Embed Size (px)
Citation preview
Queensland University of Technology
Science and Engineering Faculty
Chemistry, Physics and Mechanical Engineering
The role of marine biota on the composition and
concentration of potential cloud condensation
nuclei
Submitted in fulfilment of the requirement for the degree of
IF49: Doctor of Philosophy
Submission: 2019
Luke Cravigan
Bachelor of Applied Science/ Bachelor of Mathematics
Master of Applied Science (Research)
Supervisors
Prof. Zoran Ristovski (Principal)
Dr Branka Miljevic (Associate)
Dr Robyn Schofield (University of Melbourne)
Dr Melita Keywood (CSIRO)
2
1 Abstract
Aerosol-cloud interactions contribute the largest uncertainty to radiative forcing in atmospheric mod-
elling. Over the Southern Ocean, in particular, a strong shortwave bias has been consistently observed
in model output. The Southern Ocean bias has been attributed to poor representation of low-to-mid
level cloud, particularly over the high latitude Southern Ocean. This work reports the results from
measurements taken during four voyages over the Southern and South Pacific Oceans which aim to
characterise the aerosol sources and their contribution to cloud condensation nuclei concentrations. The
Surface Ocean Aerosol Production (SOAP) voyage took place in the highly productive water of the
Chatham Rise, east of New Zealand. Chamber measurements during the SOAP voyage characterised
the enrichment of organics to nascent sea spray aerosol and the modifications to sea spray aerosol wa-
ter uptake. The Cold Water Trial of the RV Investigator was a zonal transect of the Southern Ocean
south of Hobart in January and February 2015. The CAPRICORN voyage aimed at measuring aerosol
and cloud properties over the Southern Ocean during March and April 2016 and had a southern most
extent of 55 oS. The Ice-Edge to Equator voyage was a zonal transect from the edge of the Antarctic
sea ice to the equator in May and June 2016. The results from the Cold Water Trial, CAPRICORN
and Ice-Edge to Equator voyages have been grouped together to provide seasonal and zonal context to
the Southern Ocean aerosol.
Nascent SSA observations taken during the SOAP voyage indicated that the primary marine aerosol
was internally mixed with an organic component. Organics contributed up to 23% of the sea spray
mass for particles with diameter less than approximately 1μm, and up to 87% of the particle volume
in the Aitken mode. The organic fraction was largely composed of a polysaccharide like component,
characterised by very low alkane to hydroxyl concentration ratios, approximately 0.1 - 0.2. The en-
richment of organics was compared to the output from the chlorophyll-a based SSA parameterisation
suggested by Gantt et al. (2011) and the OCEANFILMS models. OCEANFILMS improved on the rep-
resentation of the organic fraction predicted using chlorophyll-a, in particular when the co-adsoprtion
of polysaccharides was included, however the model still under predicted the overall proportion of
polysaccharides and over predicted the lipid fraction. In addition to the impact of marine biota on
nascent SSA composition, the contribution of sea spray aerosol to the cloud condensation nuclei con-
centrations over the Southern Ocean was examined. The contribution of the nascent sea spray to cloud
condensation nuclei was lowest during the summer, averaging 48% of the cloud condensation nuclei at
0.2% SS in January and increased to 79% in May. The seasonal change in the sea spray contribution
to cloud condensation nuclei was driven by the seasonally varying secondary sulfate component. The
seasonal contributions to cloud condensation nuclei over the Southern Ocean are consistent with, and
provide valuable validation for, values determined using remote sensing techniques.
The nascent sea spray aerosol water uptake during the SOAP voyage was resilient to changes in the sea
spray aerosol organic fraction, and deviated from that expected from commonly used models assum-
ing full solubility, particularly at organic fractions greater than approximately 0.4. The water uptake
behaviour observed in this study is consistent with that observed for other measurements of phyto-
plankton blooms, and was attributed to the surface partitioning of the organic components which leads
3
to a decrease in particle surface tension and an associated increase in water uptake. The compressed
film model was used to estimate the influence of surface partitioning and it was found that when the
organic fraction was partitioned to the surface the error in the modelled hygroscopicity dropped sig-
nificantly. The difference between nascent sea spray aerosol cloud condensation nuclei concentrations
calculated using the full solubility assumption and those calculated using the compressed file model
was up to 17%. Surface tension modification was observed sporadically, suggesting that certain con-
ditions are required, these could include high organic fractions, intense phytoplankton blooms and an
intact sea surface micro-layer due to lower wind speeds. Further work should focus on determining the
specific species which drive the surfactant behaviour of sea spray aerosol and identify the regions and
conditions under which these species become important contributors to sea spray aerosol.
4
Contents
1 Abstract 3
2 Keywords 7
3 Statement of original authorship 8
4 Acknowledgements 13
5 Abbreviations and definitions 14
6 Introduction 16
7 Literature review 21
7.1 Marine aerosol size and number . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
7.2 Aerosol water uptake . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
7.2.1 Hygroscopicity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
7.2.2 Köhler theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
7.2.3 CCN activation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
7.2.4 Deliquescence and efflorescence . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
7.2.5 Mixtures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
7.2.6 Surface partitioning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
7.2.7 Water uptake measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
7.3 Primary marine aerosols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
7.3.1 Sea spray aerosol formation and measurement . . . . . . . . . . . . . . . . . . . 34
7.3.2 Sea spray aerosol water uptake . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
7.3.3 Sea spray aerosol composition . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
7.4 Secondary marine aerosols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
7.4.1 Secondary marine aerosol formation and measurement . . . . . . . . . . . . . . 42
7.4.2 Secondary marine aerosol water uptake and composition . . . . . . . . . . . . . 43
7.5 Climate influences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
7.5.1 SSA parameterisations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
7.5.2 Sources of CCN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
7.5.3 Representation in models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
7.6 Knowledge gaps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
8 Research design 52
8.1 Observation campaigns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
8.1.1 Surface Ocean Aerosol Production (SOAP) . . . . . . . . . . . . . . . . . . . . 53
8.1.2 Southern Ocean voyages (RV Investigator) . . . . . . . . . . . . . . . . . . . . . 55
8.2 Measurement instrumentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
8.2.1 SOAP instrumentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
8.2.2 RV Investigator instrumentation . . . . . . . . . . . . . . . . . . . . . . . . . . 61
5
8.3 Analysis and modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
8.3.1 SOAP analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
8.3.2 RV Investigator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
9 Results 70
9.1 SOAP results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
9.1.1 Ocean water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
9.1.2 Nascent SSA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
9.1.3 Ambient marine aerosol . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
9.2 RV Investigator results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
9.2.1 Marine biological activity and meteorology . . . . . . . . . . . . . . . . . . . . . 90
9.2.2 Aerosol concentration and size distributions . . . . . . . . . . . . . . . . . . . . 92
9.2.3 Aerosol composition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
9.2.4 Southern Ocean aerosol water uptake . . . . . . . . . . . . . . . . . . . . . . . . 102
10 Discussion and implications 105
10.1 SSA organic enrichment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
10.2 SSA organic partitioning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
10.3 SSA contribution to CCN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
11 Conclusions 117
12 References 119
13 Appendix A 137
14 Appendix B 141
6
2 Keywords
Cloud condensation nuclei, aerosol-cloud interactions, natural aerosol, marine aerosol, sea spray aerosol,
sea salt aerosol, primary organic aerosol, hygroscopicity, water uptake, compressed film, organic film,
aerosol surface tension, sea surface micro-layer, Southern Ocean, hygroscopicity, volatility.
7
3 Statement of original authorship
8
QUT Verified Signature
List of Figures
1 Sources and production mechanisms for CCN in the remote marine environment. . . . 17
2 Radiative forcing of climate between 1890 and 2011. . . . . . . . . . . . . . . . . . . . 18
3 Top of atmosphere outgoing shortwave radiation bias. . . . . . . . . . . . . . . . . . . 20
4 Log-normal approximation of the sub micrometre marine number size distributions. . . 22
5 Köhler curves for 50 - 500 nm ammonium sulfate. . . . . . . . . . . . . . . . . . . . . . 28
6 Deliquescence, efflorescence curve of 100 nm marine inorganic species. . . . . . . . . . 29
7 Compressed film modelled water uptake for ammonium sulfate seeds coated with malonic
acid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
8 Schematic of bubble bursting process generating film and jet drops. . . . . . . . . . . . 35
9 Size distributions for chamber measurements of nascent SSA. . . . . . . . . . . . . . . 36
10 OMF of sub-200nm SSA as a function of Chl-a concentration. . . . . . . . . . . . . . . 41
11 Schematic of organic enrichment process during bubble bursting. . . . . . . . . . . . . 46
12 Voyage map for SOAP study. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
13 Voyage track for RV Investigator Southern Ocean voyages. . . . . . . . . . . . . . . . . 53
14 Experimental schematic of nascent SSA chamber experiments. . . . . . . . . . . . . . . 58
15 Characterisation of biological activity for water samples used to generate SSA. . . . . . 72
16 Nascent SSA size distributions from chamber SSA measurements. . . . . . . . . . . . . 73
17 Summary of nascent SSA properties. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
18 Comparison of SSA organic fraction from volatility and from filter measurements. . . . 76
19 Inorganic and organic mass fractions from nascent SSA filter samples. . . . . . . . . . 78
20 HGF and OMF by phytoplankton bloom. . . . . . . . . . . . . . . . . . . . . . . . . . 80
21 HGF as a function or nascent SSA organic volume fraction . . . . . . . . . . . . . . . . 81
22 Measured and modelled HGF as a function of OVF. . . . . . . . . . . . . . . . . . . . 82
23 Nascent SSA deliquescence curve. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
24 Timeseries of aerosol properties measured during SOAP voyage. . . . . . . . . . . . . . 84
25 Aerosol number size distributions from SOAP voyage by bloom. . . . . . . . . . . . . . 85
26 Ambient aerosol composition during SOAP voyage. . . . . . . . . . . . . . . . . . . . . 87
27 Ambient HGF distributions for SOAP voyage. . . . . . . . . . . . . . . . . . . . . . . . 88
28 SOAP ambient computed vs measured CCN concentration. . . . . . . . . . . . . . . . 89
29 Chl-a and meteorology during the Southern Ocean voyages. . . . . . . . . . . . . . . . 91
30 Aerosol concentration during the Southern Ocean voyages. . . . . . . . . . . . . . . . . 93
31 Aerosol size distribution during the Southern Ocean voyages. . . . . . . . . . . . . . . 94
32 Aerosol size distribution by wind speed for the Southern Ocean voyages. . . . . . . . . 95
33 SSA number concentration as a function of wind speed. . . . . . . . . . . . . . . . . . 96
34 Non-volatile aerosol number fraction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
35 Aerosol volume fraction remaining. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
36 Comparison of SSA log-normal mode with volatility and composition measurements. . 100
37 ACSM aerosol composition by voyage and latitude. . . . . . . . . . . . . . . . . . . . . 101
38 HGF distributions for the Southern Ocean voyages. . . . . . . . . . . . . . . . . . . . . 103
39 Modelled and measured CCNc from the Southern Ocean voyages. . . . . . . . . . . . . 104
10
40 Filter organic mass fraction as a function of seawater chlorophyll-a concentration. . . . 106
41 Caprison of OMF modelled using chl-a and using OCEANFILMS. . . . . . . . . . . . . 106
42 Comparison of modelled and measured organic composition. . . . . . . . . . . . . . . . 108
43 Compressed film model ambient HGF error. . . . . . . . . . . . . . . . . . . . . . . . . 109
44 Compressed film mode heated HGF error. . . . . . . . . . . . . . . . . . . . . . . . . . 111
45 Modelled (ZSR and compressed film) and measured nascent SSA HGF. . . . . . . . . . 112
46 Modelled CCNc concentrations and surface tension as a function of SSA OVF. . . . . . 113
47 Seasonality in fraction of CCN (0.5% SS) from SSA. . . . . . . . . . . . . . . . . . . . 115
48 Seasonality in fraction of CCN (0.2% SS) from SSA. . . . . . . . . . . . . . . . . . . . 115
49 Fraction of CCN from SSA by latitude. . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
11
List of Tables
1 HGF, deliquescence relative humidity (DRH), hygroscopicity parameter (kappa) and
volatilisation temperature (Tv) for atmospherically relevant species.Kappa calculated
from hygroscopic growth measurements (kappaHGF) and from CCN measurements (kap-
paCCN) using kappa-Köhler model. Reproduced from Fletcher et al. 2007, Petters and
Kreidenweis 2007 and Cravigan et al. 2015. . . . . . . . . . . . . . . . . . . . . . . . . 25
2 Typical composition in the marine environment for given HGF ranges. . . . . . . . . . 33
3 Standard aerosol configuration for GLOMAP-mode. Source: Mann et al. (2010). . . . 50
4 Details of ocean water samples collected for generation of SSA. . . . . . . . . . . . . . 54
5 List of aerosol instrumentation deployed during SOAP voyage.Reproduced from Law et
al. (2017). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
6 Instrumentation deployed on the RV Investigator during the Cold Water Trial (CWT),
CAPRICORN and Ice Edge to Equator voyages . . . . . . . . . . . . . . . . . . . . . . 62
7 VH-TDMA operating parameters for the Southern Ocean campaigns onboard the RV
Investigator (Cold Water Trial, CAPRICORN and Ice-edge to Equator) . . . . . . . . 62
8 Nascent SSA log-normal parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
9 Average baseline aerosol number concentration (CN10) and CCN concentration by bloom
for the SOAP voyage. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
10 Satellite retrieved chlorophyll-a concentration averaged by voyage (representing the sea-
son) and by latitude range. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
11 Voyage average CN and CCN number concentrations for baseline conditions. . . . . . . 92
12
4 Acknowledgements
This work was supported through an Australian Government Research Training Program Scholar-
ship.
The SOLAS-SOAP experiment was funded by the NIWA programme on Oceanic Control of Atmo-
spheric Composition and supported by funding from NZ Ministry of Primary Industries. Logistic
support to enable these measurements was provided by the Royal Society of New Zealand Interna-
tional Mobility Fund Contract IMF11A-70. QUT participation was partly funded by the NZ RS&T
2011-12 International Mobility Fund. In addition I would like to thank NIWA researchers, in partic-
ular Cliff Law and Mike Harvey for their support and providing data crucial to the work reported
here.
I would like to thank AINSE Limited for providing financial assistance (Award - ALNGRA13048) to
enable the Ion Beam Analysis of elemental composition of marine aerosol filters collected during the
SOAP voyage. In addition I would like to thank the researchers at ANSTO, in particular Eduard
Stelcer and David Cohen for providing time and expertise in analysing these samples.
I would like to thank researchers at the SCRIPPS Institute of Oceanography for completing FTIR
analysis of filter samples collected during the SOAP voyage, in particular Robin Modini and Lynn
Russell.
Some of the data reported in this paper were obtained at the Central Analytical Research Facility
operated by the Institute for Future Environments (QUT). Access to CARF is supported by generous
funding from the Science and Engineering Faculty (QUT).
I would like to acknowledge the assistance of the Australian Marine National Facility, in particular the
support team on-board the RV Investigator. I would also like to thank researchers at CSIRO for the
provision of data crucial to the results reported herein, in particular Paul Selleck, Melita Keywood, Ruhi
Humphries, Jason Ward and James Harnwell. I would also like to acknowledge the researchers that
I have worked alongside during research voyages, in particular the CAPRICORN team and principal
investigator Alain Protat, I have learnt a lot from these people.
