Observations of Clouds, Aerosols, Precipitation, and Surface Radiation over the Southern Ocean: An Overview of CAPRICORN, MARCUS, MICRE, and SOCRATES

Greg M. McFarquhar Cooperative Institute for Mesoscale Meteorological Studies, and School of Meteorology, University of Oklahoma, Norman, Oklahoma

Search for other papers by Greg M. McFarquhar in
Current site
Google Scholar
PubMed
Close
,
Christopher S. Bretherton Department of Atmospheric Sciences, University of Washington, Seattle, Washington

Search for other papers by Christopher S. Bretherton in
Current site
Google Scholar
PubMed
Close
,
Roger Marchand Department of Atmospheric Sciences, University of Washington, Seattle, Washington

Search for other papers by Roger Marchand in
Current site
Google Scholar
PubMed
Close
,
Alain Protat Australian Bureau of Meteorology, Melbourne, Victoria, and Australian Antarctic Programme Partnership, Institute for Marine and Antarctic Science, University of Tasmania, Hobart, Tasmania, Australia

Search for other papers by Alain Protat in
Current site
Google Scholar
PubMed
Close
,
Paul J. DeMott Department of Atmospheric Science, Colorado State University, Fort Collins, Colorado

Search for other papers by Paul J. DeMott in
Current site
Google Scholar
PubMed
Close
,
Simon P. Alexander Australian Antarctic Division, and Australian Antarctic Programme Partnership, Institute for Marine and Antarctic Science, University of Tasmania, Hobart, Tasmania, Australia

Search for other papers by Simon P. Alexander in
Current site
Google Scholar
PubMed
Close
,
Greg C. Roberts Scripps Institution of Oceanography, La Jolla, California, and Centre National de Recherches Météorologiques, UMR3589, Toulouse, France

Search for other papers by Greg C. Roberts in
Current site
Google Scholar
PubMed
Close
,
Cynthia H. Twohy NorthWest Research Associates, Redmond, Washington, and Scripps Institution of Oceanography, La Jolla, California

Search for other papers by Cynthia H. Twohy in
Current site
Google Scholar
PubMed
Close
,
Darin Toohey Department of Atmospheric and Oceanic Sciences, University of Colorado Boulder, Boulder, Colorado

Search for other papers by Darin Toohey in
Current site
Google Scholar
PubMed
Close
,
Steve Siems School of Earth, Atmosphere and Environment, Monash University, Melbourne, Victoria, Australia

Search for other papers by Steve Siems in
Current site
Google Scholar
PubMed
Close
,
Yi Huang School of Earth Sciences, University of Melbourne, Melbourne, Victoria, Australia

Search for other papers by Yi Huang in
Current site
Google Scholar
PubMed
Close
,
Robert Wood Department of Atmospheric Sciences, University of Washington, Seattle, Washington

Search for other papers by Robert Wood in
Current site
Google Scholar
PubMed
Close
,
Robert M. Rauber Department of Atmospheric Sciences, University of Illinois at Urbana–Champaign, Urbana, Illinois

Search for other papers by Robert M. Rauber in
Current site
Google Scholar
PubMed
Close
,
Sonia Lasher-Trapp Department of Atmospheric Sciences, University of Illinois at Urbana–Champaign, Urbana, Illinois

Search for other papers by Sonia Lasher-Trapp in
Current site
Google Scholar
PubMed
Close
,
Jorgen Jensen National Center for Atmospheric Research, Boulder, Colorado

Search for other papers by Jorgen Jensen in
Current site
Google Scholar
PubMed
Close
,
Jeffrey L. Stith National Center for Atmospheric Research, Boulder, Colorado

Search for other papers by Jeffrey L. Stith in
Current site
Google Scholar
PubMed
Close
,
Jay Mace University of Utah, Salt Lake City, Utah

Search for other papers by Jay Mace in
Current site
Google Scholar
PubMed
Close
,
Junshik Um Cooperative Institute for Mesoscale Meteorological Studies, University of Oklahoma, Norman, Oklahoma, and Department of Atmospheric Sciences, Pusan National University, Busan, South Korea

Search for other papers by Junshik Um in
Current site
Google Scholar
PubMed
Close
,
Emma Järvinen National Center for Atmospheric Research, Boulder, Colorado, and Karlsruhe Institute of Technology, Karlsruhe, Germany

Search for other papers by Emma Järvinen in
Current site
Google Scholar
PubMed
Close
,
Martin Schnaiter Karlsruhe Institute of Technology, Karlsruhe, Germany

Search for other papers by Martin Schnaiter in
Current site
Google Scholar
PubMed
Close
,
Andrew Gettelman National Center for Atmospheric Research, Boulder, Colorado

Search for other papers by Andrew Gettelman in
Current site
Google Scholar
PubMed
Close
,
Kevin J. Sanchez Scripps Institution of Oceanography, La Jolla, California

Search for other papers by Kevin J. Sanchez in
Current site
Google Scholar
PubMed
Close
,
Christina S. McCluskey National Center for Atmospheric Research, Boulder, Colorado

Search for other papers by Christina S. McCluskey in
Current site
Google Scholar
PubMed
Close
,
Lynn M. Russell Scripps Institution of Oceanography, La Jolla, California

Search for other papers by Lynn M. Russell in
Current site
Google Scholar
PubMed
Close
,
Isabel L. McCoy Department of Atmospheric Sciences, University of Washington, Seattle, Washington

Search for other papers by Isabel L. McCoy in
Current site
Google Scholar
PubMed
Close
,
Rachel L. Atlas Department of Atmospheric Sciences, University of Washington, Seattle, Washington

Search for other papers by Rachel L. Atlas in
Current site
Google Scholar
PubMed
Close
,
Charles G. Bardeen National Center for Atmospheric Research, Boulder, Colorado

Search for other papers by Charles G. Bardeen in
Current site
Google Scholar
PubMed
Close
,
Kathryn A. Moore Department of Atmospheric Science, Colorado State University, Fort Collins, Colorado

Search for other papers by Kathryn A. Moore in
Current site
Google Scholar
PubMed
Close
,
Thomas C. J. Hill Department of Atmospheric Science, Colorado State University, Fort Collins, Colorado

Search for other papers by Thomas C. J. Hill in
Current site
Google Scholar
PubMed
Close
,
Ruhi S. Humphries Climate Science Centre, Oceans and Atmosphere, CSIRO, Melbourne, Victoria, Australia

Search for other papers by Ruhi S. Humphries in
Current site
Google Scholar
PubMed
Close
,
Melita D. Keywood Climate Science Centre, Oceans and Atmosphere, CSIRO, Melbourne, Victoria, Australia

Search for other papers by Melita D. Keywood in
Current site
Google Scholar
PubMed
Close
,
Zoran Ristovski School of Earth and Atmospheric Sciences, Queensland University of Technology, Brisbane, Queensland, Australia

Search for other papers by Zoran Ristovski in
Current site
Google Scholar
PubMed
Close
,
Luke Cravigan School of Earth and Atmospheric Sciences, Queensland University of Technology, Brisbane, Queensland, Australia

Search for other papers by Luke Cravigan in
Current site
Google Scholar
PubMed
Close
,
Robyn Schofield School of Earth Sciences, University of Melbourne, Melbourne, Victoria, Australia

Search for other papers by Robyn Schofield in
Current site
Google Scholar
PubMed
Close
,
Chris Fairall NOAA, Boulder, Colorado

Search for other papers by Chris Fairall in
Current site
Google Scholar
PubMed
Close
,
Marc D. Mallet Australian Antarctic Programme Partnership, Institute for Marine and Antarctic Science, University of Tasmania, Hobart, Tasmania, Australia

Search for other papers by Marc D. Mallet in
Current site
Google Scholar
PubMed
Close
,
Sonia M. Kreidenweis Department of Atmospheric Science, Colorado State University, Fort Collins, Colorado

Search for other papers by Sonia M. Kreidenweis in
Current site
Google Scholar
PubMed
Close
,
Bryan Rainwater Department of Atmospheric and Oceanic Sciences, University of Colorado Boulder, Boulder, Colorado

Search for other papers by Bryan Rainwater in
Current site
Google Scholar
PubMed
Close
,
John D’Alessandro Cooperative Institute for Mesoscale Meteorological Studies, and School of Meteorology, University of Oklahoma, Norman, Oklahoma

Search for other papers by John D’Alessandro in
Current site
Google Scholar
PubMed
Close
,
Yang Wang Cooperative Institute for Mesoscale Meteorological Studies, University of Oklahoma, Norman, Oklahoma, and Beijing Normal University, Beijing, China

