 m s-1 from the N-W and surface pressure with two modes of 980 and 995 hPa, while the SSO has a wind speed mode of 4-8 m s-1 from the S-E and a single pressure mode of 975 hPa. When the ship was close to an extratropical cyclone center, decreased surface pressure, drastic changes of surface wind, increased precipitation, and decreased Marine Boundary Layer Height (MBLH) were consistently observed. The observed MBLH shows a Pearson correlation of 0.73 with the surface horizontal wind speed, consistent with the dominance of surface turbulent mixing over the open water. The increased surface CCN number concentration (NCCN) observed south of 62°S is consistent with the presence of secondary CCN and their transport through katabatic flows above the MBL in the free troposphere so that wet scavenging is avoided. The age of these hypothesized sulfates is quantified using the ratio between the number of condensation nuclei (CN) to CCN (NCN/NCCN). Back-trajectories show that air masses containing more fresh particles (3.5 < NCN/NCCN < 16) have a higher MBLH (median of 0.8-1 km) in their history compared to the air masses (MBLH median of 0.4-0.8 km) with NCN/NCCN < 1.6.Previous studies identified that modeled MBL cloud droplet number concentration (Nd) is underestimated by a factor of 2 over the summertime SO close to the Antarctic coast. Here, comparisons between the observed CCN from MARCUS and simulated CCN from the Community Atmospheric Model 6 (CAM6) are presented. Modeled MBL NCCN is underestimated, by close to 100% at latitudes south of 55°S with the NCCN bias 1) largest close to the Antarctic coast during summer, implying the biased CCN type has seasonal and latitudinal variation and, 2) three times larger over sea ice than over open water, implying sea spray CCN are better simulated compared to secondary CCN. Assessments of aerosol size distribution indicate an underestimation of accumulation-mode-aerosols (Ac) with diameters 70 nm < D < 100 nm. CCN supersaturation spectra indicates that the observed CCN had lower hygroscopicity compared to simulated CCN, implying differences in CCN chemical composition. Because secondary aerosols including sulfate are less hygroscopic than sea salt CCN, the CCN-activation-ratio derived using bulk hygroscopicity kappa in the Abdul-Razzak function is under-calculated with an overestimation of critical supersaturation south of 62°S. For CCN serves as the nuclei on which water condense to activate, the biases reported here highlight shortfalls in simulated CCN that can be important to the well-documented underestimated Nd by Earth System Models, a key feature and uncertainty of pre-industrial conditions. Further, the overestimate of sea salt and underestimate of sulfate as MBL CCN was investigated by adjusting surface emission parameterizations. For sea salt, when updating the 10-m wind (U10) dependence from ?103.41 to ?102.8, the simulated NCCN are within a factor of 2 of those observed for 74% of times over 50-55°S (previously 45%), while tripling the surface gaseous dimethyl sulfide emissions (DMS3) increases agreement from 37% to 58% south of 55°S. In combination, these changes (?102.8 and DMS3) improve agreement between the simulated and surface-retrieved Nd. The mode of Nd increases from 40 cm-3 to 100 cm-3 after the updates, resulting in more reflective clouds along the Antarctic coast with an extra 5 ~ 12 W m-2 cloud shortwave forcing back to space. In detail, 1) the performance of the DMS*3 simulation depends not only on the absolute values of DMS emission over the SO, but also on regions and elevations where DMS is elevated; 2) the consequential increase in small sulfates and decrease in large sea salts makes the NCCN-supersaturation (SSw) spectra closer to that of observations. However, the observed shape of the spectra is not fully represented, implying this method does not completely solve the identified bias and further observations are thus needed.