I would like to thank my supervisory team, I am extremely grateful for all of the opportunities that
they have given to me. In particular I would like to thank Zoran Ristovski.
Thank you to my ILAQH and QUT friends and colleagues, in particular Joel Alroe, Marc Mallet and
Reece Brown who took measurements and maintained instrumentation during various research voyages.
I have collaborated closely with Marc and Joel, in particular, and have sincerely enjoyed working with
them and their friendship.
I would also like to sincerely thank my friends and family for all of their love and support. Most of all
I would like to thank Emily, Willa and Maeby.
13
5 Abbreviations and definitions
• ACCESS - Australian Community Climate and Earth Systems Simulator• ACSM - Aerosol Chemical Speciation Monitor• AMS - Aerosol Mass Spectrometer• ANSTO - Australian Nuclear Science and Technology Organisation• APS - Aerodynamic Particle Sizer• CAP - CAPRICORN (see below)• CAPRICORN - Clouds, Aerosols, Precipitation, Radiation, and Atmospheric Composition over the South-
ern Ocean (CAPRICORN) voyage (March-April 2016 onboard the RV Investigator)• Chl-a - Chlorophyll-a concentration• CCN - Cloud Condensation Nuclei• CDNC - Cloud Droplet Number Concentration• CN - Condensation nuclei number concentration (interchangeable with aerosol number concentration)• CN10 - Number concentration of aerosol with diameter > 10 nm• CN3 - Number concentration of aerosol with diameter > 3 nm• CPC - Condensation Particle Counter• CSIRO - Commonwealth Scientific and Industrial Research Organisation• CTD - Sonde to measure the ocean Conductivity and Temperature as a function of water Depth• CWT - Cold Water Trial voyage (January-February 2015 onboard the RV Investigator)• DMS - Dimethyl Sulfide• DRH - Deliquescence Relative Humidity• ERH - Efflorescence Relative Humidity• FTIR - Fourier Transform Infra Red spectroscopy for organic functional group concentrations• GLOMAP - Global Model of Aerosol Processes• HGF - Hygroscopic Growth Factor• HYSPLIT - Hybrid Single Particle Lagrangian Integrated Trajectory Model• IBA - Ion Beam Analysis for elemental composition• I2E - Ice-Edge to Equator voyage (May - June 2016 onboard the RV Investigator)• LPS - Lipopolysaccharide• MAAP - Multi-Angle Absorption Photometer• MBL - Marine Boundary Layer• MF - Mass Fraction• MODIS - Moderate Resolution Imaging Spectrometer satellite imaging sensor• MNF - Australian Marine National Facility (owned and operated by CSIRO)• NIWA - New Zealand National Institute of Water and Atmospheric Research• NSS - Non-Sea Salt (used to distinguish secondary, non-sea salt sulfate)• NV - Number fraction of non-volatile particles• OMF - Organic Mass Fraction• OVF - Organic Volume Fraction• PMA - Primary Marine Aerosol• RH - Relative Humidity• RV - Research Vessel• SML - Sea Surface Micro Layer• SMPS - Scanning Mobility Particle Sizer• SO - Southern Ocean
14
• SOAP - Surface Ocean Aerosol Production voyage (February - March 2012 onboard the RV Tangaroa)• SS - Sea Salt• SSA - Sea Spray Aerosol (includes both inorganic sea salt and organic component)• VF - Volatile Fraction (volume)• VFR - Volume Fraction Remaining• VH-TDMA - Volatility and Hygroscopicity Tandem Differential Mobility Analyser• UKCA - United Kingdom Chemistry and Aerosols model• ZSR - Zdanovskii, Stokes and Robinson assumption for water content of multicomponent mixtures
15
6 Introduction
Aerosols are solid or liquid particles suspended in a gas. In the context of the remote marine environ-
ment secondary aerosols are the product of chemical reactions in the atmosphere involving precursor
gases, such as dimethyl sulphide (DMS), which are emitted from the ocean. Primary marine aerosols,
or sea spray aerosols (SSA), are produced via bubble bursting from wind and wave action at the ocean’s
surface (Lewis and Schwartz 2004). Secondary marine aerosols are largely composed of non-sea salt
(nss) sulfates, and can also contain a secondary organic component, SSA is composed of sea salt and
an organic component derived from marine biota in the surface water (Blanchard 1989). A schematic
of these aerosol sources is shown in Figure 1. Anthropogenic sources are also common in the marine
atmosphere, either emitted from shipping or transported from terrestrial sources, and can significantly
contribute to the marine particle load. Polluted marine atmospheres can be detrimental to the study
of fundamental processes, for example secondary nss sulfates can have both a natural biogenic and
anthropogenic sources. Measurements are generally undertaken in pristine conditions, with pollution
periods removed (Shank et al. 2012). SSA are directly emitted into the marine boundary layer (MBL),
which is the well mixed atmospheric layer that extends from the ocean surface (Seinfeld and Pandis
2006), whereas the conversion of gaseous precursors into particles largely occurs above the MBL, in the
free troposphere (Quinn and Bates 2012). Air from the free troposphere is occasionally mixed into the
MBL, particularly during periods of atmospheric instability, and in this way the free troposphere acts
as a sources of secondary particles to the MBL. The physical and chemical properties of marine aerosols
dictate how they interact with the atmosphere and therefore how they influence climate (Seinfeld and
Pandis 2006).
Radiative forcing is the net downward radiative flux at the top of the atmosphere, that is the differ-
ence between the incoming short wave solar radiation and the outgoing long wave radiation. Radiative
forcing is used to measure the contribution of each element in models of the climate system to the
Earth’s overall radiative balance. A positive radiative forcing indicates that the atmosphere has ab-
sorbed energy, leading to a warming effect. Carbon dioxide absorbs infra-red radiation, resulting in
a positive radiative forcing and a net warming effect on the atmosphere, while certain aerosols reflect
incoming solar radiation resulting a negative radiative forcing and a net cooling effect on the atmo-
sphere. The contribution of aerosols to radiative forcing is broken up into two components (Myhre
et al. 2013):
• aerosol-radiation interactions, sometimes referred to as the direct aerosol effect, is the radiative
forcing resulting from how aerosol scatter or absorb both short and long wave radiation, and
• aerosol-cloud interactions, sometimes referred to as the indirect aerosol effect, is the radiative
forcing resulting from the alteration of cloud properties by aerosols.
Compared to other forcing agents, for example greenhouse gases, radiative forcing from aerosols is
highly variable. The breakdown of contributors to radiative forcing since pre-industrial times (Figure 2)
shows that aerosol, and in particular aerosol-cloud interactions, are by far the largest contributor to
the overall uncertainty in radiative forcing. The variability in aerosol radiative forcing is driven by
16
Figure 1: Major sources and production mechanisms for CCN in the remote marine environment.
DMS contributes to the MBL CCN population primarily via particle nucleation in the free troposphere
in cloud outflow regions with subsequent subsidence. Sea salt and organics are emitted as a result of
wind-driven bubble bursting. Source: Quinn and Bates (2012).
17
a number of factors. Firstly, aerosols display wide ranging physical and chemical properties and
therefore the direct interaction with radiation is complex. Aerosol populations are characterised by the
distribution of particle sizes, particle composition, morphology and phase state, in-turn the scattering
of incoming solar radiation depends on these properties. Secondly, aerosol particles play a key role in
the atmosphere as sites for the formation of cloud droplets, cloud condensation nuclei (CCN), acting
as CCN aerosols have impact on the radiative properties of clouds. In particular, an increase in the
CCN number concentration can increase the lifetime and brightness of clouds, sometimes referred to
as the first and second indirect effects, respectively.
Figure 2: Radiative forcing of climate between 1890 and 2011. Source: Myhre et al. (2013).
It is clear that the representation of aerosols in climate modelling is the largest contributor to global
radiative forcing uncertainty. Sensitivity testing of model parameters has been used to constrain the
radiative forcing uncertainty associated with aerosol-cloud interactions. Model parameters can be
broadly broken up into those associated with emissions and those associated with processes. Emissions
drive how much of a substances is released into the atmosphere and processes drive what happens
to these emissions as they are transported through the atmosphere. For example, in the marine at-
mosphere sea spray and gaseous precursors (such as DMS) are the most important emission sources,
these are generally parameterised using wind speed, ocean temperature and chlorophyll-a concentra-
tions. Important process variables in pristine marine environments include the minimum diameter for
aerosols to act as CCN, and variables characterising the particle size distribution. It has been shown
that emissions from the natural environment contribute approximately 45% of the global radiative
uncertainty, and a further 21% is associated with variables governing aerosol processes (Carslaw et al.
18
2013; Lee et al. 2012). Despite the dominance of anthropogenic emissions on global aerosol number
concentrations, the contribution to radiative uncertainty is largely driven by emissions and processes
in natural pristine environments. In recent years there has been a push to take observations in remote
marine environments (Meskhidze et al. 2013), which are regions of particular importance because of
the vast area, low CCN concentrations and dark surface ocean albedo which results in high sensitivity
to CCN. This work will help to constrain the model parameters that contribute to global radiative
forcing uncertainty.
The uncertainty in global radiative forcing due to aerosol cloud interactions is not uniform, some re-
gions display biases in the prediction of aerosol-cloud interactions. The ability of a model to represent
radiative forcing is validated against the outgoing radiation at the top of the atmosphere, measured
using satellite instrumentation. Top of atmosphere shortwave radiation provides information on the
amount of reflected solar radiation i.e. the earth’s albedo, while top of atmosphere long wave radi-
ation provides information on the absorption from greenhouse gases. Figure 3 shows the difference
between modelled and measured top of atmosphere shortwave radiation, positive values indicate that
atmospheric model under predicts the amount of reflected radiation.
Over the Southern Ocean, in particular, there is a bias in the amount of shortwave radiation leaving the
atmosphere (Flato et al. 2013). Most atmospheric models show consistent, strong deficits in reflected
solar radiation over the high latitude Southern Ocean (beyond approximately 55 °S), this results from
the underrepresentation of low-to-mid-level cloud, particularly during the summer months (Bodas-
Salcedo et al. 2012; Hyder et al. 2018). In turn the radiative bias influences modelled sea surface
temperature and atmospheric circulation over the southern hemisphere, with knock on effects for the
modelled southern hemisphere climate (Flato et al. 2013; Kay et al. 2016). One such example is the
predisposition of climate models towards excessive precipitation in the Southern Hemisphere tropics,
which has been linked to high energy flux over the Southern Ocean (Hwang and Frierson 2013). Aerosols
are a key component in describing the radiative effects of low level marine clouds, cloud droplet number
concentrations (CDNC) describe three quarters of the variability in cloud cooling effects (Rosenfeld
et al. 2019).
The pristine Southern Ocean air masses with large oceanic fetches and high wind speeds result in unique
place for aerosol observations, with a SSA contribution greater than for any other ocean basin. Aerosol
observations over the remote Southern Ocean are scarce and model parameterisations have largely been
developed based on measurements in the Northern Hemisphere. The seasonal cycle in marine biota
strongly influences aerosol composition and concentrations over the Southern Ocean. Summertime
nascent SSA are enriched in organics from the surface water biota (Sciare et al. 2009; McCoy et al.
2015), and the emissions of precursor gases (in particular DMS) increase during the biologically active
summer months resulting in higher production of secondary marine aerosols (in particular non-sea
salt sulfates). Biological activity (Raes et al. 2018), and secondary aerosol production (McCoy et
al. 2015), is highest in the lower latitude Southern Ocean near the oceanic subantarctic front (SAF)
where warm subtropical waters mix with cooler nutrient rich subantarctic waters at approximately 45
- 50 °S. Aerosols over the Southern Ocean are also influenced by atmospheric features, in particular
the boundary between the Ferrel and Polar Cells is a low pressure region in which the mixing of
19
cold Antarctic and warmer lower latitude air masses cause instability and frequent production of low
pressure systems. Atmospheric instability from the polar front results in enhanced mixing between the
free troposphere and the MBL, and influences the Southern Ocean aerosol. Validating the radiative
balance over the Southern Ocean requires a good representation of the sources of CCN, which is
largely driven by marine biota (McCoy et al. 2015). The biogeochemical processes that drive aerosol
production and aerosol-cloud interactions over the Southern Ocean will shift in a changing climate,
therefore understanding aerosol production and processing over this region is essential for predicting
atmospheric circulation and precipitation as well as sea surface temperature.
Figure 3: Difference between modelled and observed top of atmosphere outgoing shortwave radiation
(Wm-2). Modelled data from CMIP5 models compared against the Clouds and the Earth’s Radiant
Energy System Energy Balanced and Filled 2.6 (CERES EBAF 2.6) satellite observations. Source:
Flato et al. (2013).
The key research questions for this research program are:
1. How does seawater composition in the Southern and South West Pacific Oceans influence the
concentration and composition of SSA and NSS sulfates and therefore the CCN number concen-
tration?
2. How can the representation of marine aerosol sources over the Southern and South Pacific Oceans
in current chemical transport and earth system models be improved to better represent observa-
tions and reduce model uncertainties?
20
7 Literature review
The following sections outline the key components from the currently available research relating to
marine biogenic aerosol as a source of cloud condensation nuclei. The first two sections, Chapter
7.1 (Marine aerosol size and number) and Chapter 7.2 (Aerosol water uptake), describe important
concepts that are closely linked to analysis techniques applied in this work. Chapter 7.3 and Chapter
7.4 characterise the existing research relating to sea spray aerosol and secondary marine aerosols,
respectively. The final section, Chapter 7.5, describes how remote marine aerosol influences climate,
and how uncertainties in the current characterisation of remote marine aerosol are propagated into
climate modelling uncertainties.
7.1 Marine aerosol size and number
Remote marine aerosol number concentrations are relatively low, ranging from approximately 200 - 300
cm-3 in the high latitudes, up to 1000 cm-3 in the tropics and mid-latitudes (Heintzenberg, Covert, and
Dingenen 2000). A distinct seasonality is observed in the particle number concentration in both the
southern and northern hemisphere. Particle number concentrations of 200 - 230 cm-3 during the winter
have been reported at Mace Head in the North Atlantic ocean and 442 - 493 cm-3 during the biologically
active summer months (Yoon et al. 2007). A similar pattern is observed at the Cape Grim baseline
air pollution station, with summertime concentration of approximately 400 cm-3, approximately 3 -
4 times that of wintertime concentrations (Fletcher et al. 2007; Gras 1995). The seasonal cycle in
particle number concentrations is thought to be associated with biological activity and in particular
the release of dimethyl sulfide (DMS), a precursor for particle formation (Ayers et al. 1995; Ayers and
Gras 1991; Jimi et al. 2007), the formation of non-sea salt sulfate aerosol will be discussed in more
details in Chapter 7.4.1.
The concentration of atmospheric aerosols as a function of diameter can be modelled using a sum of
log-normal distributions, where each log-normal distribution characterises a production or atmospheric
processing mechanism (Seinfeld and Pandis 2006). Log-normal distributions are used to model aerosol
size distributions because the variability arises from successive random dilutions, which is a multiplica-
tive process (Ott 2012; Limpert, Stahel, and Abbt 2001). Equation 1 shows the log-normal function,
where Nt,i is the total number concentration, μi is the mean diameter and σi is the geometric standard
deviation for the log-normal mode i.