Search for other papers by Yang Wang in
Current site
Google Scholar
PubMed
Close
,
Wei Wu Cooperative Institute for Mesoscale Meteorological Studies, University of Oklahoma, Norman, Oklahoma

Search for other papers by Wei Wu in
Current site
Google Scholar
PubMed
Close
,
Georges Saliba Scripps Institution of Oceanography, La Jolla, California

Search for other papers by Georges Saliba in
Current site
Google Scholar
PubMed
Close
,
Ezra J. T. Levin Department of Atmospheric Science, Colorado State University, Fort Collins, and Handix Scientific, Boulder, Colorado

Search for other papers by Ezra J. T. Levin in
Current site
Google Scholar
PubMed
Close
,
Saisai Ding Peking University, Beijing, China

Search for other papers by Saisai Ding in
Current site
Google Scholar
PubMed
Close
,
Francisco Lang School of Earth, Atmosphere and Environment, Monash University, Melbourne, Victoria, Australia

Search for other papers by Francisco Lang in
Current site
Google Scholar
PubMed
Close
,
Son C. H. Truong School of Earth, Atmosphere and Environment, Monash University, Melbourne, Victoria, Australia

Search for other papers by Son C. H. Truong in
Current site
Google Scholar
PubMed
Close
,
Cory Wolff National Center for Atmospheric Research, Boulder, Colorado

Search for other papers by Cory Wolff in
Current site
Google Scholar
PubMed
Close
,
Julie Haggerty National Center for Atmospheric Research, Boulder, Colorado

Search for other papers by Julie Haggerty in
Current site
Google Scholar
PubMed
Close
,
Mike J. Harvey National Institute of Water and Atmospheric Research, Wellington, New Zealand

Search for other papers by Mike J. Harvey in
Current site
Google Scholar
PubMed
Close
,
Andrew R. Klekociuk Australian Antarctic Division, and Australian Antarctic Programme Partnership, Institute for Marine and Antarctic Science, University of Tasmania, Hobart, Tasmania, Australia

Search for other papers by Andrew R. Klekociuk in
Current site
Google Scholar
PubMed
Close
, and
Adrian McDonald Gateway Antarctica, and School of Physical and Chemical Sciences, University of Canterbury, Christchurch, New Zealand

Search for other papers by Adrian McDonald in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

Weather and climate models are challenged by uncertainties and biases in simulating Southern Ocean (SO) radiative fluxes that trace to a poor understanding of cloud, aerosol, precipitation, and radiative processes, and their interactions. Projects between 2016 and 2018 used in situ probes, radar, lidar, and other instruments to make comprehensive measurements of thermodynamics, surface radiation, cloud, precipitation, aerosol, cloud condensation nuclei (CCN), and ice nucleating particles over the SO cold waters, and in ubiquitous liquid and mixed-phase clouds common to this pristine environment. Data including soundings were collected from the NSF–NCAR G-V aircraft flying north–south gradients south of Tasmania, at Macquarie Island, and on the R/V Investigator and RSV Aurora Australis. Synergistically these data characterize boundary layer and free troposphere environmental properties, and represent the most comprehensive data of this type available south of the oceanic polar front, in the cold sector of SO cyclones, and across seasons. Results show largely pristine environments with numerous small and few large aerosols above cloud, suggesting new particle formation and limited long-range transport from continents, high variability in CCN and cloud droplet concentrations, and ubiquitous supercooled water in thin, multilayered clouds, often with small-scale generating cells near cloud top. These observations demonstrate how cloud properties depend on aerosols while highlighting the importance of dynamics and turbulence that likely drive heterogeneity of cloud phase. Satellite retrievals confirmed low clouds were responsible for radiation biases. The combination of models and observations is examining how aerosols and meteorology couple to control SO water and energy budgets.

Supplemental material: https://doi.org/10.1175/BAMS-D-20-0132.2

© 2021 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Greg M. McFarquhar, mcfarq@ou.edu

Abstract

Weather and climate models are challenged by uncertainties and biases in simulating Southern Ocean (SO) radiative fluxes that trace to a poor understanding of cloud, aerosol, precipitation, and radiative processes, and their interactions. Projects between 2016 and 2018 used in situ probes, radar, lidar, and other instruments to make comprehensive measurements of thermodynamics, surface radiation, cloud, precipitation, aerosol, cloud condensation nuclei (CCN), and ice nucleating particles over the SO cold waters, and in ubiquitous liquid and mixed-phase clouds common to this pristine environment. Data including soundings were collected from the NSF–NCAR G-V aircraft flying north–south gradients south of Tasmania, at Macquarie Island, and on the R/V Investigator and RSV Aurora Australis. Synergistically these data characterize boundary layer and free troposphere environmental properties, and represent the most comprehensive data of this type available south of the oceanic polar front, in the cold sector of SO cyclones, and across seasons. Results show largely pristine environments with numerous small and few large aerosols above cloud, suggesting new particle formation and limited long-range transport from continents, high variability in CCN and cloud droplet concentrations, and ubiquitous supercooled water in thin, multilayered clouds, often with small-scale generating cells near cloud top. These observations demonstrate how cloud properties depend on aerosols while highlighting the importance of dynamics and turbulence that likely drive heterogeneity of cloud phase. Satellite retrievals confirmed low clouds were responsible for radiation biases. The combination of models and observations is examining how aerosols and meteorology couple to control SO water and energy budgets.

Supplemental material: https://doi.org/10.1175/BAMS-D-20-0132.2

© 2021 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Greg M. McFarquhar, mcfarq@ou.edu

The Southern Ocean (SO) surrounding Antarctica and consisting of parts of the southern Atlantic, Pacific and Indian Oceans, is one of the cloudiest places on Earth. The fractional cover of low clouds (below 3-km altitude) prevalent in the warm and cold sectors of frequent extratropical cyclones reaches nearly 80% year-round (Mace et al. 2009; IPCC 2013). Relative to more easily sampled locations, there is a dearth of in situ observations of aerosols, clouds, and precipitation over the SO, especially south of 60°S. This makes it difficult to evaluate remote sensing retrieval products. General circulation models (GCMs) have difficulty with simulating the present-day aerosol, cloud coverage and cloud phase over the SO, with implications for anthropogenic aerosol impacts and cloud feedbacks on climate (e.g., Trenberth and Fasullo 2010; Tan et al. 2016), two key uncertainties in interpreting the historical climate record and projecting future climate change.

Numerical weather prediction (NWP) and GCMs have struggled to correctly simulate the radiative budget over the SO due to low cloud biases. Most Coupled Model Intercomparison Project phase 5 (CMIP5) models predict too much shortwave (SW) radiation absorbed over the SO region (Bodas-Salcedo et al. 2014, 2016; Naud et al. 2014), with impacts on ocean temperature, the Southern Hemisphere (SH) jet (Ceppi et al. 2014), Antarctic sea ice trends (Flato et al. 2013) and tropical rainfall (Hwang and Frierson 2013). Comparisons with satellite data indicate that model radiative biases are due primarily to a lack of low- and midlevel clouds in the cold sectors of cyclones (e.g., Flato et al. 2013; Bodas-Salcedo et al. 2014). It was hypothesized on the basis of limited observations, mainly from satellites, that GCMs might be glaciating what are in reality persistent supercooled liquid clouds. Indeed, GCM simulations in which convective parameterizations have been forced to produce greater amounts of supercooled liquid water (SLW) have reduced SW biases (Kay et al. 2016).

A related motivating issue is the apparent paucity of ice nucleating particles (INPs) over the SO (Bigg 1973; Burrows et al. 2013), due to it being far removed from any continental air sources; INP parameterizations are based mostly on Northern Hemisphere (NH) observations. Satellite retrievals of cloud-top phase indicate that SLW is more prevalent over the SO than at equivalent latitudes in the NH (Choi et al. 2010; Hu et al. 2010; Morrison et al. 2011). This could be because SO supercooled clouds are starved for INPs, as hypothesized by Kanitz et al. (2011) and Vergara-Temprado et al. (2018).