Ni(Dp) =Nt,i√
2πlog10σg,ie− (log10Dp−log10μi)
2
2(log10μi)2 (1)
A set of consistent log-normal modes have been identified from oceanic number size distribution mea-
surements, as shown in Figure 4, and as such distributions are generally divided into 3 modes, the
Aitken mode with an upper limit of approximately 80 nm, the accumulation mode with an upper limit
of approximately 600 nm and the coarse mode which extends to particle diameters greater than 1 μm
(Heintzenberg, Covert, and Dingenen 2000; Hoppel et al. 1990; Modini et al. 2015; Quinn et al. 2017;
21
Seinfeld and Pandis 2006). The minimum between the Aitken and accumulation mode is known as the
Hoppel minimum and its location is driven by in-cloud processing of particles (Hoppel et al. 1990),
therefore 80 nm is not a strict limit on the boundary for these modes. The mechanisms driving the
formation of a Hoppel minimum are discussed further in Chapter 7.4.1.
Figure 4: Median log-normal approximation of the four-modal sub micrometre marine number size
distributions. Probabilities of number concentrations about the median distributions are given ac-
cording to a log-normal modal for the range 0.1 to 10 times the median concentrations. Source:
Heintzenberg et al. (2011).
A somewhat consistent picture of the aerosol sources, and resulting composition, for the log-normal
modes has also developed. The coarse mode is consistently observed to be made up of primary sea
spray particles, and is commonly referred to as the nascent SSA or primary marine mode. In-situ and
laboratory observations have been used to constrain the mean diameter of the nascent SSA mode to
200 nm±30%, and the σg to 2.5 to 3 (Modini et al. 2015). Typically sub-micrometer size distributions
have a σg less than 2, the broad nature of the SSA mode is thought to be due to the superposition
of a number of log-normal modes corresponding to primary marine particles of different composition.
Measurements of chamber (Fuentes et al. 2010b) and ambient (Ovadnevaite et al. 2014) SSA have fitted
up to five narrow log-normal distributions driven by different production mechanisms and displaying
different composition. Extracting the SSA log-normal modes that fall in the Aitken and accumulation
modes is often complicated by large number concentrations of secondary particles in ambient size
distributions. The broad SSA size distribution described above is used to approximately the total SSA
distribution and therefore may not entirely represent all of the SSA modes in some circumstances.
22
The contribution of the nascent SSA mode to the overall number concentration is low, generally less
than 15%, but correlates with wind speed (Modini et al. 2015; Quinn et al. 2017). The production of
SSA is discussed further in Chapter 7.3. The Aitken and Accumulation mode number concentrations
have consistently been observed to be driven by non-sea salt sulfates (Blot et al. 2013; Clarke et al.
2013; Fossum et al. 2018; Modini et al. 2015; Quinn et al. 2017). The characteristics of the Aitken and
accumulation modes are governed by atmospheric processing, not by a particle formation mechanism,
therefore the composition of these modes may vary, for example aqueous phase reactions in clouds can
add mass to and change the accumulation mode composition (Hoppel et al. 1990; Seinfeld and Pandis
2006; Zheng et al. 2018).
It should be noted here that the size distributions above are presented with respect to aerosol number
concentrations, data are most commonly presented in this way when examining aerosol cloud inter-
actions (indirect effect), because the number concentration of CCN are of primary interest. Volume
and mass distributions are skewed toward larger sizes, for example particles with diameter greater
than 1 μm can contribute approximately 95% of the particle mass, but only make up approximately
5% of the particle number concentration (Fitzgerald 1991). Generally the surface area is dominated
by the accumulation mode and the volume is dominated by the coarse mode. The accumulation and
coarse modes are important for aerosol-radiation interactions (direct effect) because light scattering is
proportional to the aerosol surface area (Lewis and Schwartz 2004; Marshall et al. 2005). The overall
contribution of coarse mode SSA to CCN may not be high, but they may have an important role in
cloud microphysics, and are sometimes referred to as giant CCN (Andreae and Rosenfeld 2008). The
activation of aerosol to CCN is discussed in Chapter 7.2.2.
Aerosol particle morphology or shape is also an important consideration when determining aerosol
size distributions. The most commonly used instrumentation for measuring aerosol size distributions
is a scanning mobility particle sizer (SMPS), this instrument sizes particles based on their electrical
mobility (dependent on electrical charge of particles as well as their drag), and gives the electrical
mobility diameter, dm. The volume equivalent diameter (dve) is the diameter of a sphere with the
same volume as the particle of consideration, and can be related to dm using Equation 2, where χ is
the dynamic shape factor, and Cc is the Cunningham slip correction factor (DeCarlo et al. 2004). Cc
is a diameter dependent correction on drag, the drag decreases for particles with a larger diameter,
and χ is a correction due to the increase in drag for non-spherical particles. Spherical particles have a
χ = 1, therefore dm = dve. In the context of marine aerosols, secondary particles are close to spherical,
for example ammonium sulfate has χ = 1.02 (Biskos et al. 2006a, 2006b), and sea salt has a χ = 1.05
at 200 nm and χ = 1.1 at 20 nm (Zieger et al. 2017).
dmCc(dm)
=dve.χ
Cc(dve)(2)
SMPS systems generally measure aerosols up to 1 μm in diameter, above this aerodynamic or optical
particle sizers are often used. The aerodynamic diameter, da, can be related to dve using Equation 3,
where ρp is the particle density, ρ0 is unit density (1 g/cm-3). Note that all of the diameters and
relationships here assume that the particle is dry, if this is not the case a relative humidity dependent
23
correction for the particle diameter is required and the χt is assumed to be 1 for a droplet.
da = dve
√ρpρ0χ
(3)
7.2 Aerosol water uptake
7.2.1 Hygroscopicity
Hygroscopicity describes how aerosols of a particular composition and diameter uptake water from the
surrounding atmosphere. A common empirical measure of hygroscopicity is the hygroscopic growth
factor (HGF), which is the particle diamater at a reference relative humidity (usually 90%) divided by
the dry particle diameter, Equation 4. HGF is used to describe aerosol water uptake in sub-saturated
conditions (RH < 100%) and is commonly measured using a H-TDMA (see Chapter 7.2.7 for more
details). Hygroscopic particles uptake water readily and have a high hygroscopic growth factor (up to
approximately 2.4 at 90% RH), whereas hydrophobic particles do not uptake water and have a HGF
of 1. Table 1 outlines the HGFs for a number of common atmospheric species.
HGF =dRH
ddry(4)
Another commonly applied measure of hygroscopicty is the hygroscopicity parameter, κ, which is the
volume of water per unit volume of dry particle solute. Note that κ is scaled differently to HGF, a
hydrophobic particle will have no water associated with it and therefore κ = 0, the most hygroscopic
particles (HGF = 2.4 at 90% RH) has κ = 1.4 and therefore has a higher volume of water than dry
particle solute. Table 1 outlines the κ values for a number of common atmospheric species. κ can be
computed using Equation 5, where aw is the water activity, Vw is the volume of water and Vs is the
volume of solute (dry particle volume). Water activity is closely related to the relative humidity, and
scaling by this factor in Equation 5 means that kappa is independent of the RH and can be computed
from measurements in the sub or super-saturated regimes (see Chapter 7.2.7 for further details). Kappa
can be thought of as a property of the dry particle solute, whereas the HGF is a property of the solute-
water system. Further detail on the application of κ are outlined in Chapter 7.2.2.
κ = (1
aw− 1)
Vw
Vs= (
1
aw− 1)(HGF 3 − 1) (5)
7.2.2 Köhler theory
The uptake of water onto aerosol particles is the product of an equilibrium between water vapour in
the gas phase and liquid water, either in the form of aqueous solution (sub-saturated conditions) or
as a cloud droplet. To maintain water vapour equilibrium, potential energy in the form of Gibbs free
24
Table 1: HGF, deliquescence relative humidity (DRH), hygroscopicity parameter (kappa) and volatili-
sation temperature (Tv) for atmospherically relevant species.Kappa calculated from hygroscopic growth
measurements (kappaHGF) and from CCN measurements (kappaCCN) using kappa-Köhler model. Re-
produced from Fletcher et al. 2007, Petters and Kreidenweis 2007 and Cravigan et al. 2015.
Substance HGF DRH[%]
kappa
HGF
kappa
CCN
TV
[°C]Sources
Sulfuricacid
1.68 0a 1.19b 0.9 139 Berg, Swietlicki, and Krejci 1998; Clegg, Brimblecombe,and Wexler 1998; Johnson, Ristovski, and Morawska2004
Ammo-niumbisulfate
1.79 40 - - 202 Tang and Munkelwitz 1994; Berg, Swietlicki, and Krejci1998; Johnson et al. 2008
Ammo-niumsulfate
1.7 79-80
0.53 0.61 205/180(40nm)
Clegg, Brimblecombe, and Wexler 1998; Prenni,DeMott, and Kreidenweis 2003; Johnson, Ristovski, andMorawska 2004; Koehler et al. 2006
Methane-sulfonicacid
1.57 0a - - 171 Johnson, Ristovski, and Morawska 2004
Ammo-niumnitrate
1.5 61.5 - 0.67 75 Lightstone et al. 2000; Johnson, Ristovski, andMorawska 2004
Sea salt 2.2-
2.35
75 - - >500/290 (27
nm)
Modini, Harris, and Ristovski 2010; Fuentes et al. 2011
Sea salt 2.09-
2.11
73.5 0.96-
1.14
- - Zieger et al. 2017
Sodiumchloride
2.45 75 0.91-
1.33
1.28 >600 Clegg, Brimblecombe, and Wexler 1998; Biskos et al.2006a; Koehler et al. 2006; Modini, Harris, andRistovski 2010
Succinicacid
1.01 - 0.006 0.231 - Chan and Chan 2003; Hori et al. 2003; Prenni, DeMott,and Kreidenweis 2003
Oxalicacid
1.43 - - - - Peng, Chan, and Chan 2001; Prenni, DeMott, andKreidenweis 2003
Malanoicacid
1.73 - 0.44 0.227 - Peng, Chan, and Chan 2001; Pradeep Kumar,Broekhuizen, and Abbatt 2003; Prenni, DeMott, andKreidenweis 2003; Koehler et al. 2006
Secondaryorganicaerosol
1.06-
1.1
0a 0.009-
0.07
0.014-
0.229
75 -150
VanReken et al. 2005; Hartz et al. 2006; Prenni et al.2007
a Continuous growth. b modelled HGF.
25
energy is transferred between the vapour and liquid until the potential in the two phases is equal.
The droplet diameter for which water vapour equilibrium is achieved is described using Köhler theory
in both the sub-saturated and super-saturated regimes. Köhler theory is a combination of the solute
effect and the Kelvin effect.
The solute effect dictates that the presence of particles (solutes) lowers the required saturation ratio to
maintain water vapour equilibrium, the larger the molar fraction of the aerosol component the lower
the required water vapour saturation. The solute effect is a application of Raoult’s law, the partial
vapour pressure of water is given by the saturation vapour pressure of water multiplied the molar
water fraction, as shown in Equation 6, where pw is the partial pressure of water vapour, pwo is the
saturation vapour pressure of water and xw is the molar fraction of water in the liquid phase. In Köhler
theory the solute effect is represented using the water activity, aw, as shown in Equation 6. The water
activity parameterises solute concentration and composition, the larger the particle the lower the water
activity and the larger the molar mass the lower the water activity.
S =pwp0w
= xw = aw (6)
The Kelvin effect (Equation 7) implies that the smaller the droplet diameter the higher the vapour
pressure required to maintain equilibrium (Seinfeld and Pandis 2006). An explanation for this is that
the greater curvature of droplets with a smaller diameter results in fewer neighbouring molecules. The
Kelvin effect is important for particles less than 200 nm in diameter, although this depends on particle
composition (Seinfeld and Pandis 2006). Water vapour supersaturations of less than 1% are sufficient
to form droplets in the ambient atmosphere, while observations and theory show that for droplets
to form homogeneously supersaturations of approximately 10% are required. This indicates that the
Kelvin effect alone does not describe droplet formation in the atmosphere.
S = e4σVRTD (7)
Combining the influence of Raoult’s law and the Kelvin effect on supersaturation gives the Köhler
equation, Equation 8, which describes the equilibrium water vapour saturation required for droplet
formation from a particle.
S = aw.e4σVRTD (8)
Parameterisations for the Raoult effect have been developed to overcome shortfalls in available in-
formation on the physiochemical properties of solutes. The κ-Köhler model is a commonly used pa-
rameterisation based on the water uptake of particles (Petters and Kreidenweis 2007, 2013). It is an
empirical model that uses the hygroscopicity parameter, κ, which is the volume of water associated
with each unit volume of dry particle. The parameterisation κ is defined through its effect on aw as
shown in Equation 9. Vs is the volume of solute (i.e. volume of the dry particle) and Vw is the volume
26
of water.
1
aw= 1 + κ
Vs
Vw(9)
Substituting Equation 9 into Equation 8 yields the κ-Köhler relationship, Equation 10, where D is the
diameter of the droplet and Dd is the diameter of the dry particle.
S =D3 −D3
d
D3 −D3d(1− κ)
e4σVRTD (10)
Equation 10 can also be represented in terms of the HGF, as shown in Equation 11.
S =HGF 3 − 1
HGF 3 − (1− κ)e
4σVRTD (11)
The hygroscopicity parameter, κ, parameterises the composition of the particle via its water uptake
properties under both sub-saturated and supersaturated conditions. In this way measurements of water
uptake under sub-saturated conditions can be used to infer CCN concentrations (Good et al. 2010;
Petters and Kreidenweis 2007).
7.2.3 CCN activation
The supersaturation required for the growth of a droplet is represented in Figure 5. The peak of the
Köhler curves indicate the critical supersaturation (Sc) beyond which particles of a given composition
and dry diameter (referred to as the critical diameter Dc) will activate to form cloud droplets, acting
as cloud condensation nuclei (CCN). For example, Figure 5 shows that at 0.15% supersaturation, am-
monium sulfate particles with a dry diameter of 100 nm or greater (i.e. Dc = 100 nm) will be activated
and grow into droplets via condensation (Andreae and Rosenfeld 2008; Biskos et al. 2006a; Seinfeld
and Pandis 2006). The critical diameter in the marine environment is generally under approximately
200 nm, hence the focus on this size range.
The repeated activation of marine aerosols into cloud droplets and subsequent evaporation drives
the persistent Hoppel minimum observed in marine size distributions. Cloud droplet aqueous phase
reactions convert gas phase species into non-volatile components which persists in the particle phase
upon droplet evaporation. Additionally the collision of unactivated smaller particles with cloud droplets
can occur. The result of cloud processing is an increase in the activated particle diameter and an
associated minimum in the concentration of particles with diameters around the critical diameter for
droplet activation. The cloud activation and evaporation cycle is repeated a number of times, estimates
of approximately 10-20 cycles for a given particle lifetime have been reported, and result in the observed
Hoppel minimum (Hoppel et al. 1990).
27
Figure 5: Köhler curves indicate the equilibrium water vapour saturation ratio at 293 K for droplets
of pure water shown by the dotted black line. Droplets containing ammonium sulfate particles with
50 nm, 100 nm and 500 nm dry diameters (and thus various masses) shown in red, green and blue,
respectively. Water vapour saturation ratio, S(%)=100×(p/(p0-1). Where p is the partial pressure
of water vapour over the droplet and p0 is the saturated vapour pressure over a flat surface. Source:
Seinfeld and Pandis (2006); Andreae and Rosenfeld (2008).
28
7.2.4 Deliquescence and efflorescence
Aerosol water uptake is not continuous for many crystalline species, for example sodium chloride and
ammonium sulfate. These species undergo deliquescence at a precise relative humidity (DRH). At the
DRH the particle begins to absorb water and the HGF increases sharply, below this value the HGF
is approximately one. At the DRH (and above) the salt is in an aqueous state (Biskos et al. 2006a;
Mifflin, Smith, and Martin 2009; Seinfeld and Pandis 2006).
The RH at which the particle returns to a solid crystalline state is known as the efflorescence relative
humidity (ERH), this is generally lower than the DRH. Not all substances deliquesce, for example
continuous growth with increasing RH is observed for many organics in the marine environment.
Species displaying deliquescence may be identified using their characteristic DRH. Table 1 lists the
DRH for a number of common atmospheric species.
Figure 6: Deliquescence, efflorescence curve of 100 nm diameter common marine inorganic species.
Source Modini (2010).
7.2.5 Mixtures
Thus far only the water uptake to particles consisting of a single species has been discussed, however
in the ambient marine environment predominantly internally mixed (or multicomponent) particles are
observed. Mixing state refers to how the aerosol composition in an air mass is distributed between
29
particles. An internal mixture refers to a population of particles in which each particle contains more
than one species, for example particles composed of nss-sulfates and organics are internally mixed.
Externally mixed particles are made up of two (at least) populations of particles, each population dis-
playing a different chemical composition, for example an air mass containing both nss-sulfate particles
and sea spray particles is an external mixture. In the case of externally mixed particles the hygroscopic
properties can be determined for each component individually, however for internally mixed particles
interactions may occur between the species and a rationale for assigning water uptake properties needs
to be determined.
The amount of water absorbed by internally mixed particles at equilibrium water vapour pressure can
be described by the Zdanovskii, Stokes and Robinson (ZSR) relationship (Chen et al. 1973; Stokes and
Robinson 1966). The ZSR rule states that the volume of water absorbed by a particle is simply the sum
of the water absorbed by each component of the particle, and as a consequence the HGF of the particle is
given by the volume fraction weighted sum of the component HGFs. The ZSR assumption is commonly
used to calculate the HGF for internally mixed particles of known composition in hygroscopic closure
studies (Hersey et al. 2009; Putaud et al. 2000; Quinn and Coffman 1998; Swietlicki et al. 1999).
HGF 3 =∑i
HGF 3i .εi (12)
Similarly, the ZSR assumption can be used for κ as shown in Equation 13.
κ =∑i
κi.εi (13)
In Equation 12 εi is the proportion of component i by volume and HGFi is the HGF of component i.
The ZSR rule provides a guide to the HGF of internally mixed particles, however the absorption of
water onto the surface of a particle can be influenced by particle composition via changes in the surface
tension (Nozière, Baduel, and Jaffrezo 2014). In some cases the organic component of internally mixed
particles in the marine environment have been shown to act as a surfactant, resulting an an overall
HGF different from that given by the ZSR assumption (Facchini et al. 2000, 1999; Modini 2010; Modini
et al. 2010).
7.2.6 Surface partitioning
A number of alternative models have been developed, in particular to improve the representation of
surface tension on aerosol water uptake. Here we will focus on the compressed film model. The
compressed film model creates an organically enriched surface layer, which acts to suppress surface
tension, and a bulk solution which is a mixture of organic species, inorganic species and water. The
relationship between the concentration of organics at the surface and the particle surface tension is
given by the compressed film equation of state, Equation 14. A is the molecular area and A0 is the
critical molecular area. At A < A0 an organic mono-layer is formed resulting in a suppression of
30
surface tension (σ) below of that of the surface tension of water (σw). σmin is an imposed lower limit
for the surface tension, and mσ relates surface tension to the organic surface concentration.
σmin = min(σw,max(σw − (A0 −A)mσ, σmin)) (14)
The isotherm for the compressed film model is given in Equation 15, Cbulk and C0 are the solution
concentrations in the bulk and at the phase transition, respectively, R is the gas constant, NA is
Avogadro’s number and T is temperature.
lnCbulk
C0=
(A20 −A2)mσNA
2RT(15)
The bulk organic concentration and the molecular surface area can be expressed in terms of droplet,
particle inorganic seed, and organic coated diameters (dwet, dseed, dcoat) and the fraction of organic
molecules partitioned to the surface (fsurf), Equation 16 and Equation 17. Vorg and Vw are the
molecular volumes of the organic species and water respectively.
Cbulk =(1− fsurf )(D
3coat −D3
seed)Vw
D3wetVorg
(16)
A =6VorgD
2wet
fsurf (D3coat −D3
seed)NA(17)
The compressed film model equations can be used alongside the Köhler equation to compute dwet as
a function of RH, and the associated and fsurf for a particular organic surfactant characterised by A0
and C0. This model characterises the water uptake to the particle, as well as tracking the organic
partitioning and subsequent impact on surface tension. Figure 7 shows the impact of surface tension
effects on droplet growth under the compressed film model. Under sub-saturated conditions there is
a complete organic surface layer and the surface tension is at the minimum value. As the droplet
grows the organic component is spread more thinly and the surface tension increases until the organic
component is diluted and any surface activity become negligible, the Köhler model and the compressed
film model converge. A decrease in the critical supersaturation in the compressed film model can be
observed, indicating a shift in the critical diameter for CCN activation. Comparing the ZSR assumption
with output from the compressed film model provides a good constraint on the influence of surface
tension on water uptake.
7.2.7 Water uptake measurements
Hygroscopicity Tandem Differential Mobility Analysers (H-TDMA) are used to measure the hygroscopic
growth of particles, generally in the Aitken and accumulation modes (McMurry and Stolzenburg 1989;
Rader and Mcmurry 1986; Swietlicki et al. 2008) . Although the native hygroscopicity variable of
H-TDMA instruments is the HGF, the κ-Köhler equation (Equation 10) can be used to calculate κ for
31
Figure 7: Modelled water uptake for 50nm diameter ammonium sulfate seeds coated with malonic
acid up to total diameter of 150 nm. Köhler curves (top panel), fraction of organics partitioned to
the surface (middle panel) and surface tension (bottom panel) as a function of droplet diameter for
modelled aerosol water uptake using the Köhler theory (blue lines) and the compressed film model (red
lines). Dashed vertical lines bound the region for which the droplet is partially covered by the surface
film. Reproduced from Ruehl et al. (2016).
32
Table 2: Typical composition in the marine environment for given HGF ranges.
HGFclass
HGFrange
(approx.)
Likely composition Sources
Nearlyhy-dropho-bic
1.0 - 1.1 marine organics or areinfluenced by anthropogenicsources
Berg et al. [1998]; Swietlicki et al. [2000]; [2008];Bialek et al. [2012]
Less hy-groscopic
1.1 - 1.3 marine organics andmodified NSS sulfates
Ovadnevaite et al. [2011b]; Bialek et al. [2012]
More hy-groscopic
1.35 - 1.85 NSS-sulfates Swietlicki et al. [2000]; Ovadnevaite et al. [2011b];Bialek et al. [2012]; Cravigan et al. [2015]
Sea saltaerosols
1.85 - 2.4 inorganic sea salt Fuentes et al. [2010a]; Modini et al. [2010]; Bateset al. [2012]; Bialek et al. [2012]; Cravigan et al.[2015]
a given HGF and dry particle diameter. The operation of the H-TDMA is outlined in Chapter 8.2. The
H-TDMA technique is generally able to distinguish between the components of marine aerosols and
provides information on the mixing state of the particle population as well, this is an important variable
for determining CCN activation (McFiggans et al. 2006). Existing H-TDMA studies have broken up
the HGF space into a number of HGF categories with similar particle composition, as shown in Table 2.
This provides a guide to the components that are likely to be in the aerosol. Caution should be taken
when applying these HGF classes, as internally mixed particles may have a HGF that could lead to
misattribution of the composition. HGFs should be considered along with other measurements to
obtain a full picture of particle composition.
The fraction of an internally mixed component can be calculated from hygroscopicity measurements
using the ZSR assumption, Equation 12. For example, for an internal mixture of sea salt (ss) and
organics (org), with known HGFs, Equation 18 can be solved to give the volume fraction of the organic
component. The inorganic sea salt HGF is known, Table 1, and the organic HGF can be estimated
based on the HGFs of known organic compounds.
HGF 3SSA = (1− εorg).HGF 3
ss + εorg.HGF 3org (18)
Volatility and hygroscopicity measurements can also be combined to measure the water volume uptake
(WVU), a positive water volume uptake suggests that the volatile component suppressed the HFG of
the non-volatile component as shown in Equation 19, where T indicates the temperature to which the
33
sample is heated and 0 indicates sampling at ambient temperature.
WV U =d3RH,T − d3dry,Td3RH,0 − d3dry,0
(19)
Cloud condensation nuclei counters (CCNc, Droplet Measurement Technologies, Longmont, CO) mea-
sure the total number concentration of particles that activate into cloud droplets at a given supersatu-
ration (Roberts and Nenes 2005). The CCN concentration measured in this way quantifies the number
of particles that could potentially be activated into cloud droplets, and does not directly reflect the
number of cloud droplets for a given region. Herein CCN concentrations refer to the concentration
of particles that are available as CCN at some measured (or calculated) supersaturation unless other
wise mentioned. The vast majority of these measurements have been taken at the surface, not in
cloud, and as such are seperate from the the cloud droplet number concentration (CDNC). The CCNc
can be operated at a single supersaturation and/or scanning across a range of supersaturations pro-
viding continuous data on how CCNc varies over time. In this mode broad information on CCN for
a particular measurement location is acquired. Alternatively the CCN concentration is measured for
pairs of pre-selected particle mobility diameters and CCNc supersaturations, these data are used to
determine the critical supersaturation, which is then used to calculate κ via the κ-Köhler equations,
Equation 10.
Kappa can be computed from measurements in the sub-saturated regime, using H-TDMA methods,
and from measurements in the super-saturated regime, using a CCNc. The degree to which these two
methods agree provides information on the continuity of the particle chemical characteristics during
droplet formation, of particular interest is the surface activity of the aerosol sample.
7.3 Primary marine aerosols
7.3.1 Sea spray aerosol formation and measurement
SSA is produced from the impact of wind and wave action on the surface of the ocean (Blanchard
1989; Lewis and Schwartz 2004). The primary mechanism for the production of SSA is the bursting
of entrained bubbles, film drops are produced when the bubble-cap bursts and jet drops are produced
from fragments of the jet resulting from the collapse of the bubble cavity, as shown in Figure 8 (Lewis
and Schwartz 2004; Wang et al. 2017). The excess water in the jet and film drops quickly evaporate
leaving aqueous SSA particles. Film drops generally produce SSA particles with a smaller diameter,
less than 300 nm dry diameter, than jet drops, greater than 1 μm, however it should be noted that
recent observations suggest that jet drops can contribute up to 43% of the sub micrometer SSA number
concentration (Wang et al. 2017). The ratio of jet to film drops is dependent on the composition of
seawater, in particular marine organics act as surfactants and stabilise the bubble cap allowing it
to last longer, drain more, become thinner and produce fewer film drops . A greater proportion of
jet drops were observed by Wang et al. ( 2017) during periods of high biological activity, however
increased SSA production fluxes were also observed for water samples with higher Chl-a (Fuentes et al.
34
2010b; Long et al. 2014). Spume droplets are also created from by tearing droplets from the crests of
waves at high wind speeds, however these are greater than 10 μm and are quickly removed from the
atmosphere.
SSA should not be confused with sea salt, and is comprised of an inorganic component from sea salt
and an organic component. The organic component of SSA is produced from marine biota, primarily
phytoplankton, which can be emitted from the surface ocean during bubble bursting and contribute to
the SSA mass (Blanchard 1989; Fuentes et al. 2011; Modini, Harris, and Ristovski 2010; Quinn et al.
2014). The organic fraction of SSA is generally considered to be greater for film drops than for jet
drops. Surface active (water insoluble) organics are more likely to reside at the ocean surface layer
(micro-layer) from which the film drop media is derived, whereas the source of jet drops is sub-surface
water and contain more salts and water soluble organic species (Wang et al. 2017).
Figure 8: Schematic of bubble bursting process generating film and jet drops. Reproduced from
Lewis and Schwartz (2004).
There is considerable uncertainty in the bubble mediated SSA source over the ocean, with sub-200 nm
SSA fluxes varying by approximately 2 orders of magnitude (Leeuw et al. 2011; Lewis and Schwartz
2004; Ovadnevaite et al. 2014). Low concentrations and atmospheric processing hinder ambient nascent
SSA measurements (Cravigan et al. 2015; Frossard et al. 2014b; Laskin et al. 2012; Shank et al. 2012)
and SSA measurements are scarce in the Southern Hemisphere (Cravigan et al. 2015). Chamber
measurements are often used to simulate nascent SSA, providing a clearer picture of the relationship
between water phase composition and SSA composition. Laboratory generation of SSA aims to recreate
wave and wind induced bubble bursting processes that occur on the surface of the ocean. This is
generally achieved using one of three techniques.
1. Sintered glass - Clean air is passed through porous sintered glass submerged in a seawater sample.
2. Plunging water - Water is impinged into the surface of the water sample from above, entraining
air and forming bubbles.
3. Wave chamber - Breaking waves are induced using a hydraulic paddle.
The methods above vary in their representation of observed surface ocean bubble size and ambient SSA
size distributions, Figure 9 (Cravigan 2015; Fuentes et al. 2010b). Observed ambient SSA distributions
contain both an Aitken and accumulation mode (Sellegri et al. 2006). Wave chamber (Collins et al.
2014; Prather et al. 2013; Stokes et al. 2013) and plunging water (Facchini et al. 2008; Fuentes et al.
2010a; King et al. 2013; Prather et al. 2013; Sellegri et al. 2006) methods are thought to best represent
35
the surface ocean as they generally produce a broad bubble size distribution, in particular the bubble
spectra from wave chambers extend to diameters > 1mm and better represent measurements from open
ocean wave breaking (Deane and Stokes 2002; Prather et al. 2013). In turn the aerosol size distribution
is much broader for wave breaking and plunging water methods, with a mean of approximately 200 nm
and a geometric standard deviation of over 2.5, whereas the sintered glass method has been observed
to produce a single mode at 40 - 100 nm (Fuentes et al. 2010a; Keene et al. 2007; Mallet et al.
2016). Observations have indicated that SSA composition (King et al. 2013; Prather et al. 2013) and
morphology (Collins et al. 2014) are dependent on the method of SSA generation used, however the
direction and magnitude of these changes are not consistent between studies.
Figure 9: Size distributions for chamber measurements of nascent SSA using glass filters, plunging
water and wave breaking techniques. Size distributions normalised to the maximum dN/dlogDp. SSA
log-normal mode computed assuming a mean diameter of 200nm and a geometric standard deviation
of 3. Source: Cravigan (2015).
The characteristics from wave chamber size distributions have been used to constrain the log-normal
parameters used to fit the PMA/SSA log-normal mode to ambient size distribution data as described
in Chapter 7.1. The mean diameter of the fitted SSA log-normal mode is 200 nm ± 30 %, and the
σg range is 2.5 - 3. SSA distributions are also only fitted to the coarse mode of the size spectra,
greater than 500 nm, which effectively constrains the number concentration and minimises the impact
from non-SSA particles at lower diameters (Modini et al. 2015; Quinn et al. 2017). It is worth noting
that this SSA distribution is an approximation of the total of a number of SSA log-normal modes as
described in Chapter 7.1.