A final overarching question is how droplet concentrations are regulated in SO boundary layer (BL) clouds in a synoptically active environment with high winds over a biologically productive ocean. The SO is a biologically unique marine aerosol environment, its pristine nature is as close to preindustrial conditions as exists on Earth, and thus represents a natural laboratory to study anthropogenic aerosol indirect radiative forcing (Carslaw et al. 2013; Ghan et al. 2013). Hoose et al. (2009) showed that GCMs with prognostic aerosols tended to simulate SO clouds with too few droplets compared to satellite observations, making them overly susceptible to human aerosol perturbations. One hypothesis is that these models underestimate marine biogenic production of cloud condensation nuclei (CCN). Satellite retrievals and some previous field observations show the SO has a strong summertime maximum in cloud droplet concentration Nc (Boers et al. 1996, 1998), CCN (Ayers and Gras 1991), and aerosol concentrations Na (Sciare et al. 2009) correlated with phytoplankton productivity. Quinn et al. (2017) found that except for the high southern latitudes, sea spray contributes less than 30% to the total CCN. However, observations in the Aerosol Characterization Experiment 1 (ACE-1) campaign suggested that copious sulfate aerosols can be produced in the outflow of shallow precipitating cumulus clouds from nucleation of marine biogenic gases (Hudson et al. 1998; Clarke et al. 1998; Russell et al. 1998).

Thus, there is a clear need for observations to help better model the natural aerosol life cycle and mixed-phase BL cloud over the SO. Prior to the campaigns described here, cloud and aerosol measurements over the SO included those listed in Table 1. But, further observations on cloud and aerosol concentrations over cold waters poleward of 60°S are critical for understanding cloud processes over the SO. To understand the transition of aerosols to CCN over the remote oceans, it is necessary to quantify particle sources and sinks as well as processes related to their aging, including the role of new particle formation in the free troposphere, generation from breaking waves over the ocean, generation of biogenic particles from gas phase oceanic emissions, the role of precipitation scavenging, and the effects of updrafts and dynamics on clouds.

Table 1.

Previous field campaigns and data collection activities over the SO.

Table 1.

Climate model evaluation, and much current knowledge of SO clouds, aerosols, precipitation, and surface radiation properties is based on satellite retrievals. Satellite studies have found that cloud-top SLW is more frequent over the SO (Hu et al. 2010; Choi et al. 2010; Huang et al. 2012a,b; Kanitz et al. 2011; Morrison et al. 2011; Protat et al. 2014; Huang et al. 2015a, 2016) and Antarctic (Grosvenor et al. 2012) than over the NH, but there are significant variations between satellite retrieval products in the frequency of cloud-top SLW (Delanoë and Hogan 2010; Huang et al. 2015a) and these retrievals tell us little about the phase of condensate below cloud top. However, potential errors in cloud retrievals, particularly those related to large solar zenith angles (Grosvenor and Wood 2014) and three-dimensional effects (Wolters et al. 2010; Zeng et al. 2012; Cho et al. 2015) remain a concern. Additional ground-based and airborne remote sensing, and airborne in situ measurements, are therefore needed to evaluate satellite retrievals.

A 2014 community workshop at the University of Washington discussed these issues, recognizing the need for a large international multiagency effort to improve the understanding of clouds, aerosols, precipitation and their interactions over the SO (Marchand et al. 2014). The workshop served as a motivation for the proposals of separate, but integrated, projects to various funding agencies in the United States and Australia. These four collaborative projects were 1) the Clouds Aerosols Precipitation Radiation and atmospheric Composition over the Southern Ocean (CAPRICORN) I and II research voyages of the Research Vessel (R/V) Investigator, led by the Australian Bureau of Meteorology (BoM), that made extensive in situ and remote sensing measurements in 2016 and 2018, respectively; 2) the 2017–18 Measurements of Aerosols, Radiation and Clouds over the Southern Ocean (MARCUS) project, during which the United States Department of Energy (DOE) Atmospheric Radiation Measurement (ARM) Program Mobile Facility 2 (AMF2) was deployed on the Australian icebreaker Research Supply Vessel (RSV) Aurora Australis (AA) as it made resupply voyages to Australian Antarctic bases; 3) the 2016–18 Macquarie Island Cloud Radiation Experiment (MICRE) acquiring surface in situ and remote sensing observations using equipment from DOE ARM, the BoM, and the Australian Antarctic Division (AAD); and 4) the 2018 Southern Ocean Cloud Radiation and Aerosol Transport Experimental Study (SOCRATES) using the NSF–NCAR G-V aircraft to sample clouds, aerosols, and precipitation from Hobart, Australia, to within approximately 650 km of the Antarctic coast. Although each project was a separate effort and no formal steering committee coordinated the projects, many investigators served on the advisory board of several of the projects and there was much collaboration between the campaigns. There was one integrated planning workshop (2017 Boulder) and two integrated data workshops after the completion of the projects (2018 Boulder, 2019 Hobart). Data have been freely exchanged among participants, and a special collection of papers in the Journal of Geophysical Research/Geophysical Research Letters covering all four projects has been established and is expected to grow substantially over the next few years. This collaboration is essential to maximize the projects’ impacts. Synergistically these data provide the best available measurements of the BL and free troposphere structure, together with vertical distributions of liquid and mixed-phase clouds and aerosols properties, over cold SO waters where SLW and mixed-phase BL clouds are frequent.

Overview of field campaigns

In this section, the campaigns are introduced, detailing the scientific objectives, the time period of the observations, the instruments and platforms used to acquire the observations, the manner in which the observations were obtained, and a broad overview of the meteorological conditions sampled. The majority of the observations was obtained in a north–south curtain extending from Hobart, Australia, to the Antarctic coast in the Australasian sector of the SO. Figure 1 shows the ship tracks from CAPRICORN I, II and MARCUS, as well as the G-V flight tracks during SOCRATES and the location of the ground-observing site at Macquarie Island during MICRE.

Fig. 1.
Fig. 1.

(top) Ship tracks from CAPRICORN I (dark blue), CAPRICORN II (light blue), and MARCUS voyages (green colors), as well as the SOCRATES G-V flight tracks (orange) and the location of the ground-observing site at Macquarie Island during MICRE (red). The locations of Mawson, Davis, and Casey stations are also shown (gray stars). (bottom) The years and seasons corresponding to each campaign.

Citation: Bulletin of the American Meteorological Society 102, 4; 10.1175/BAMS-D-20-0132.1

MICRE.

The DOE ARM program, the AAD and the BoM collaborated in deploying ground instrumentation to Macquarie Island between March 2016 and March 2018. Macquarie Island is located at 54.5°S, 158.9°E (north of the oceanic polar front, Fig. 1) and has a small research station operated by the AAD that is staffed year-round, in part by BoM. The station supports a variety of research activities and includes a long history of surface weather and radiosonde observations (Hande et al. 2012; Wang et al. 2015).

The primary objective of MICRE was to collect surface-based observations of radiation, precipitation, BL clouds, and aerosol properties in order to evaluate satellite datasets and to improve knowledge of diurnal and seasonal variations, especially with regards to the vertical structure of BL clouds. Instrumentation deployed during MICRE is listed in Table ES1 (in the online supplement; https://doi.org/10.1175/BAMS-D-20-0132.2), along with time periods for which high-quality observations are available for each instrument in Table ES2. The data include (i) passive surface radiation (solar, longwave, microwave); (ii) surface precipitation rain rates, types and particle sizes; (iii) cloud radar reflectivity and Doppler velocity profiles, ceilometer and lidar backscatter (including depolarization) measurements (that provide information on cloud occurrence, cloud-base and cloud-top height, precipitation particle size and phase, and some vertically resolved aerosol optical properties in cloud-free conditions); (iv) ground-based number concentrations of total aerosol and CCN; and (v) ground-based INP number concentration and type (via filter sample analyses).

CAPRICORN.

CAPRICORN was a sea-based field study using the Australian Marine National Facility (MNF) R/V Investigator, designed to better understand interrelated aerosol–cloud–precipitation–radiation processes responsible for surface SW radiation biases in global models and discrepancies between satellite rainfall measurements south of 40°S (e.g., Skofronick-Jackson et al. 2017; Protat et al. 2019a,b). The objectives were to (i) characterize cloud, aerosol, and precipitation properties, BL structure, biological production and cycling of dimethyl sulfide (DMS) in the upper ocean, atmospheric composition, and surface energy budget, as well as their latitudinal variability; (ii) evaluate and improve satellite products [with a focus on the NASA A-Train and NASA/JAXA Global Precipitation Measurement (GPM) mission cloud and precipitation products, and surface heat flux products]; and (iii) evaluate and improve the representation of these properties in the regional and global versions of the Australian Community Climate and Earth-System Simulator (ACCESS) model (Puri et al. 2013). A second voyage, CAPRICORN II, occurred simultaneously within the same overall region (south of Australia) as SOCRATES, and included four flights where the NCAR G-V aircraft passed over or near the R/V Investigator.