36
7.3.2 Sea spray aerosol water uptake
Reported hygroscopicity for inorganic sea salts are variable and strongly dependent on the applied
shape factor (Zieger et al. 2017) as shown in Table 1. Further to this variability, SSA may be enriched
in organics which can act to reduce the HGF. Chamber studies have shown organic enriched SSA HGFs
are suppressed by 4-17% (Bates et al. 2012; Fuentes et al. 2011; Modini, Harris, and Ristovski 2010;
Sellegri et al. 2006), and concluded that their SSA is comprised of an internal mixture dominated by sea
salt. In contrast, microcosm experiments have indicated that during periods when sample seawater was
likely to be highly enriched in organic matter (Deane et al. 2018) a relative high SSA hygroscopicity
parameter (κ) was observed, greater than 0.7 (Collins et al. 2016). The consistently high values led
the authors to conclude that organic enrichment to the aerosol phase didn’t necessarily influence water
uptake.
Ambient HGF measurements indicate that SSA comprises a relatively small and sporadic fraction of
the ambient number concentration in the remote marine environment (Berg, Swietlicki, and Krejci
1998; Bialek et al. 2012; Cravigan et al. 2015; Fletcher et al. 2007; Lawler et al. 2014; Maßling et al.
2007; Zhou et al. 2001), but is more frequently observed at diameters greater than 100 nm. Summer
and spring time ambient observations over the Southern Ocean show that SSA was observed in only
a small fraction of measurements, and when it was observed SSA comprised 15 to 39% of the number
fraction of 50 to 150 nm particles (Berg, Swietlicki, and Krejci 1998; Fletcher et al. 2007). A sub-100
nm SSA fraction of up to 69% was observed at the Cape Grim Baseline Air Pollution Station based on
hygroscopic growth measurements, however this was over a period of approximately 12 hours (Cravigan
et al. 2015). When observed in the West Atlantic SSA comprised 11-14% of the sub-100nm number
fraction (Swietlicki et al. 2000).
H-TDMA observations at Mace Head, in the North Atlantic show an annual average SSA number
fraction of 11-40% (Bialek et al. 2014). During periods of high biological activity a primary mode with
a suppressed growth factor of 1.3 to 1.5, consistent with SSA with a large organic fraction has been
observed (Bialek et al. 2012; Dall’Osto et al. 2010; O’Dowd et al. 2004; Ovadnevaite et al. 2011a). This
mode is enhanced at smaller sizes (Bialek et al. 2012), consistent with the prediction that organics are
favourably partition onto aerosols at smaller sizes (Oppo et al. 1999). SSA modes have been identified
more readily for particle diameters greater than 100 nm, this might be explained by a large organic
fraction resulting in the incorrect attribution of sub-100 nm SSA based on hygroscopicity.
Little variation between the hygroscopicity parameter of nascent chamber SSA derived using sub-
saturated HGF measurements and those derived using CCN measurements has been observed in a
number of studies (Bates et al. 2012; Fuentes et al. 2011; Wex et al. 2010), which indicates that
the organic fraction is low and/or is dissolved into the bulk. Similarly, in-situ nascent SSA chamber
studies have shown a strong dependence of κ on the SSA organic fraction (Quinn et al. 2014; Schwier
et al. 2015). At Mace Head, however, late summer observations have identified a population of SSA
particles with low HGF (∼1.25) and high CCN activity (0.83 at 0.25% SS) (Ovadnevaite et al. 2011b),
indicating suppression of surface tension by water insoluble organics to overcome the solute effects. The
suppression of surface tension has also been identified as a potential role in κ values observed during
37
nascent SSA microcosm experiments, which were persistently high even with high marine biological
activity and high SSA organic fractions (Collins et al. 2016). Studies using seawater analogues, such as
binary salt-organic mixtures, generally indicate that the organic component can be considered dissolved
into the bulk (Forestieri et al. 2018; Fuentes et al. 2011; Petters and Petters 2016; Prisle et al. 2010).
The surface tension suppression is far from consistent suggesting that regional differences and the
richness in organic matter and/or microbial composition are important.
7.3.3 Sea spray aerosol composition
The composition of ultra fine marine aerosols can be inferred using volatility methods, or measured
directly via chemical analyses. Aerosol mass spectrometers (AMS) are commonly used to give on-line/
real time chemical composition of non-refractory aerosol species, however, this is generally a sub-micron
average of composition and may not be representative of the Aitken mode. The components of SSA
are largely refractory, as such the AMS signal is diminished, and specialised techniques are require
to extract the SSA concentrations (Ovadnevaite et al. 2011a). Samples are also commonly collected
on substrates for subsequent analysis which can provide size resolved measurements, for example with
the use of a Micro-Orifice Uniform Deposit Impactor (MOUDI) [MSP Corporation, Shoreview, MN],
however the time resolution is of the order of 6 hourly to daily. Similarly, electron microscopy techniques
for aerosol composition, such as energy dispersive X-ray spectroscopy (EDX), are used for SSA analysis
but generally suffer from poor time resolution. The use of offline measurement techniques is not ideal
for ambient monitoring and as such their use is concentrated in chamber studies of nascent SSA.
Chapter 13 provides a summary of SSA chamber studies, and the associated organic fractions.
The inorganic composition of SSA has largely been assumed to mirror the inorganic composition of
seawater, suggesting that salts dissolved in the seawater remain well mixed during the bubble bursting
and droplet formation process (Lewis and Schwartz 2004), however a few important exceptions have
been identified. The enrichment of Ca2+ in SSA, relative to seawater, has been identified in a number
of ambient marine (Leck and Svensson 2015; Sievering et al. 2004) and nascent SSA chamber studies
(Cochran et al. 2016; Keene et al. 2007; Salter et al. 2016; Schwier et al. 2017). Cation enrichment,
has also been observed in the form of Mg2+ and K+(Ault et al. 2013a; Schwier et al. 2017), and is also
associated with deficits in Cl- (Ault et al. 2013b; Prather et al. 2013; Schwier et al. 2017). Ca2+ has
been observed to be enriched by up to 500% (Schwier et al. 2017), relative to seawater, with stronger
enrichment for smaller diameter SSA. Enrichment of Ca2+ in the SML (or close to the ocean surface)
is thought to drive SSA enrichment, however whether the Ca2+ is complexed with organic matter,
or is the product of precipitation of CaCO3, which can be a product of photosynthetic reactions, is
unknown. Observation of organic carbon particles that contain inorganic cations (such as Ca2+, Mg2+
and K+) but no Cl-, has also been reported, suggesting that in some cases these species complex with
organic material (Ault et al. 2013b) .
Volatility methods use the temperature at which a species evaporates or dissociates to infer composi-
tion. The volatilisation temperatures for some common atmospheric constituents are listed in Table 1.
Volatility methods can be used in conjunction with hygroscopic growth to provide added information
38
with which to infer composition. For example the volatile component of SSA can be computed by com-
paring the volume fraction remaining (VFR) of an SSA sample with the VFR of inorganic sea salt at a
range of temperatures. The HGF before and after the volatile (non-refectory) component is evaporated
provides water uptake information on both species. The VFR is calculated using Equation 20, where
dT is the particle diameter at temperature T and d0 is the ambient particle dry diameter. Previous
studies have taken the difference between the VFR of a nascent SSA sample generated from natural
seawater in a chamber and compared this to the VFR of a laboratory generated sea salt salt sample
to infer the organic volume fraction (Cravigan 2015; Mallet et al. 2016; Modini, Harris, and Ristovski
2010). The volatile organic volume fraction (vOVF) was assumed to be equal to the difference in the
VFR between the natural SSA sample and the sea salt sample, this assumes that the volatile organic
component is the only cause for the difference in volatility. It has been subsequently noted that com-
ponents of sea salt aerosol retain hydrates even when dried to very low RH (<5%), such as MgCl2 and
CaCl2 (Rasmussen et al. 2017). Sea salt hydrates need to be taken into account when estimating the
SSA OVF using volatility, and previous estimations should be treated with caution.
V FR =
(dTd0
)3
(20)
Chamber observations of nascent SSA universally indicate the presence of an internally mixed organic
fraction, as shown in Chapter 13. The organic fraction of SSA is made up of components such as
polysaccharides, lipids and proteins exuded from algal species (Frossard et al. 2014b; Russell et al.
2009) and varies widely between sets of observations. The majority of observation show that the
organic fraction increases with decreasing particle diameter (Facchini et al. 2008; Keene et al. 2007;
Prather et al. 2013; Quinn et al. 2014). The organic fraction of SSA appears to be comprised of a volatile
component which evaporates at approximately 150 - 200°C and comprises of the order of 10% of the SSA
volume (Modini, Harris, and Ristovski 2010; Quinn et al. 2014) and a non-volatile component (Quinn et
al. 2014), which has not been observed in all studies, for example in the South-West Pacific (Modini,
Harris, and Ristovski 2010). Externally mixed organics have also been observed to comprise the
majority of the number concentration in some studies (Collins et al. 2014; Collins et al. 2013; Prather
et al. 2013). Estimates of the SSA organic fraction based on water uptake methods (Fuentes et al.
2011; Modini, Harris, and Ristovski 2010) are generally lower than those using direct measurements of
chemical composition (Facchini et al. 2008; Keene et al. 2007; Prather et al. 2013), although exceptions
include Quinn et al. (2014), who observed OVF of up to 0.8 using CCNc measurements.
Offline measurements of ambient SSA at Mace Head (Cavalli et al. 2004; Ceburnis et al. 2008; Facchini
et al. 2008; O’Dowd et al. 2004; Rinaldi et al. 2009) indicate internal organic volume fractions of up
to 80 %. There is a distinct seasonality to the marine aerosol composition at Mace Head, with a
dominant sub-micron organic mass fraction during high biologically active periods (spring to autumn)
replacing a dominant sea salt fraction observed during winter (Cavalli et al. 2004; Facchini et al. 2008;
O’Dowd et al. 2004; Yoon et al. 2007). Accumulation mode (125 - 500 nm impactor stage) SSA organic
mass fractions of up to 65% and up to 80% for the 60 - 125 nm stage have been measured during the
summer months (Cavalli et al. 2004; O’Dowd et al. 2004). During winter sea salt comprised 74% of
the accumulation mode, with an organic fraction of 15% (Cavalli et al. 2004; O’Dowd et al. 2004; Yoon
39
et al. 2007). There is a strong size dependance to the observed SSA organic enrichment, similar to that
observed in chamber studies, with larger organic fractions for smaller particle diameters. Measurements
from the North Atlantic indicate that the internally mixed organic component of SSA is water insoluble
organic carbon (WIOC), largely lipopolysaccharides, with an upward flux, indicating a source at the
ocean surface and a striking similarity to the smaller size fraction of particulate organic carbon (POC)
and colloidal material in seawater (Ceburnis et al. 2008; Facchini et al. 2008).
Further ambient measurements during periods of high biological activity in the Southern Indian, North
Pacific and Southern Oceans showed sub-micron organic mass fractions of 26%, an approximately
13% WIOM fraction and sea salt mass fraction of approximately 45% (Claeys et al. 2010; Miyazaki,
Kawamura, and Sawano 2010; Sciare et al. 2009). This is similar to that observed at Mace Head during
periods with lower SSA fractions (Dall’Osto et al. 2010; Rinaldi et al. 2010). Over the North Atlantic,
Arctic and Southeast Pacific oceans (Russell, Bahadur, and Ziemann 2011; Russell et al. 2009) organics
accounted for 15-47% of the sub-micrometre particle mass, with hydroxyl groups comprising 44 - 61%
of the organic mass. External mixtures of sea salt and organics have also been identified using X-ray
analysis of transmission electron microscopy (TEM) samples over the Southern Ocean (Bigg 2007; Bigg
and Leck 2008; Leck and Bigg 2010). In the high Arctic authors argue that the surface micro layer is
an important source of CCN (Leck and Bigg 2005b, 2005a; Orellana et al. 2011).
The relationship between water phase organic composition and SSA organic composition is not straight
forward, with periods of both high and low biological activity yielding high SSA organic mass frac-
tions (Facchini et al. 2008; Gantt and Meskhidze 2013; Keene et al. 2007; Quinn et al. 2014). In
addition relatively low organic mass fractions have been observed during periods of high water phase
organic composition (Fuentes et al. 2011; Modini, Harris, and Ristovski 2010; Sellegri et al. 2008). A
number of studies have observed the change in SSA composition over the lifetime of a phytoplankton
bloom (Collins et al. 2013; Lee et al. 2015; Schwier et al. 2015, 2017). As phytoplankton numbers
grow during a bloom, they reduce CO2, increasing Chl-a and dissolved organic carbon concentrations
in the water, upon phytoplankton death particulate organic matter (lipids and polysaccharides) are
released into the seawater. Organic matter is further processed by enzyme digestion and processing by
heterotrophic bacteria, releasing saccharide fragments from polysaccharides and fatty acids from larger
lipids (Cochran et al. 2017; Lee et al. 2015). The lifecycle of a phytoplankton bloom is of the order of
2 weeks, and a lagged correlation between the peak in Chl-a and organic enrichment of SSA has been
observed (O’Dowd et al. 2015; Rinaldi et al. 2013), although not in all cases (Schwier et al. 2015). The
SSA organic fraction has been associated with the phytoplankton decline due to bacterial grazing and
viral infection, which release material predisposed to SSA enrichment (O’Dowd et al. 2015).
Chlorophyll-a (Chl-a) is most commonly used to parameterise primary marine organic emissions (Long
et al. 2011; O’Dowd et al. 2004; Vignati et al. 2010). The correlation between SSA organic enrichment
and Chl-a concentrations is imperfect and some authors suggest that such metrics of bulk organic
composition do not adequately represent organic enrichment of SSA (Frossard et al. 2014b; Quinn
et al. 2014). A wide variety of organic enrichment for most relevant chlorophyll-a concentrations have
been observed, as show in Figure 10. Although Chl-a may not represent the molecular processes that
drive enrichment it is the best observable metric currently available over the entire globe (Rinaldi et al.
40
2013), and it has been shown that over monthly timescales correlates well with SSA organic fraction
(O’Dowd et al. 2015).
Figure 10: Organic mass fraction (OMF) of sub-200nm SSA as a function of Chl-a concentration
for a number of nascent SSA observations, primarily from chamber measurements however ambient
observations also included (Gantt et al. 2011). Organic mass fractions computed from volume fractions
assuming an inorganic sea salt density of 2.017 (Zieger et al. 2018) and an organic density of 1.1. PM1
observations reported by Gantt et al. 2011 at Mace Head converted to 100 nm OMFs using the size
dependent organic enrichment reported by Gantt et al. 2011.
A range of molecular types have been associated with the organic fraction of SSA, the two most
commonly observed and discussed are:
• Polysaccharides (or carbohydrates) are the longest lived class emitted by phytoplankton and
have been observed in SSA in the Arctic (Gao 2012; Russell, Bahadur, and Ziemann 2011),
North Atlantic (Facchini et al. 2008) and Pacific (Bates et al. 2012; Cochran et al. 2016; Hawkins
et al. 2010; Quinn et al. 2014) and Southern Ocean (Sciare et al. 2009). Generally natural
polysaccharides are water-soluble and not surface active (Burrows et al. 2014), however water
insoluble lipopolysaccharides (LPS) have been observed in SSA (Cochran et al. 2017; Facchini et
al. 2008) and may depress water solubility and enhance surface tension effects. Samples observed
in the Southern Ocean (Sciare et al. 2009) were dominated by seasonally varying WIOC, but also
contained WSOC.
• Fatty acids are a short lived (of the order of 10 days) class of molecules, with concentration
linked to the rate of phytoplankton production (Burrows et al. 2014) and have been observed
41
in SSA samples from the North Pacific (Ault et al. 2013a; Cochran et al. 2017) and the South
Atlantic (Schmitt-Kopplin et al. 2012). Fatty acids are strongly surface active and therefore their
enrichment in SSA could have important consequences for CCN activation (Forestieri et al. 2018;
Ruehl, Davies, and Wilson 2016).
Fourier Transform Infrared (FTIR) spectroscopy allows for the analysis of organic functional groups,
that is hydroxyl, alkane, carbonyl, amine and carboxylic acid groups (Frossard et al. 2014b; Russell,
Bahadur, and Ziemann 2011; Russell et al. 2009). SSA organics are characterised by a dominant
hydroxyl fraction, smaller contributions from alkanes and smaller again from amines and carboxylic
acid groups (Russell, Bahadur, and Ziemann 2011). The hydroxyl functional group is consistent with
saccharides and has been observed to correlate with Na+ concentrations and wind speed, further
evidence of a primary source (Russell et al. 2009). The distribution of functional groups observed from
FTIR can also be used to estimate the fraction of lipids, polysaccharides and proteins, for example
the higher the ratio of alkane to hydroxyl groups the less oxygenated and more lipid like the organic
will be. The alkane to hydroxyl ratio is likely to be highest during strong phytoplankton blooms,
a ratio of 0.24 has been observed in the Northern Pacific and 0.38 in the North Atlantic. Increasing
alkane to hydroxyl ratios have been observed with increasing chlorophyll-a concentration in biologically
productive waters (Frossard et al. 2014b).
Further observations are required to determine how best to parameterise primary marine organics over
the Southern and South-West Pacific oceans, which are poorly represented. The parameterisations for
SSA organic enrichment are outlined in Chapter 7.5.1.
7.4 Secondary marine aerosols
7.4.1 Secondary marine aerosol formation and measurement
Secondary particles are formed via precursor gas molecules clustering together to form new particles
(nucleation) or via condensation of gases onto existing particles. In the marine environment precursor
gases are attributed to emissions from phytoplankton at the oceans surface (Andreae and Rosenfeld
2008). The major contributor to remote marine CCN has long thought to be nss-sulfates produced from
the emission of DMS by phytoplankton (Charlson et al. 1987) and other biogenics, such as coral (Raina
et al. 2013). DMS is transformed into particle phase nss-sulfates via oxidation to SO2, or to MSA via
oxidation to methane sulfinic acid. The oxidation to SO2 and nss-sulfates is the dominant pathway
and Aitken mode particle composition consistently identify nss-sulfates as the dominant species in the
number concentration (Good et al. 2010; Swietlicki et al. 2008). The CLAW hypothesis (Charlson
et al. 1987) postulates that the emission of DMS from phytoplankton in the surface ocean results in
new particle formation in the marine boundary layer (MBL) and an increase in the CCN concentration.
In turn aerosol-cloud interactions increase the cloud droplet number and cloud albedo, decreasing the
incoming solar radiation. This results in reduced DMS emissions from the phytoplankton, setting up
a feedback loop.
Observational and modelling studies subsequent to the proposal of the CLAW hypothesis indicate that
42
DMS emissions from phytoplankton are not oxidised and nucleated into the particle phase quickly
enough to create a feed back loop (Quinn and Bates 2012; Simpson et al. 2014). Nucleation in the
marine boundary layer requires the removal of existing aerosol via precipitation, and even in these cases
growth of new particles to CCN sizes is unlikely (Quinn and Bates 2012). Above the marine boundary
layer in the free troposphere there is strong evidence to suggest that new particle formation occurs
regularly (Clarke et al. 2013; Hoppel et al. 1990). Long range transport of DMS and nss-sulfates
in the free troposphere and subsequent entrainment into the boundary layer has been identified as
the dominant nss-sulfate, and therefore CCN, source in the marine environment (Clarke et al. 2013;
Koponen et al. 2002; Rose et al. 2015, 2017; Simpson et al. 2014; Woodhouse et al. 2013). For example
over the Southern Ocean Aitken and accumulation mode particles during ACE-1 were dominated
by ammonium sulfate particles, consistent with observations and nss-Sulfates have consistently been
observed to dominate the summertime CN at Cape Grim (Fletcher et al. 2007; Gras and Ayers 1983;
Gras and Keywood 2017).
Questions remain about the transport of DMS and nss-sulfates in the atmosphere. In this study the
focus will been on the sources and water uptake properties of nss-sulfates over the remote Southern
and South-West Pacific Oceans.
7.4.2 Secondary marine aerosol water uptake and composition
NSS-sulfates are neutralised by the ammonium ion (NH4+), the ratio of ammonium ion charges to
sulfate ion charges is the the level of neutralisation and ranges from 0 to 1. Sulfuric acid has neu-
tralisation value of 0, ammonium bisulfate 0.5, letovicite 0.75 and ammonium sulfate 1. NSS sulfates
species display distinct water uptake properties (Mifflin, Smith, and Martin 2009), in particular the
DRH can be used to identify the presence and relative abundance of these species in ambient air, as
seen in Table 1. In some cases volatility can also be used to identify the relative abundance of aerosols
of varying neutralisation states in ambient air (Alroe et al. 2018).
The complex mixture of gas phase emissions from phytoplankton also results in the secondary pro-
cessing of remote marine aerosols. HGFs widely observed over the remote ocean are consistent with
a dominant number fraction of NSS-sulfate particles with a small variable organic component (Good
et al. 2010; McFiggans et al. 2006; Swietlicki et al. 2008). Secondary processing has been observed to
change the hygroscopic properties of aged NSS-sulfate and SSA. HGFs lower than that obtained using
pure inorganic NSS sulfates are commonly observed over the remote ocean, this is attributed to the
condensation of organic species, which lower the HGF of the overall mixture (Allan et al. 2009; Bialek
et al. 2012; Fletcher et al. 2007; Good et al. 2010; Modini et al. 2010).
Secondary processing of SSA also occurs in the marine environment, which can result in changes to
the particle morphology, in particular the SSA tends to become less cubic over time (Laskin et al.
2012) . SSA aging commonly refers to the creation of an internally mixed NSS-sulfate/SSA particle,
this may occur from the condensation of DMS or via aqueous chemistry during cloud processing. The
NSS-sulfate component, in particular sulphuric and methane sulfonic acid (MSA), react with sea salt
producing volatile HCl gas and leaving the aged SSA depleted in chlorine, which is effectively displaced
43
by sulfate (Laskin et al. 2012). As with NSS sulfates the condensation of organic vapours on SSA can
lower the HGF (Miyazaki, Kawamura, and Sawano 2010; Sellegri et al. 2008).
Secondary organic components of NSS sulfate aerosols have been associated with discrepancies in sub-
saturated and supersaturated hygroscopicities (Prenni et al. 2007; Wex et al. 2009). Observations of
Aitken mode internally mixed NSS-sulfate/organics at Mace Head displayed substantially enhanced
CCN concentrations compared to that predicted using the surface tension of water (Ovadnevaite et al.
2017). Authors argued that the hygroscopic core was engulfed by a hydrophobic organic rich phase,
which acted to lower the surface tension, overcome the Raoult effect and enhance the CCN number
concentration. The approach taken to model this phase separation was similar to the compressed film
model (Ruehl, Davies, and Wilson 2016) discussed in Chapter 7.2.6. Conversely, observations from the
tropical Atlantic showed that the CCN number concentration estimated from HGF measurements were
systematically under-predicted (Good et al. 2010). The under prediction was attributed to either the
presence of low solubility compounds in the secondary organic or surface tension effects. A supersat-
uration dependence was also observed, with under-prediction of 10-20% above 0.2% supersaturation,
which increased to a maximum of around 80% at supersaturations less than 0.2%. Measurements at
Cape Grim showed an approximately 25% over-prediction of CCN concentrations (Blot et al. 2013;
Covert et al. 1998; Good et al. 2010; Twohy et al. 2013). Volatility and hygroscopicity observations
of 30-40 nm particles on the east coast of Australia indicated that a less hygroscopic component sup-
pressed the uptake of water by the NSS-sulfate fraction (Modini et al. 2010). In order to evaluate
the influence of secondary organics on CCN concentrations over the Southern and South West Pacific
oceans, measurements of the composition and water uptake, preferably under sub and supersaturated
conditions, of atmospherically relevant organics need to be considered.
7.5 Climate influences
7.5.1 SSA parameterisations
SSA source functions vary by orders of magnitude and tend to over predict the boundary layer number
concentrations for sub-200 nm particles (Leeuw et al. 2011). The total SSA flux is generally dependent
on wind speed, and more recent studies have used the ocean wave state and temperature (Ovadnevaite
et al. 2014; Grythe et al. 2014). A number of emissions schemes have been developed to partition the
composition of SSA between primary organic and inorganic components (Fuentes et al. 2010a; Gantt
et al. 2012b; Long et al. 2011; Spracklen et al. 2008; Vignati et al. 2010). An increase in the number
flux of SSA from seawater high in biological activity has also been observed (Fuentes et al. 2010b;
Long et al. 2014), and is thought to be related to the presence of biogenic surfactants. Observations of
enhanced fluxes due to biological activity are relatively scarce and have not been widely implemented
in modelling.
Chlorophyll-a concentrations, which are available via satellite retrieval for the entire globe, are used to
estimate the enrichment of organics in SSA. An additional wind speed term may be used to parameterise
the coverage of the SML (Gantt et al. 2012a), at higher wind speeds the SML coverage is reduced,
44
thereby reducing the SSA OMF. There are still questions regarding the validity of using chlorophyll-a
as an indicator of primary marine organics over the entire globe (Quinn et al. 2014, 2015), however it is
likely to remain the most appropriate metric for seasonal variation in marine organics (O’Dowd et al.
2015). Emissions schemes based on Chl-a also assume a uniform primary marine organic composition,
ignoring the possible spatial and seasonal variation, which could significantly influence the CCN activity
(Frossard et al. 2014b; Quinn et al. 2015). Only a small number of long term datasets are available for
validation, therefore applicability of these schemes globally, in particular in the Southern Hemisphere,
is relatively unknown (Quinn et al. 2015).
A model based on the adsorption of organics at bubble surfaces has been developed to overcome
the limitations faced by using Chl-a as a proxy (Burrows et al. 2014; Burrows et al. 2016), this
model is named OCEANFILMS (Organic Compounds from Ecosystems to Aerosols: Natural Films
and Interfaces via Langmuir Molecular Surfactants). Marine organics are broken up into classes and
assigned properties which are used to determine the likelihood that they will adsorb to the bubble
surface. The molecular classes are a lipid like class which is made up of labile DOC, a polysaccharide
like class which is made up of semi-labile DOC, a protein class with intermediate ocean lifetimes,
a processed class which is made up of long lived surface DOC and a humic like mixture from deep
upwelled water. Initially organics are adsorbed into the SML, and then into SSA via bubble bursting,
Figure 11. The presence of surfactants stabilises the bubble surface allowing the surface to drain more,
making it thinner (Modini et al. 2013), and increasing the proportion of organics in the film drops
(Burrows et al. 2014; Frossard et al. 2014b).
The surfactants are assumed to coat the film surface according to Equation 21 (Burrows et al. 2014)
where Θi is the fraction of the surface coverage of the monolayer from the ith molecular class, Ki,m is
the Langmuir adsorption coefficient, which defines the preferential adoption of surfactant like molec-
ular classes, Ci is the concentration in solution of the ith molecular class. The Langmuir adsorption
coefficient is derived based on observations of representative reference molecules for each molecular
class (Burrows et al. 2014). The surface coverage representation is further extended by considering
the interaction between polysaccharides and more surface active molecular classes, which results in
co-adsorption of more soluble polysaccharides (Burrows et al. 2016). In Equation 22 σi is the surface
coverage of the film lower surface by polysaccharides. The implementation of this co-adsorption has
been put forward as a possible explanation for the large proportion of polysaccharides observed in SSA
(Burrows et al. 2016; Frossard et al. 2014a).
Θi =Ki,1Ci
1 + ΣiKi,1Ci(21)
σi = ΣiΘiKpoly,2Cpoly
1 +Kpoly,2Cpoly(22)
The dry mass fraction of each molecular class in the nascent SSA is given by Equation 23, where mi is
the mass of molecular class i in SSA, mSS is the mass of salt in SSA, nj is the number of sides of the
45
Figure 11: Schematic of organic enrichment process during bubble bursting. Source: Burrows et al.
(2014).
46
film covered, θi is the fractional surface coverage from Equation 21, Mi is the molar mass of the organic
molecular class, ai is the specific area of the organic molecular class, ρSW is the density of ocean water,
s is the ocean water salinity and lj is the thickness of the film, where j represents the film layer (bubble
or SML). The film thickness for the SML ranges from approximately 20 - 400 μm and the bubble film
thickness ranges from approximately 0.01 - 1 μm (Burrows et al. 2014; Modini et al. 2013). The effect
of the bubble thickness is to change the ratio of organics to salt in the film, and therefore the OMF of
the SSA, but doesn’t alter the distribution of organic molecular classes.
(mi,bub
mi,bub +mSS
)=
njθi,bubMi
ai
njθi,bubMi
ai+ ρSW .lj .s
(23)
Implementation of OCEANFILMS into atmospheric models is in its infancy, however early results
indicate that adding primary organics as an internal mixture with SSA best represents the observed
seasonality (Burrows et al. 2018). Modelling indicates that the addition of SSA organics adds up to
30 cm-3 to MBL CCN over the Southern Ocean (SS = 0.1%), which in turn enhances the cloud forcing
in the summertime by -3.5 to -4 W/m2. It should be noted that this work is still under review. The
authors note that further work should be done to characterise the impact of surface active organic
emissions on the water uptake of SSA. OCEANFILMS characterises the SSA organic composition
and therefore may have the capacity to explain some of the observed regional variability and apparent
contrary results from previous nascent SSA experiments (Burrows et al. 2018; Elliott et al. 2018).
7.5.2 Sources of CCN
Observations and modelling indicate that entrainment from the free troposphere is a key source of
CCN to the marine boundary layer (Clarke et al. 2013; Merikanto et al. 2009; Quinn and Bates 2012;
Simpson et al. 2014). Observations over the remote equatorial Pacific indicated that particles from the
free troposphere contribute 65% to the marine boundary layer CCN immediately, and after growth in
the marine boundary layer another 25%. Refractory sea salt (measured at 360°C) were the final 10%
of marine boundary layer CCN (Clarke et al. 2013). The contribution of sea salt to the CCN number
concentrations was approximately 20% in a related study which also used volatility measurements (Blot
et al. 2013). Studies from Cape Grim have shown that following the passage of cold fronts nanoparticle-
rich air from the free-troposphere is mixed into the boundary layer, resulting in enhancements of 100
- 500 cm-3 in particle concentrations and contributing up to 30% of the Aitken mode concentration
(Gras 2009; Gras et al. 2009). Quinn et al. (2017) used size distribution and compositional data from
seven voyages in the Atlantic, Pacific and Southern Oceans across a number of seasons to estimate
the source contributions to CCN and found that between 70°S and 80°N meteorological features drive
the entrainment of NSS sulfates from the free troposphere, which primarily make up over 80% of
CCN.