CAPRICORN I, held 13 March–15 April 2016 south of Tasmania, used the instruments listed in Table ES3. All instruments operated near 100% of the time, and characterized the basic atmospheric state (∼1 radiosonde per day), vertical cloud structure, including integrated liquid water and water vapor contents, cloud phase, and microphysical properties (based on cloud radar, lidar, and microwave radiometer measurements), and rainfall rates and drop size distributions (from disdrometer and micro rain radar measurements). The aerosol size distributions, morphologies and compositions, size-resolved chemical compositions and hygroscopic growth factors, cloud nuclei, CCN and INP concentrations, and some gaseous atmospheric compositions including DMS and VOCs, were measured. Bioaerosol size distributions, air–sea bulk and turbulent fluxes and surface energy budgets, and subsurface oceanic properties were also measured. Three CloudSatCALIPSO overpasses were successfully intersected by the ship. The online supplement contains more details about the CAPRICORN I voyage, including dates and locations of five cases when the R/V Investigator was in the cold sector of major cold fronts in Table ES4.

The main limitation of CAPRICORN I was its latitude span, with no measurements collected south of 55°S (Figs. 1 and 2), and the period (late austral summer–early fall, thereby not providing observations in the summer season where the largest surface radiation bias is found in GCMs). This motivated CAPRICORN II, where the same comprehensive set of data as CAPRICORN I was collected south of 55°S during summer. CAPRICORN II was held from 11 January to 21 February 2018, in combination with a major oceanographic project on the R/V Investigator aimed at quantifying changes in water properties and circulation of the SO, and measuring distributions of trace metals and isotopes in the SO and the physical, chemical, and biological processes controlling their evolving distributions. The objectives were similar to CAPRICORN I, with the additional aim to collect precipitation measurements within the swath of the GPM dual-frequency radar. The instrumentation was similar (Table ES5), with notable additions of the C-band dual-polarization Doppler radar (which did not operate during CAPRICORN I) and the NSF-funded contributions as part of SOCRATES that included radiosonde launches every 6 h, remote sensing instruments, and INP and bioaerosol measurements. More details about the voyage and cloud types sampled are included in the online supplement. Seventeen cases of collocated GPM observations were collected with rain, snow, and mixed-phase precipitation (Table ES6). The number of identified cold sectors and cold fronts traversed by the R/V Investigator during CAPRICORN II are listed in Table ES7.

Fig. 2.
Fig. 2.

Normalized frequency of sampling in each 1° latitude bin sampled during MARCUS (green), CAPRICORN I (dark blue), CAPRICORN II (light blue), and SOCRATES (yellow). These are based on the following total number of minutes that each campaign sampled south of 43°S: SOCRATES: 6,319 min; MARCUS: 182,470 min; CAPRICORN I: 44,408 min; CAPRICORN II: 60,060 min.

Citation: Bulletin of the American Meteorological Society 102, 4; 10.1175/BAMS-D-20-0132.1

MARCUS.

During MARCUS the DOE AMF2 instrument package, including the Aerosol Observing System (AOS) was installed on the AA as it made routine transits between Hobart, Australia and the Australian Antarctic stations of Mawson, Davis, and Casey, as well as Macquarie Island between 21 October 2017 and 23 March 2018. MARCUS observations enhance the CAPRICORN observations in that they were collected over a 5-month period centered upon the austral summer, allowing transitions from spring to fall to be observed across the 80 days of the MARCUS voyages. Because the data were collected during resupply voyages, the science team had no control on the timing of the voyages, nor could specific cloud types be targeted. Thus, a range of synoptic settings was sampled, providing knowledge of temperature-dependent distributions of cloud properties under a variety of aerosol and cloud conditions.

Specific objectives proposed for MARCUS were to 1) understand the synoptically varying vertical structure of SO BL clouds and aerosols; 2) quantify sources and sinks of SO CCN and INPs, including the role of local biogenic sources over spring, summer and fall; 3) quantify mechanisms controlling SLW and mixed-phase clouds; and 4) advance retrievals of clouds, precipitation and aerosols over the SO. Parameterization development and model evaluation requirements were integrated in MARCUS’s design so that systematic confrontation and improvement of GCMs and NWP is possible. Instrumentation deployed during MARCUS, listed in Table ES8, included active and passive remote sensing instrumentation, in situ measurements of aerosols, bioaerosols, and INPs, trace gas measurements, and meteorological measurements including 6-hourly radiosonde launches, rain gauges, and disdrometers. The conditions sampled are listed in the online supplement including passages through cold fronts (Table ES9).

SOCRATES.

SOCRATES used the NSF–NCAR G-V aircraft to sample clouds, aerosols and precipitation along (primarily) north–south transects south of Hobart, Australia, reaching as far south as 62°S, from 15 January to 26 February 2018. The G-V made in situ measurements within the BL and free troposphere, and included remotely sensed measurements using a cloud radar and lidar. The G-V flight tracks, shown in Fig. 1, were designed to target the cold sectors of cyclones where models have the most trouble producing SLW, and thus were not the same for each day.

The overarching objectives of SOCRATES were similar to those of MARCUS, MICRE and CAPRICORN. In particular, the G-V was tasked to obtain a dataset characterizing the structure of the MBL and free troposphere over the SO, including observations of the vertical distribution and properties of clouds and aerosols, including CCN and INPs, so that possible mechanisms to explain the excessive absorbed SW radiation in models could be tested. The instruments and flight paths were designed to gather statistics on aerosols and clouds as a function of latitude, and included measurements over both the R/V Investigator during CAPRICORN II and Macquarie Island as explained in the online supplement. Table ES10 lists the instrumentation installed on the G-V including in situ cloud and aerosol probes and remote sensing devices. The online supplement also provides information about the sampling strategy that was used to execute flights collecting both in situ and remote sensing data along with a list of all the research flights (RFs) in Table ES11.

Figure 2 shows the normalized fraction of observations made at each latitude during the four campaigns. Apart from time spent at the Australian Antarctic stations and Macquarie Island for resupplying during MARCUS, there is an even sampling of latitudes during both CAPRICORN II and MARCUS. Both the R/V Investigator and RSV Aurora Australis spent a large time south of 60°S, providing a very rare and invaluable set of data over cold waters poleward of the oceanic polar front. MARCUS data are unique because they provide observations over the sea ice and cover more of a seasonal cycle (October–March) than CAPRICORN II, whereas CAPRICORN I and II included more thorough aerosol, oceanographic, and surface energy budget measurements to put cloud observations in context, and MICRE provides the longest seasonal cycle at a single location. SOCRATES provides the in situ observations that are critical for process studies and evaluation of remote sensing retrievals, and they are the only direct observations of aerosols below, inside and above cloud. Thus, the combination of CAPRICORN, MICRE, MARCUS, and SOCRATES data are synergistic in their characterization of the latitudinal and seasonal variabilities of aerosol–cloud–precipitation–radiation processes over the SO.

Preliminary findings

Much of the initial effort since completing the projects has focused on evaluation of data quality and development of higher-level data products, as well as characterizing cloud and aerosol conditions over the SO. Some of the more noteworthy findings are discussed here. Integration of the datasets and comparison with model simulations and satellite retrievals is starting, a necessary step to evaluate mechanisms responsible for the excess absorption of solar radiation over the SO, which is the overarching objective of these projects.

Latitudinal dependence/composition of surface aerosols.

Information on the composition and latitudinal dependence of aerosols is required to understand the origin of aerosols and the role of biological aerosols and sea salt on droplet nucleation in different locations and seasons. Surface aerosol volatility and hygroscopicity were measured during CAPRICORN I at diameters of 40, 100, and 150 nm to provide information about the composition of the Aitken and accumulation modes. Figure 3 shows that the daily averaged number fraction of low volatility aerosol (persisting at 250°C) in the Aitken mode was 0.22 ± 0.2 (mean ± 1σ), which indicates that the Aitken mode was largely composed of secondary non–sea salt sulfates. In the accumulation mode, the mean number fraction of low volatility particles was 0.79 ± 0.2, indicating most particles contained a primary sea spray sourced fraction. Low volatility sea spray particle number fractions, particularly the Aitken mode, increased at higher SO latitudes and were associated with higher wind speeds and generally lower particle number concentrations. The proportion of primary sea spray particles observed from volatility measurements during CAPRICORN I was larger than that observed from BL measurements via scanning transmission electron microscopy (STEM) during SOCRATES in the summer. Further information about aerosol composition and hygroscopicity measurements is provided in the online supplement, which include chemical characterization of single particle composition by scanning transmission X-ray microscopy by near-edge X-ray absorption fine structure (STXM-NEXAFS). These measurements show that particles in-cloud and below-cloud have very similar organic functional group compositions (Fig. ES6).

Fig. 3.
Fig. 3.

Daily mean number fraction of low volatility particles measured using the V-TDMA during CAPRICORN I. Preselected particle diameters were 40 nm (blue), 100 nm (yellow), and 150 nm (red). Error bars represent the standard deviation in the daily mean for each preselected particle size.

Citation: Bulletin of the American Meteorological Society 102, 4; 10.1175/BAMS-D-20-0132.1

The average CN number concentrations (diameter greater than 3 nm) during CAPRICORN I were 290 ± 170 cm−3, below typical summertime maxima (Gras and Keywood 2017; McCoy et al. 2015) and hence consistent with the seasonal cycle observed at Cape Grim, with summertime maxima of approximately 500–550 cm−3 and wintertime minima of approximately 150 cm−3. The seasonal cycle in SO aerosol number is largely driven by enhanced secondary sulfate production in the summer months (Gras and Keywood 2017; McCoy et al. 2015).

Information on surface fluorescent biological aerosol particles (FBAPs) was provided by the WIBS-4. It measures the fluorescence from single aerosol particles in three excitation/emission channels for particle sizes between 0.8 and 13 μm (Toprak and Schnaiter 2013) to deduce fluorescent (i.e., biological) and total aerosol number concentrations and size distributions. During MARCUS, the FBAP aerosol number concentration was rather low with a median value of 0.43 L−1 giving an average FBAP fraction of about 0.3% in the MBL at latitudes from 46° to 68°S consistent with WIBS-4 measurements in other projects. The total number concentration varied strongly with latitude while the FBAP concentration was rather stable with indicated minimum around –56° latitude and increasing concentrations toward the north and south. Implications of these results on the sources and sinks of aerosols over the SO are being examined in several publications under preparation.

BL aerosol and CCN vary according to origin.

Aerosol measurements in the BL but above the surface give more information about sources and sinks of aerosols, and their role in droplet nucleation. Ambient aerosols 150 m above the ocean were collected through a CVI inlet on the G-V, but without the counterflow airstream that excludes small particles. Particles in two dry diameter ranges were impacted onto carbon-coated nickel grids or silicon nitride windows and stored frozen for subsequent analysis by analytical STEM and X-ray spectroscopy that produces elemental inorganic composition of individual aerosol particles. The size ranges were about 0.1–0.5- and 0.5–5-μm diameter (50% cut size) for particle densities of 2 g cm−3 at 1,000 mb (1 mb = 1 hPa). Based on size distributions from the UHSAS, the 0.1–0.5-μm size range comprised between 54% and 93% of the aerosol accumulation-mode number concentration (above the Hoppel minimum), and aerosol concentrations > 0.1 μm were similar to nearby cloud droplet concentrations Nc. Thus, particles in this size range would be expected to be representative of the composition of most CCN for the cases analyzed. For the data presented here, heaters on the titanium inlet and stainless steel sample line were turned off to minimize losses of volatile species.

Figure 4 shows STEM results for six flights after grouping particles into different types based on elemental composition and morphology (Twohy and Anderson 2008). Figure 4a shows results for each flight, while Fig. 4b shows the overall mean composition. Particles 0.1–0.5 μm in diameter were dominated by sulfur-based particles (mean 69% by number). Based on the ionic composition measured on the R/V Investigator during CAPRICORN II these particles were primarily acidic sulfate, likely with a small contribution from methanesulfonic acid (MSA) and other organics (Twohy et al. 2021). The second-most frequent particle type in this size range (mean 28% by number) was salt-based sea spray. Figure 4b shows different types of sea spray, which were dominated by unprocessed, sodium chloride–based sea spray particles. However, about 40% of sea spray particles were enriched in sulfur and depleted in chlorine through uptake and condensation of sulfur gases (McInnes et al. 1994), and a small percentage (3%) were salts enriched in calcium or magnesium. Crustal and metallic particles and externally mixed organics were also detected in the <0.5-μm population in approximately equal proportions, but were together only about 3% by number. Overall these data indicate that 0.1–0.5-μm particles in the BL were dominated by biogenic sulfates, with a smaller but significant contribution from sea spray. Particles > 0.5 μm (not shown) were dominated by sea spray, with only about 2% other aerosol types. Further, many sea spray particles in the larger size fraction had detectable carbonaceous coatings, which may be important in ice nucleation in the marine environment (McCluskey et al. 2018a). More information about the chemical composition of the organic compounds is shown in the supplement.

Fig. 4.
Fig. 4.

(a) Compositional fraction of total particles by number in the 0.1–0.5-μm dry size range for six 150-m samples on SOCRATES flights 11–15. (Sample times in UTC: 11-D: 0441:30–0446:30; 12-D: 0453:00–0455:00; 13-G: 0426:20–0431:20; 13-I: 0526:00–0530:00; 14-C: 0422:30–0426:00; 15-F: 0559:40–0604:50.) Categories: S-containing: with S (and sometimes, O) primary elements. Sea spray: Salts of Na, Cl, Mg, S, K, Ca, sometimes with organic coatings. Other types: Includes crustal dust (silicates and carbonates), metals (Al, Fe, Cr, Ti, Mn, Co, Zn, Cu, O, etc.), primary organics (C and sometimes O), and combustion particles (high S, C, O with K). (b) Average composition by number for all six samples; smaller inset further subdivides sea spray into different types of salts.

Citation: Bulletin of the American Meteorological Society 102, 4; 10.1175/BAMS-D-20-0132.1

Direct observations of CCN make it possible to understand how aerosols act as CCN. To investigate controls of CCN, the variability in CCN spectra in the BL was characterized using a k-means clustering to group into four clusters associated with the observed bimodality in CN and CCN concentrations. Minima in the bimodal frequency distributions of number concentrations occurred at approximately 750 cm−3 for CN and was dependent on supersaturation for CCN (Fig. 5a). The four clusters were characterized as follows: 1) low CN/high CCN—southerlies influenced by Antarctic coastal biological productivity; 2) high CN/low CCN—westerlies over the SO characteristic of recent particle formation (RPF) events with low accumulation mode concentrations due to recent precipitation; 3) high CN/high CCN—similar characteristics as high CN/low CCN but with condensational growth of recently formed particles to CCN sizes; 4) low CN/low CCN—aerosol populations scavenged by precipitation and lack of RPF.

Fig. 5.
Fig. 5.

(a) CCN spectra measured in the marine boundary layer, clustered by CN and CCN concentrations. Error bars represent the standard error. (b) Measured below-cloud CCN concentrations at 0.3% supersaturation vs the observed in-cloud droplet number concentration (CDNC). The black line is a linear fit with r = 0.75. The same color scheme used to identify the cluster is used in both figures. The white points in (b) represent measurements that were not clustered due to missing data.

Citation: Bulletin of the American Meteorological Society 102, 4; 10.1175/BAMS-D-20-0132.1

The CCN concentrations (at 0.3% supersaturation) correlated well (Fig. 5b) with the overlying Nc indicating large variations in CCN over the SO exist and have an important influence on cloud microphysics. The large variability in CCN led to larger than expected variability in Nc, which ranged from 10 to 449 cm−3. The variation in CN concentration was also notable, ranging from 115 to 1,153 cm−3. To understand this variability, HYSPLIT (Stein et al. 2015; Rolph et al. 2017) back trajectories were performed to identify differences in source location and transport history. The back trajectories for the low CN/high CCN were consistently from the south (Fig. 6d) along the Antarctic coast. This source location is associated with upwelling and marine biological productivity that produces biogenic gases such as DMS, which can oxidize and condense to form CCN-active particles (Hegg et al. 1991; Covert et al. 1992; Andreae et al. 1995; Read et al. 2008; Sanchez et al. 2018). The two clusters with westerly back trajectories (Figs. 6a,b) contained the highest CN. High CN over the pristine SO are likely due to RPF aloft (“New particle formation in free troposphere” section) and mixed downward into the MBL (Sanchez et al. 2018). The high CN/low CCN cluster contained low concentrations of accumulation mode particles (and consequently, small total aerosol surface area) making conditions ideal for particle formation leading to high CN concentrations (Warren and Seinfeld 1985; Clarke 1993; Pirjola et al. 2000). While the high CN/high CCN cluster did not have low accumulation mode concentrations, the spikes in CCN concentrations at the highest supersaturations (Fig. 5a, >0.6%) are consistent with RPF where some particles grow to CCN sizes, typically through condensational growth during long residence times over the ocean (Russell et al. 1998; Bates et al. 2000; Kulmala et al. 2004; Rinaldi et al. 2010; Zhang et al. 2014). Sanchez et al. (2021) and the online supplement offers more information on how the back trajectories were combined with ECMWF reanalysis to identify relations between BL cloud fraction and particle concentration.