Observations in the summertime North Pacific (Modini et al. 2015) suggested that SSA contributed less
than 10% to CCN number concentrations (at <0.3% SS) during low wind speed periods and averaged
16 - 28 % (but was as high as 63%) during high wind speed periods. Similar dependence on wind
47
speed was observed during summertime measurements over the Southern Ocean (Fossum et al. 2018),
for wind speed less than 16 m/s SSA contributed 2 - 13 % of CCN concentrations (0.2 - 0.3 % SS), at
higher wind speeds SSA contributed up to 100% of the CCN number concentration. Observations at
Cape Grim have also shown a strong dependence on the coarse mode contribution to CCN with wind
speed, which is absent for the Aitken and accumulation modes (Gras and Keywood 2017). Wind speed
is clearly an important variable for the determining SSA as CCN, this is modelled via the wind speed
dependent SSA flux.
Observations highlight the importance of seasonality to the CCN contributions, with a seasonally
varying NSS sulfate contributions generally overlaying a relatively stable contribution from SSA (Gras
and Keywood 2017; McCoy et al. 2015; Sanchez et al. 2018). In the summertime, when marine
productivity and DMS fluxes are high, the SSA contribution to CCN is at its lowest, for example in
the North Atlantic SSA were observed to contribute 55% of the CCN in late autumn, but only 4%
in late spring (Sanchez et al. 2017). At Cape Grim the contribution from SSA in summer has been
observed to be 21% (0.23% SS) and during winter increased to 38.5% (Gras and Keywood 2016).
Remote sensing over the Southern Ocean also shows that during the summer months the contribution
from SSA to CDNC is of the order of 50 - 60%, increasing to 80 - 90% over the winter months (McCoy
et al. 2015).
Over the Southern Ocean the contribution of SSA is understandably higher than for other regions due
to higher than average wind speeds. Of interest is the observation that within the Southern Ocean
there is variability in the proportion of CCN sourced from SSA and from NSS sulfates, in particular
further south the lower the NSS-sulfate number concentrations and therefore the greater the proportion
of SSA (McCoy et al. 2015; Quinn et al. 2017). Quinn et al. (2017) highlighted that the only region in
which their observations showed a significant contribution from SSA was in the high latitude Southern
Ocean. McCoy et al. (2015) also noticed a correlation between CDNC and the organic fraction in
SSA, suggesting that summertime enrichment of organics in SSA enhances CDNC in the high latitude
Southern Ocean. The mechanism for this is unknown, but could related to changes in surface tension
with an increased SSA organic fraction, or an enhanced number flux in organically enriched waters. It
should be noted that the contributions to CCN from McCoy et al. (2015) are higher than those from
in-situ studies. The pattern of these observations correlates with the strength of the observed radiative
bias, which is strongest in the high latitude Southern Ocean (Flato et al. 2013).
SSA has generally been considered an important contributor to remote marine CCN due to the high
hygroscopicity of sea salt. It is worth noting, however, that under conditions in which the water vapour
is limited more CCN doesn’t necessarily lead to a greater number of cloud droplets (Fanourgakis et al.
2019). An important example of this is coarse mode SSA, which can rapidly uptake water and decrease
in cloud supersaturation, which in turn reduced the number of particles activated into cloud droplets.
This has an important impacts for the cloud lifetime and precipitation (O’Dowd, Lowe, and Smith
1999).
Key uncertainties surround the size resolved removal, growth and compositional change during long
range transport and entrainment of potential CCN (Merikanto et al. 2009; Spracklen et al. 2008). In
48
particular the mixing state of aerosols and how this changes over time can have a large impact on CCN
concentrations (Wex et al. 2010). The large cloud bias observed over the Southern Ocean, in particular
during summer, is an indicator that attention needs to be paid to the seasonally varying contribution
of CCN in this sparsely observed region. Constraining variables associated with uncertain transport
processes requires observations of size resolved ambient marine particle composition and concentration
measurements (Quinn and Bates 2012; Simpson et al. 2014).
7.5.3 Representation in models
Atmospheric dynamics, including cloud thermodynamic and microphysical properties and boundary
layer structure, are poorly observed over the Southern Ocean. As a consequence climate and numerical
weather prediction models do not adequately represent cloud and precipitation fields in this region.
Models show an excess in the downwelling surface radiation and a deficit in the downwelling long
wave radiation at the surface (Bodas-Salcedo et al. 2012; Huang et al. 2014). Bright low to mid level
postfrontal cloud has been identified as being under predicted in a number of models including the Met
Office Unified Model, leading to the downwelling radiation biases (Bodas-Salcedo et al. 2012; Huang
et al. 2014; Webb et al. 2001; Williams and Webb 2008). The sensitivity of cloud radiative effects
to CDNC, relative to parameters controlling cloud thermodynamics, is an area of ongoing research.
It should be noted that recent observations have pointed to a strong dependence of cloud radiative
properties on cloud droplet number concentration (Glassmeier et al. 2019; Rosenfeld et al. 2019; Russell
et al. 2013), suggesting that aerosols describe three quarters of the low level marine cloud cooling
effects. Including high CDNC sensitivity in global circulation models would result in unrealistic global
cooling, therefore the CDNC sensitivity has been tuned down in some models, for example by setting
a minimum CDNC (Rosenfeld et al. 2019). An alternative is that the CDNC sensitivity is realistic
and that there is another unrepresented positive forcing due to aerosols, for example via deep clouds.
Further remote marine observations, particularly over the Southern Ocean are required to answer the
persistent and significant questions on the impact of aerosol-cloud interactions and to improve model
performance.
The implementation of aerosols into atmospheric models varies widely, so an example of some of the
issues surrounding implementation of a model commonly used in the Australian research community
is presented here. The Australian Community Climate and Earth-System Simulator is coupled with
the United Kingdom Chemistry and Aerosol (UKCA) model. UKCA uses an updated version of the
Global Model of Aerosol Processes (GLOMAP) model (Spracklen et al. 2005) known as GLOMAP-
mode (Mann et al. 2010). GLOMAP-mode simulates a number of aerosol components broken up into
7 modes as shown in Table 3 and simulates particle coagulation and condensation of sulfuric acid and
secondary organics. Over time insoluble modes can age and become partly soluble via coagulation
with soluble modes or condensation of sulfuric acid and secondary organics. The model deals with
the hygroscopic growth of the resultant mixtures using the ZSR assumption for each aerosol mode.
Particulate organics in the soluble modes are assumed to be either secondary or aged primary organics
and are assigned a water uptake 65% of sulfate, representing a moderate hygroscopicity (Mann et al.
2010). Primary marine organics and Aitken mode SSA are not represented in GLOMAP-mode, neither
49
Table 3: Standard aerosol configuration for GLOMAP-mode. Source: Mann et al. (2010).
Mode name Size range Composition
Nucleation-soluble up to 10 nm Sulfate, Particulate Organic Matter
Aitken-soluble 10 to 100 nm Sulfate, black carbon, particulate organic matter
Accumulation-soluble 100 nm to 1 μm Sulfate, black carbon, particulate organic matter, sea salt, dust
Coarse-soluble >1 μm Sulphate, black carbon, particulate organic matter, sea salt,dust
Aitken-insoluble 10 to 100 nm Black carbon, particulate organic matter
Accumulation-insoluble
100 nm to 1 μm Dust
Coarse-insoluble >1 μm Dust
is nucleation scavenging (CCN activation) of the Aitken mode or SSA chlorine depletion (Mann et al.
2010).
The model has been validated against a set of observational data sets (Ayers and Gras 1991; Heintzen-
berg, Covert, and Dingenen 2000), with the model overall capturing surface sulfate, sea-salt and dust
mass concentrations and under predicting black carbon and particulate organic matter (Mann et al.
2010). It is worth noting that there are few observational data sets for comparison over the Southern
Ocean. There is regional variation in model skill for CN and CCN concentrations. At Cape Grim
(and 2 other marine boundary layer sites) the modelled CN is strongly biased low, with normalised
mean bias (b) of -0.75. In addition a strong low bias (b = -0.70) was reported for Aitken mode number
concentrations, driven by under-prediction in the southern hemisphere (Ayers and Gras 1991; Heintzen-
berg, Covert, and Dingenen 2000). A more modest low bias for the accumulation mode was observed
(b = -0.46), but is nevertheless driven by under-prediction in the southern hemisphere polar region
and mid-latitudes. CCN concentrations at 1.2% supersaturation are well represented at Cape Grim,
however at 0.23% supersaturation there is a high bias in the modelled CCN concentrations (b=0.78),
particularly in spring and summer. This is of note because 0.23% supersaturation is within the range
expected for typical marine stratiform cloud (Twohy et al. 2013).
7.6 Knowledge gaps
1. The composition of SSA from both chamber and ambient measurements is not well characterised
for the Southern and South Pacific oceans. In particular the organic enrichment of SSA and
the resulting water uptake is scarcely measured over the Southern Ocean, over which SSA is an
important source.
50
2. The seasonal contribution of SSA to CCN over the Southern Ocean has been observed using
remote sensing techniques, however in-situ data are required.
3. The mixing state and impact of cloud processing of SSA and NSS-sulfates on CCN concentrations
is poorly characterised over the Southern and South Pacific oceans.
4. The suitability of SSA organic emissions parameterisations for the southern hemisphere is poorly
characterised, and therefore the SSA contributions to modelled summertime cloud biases over
the Southern and South-Pacific oceans is unknown.
51
8 Research design
Observations of aerosol physiochemical properties from a total of four research voyages over the South
Pacific and Southern Oceans were used to address the knowledge gaps outlined above and will be
reported on here. The first research voyage was the SOAP voyage over the South West Pacific as
shown in Figure 12. SOAP was primarily aimed at characterising the organic enrichment of nascent
SSA and the resulting impact on water uptake, this work addressed the first and fourth knowledge
gaps identified in Chapter 7.6.
Figure 12: Voyage map for SOAP study, coloured by bloom periods.
In addition to the SOAP voyage, three Southern Ocean voyages will be reported on here, the Cold
Water Trial of the RV Investigator, The CAPRICORN voyage and the Ice-Edge to Equator voyage,
maps for which are shown in Figure 13. Aerosol composition and water uptake data during these
voyages characterise the contributions to CN and CCN from primary and secondary marine aerosol
and the aerosol mixing state, these observations address knowledge gaps two and four. The three
Southern Ocean voyages span summer to winter seasons and provide zonal transects of the Southern
Ocean, therefore results from these voyages will be presented together.
52
Figure 13: Voyage track for RV Investigator Southern Ocean voyages examined in this study.
8.1 Observation campaigns
8.1.1 Surface Ocean Aerosol Production (SOAP)
The SOAP voyage examined air-sea interaction over the biologically-productive frontal waters of the
Chatham Rise, east of New Zealand in February and March 2012. The Chatham Rise couples pristine
marine air masses with high biological activity due to the mixing of warm subtropical water, with
relatively low macronutrient levels, with cool Southern Ocean waters which are depleted in iron, but
not in macronutrients (Law et al. 2017b). Phytoplankton blooms were identified via satellite ocean
colour images and further mapped using a suite of underway sensors (Chl-a, β 660 backscatter, pCO2,
DMSsw). Three broad bloom periods were defined as shown in Figure 12, 12 hours into the voyage
the first bloom (B1) was driven by dinoflagellates and displayed elevated Chl-a and seawater DMS, 7
days into the voyage a weakening bloom (B2) was driven by coccolithophores, and a final bloom (B3)
displayed a mixture of phytoplankton communities. Bloom 3 was subdivided into B3a and B3b due to
changes in the surface water characteristics following the passage of a storm.
23 seawater samples were collected throughout the voyage from 20 sampling locations for the purpose of
generating nascent SSA. Seawater was primarily collected from the ocean surface (10 cm depth) during
workboat operations or from the mixed layer (3 - 12 m depth, always less than the measured mixed
layer depth) from CTDs, however a number of deep water samples were collected for comparison. Due
to the sampling method used these seawater samples are not necessarily representative of the SML.
Samples also spanned the variability in ocean biological conditions observed throughout the the SOAP
voyage. Table 4 provides a description of the seawater samples that were taken.
53
Table 4: Details of ocean water samples collected for generation of SSA.
Watersample
Depth[m]
Depthclass
Date/Time[UTC+12]
Bloom Lat/Lon[S/E]
Chl-a[g/L]
DVH-TDMA[nm]
Filter
Workboat1
0.1 Surface 15/2/12 8:05 1 44.621/174.772
0.985 50 Y
Workboat4
0.1 Surface 17/2/12 8:02 1 44.587/174.690
1.405 50 Y
CTDU7505
50 Deep 18/2/12 9:16 1 44.574/174.735
0.974 50 Y
Workboat5
0.1 Surface 18/2/12 8:04 1 44.590/174.685
1.160 50 Y
CTDU7506
2.79/505.5
Mixed/Deep
19/2/12 7:30 - 44.335/175.246
0.91/- 50 Y
CTDU7507
4.09 Mixed 20/2/12 7:15 - 45.960/173.646
0.880 50/100 N
CTDU7508
2.07/400
Mixed/Deep
21/2/12 7:55 - 43.741/176.966
0.670/-7 50
N
CTDU7510
3.13 Mixed 22/2/12 9:22 2 43.717/-178.156
1.520 50 Y
Workboat6
0.1 Surface 22/2/12 8:27 2 43.715/-179.860
1.530 50 Y
Workboat7
0.1 Surface 24/2/12 13:03 2 43.585/-179.753
0.490 50 Y
CTDU7518
254.7 Deep 24/2/12 15:15 2 43.599/-179.767
- 30/50 Y
CTDU7520 a
211.62 Mixed 25/2/12 14:30 2 43.630/179.741
0.630 50 Y
CTDU7521
2.48 Mixed 26/2/12 6:52 - 43.962/179.308
- 50 Y
Workboat8
0.1 Surface 27/2/12 14:39 3a 44.110/175.140
0.530 50/100 Y
CTDU7524
11.21 Mixed 28/2/12 13:10 3a 44.542/174.873
0.460 50 Y
Workboat9 b
30.1 Surface 29/2/12 8:03 3a 44.600/174.870
0.290 50/130 Y
CTDU7528
9.17 Mixed 2/3/12 14:59 3b 44.191/174.944
0.450 50 Y
CTDU7530
10.31/810.5
Mixed/Deep
3/3/12 14:45 3b 44.781/174.650
0.490/- 50 Y
CTDU7532
8.96 Mixed 4/3/12 15:25 3b 44.243/174.523
1.010 50 Y
Workboat10
0.1 Surface 5/3/12 9:04 3b 44.185/174.295
- 50 Y
a Thermodenuder placed upstream of H-TDMA and UFO-TDMA. No VH-TDMA organic fractions available. b DRH
measured for Workboat 9.54
8.1.2 Southern Ocean voyages (RV Investigator)
The results from in-situ aerosol measurements taken during three voyages on the Southern Ocean will
be reported in this study. The Cold Water Trial took place in the 2015 Austral summer, CAPRICORN
took place in Autumn 2016 and Ice-edge to equator took place in Autumn - Winter 2016. Together
the three voyages reported herein provide a good zonal transect over the Southern Ocean and allow
the examination of the influence of seasonality on the Southern Ocean aerosol. In-situ measurements
over a number of seasons is valuable because the cloud bias over the Southern Ocean is particularly
strong during the high biologically active summer months.