Fig. 6.
Fig. 6.

HYSPLIT 5-day back trajectories for the four clusters shown in Fig. 5, (a) high CN/high CCN, (b) high CN/low CCN, (c) low CN/low CCN, and (d) low CN/high CCN. The magenta circles represent the HIAPER G-V location used to initialize the back trajectory.

Citation: Bulletin of the American Meteorological Society 102, 4; 10.1175/BAMS-D-20-0132.1

New particle formation in free troposphere.

Analysis of free-tropospheric (3–6 km) aerosol measurements from the G-V identified signatures of RPF events occurring frequently across the SO, often in association with synoptic uplift. It is hypothesized that air masses rich in precursor gasses (i.e., emissions from phytoplankton at the surface) undergo rapid synoptic uplift, are processed through the associated convection, cleansed of coarse and accumulation mode aerosol, and released into the free-tropospheric, low-aerosol surface area environment where gas-to-particle conversion is favored (McCoy et al. 2021). This synoptic uplift mechanism is complementary but independent from RPF occurring in the outflow of SO shallow cumulus clouds documented during ACE-1 (Clarke et al. 1998). It is likely both contribute to the widespread observations of high Aitken aerosol number concentrations throughout the SO free troposphere.

A free-tropospheric sample from RF09 is used to illustrate the synoptic-uplift mechanism (Fig. 7). During RPF events, simultaneously low accumulation mode aerosol number concentrations (from the wing-mounted UHSAS, 100 < D < 1,000 nm) and high total number concentrations (from the CN counter, D > 11 nm) occur, indicating presence of large Aitken mode concentrations. High concentrations and rapid spatial variability in CN suggest sampling of particle formation bursts or air masses at different stages of nucleation (Clement et al. 2002). RPF occurrences were prolific during RF09 due to a warm conveyor belt occurring west of Australia and propagating south-east toward Antarctica. For statistical airmass evolution analysis, HYSPLIT (Stein et al. 2015) 72-h back trajectories initiated at 10-min intervals along the flight path are identified by maximum CN into RPF (CNMax ≥ 2,500 mg−1) and non-RPF events (CNMax < 2,500 mg−1) (Fig. 7b). Standard temperature and pressure corrected units (mg−1) are used to enable altitude invariant analysis across the campaign. The majority of these RF09 air masses are RPF and have undergone recent synoptic uplift (ascent exceeds characteristic vertical velocity for synoptic events, ∼1 cm s−1; Holton and Hakim 2013) in the previous 20–30 h. In the 72-h before sampling, the majority of these air masses have access to the surface (Z < 1 km) and the precursor gases necessary for generating new particles in a low aerosol surface area environment.

Fig. 7.
Fig. 7.

(a) Time series of total CN (D > 11 nm) and accumulation mode (wing mounted UHSAS, 100 < D < 1,000 nm) aerosol number concentrations sampled in the free tropospheric survey leg (∼6 km) of RF09 as the G-V flew south. (b) HYSPLIT 72-h back trajectories of air masses initiated in 10-min intervals [starred locations in (a) with color-coded matched time in (b)] along the G-V flight path. Trajectories dominated by RPF events are identified by where maximum CN over the corresponding 10 min exceeds 2,500 mg−1 [dotted line in (a)]. In this case, only one trajectory does not satisfy the criteria for RPF events [dashed line in (b)]. The majority of trajectory ascents exhibit synoptic uplift (3–6 cm s−1) within 20–30 h of SOCRATES sampling (star) and are in proximity to phytoplankton emissions (Z < 1 km) in the prior 72 h.

Citation: Bulletin of the American Meteorological Society 102, 4; 10.1175/BAMS-D-20-0132.1

RF09 is characteristic of RPF events during SOCRATES and their connection to synoptic uplift. The two most frequent large-scale uplift mechanisms associated with RPF events are warm conveyor belts and subpolar vortices. Volatility analysis via comparison of heated to unheated CN concentrations confirms that the particles sampled during RPF events are likely composed mostly of H2SO4, a prominent aerosol precursor gas arising from phytoplankton emissions. It is likely that the high concentrations of Aitken-mode aerosol particles produced above cloud by these RPF events are brought into the BL (Covert et al. 1996) and influence the subcloud CN and CCN concentrations (McCoy et al. 2021; Sanchez et al. 2018) (“BL aerosol and CCN vary according to origin” section). This source of Aitken mode aerosol above cloud may help to buffer SO clouds against precipitation removal, sustaining higher than expected Nc (on the order of 80–100 cm−3 between 45° and 62°S), and explains the larger contribution of sulfur-based particles to subcloud CCN compared to sea spray (“BL aerosol and CCN vary according to origin” section; Twohy et al. 2021; McCoy et al. 2021; Sanchez et al. 2021). Evidence supporting this hypothesis, the broader implications for SO cloud–aerosol interactions, and a more detailed assessment of the synoptic uplift mechanism are presented in McCoy et al. (2021).

Low INP concentrations over SO.

To investigate the processes giving rise to extensive SLW over the SO, not only is information about CN and CCN needed, but also about INPs. INP measurements were conducted during the various SO projects to define the spatial and temporal distributions of INPs over the region for the first time since the comprehensive measurements of Bigg (1973). A summary of campaigns, dates and INP sampling methods are given in the online supplement and in Table ES12. Wide regions of the surface marine BL were sampled south of 45°S, while INP measurements on the G-V were tailored to the standard flight patterns.

Figure 8 gives a broad overview of the INP datasets by focusing on the IS data collected during the four ship campaigns. Key findings are the large variability of, but generally very low, INP concentrations at any particular latitude, a weak overall latitudinal dependence, with highest concentrations near landmasses (especially toward Australia), and the large discrepancy with historical measurements over the region, first pointed out in the CAPRICORN I study by McCluskey et al. (2018b). McCluskey et al. (2018b) demonstrated that INP concentrations were up to 100 times lower during CAPRICORN I than measured by Bigg (1973) over some of the same regions, that INPs were (excepting episodic events) often predominately organic in nature with contributions of both heat labile and more stable organics, and that the INP content of Austral summer SO seawater samples were lower than those found in Arctic seawater. These results are consistent with a primary ocean sea spray source of SO BL INPs and also lower derived INP site densities (INPs per aerosol surface area) for immersion freezing in SO air compared to North Atlantic air masses. Using CAM5 with constrained meteorology, McCluskey et al. (2019) demonstrated that using parameterizations linking the number concentrations of mineral dust and surface area of sea spray aerosols in the global aerosol model could predict the magnitude of INPs observed in CAPRICORN I, and that sea spray organic INPs dominated on average, but that episodic incursions of inorganic mineral dust INPs present in the middle troposphere could occur and then dominate ice nucleation in the MBL. This vertical structure of compositions is demonstrated in analyses of collected aerosol compositions above, below and within clouds during SOCRATES (Twohy et al. 2021).

Fig. 8.
Fig. 8.

INP number concentrations per volume of air at −20°C over the SO region for a selection of the studies (MARCUS, CAPRICORN I, CAPRICORN II, TAN1502) listed in Table ES12. Each data point represents the midpoint position of a single filter collection. Historical data from Bigg (1973) are shown at right for context (each is the mean of numerous measures at that latitude); all are the same color since all were >0.1 per standard liter (sL−1).

Citation: Bulletin of the American Meteorological Society 102, 4; 10.1175/BAMS-D-20-0132.1

The INP datasets remain to be fully explored to investigate spatial, temporal and compositional variabilities, through aligning with aerosol data including real-time bioaerosol and next-generation DNA sequencing of bacteria. Those bacterial sequencing analyses have been completed for aerosol samples collected on equivalent filters to the INP units during CAPRICORN II. Results reported in Uetake et al. (2020) indicate the predominance of marine bacteria in the MBL during the ship campaign, confirming the pristine marine source of aerosols, and thus INPs, under most circumstances in this region. Comprehensive INP data from all SO studies will ultimately be normalized for use in parameterization development (see, e.g., McCluskey et al. 2018c, 2019; Vignon et al. 2021), and will serve as a basis for constraining primary ice nucleation for comparison with observations of ice formation and numerical model simulations of SO clouds.

Clouds: In situ observations of variability in liquid cloud droplet number concentration.

In situ G-V observations allow for process studies to investigate aerosol–cloud interactions and processes controlling distributions of SLW. For example, using data obtained during four ramped ascents and descents through BL clouds, Fig. 9 shows Nc measured by the CDP as a function of altitude. Although all profiles were collected in a similar geographical area on two different days, there is considerable variability in Nc, ranging from less than 50 cm−3 near cloud top on RF08 at latitude 55.8°S to greater than 450 cm−3 near the top and in midcloud layer for the same flight further south at 58.7°S. Although some lower Nc, such as concentrations of about 50 cm−3 seen on RF04, were associated with lower wind speeds averaging 5.5 m s−1, and some higher concentrations of 250 cm−3 on RF08 at 59.9°S and up to 450 cm−3 on RF08 at 58.7°S were associated with larger wind speeds averaging 20.5 and 22.0 m s−1, respectively, correlation with wind speed was not always the case (e.g., low Nc of less than 50 cm−3 on RF08 at 55.8°S occurred when wind speeds were 26.6 m s−1) as updrafts, dynamics, turbulence, and coupling of the cloud with the surface layer can also affect Nc. Thus, while generation of sea salt CCN caused by breaking waves associated with high winds likely contribute to variations in Nc, other factors also contribute significantly, such as the influence of source regions with different bioactivity on the production of CCN and the degree of coupling between the surface and cloud.

Fig. 9.
Fig. 9.

Cloud droplet concentration measured as a function of altitude for four ramped ascents/descents through boundary layer cloud for the days indicated in the legend. The time period, average latitude, and average wind speed of each ascent or descent are indicated. Each circle represents a 2-s average.

Citation: Bulletin of the American Meteorological Society 102, 4; 10.1175/BAMS-D-20-0132.1

Variable but prevalent supercooled water observed in situ.

Although SLW dominated many BL clouds observed during SOCRATES, information about ice crystals, when present, is important for understanding SLW persistence and cloud glaciation. During SOCRATES ice particle number concentrations and high-resolution images were acquired over a large range of temperature with optical array probes and the PHIPS probe. Using cloud phase determined with a combination of in situ cloud probes (J. D’Alessandro et al. 2021, manuscript submitted to J. Geophys. Res. Atmos.), Fig. 10 shows the distributions of phases as a function of temperature. In some instances, even the identification of phase is poorly defined (e.g., Korolev et al. 2017) as, for example, there is no consensus on how many ice crystals need to be mixed within a sample volume of water drops to be mixed- rather than a liquid-phase cloud. For analysis of in situ data, the term mixed-phase refers to the occurrence of a liquid mass fraction between 0.1 and 0.9 in a 1-s time period as calculated using data from a combination of size-resolved and bulk mass in situ probes. In addition to the frequent presence of clouds made exclusively of SLW at very low temperatures, another notable feature was the frequent observation of glaciated clouds at relatively high temperature corresponding to the Hallett–Mossop (H-M) range of −2°C < T < −8°C. Figure 11 shows a collection of representative ice particles images captured by the PHIPS during RF02 between 0° and −5°C. Typical ice particle habits were needles that were frequently rimed, thus acting as possible rime splintering sources in the H-M process. Smaller (D < 100 mm) pristine hexagonal columns and plates were also observed that possibly grew from ice splinters (Korolev et al. 2020) (Fig. 11, first row). Some of the pristine small particles were observed to have been scavenged by the larger needles and needle aggregates (Fig. 11 third row, second needle from the left). Occasionally, frozen drizzle droplets were detected—either as complete or sometimes as fractured particles (Fig. 11, second row).

Fig. 10.
Fig. 10.

Relative occurrence frequency of different phases derived from suite of in situ probes as a function of temperature (adapted from J. D’Alessandro et al. 2021, manuscript submitted to J. Geophys. Res. Atmos.). The black line indicates the number of samples, giving some information about statistical significance of results.

Citation: Bulletin of the American Meteorological Society 102, 4; 10.1175/BAMS-D-20-0132.1

Fig. 11.
Fig. 11.

Collection of PHIPS images of ice particles from RF02 sampled in the temperature range from 0° to −5°C.

Citation: Bulletin of the American Meteorological Society 102, 4; 10.1175/BAMS-D-20-0132.1

The presence of small horizontal scale generating cells were noted near the tops of BL stratocumulus and higher cloud layers in the free troposphere. Such cells are small regions of high reflectivity that frequently produce precipitation streaks below. Although such cells have been observed in other environments, such as the Arctic (McFarquhar et al. 2011) and midlatitudes (Plummer et al. 2014), the cells observed over the SO had smaller horizontal scales and thus their structure and properties need to be determined to understand precipitation development and cloud life cycles. Wang et al. (2020) provide this characterization using times when the G-V was flying near cloud top. Figure 12 shows an example of their approach whereby the probability distribution functions of liquid water content, total concentration, and ice water content were compared inside and outside of generating cells identified by the HCR. All three parameters are higher to a statistically significant degree inside the cells, but substantial liquid water and numbers of particles also occur outside the cells. This shows that although the cells provide a favorable environment for particle nucleation and growth, turbulent mixing at cloud top reduces the gradients inside and outside of the cells. The online supplement gives extra information on how the combination of in situ and remote sensing measurements can be used to identify and characterize the finescale structure of SLW, SLD, drizzle, and ice crystal type.

Fig. 12.
Fig. 12.

(a) Altitude–time cross section of HCR equivalent reflectivity factor Ze over the time period between 0030 and 0034 UTC 8 Feb 2018. The black line shows the flight level of the G-V aircraft, and the shadows represent the location of the generating cells identified by the method of Wang et al. (2020). Other plots show statistical analysis of data collected over this time as shown by kernel probability distribution functions of properties inside (pink) and outside (blue) of generating cells for (b) Nt, (c) LWC, and (d) IWC. Black boxplots show 5th, 25th, 50th (red line), 75th, and 95th percentiles of data. White points indicate the mean value. The width of the red and blue shaded area represents the portion of data located at particular value.

Citation: Bulletin of the American Meteorological Society 102, 4; 10.1175/BAMS-D-20-0132.1

The location of liquid water is of interest not only for understanding cloud microphysical process and radiative properties, but also for assessing the potential hazard posed by aircraft icing when the liquid is supercooled. Icing is a significant hazard for aviation, especially over the SO, and is most concerning as the droplets become large enough to impact on areas not typically protected by anti-icing or de-icing systems. Small cloud droplets (D < 50 μm) tend to impact on forward edges of aircraft as seen in Fig. 13, while larger drops tend not to freeze on impact and instead flow back farther before freezing or else are heavy enough to be somewhat independent of the airflow and actually impact the aircraft behind the forward edges (Fig. 13b) (FAA 2015; Cober and Isaac 2012). Known as supercooled large drops (SLD), they can accrete on the wing and other important control areas of the aircraft which are outside the heated surfaces. Freezing drizzle and freezing rain are examples of SLD.

Fig. 13.
Fig. 13.

(a) The nose cone of the NSF–NCAR G-V aircraft large wing pod during an encounter with small supercooled cloud droplets during SOCRATES. Ice has accreted on the tip of the pod. (b) The nose cone of the G-V wing pod under conditions where larger supercooled droplets were encountered. Notice the drops have impacted and run back before freezing further back on the pod.

Citation: Bulletin of the American Meteorological Society 102, 4; 10.1175/BAMS-D-20-0132.1

Observing secondary ice production (rime splintering) over SO.

Two research flights (RF11 and RF15) during SOCRATES were dedicated to the sampling of shallow cumulus clouds in the cold sector of extratropical cyclones to understand the possible maintenance of SLW in those clouds. Mossop (1970) had found ample evidence of secondary ice production by rime splintering in cumuli sampled off the western and eastern coasts of Tasmania. Because of the need to focus sampling at multiple levels in the same cumulus field, there was insufficient time to sample the cumuli using the standard curtain flight pattern to 60°S. Thus a population of cells as far south as possible, near 55°S, was identified for sampling. Thereafter the G-V flew a series of constant altitude legs about 15 min long targeting the tops of actively growing cells, and also sampling at and below cloud bases, and above the cloud tops, to measure aerosol, CCN and INPs. These two SOCRATES flights provided clear evidence of rime splintering, farther away from land sources than documented before. Of the 34 sampled shallow cumuli occupying temperatures where rime splintering can act (from −3° to −9°C), 47% contained regions where ice crystals were orders of magnitude more than the INP observed (Scott 2019). The SOCRATES airborne radar data captured the cloud macrostructure needed to place the in situ microphysical data collected near the cloud tops in context (Fig. 14). A complex, multithermal structure was common in clouds exhibiting the features of rime splintering, and lacking in clouds that only contained SLW. These new data are being used to guide and constrain detailed process-level numerical modeling, to understand why some SO cumuli glaciate by this mechanism, while others do not.

Fig. 14.
Fig. 14.

Aircraft pass on 17 Feb 2018, through the tops of two closely spaced cumuli at a temperature of −9°C, as shown by the downward-pointing airborne Doppler radar on the G-V aircraft [(top) radar reflectivity and (middle) vertical velocity; positive values denote upward motion of particles]. The black horizontal line at the top shows the aircraft location as it passed through the cloud tops. Shaded gray boxes demarcate the two clouds, with shaded blue boxes identifying 10-s regions with the highest maximum ice number concentrations (corrected for possible shattering artifacts) as labeled. (bottom) Snapshot from forward video also shows the multiple thermal structure of these cumuli as the aircraft approached. (inset) Particle images from PHIPS probe indicated rimed ice and liquid drops were present, necessary for rime splintering to be active, as well as the expected products of rime splintering, pristine, and rimed columns.

Citation: Bulletin of the American Meteorological Society 102, 4; 10.1175/BAMS-D-20-0132.1

Himawari-8 retrievals consistent with field observations.

Satellite data provide both a large-scale context for interpretation of finer-resolution remote sensing data and in situ measurements. The Himawari-8 satellite, developed and operated by the Japan Meteorological Agency, has provided a significant advance in geostationary satellite capability over the Asia–Oceania region. It provided rapid updates on meteorological conditions and cloud systems throughout the SO campaigns which were especially critical for aircraft operations.

Figure 15 shows the frequency of occurrence of Himawari-8 cloud type (Pavolonis 2010) as a function of cloud-top temperature (Heidinger 2011) for the 15 SOCRATES flights. For the duration of each flight, data are taken from a rectangular area that extends from 45° to 63°S and covers the entire width of the flight track. The statistics represent the overall atmospheric and cloud conditions sampled during both the in situ and remote sensing sampling legs. Figure 15 highlights the prevalence of SLW cloud tops for 0°C > T > −25°C. More details of this Himawari-8 cloud classification can be found in Huang et al. (2019), and information about how the in situ cloud properties are being used to evaluate cloud microphysical properties is included in the online supplement.

Fig. 15.
Fig. 15.

Frequency of occurrence of Himawari-8 cloud type as a function of cloud-top temperature for the 15 SOCRATES missions (includes times for both outbound and inbound legs).

Citation: Bulletin of the American Meteorological Society 102, 4; 10.1175/BAMS-D-20-0132.1

Radiative fluxes confirm bias in climate models.

Many studies involving surface radiative fluxes rely on fluxes retrieved from satellites, primarily from the Clouds and the Earth’s Radiant Energy System (CERES) instruments or derived from spaceborne cloud radar and lidar observations (CloudSatCALIPSO). Based on CERES data, most climate models participating in CMIP5 had excessive SW radiation reaching the surface over the SO (Zhang et al. 2016). An evaluation of CERES Synoptic (SYN) and Energy Balanced and Filled (EBAF) edition 4 and CloudSat retrieved surface SW and longwave (LW) downwelling fluxes against surface observations collected during MICRE (Hinkelman and Marchand 2020) finds that the overall biases in the CERES-surface fluxes are modest, but slightly larger at Macquarie Island than at most other locations, approximately +10 W m−2 for the SW and −10 W m−2 for the LW in the annual mean. The SW bias is positive meaning that climate model biases in downwelling SW fluxes are, if anything, slightly larger than previous studies suggest because CERES downwelling fluxes may be a bit too large and model fluxes are larger yet. However, both the SW and LW bias have significant seasonal and diurnal variations, with SW biases being near +20 W m−2 during the SH summer. Biases in LW fluxes are much larger at night (−16 W m−2) than during the day (<2 W m−2) with significant seasonal variations controlled by the relative ratio of daytime versus nighttime, and consequently are largest during the SH winter. This thus confirms that the climate model biases that motivated the projects are indeed real.

Low clouds responsible for much of climate model bias.

Understanding the contributions of different cloud types to the surface SW radiation bias in models is a major objective of these field campaigns, which complements the analysis of large-scale environments most conducive to such biases. Figure 16 shows the observed and modeled surface cloud radiative effect (CRE) during CAPRICORN I for different cloud cover types using the BoM ACCESS-C3 numerical weather prediction system (4-km horizontal resolution, no data assimilation, downscaled from the regional 12-km-resolution model), which was run for the campaign period. The observed CRE is the difference between the measured downwelling radiative flux at the surface and the simulated clear-sky downwelling radiative flux, accounting for ocean albedo and the broadband infrared emissivity of seawater (e.g., Protat et al. 2017). Over the CAPRICORN I period, the mean SW CRE was −66.7 W m−2, partially offset by a mean LW CRE of 44.4 W m−2, resulting in a mean net CRE of −22.3 W m−2 (Fig. 16d). A 1-min merged cloud radar–lidar product from ship-based measurements was used to classify the observed cloud profiles into different cloud cover types (Noh et al. 2019) at 1-h resolution to compare with model outputs. Hours that contained only clear skies were classified as clear. Hours that contained more than 30 min of clear skies were classified as “mostly clear.” Because the cloud cover types containing clouds overlapping low-, mid-, and high-altitude slabs made up only 5% of all the observations, they were grouped into a “thick” cloud type classification. Hours that contained at least 15 min of precipitating clouds were classified as precipitating, even if one of the other conditions was met. Last, hours that did not meet any of these conditions were classified as “mixed.”

Fig. 16.
Fig. 16.

The mean CRE biases [(a) shortwave, (b) longwave, and (c) net] in ACCESS-C3 relative to the mean measured CRE for different cloud types over the Southern Ocean during CAPRICORN I. Vertical error bars represent the standard deviation of each CRE bias for each cloud type. The size of the points is proportional to the number of sampling hours of each cloud type. (d) The decomposed CRE and CRE biases, weighted by the relative frequency of occurrence for each cloud type. The black horizontal bars represent the total CRE and CRE biases for the whole campaign.

Citation: Bulletin of the American Meteorological Society 102, 4; 10.1175/BAMS-D-20-0132.1

During CAPRICORN I, 51% of the 697 observation hours were characterized by low clouds, followed by multilayer (14%), precipitating (10%), mostly clear (10%), mixed (8%), thick (5%), and clear conditions (2%). Large negative SW CREs are observed for precipitating, multilayer and low cloud categories and these correspond to a mean positive SW CRE bias for all three clouds types, meaning too much SW flux is reaching the surface in the model under these conditions (Fig. 16a). The negative SW CRE and positive SW CRE bias was partially offset by positive LW CREs and a negative LW CRE bias for these cloud types (Fig. 16b). This resulted in a net negative CRE and a positive net CRE bias for precipitating, multilayer and low clouds (Fig. 16c). While smaller negative (positive) SW (LW) CREs were observed for the other cloud types, the CRE biases for these had little impact on the overall CRE and CRE bias once weighted by their respective frequency of occurrence (Fig. 16d). For the measured SW, LW, and net CRE, low clouds were responsible for just over half of the total contribution during CAPRICORN I, with most of the remaining contributions from multilayer and precipitating clouds. Interestingly, however, low clouds were responsible for nearly all of SW, LW, and net CRE biases in ACCESS, highlighting again the need to focus our attention on better understanding these low clouds. This work is being extended to include the MARCUS, MICRE, and SOCRATES observations.

Remote sensing data also show prevalence of supercooled water.

The online supplement summarizes studies that have used the CAPRICORN and MARCUS data to determine the frequency of and sources of SLW over the SO. To further understand processes responsible for the production and maintenance of SLW over the SO, and to understand the seasonal and latitudinal dependence of cloud properties, the MARCUS cloud retrievals were combined with a value added product developed to describe the environmental quantities at the position of the AA at 10-min resolution. Parameters examined include ship navigation parameters, local meteorological conditions, SST, location of the AA relative to the oceanic polar front, lower tropospheric stability, marine cold air outbreak index, inversion height, lifting condensation level, location relative to the center of the nearest cyclone, warm front and cold front and location of air parcels 72 h prior to their arrival at the ship computed from HYSPLIT. Consistent with prior satellite retrievals, the MARCUS data show that low-level liquid water clouds are ubiquitous over the SO and that much of the water is supercooled. For instance, south of 60°S over 49% of nonprecipitating clouds had cloud base T < 0°C and mean liquid water paths greater than 50 g m