Cold Water Trial
The Cold Water Trial (CWT) of the research vessel (RV) Investigator (Marine National Facility,
Hobart) was undertaken in January and February 2015. Ambient atmospheric data was obtained
during the 21 day return transect across the Southern Ocean south from Hobart adjacent to the
longitude 147.5 ° East as shown in Figure 14. The CWT was a test voyage for the RV Investigator
as a scientific platform for remote marine research, particularly in the cold sub-Antarctic waters.
Despite the opportunistic nature of the voyage a comprehensive suite of aerosol and gaseous precursor
measurements were made over a complete zonal transect of the Southern Ocean during the period of
highest biological activity.
Cloud-aerosol-precipitation processes over the Southern Ocean (CAPRICORN)
The Clouds, aerosols, precipitation, radiation and atmospheric composition over the Southern Ocean
(CAPRICORN) study aimed to improve the representation of Southern Ocean cloud for satellite re-
trieval and global atmospheric models. CAPRICORN largely focused on the measurement of aerosol-
cloud interactions, and in particular the objectives of the study, as stated in Mace and Protat 2018a
and Protat et al. 2017, were to:
1. characterise the cloud, aerosol, and precipitation properties,
2. characterise the boundary layer 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,
3. evaluate and improve satellite estimations of these properties,
4. evaluate and improve the representation of these properties in the Australian ACCESS regional
and global model.
The CAPRICORN voyage took place from 12 March to 15 April 2016 onboard the RV Investigator, and
was part of a larger series of investigations, including the CAPRICORN-2 voyage and the associated
Southern Ocean Clouds, Radiation, Aerosol Transport Experiment Study (SOCRATES) in January -
55
February 2018 (Protat et al. 2017). Analyses and interpretation of the results from these observations
is still ongoing, however in this study we will focus on the results from the initial CAPRICORN voyage,
particularly the aerosol measurements.
Two projects ran supplementary to the CAPRICORN project on the RV Investigator, the Southern
Ocean Time Series (SOTS) automated moorings for climate and carbon cycle studies and the eddy
physics and biogeochemistry in the Antarctic Circumpolar Current (eddies) study. The SOTS and
eddies projects are described briefly here because they dictated the voyage track, and in the case of the
eddies project provide some valuable ocean biogeochemical data that provides contextual information
to the aerosol properties. The SOTS study involved the retrieval and replacement of three automated
moorings at approximately 46.8 °S and 141 °W, these activities dominated the first 10 days of the
voyage. As part of the eddies project a 190km diameter cold core cyclonic eddy that had detached
from the Subantarctic Front (SAF) was sampled in detail, 3 transects of the eddy were completed,
measuring the surface water pCO2, Chlorophyll-a and nutrient concentrations as well as sea surface
temperature and salinity (Moreau et al. 2017). Oceanographic measurements were also made during
transects across the SAF. As a result of the time constraints imposed by supplementary projects
CAPRICORN voyage captured data only to 55 °S, which is not as far south as the CWT and Ice-edge
to Equator voyages.
Ice-edge to equator voyage
The Ice-edge to Equator voyage ran from 23 April to 29 June 2016, and the primary voyage track
was a 7000 km northerly transect from 65 °S to the equator. The primary purpose of this voyage was
to capture temperature, salinity, pressure, oxygen, fluorometry, major nutrients and phytoplankton
information by performing full depth CTD casts at high spatial resolution, every 0.5°of latitude (Raes
et al. 2018). The measurements performed on this voyage repeat previous measurements, made 10
years ago, and will provide excellent context for the influence of the marine biota and organics on
aerosol properties. The atmospheric component of this voyage was supplementary, however the transect
provided a good opportunity to examine the zonal variability in CCN contributions.
8.2 Measurement instrumentation
8.2.1 SOAP instrumentation
During the Surface Ocean Aerosol Production (SOAP) campaign ambient marine aerosol measurements
were made from a common aerosol inlet line which provided clean air from the bridge, a height of
17.5 m above sea level (Law et al. 2017b). In addition water samples were collected and nascent SSA
measurements were made from a chamber inside the aerosol laboratory. The VH-TDMA, UFO-TDMA
and SMPS were switched between these two sampling methods as noted in Table 5.
Nascent SSA was generated in-situ in a 0.45 m3 cylindrical polytetrafluoroethylene chamber housing
4 sintered glass filters with porosities between 16 and 250 μ m (Cravigan 2015; Mallet et al. 2016,
56
Table 5: List of aerosol instrumentation deployed during SOAP voyage.Reproduced from Law et al.
(2017).
Instrument Measurement Common aerosolinlet
SSAchamber
CPC-3772 Aerosol concentration > 10 nm Yes -
CPC-3010 Aerosol concentration > 10 nm Yes -
SMPS Aerosol size distributions (10 - 300 nm) Yes Yes
OPC Aerosol size distribution (0.5 - 30 m) Yes -
CCNc Cloud condensation nuclei concentration (0.5% SS) Yes -
VH-TDMA
Aerosol volatility and hygroscopicity Yes Yes
UFO-TDMA
Aerosol alcohol uptake Yes Yes
Aethalome-ter
Black carbon mass concentration Yes -
PicarroCRDS
CO2 and methane concentration Yes* -
PTR-MS VOC concentration (DMS, isoprene, monoterpenes,acetaldehyde, and others)
Yes* -
*Piccarro and PTR-MS were connected to gas sampling line not common aerosol inlet, but measured ambient air.
57
2016). Generation of SSA using sintered glass filters isn’t the most representative of open ocean SSA
production, as discussed in Chapter 9.1.2.The Dried and filtered compressed air was passed through the
glass filters at a flow rate of 15.5 ± 3 L/min and resulting SSA was sampled from the headspace of the
chamber. A diffusion drier was used to dry the sample flow to 20 ± 5 % RH prior to characterisation.
Figure 14 shows the sampling set-up used to generate, condition and measure nascent SSA.
Figure 14: Experimental schematic of nascent SSA chamber experiments during SOAP voyage. The
VH-TDMA (grey) contains an RH controlled region (blue) used for water uptake measurements.
Size distributions and number concentrations of 10 to 300 nm diameter SSA were measured using a TSI
3080 scanning mobility particle sizer (SMPS), coupled with a 3071 differential mobility analyser and a
3010 condensation particle counter (CPC) (TSI, Shoreview, MN), with an aerosol sample flow rate of 1
L/min and a sheath flow rate of 10 L/min. In addition, the ambient marine aerosol was characterised
by a cloud condensation nuclei counter (CCNc) (DMT, Longmont, CO) operated at a continuous
supersaturation of 0.5%. The CCNc was coupled with a condensation particle counter (CPC), TSI
3010, which measures the particle number concentration greater than 10 nm (CN10). A second CPC,
TSI 3772, also measured CN10. An optical particle counter (OPC), GRIMM EDM107 environmental
dust monitor (GRIMM, Airnring, Germany) was used to measure the particle size distribution of
ambient marine aerosol from 0.4 to 30 μm, therefore extending the range of the SMPS measurements to
cover the coarse mode. An Aethalometer (AE22, Magee Scientific, Berkeley, CA) was used to measure
the mass concentration of black carbon, which is a proxy for anthropogenic combustion, primarily
from the ship itself. A range of gas sampling was also conducted during the SOAP voyage, of primary
interest here is the Picarro cavity ring down spectrometer (Picarro, Santa Clara, CA), which measures
the concentration of CO2 and methane, which can also be useful for identifying air masses that have
been influenced by anthropogenic emissions. The proton transfer reaction mass specrometer (PTR-MS,
Ionicon, Innsbruck, Austria) measures volatile organic compounds in the gas phase, including DMS,
and is of interest for identifying periods when emissions of gaseous precursors are elevated.
58
The volatility and hygroscopicity of nascent SSA and ambient marine aerosol was determined with
a volatility and hygroscopicity tandem differential mobility analyser (VH-TDMA) (Alroe et al. 2018;
Johnson et al. 2008; Johnson, Ristovski, and Morawska 2004). The VH-TDMA selects particles based
on mobility diameter, conditions them, and measures the resulting particle size distributions using
parallel SMPSs (V-SMPS and H-SMPS in Figure 14), each with a TSI 3010 condensation particle
counter (CPC). The aerosol sample flow rate for each SMPS was 1 L/min, resulting in a total inlet
flow of 2 L/min, the sheath flow for the pre-DMA, V-DMA and H-DMA were 11, 6 and 6 L/min,
respectively. The VH-TDMA can be operated in a number of sampling modes, in general the instrument
is designed to observe the water uptake at ambient temperature and subsequently observe the water
uptake of a non-volatile component (at some elevated temperature), which is used to infer the water
uptake of the volatile component. The following two paragraphs describe the VH-TDMA sampling
method used for nascent SSA measurement and ambient marine aerosol measurements during SOAP,
respectively.
• Nascent SSA with a mobility diameter of 50 nm were preselected for each water sample, a
number of water samples were also analysed with a preselected diameter of 30, 100 or 130 nm
for comparison. Table 4 details the VH-TDMA analysis pre-selected particle size. The SSA
volatile fraction was determined by measuring the diameter of preselected SSA upon heating
by a thermodenuder up to 500 °C, in temperature increments of 5°C - 50°C. Subsequent to
heating the SSA was exposed to 90% RH and the hygroscopic growth factor was measured.
The dependence of HGF on RH at ambient temperature was measured for one water sample
(workboat 9) to provide the deliquescence relative humidity (DRH).
• Ambient marine aerosol with a mobility diameter of 30, 50 or 100 nm was preselected and the
ambient HGF was measured at a 3 minute time resolution. On 21 occasions throughout the
voyage the thermodenuder temperature was increased, generally to 300 °C in increments of 5 -
20 °C, and the volatility and HGF of the refractory component was measured.
The heater used in the VH-TDMA has a relatively short residence time of 3 seconds and no cooling
stage, it is therefore possible that there could be some residual semi-volatile components remaining
after heating.
Size distributions, volatility profiles and HGFs have also been measured for laboratory sea salt sam-
ples, which were generated using the same glass filters and chamber. In addition laboratory sea salt
Transmission Electron Microscopy (TEM) samples were collected using a TSI 3089 Nanometer Aerosol
Sampler and analysed using X-ray dispersive spectrometry (TEM-EDX). TEM data were collected on
a JEOL2100 transmission electron microscope operating at 200 kV coupled with a Gatan high angle
annular dark field (HAADF) detector.
The ultrafine organic tandem differential mobility analyzer (UFO-TDMA) was used to calculate mod-
erately oxidized organic volume fractions of SSA particles by measuring how much the particle grows
in sub-saturated (82% ± 2%) ethanol vapour. The growth factor of the SSA samples in ethanol vapour
was measured at pre-selected mobility diameters of 15, 30, 40 and 50 nm using the UFO-TDMA. The
organic volume fraction from UFO-TDMA measurements were computed using Zadanovkii-Stokes-
59
Robinson (ZSR) approximation (Chen et al. 1973; Stokes and Robinson 1966), the ethanol growth
factor for primary organics and sea salt were assumed to be 1.55 and 1, respectively (Vaattovaara et al.
2006). UFO-TDMA and H-TDMA measurements of sample U7520 were pre-treated using a thermod-
enuder (heated in 10°C steps up to 500°C), this allowed examination of the contribution of volatile
components to particle growth in ethanol vapour, but excluded the estimation of organic fractions from
VH-TDMA measurements.
SSA generated from 23 ocean water samples was collected on filters for further compositional analysis
using transmission Fourier Transform Infra Red (FTIR) and Ion Beam analysis (IBA). SSA was sampled
through a 1 μm sharp cut cyclone (SCC 2.229PM1, BGI Inc., Waltham, Massachusetts) and collected
on Teflon filters, with the sample confined to deposit on a 10 mm circular area. Back filter blanks were
used to characterise the contamination during handling, and before analysis samples were dehydrated
to remove all water, including SSA hydrates, as described in Frossard and Russell 2012. Filter blanks
were under the detection limit for the FTIR and Si was the only compound with blank measurements
above the IBA detection limit. FTIR measurements were carried out according to previous marine
sampling using this technique (Maria et al. 2003; Russell, Bahadur, and Ziemann 2011; Russell et
al. 2009) and characterised the functional groups associated with major carbon bond types, including
saturated aliphatic (alkane) groups, alcohol (used here to include phenol and polyol) groups, carboxylic
acid groups, non-acidic carbonyl groups, and primary amine groups. FTIR measurements are non-
destructive, therefore subsequent to FTIR analysis filter samples underwent simultaneous particle
induced X-ray emission (PIXE) and gamma ray emission (PIGE) analysis (Cohen et al. 2004). The
elements discussed herein, of interest for SSA, are Na (from PIGE) and Mg, Si, S, Cl, K, Ca, Zn, Br
and Sr (from PIXE). It should be noted that Rutherford backscattering and particle elastic scattering
analysis did not yield useful results for the analysis of C, N, O, and H. The SSA organic concentrations
were instead obtained solely from FTIR analysis.
A large number of parameters characterising the ocean water were taken, broadly these characterise
the physical properties, nutrient concentration, the phytoplankton population, bloom productivity and
the concentration of molecular classes important for SSA e.g. fatty acids, proteins and carbohydrates.
A detailed list of ocean water measurements undertaken during the SOAP voyage is contained in Law
et al. 2017a. The parameters of interest here are the concentrations of Chlorphyll-a, high molecular
weight proteins and sugars, alkanes and the fatty acid concentration. It should be noted that the
protein and carbohydrate measurements include both dissolved and particle components. Fatty acid
measurements are made up of measurements from 34 individual fatty acid species, which can be broken
up into saturated (14 species), monounsaturated (9 species) and polyunstaurated (11 species). Alkanes
were also speciated, with carbon numbers ranging from 13 to 36. In addition, meteorological variables
such as wind speed, wind direction, pressure and solar irradiance were also collected and used to
characterise the air masses.
60
8.2.2 RV Investigator instrumentation
A variety of aerosol and gas phase instrumentation was deployed on the three observation campaigns
on-board the RV Investigator reported in this study, as shown in Table 6. It should be noted that
Table 6 displays only the instrumentation on-board the RV Investigator that was relevant to this
study, a subset of all of the instrumentation, and that all data are from ambient measurements i.e.
there were no nascent SSA measurements. Size distributions from 4 to 700 nm are available for all three
voyages, however there was no APS for the Cold Water Trial voyage, therefore coarse mode aerosol will
have to be inferred from SMPS measurements i.e. 500 - 700 nm diameter bins. The MAAP provided
black carbon mass concentration and can be used as a marker for combustion emissions, in particular
emissions from the ship. Similarly to the SOAP voyage, the PTR-MS provided information on the gas
phase precursors. Radon is a radioactive trace gas that is emitted from land, and therefore was used
as a proxy for air masses that are of terrestrial origin (Zahorowski, Chambers, and Henderson-Sellers
2004). The ACSM provided real time speciated mass composition for non-refractory PM1. Species of
interest are sulfates, MSA, ammonium, organics, chloride and sea salt. Note that even though sea salt
is refractory the concentration can be estimated by scaling up some of the known sea salt fragments,
this will be briefly discussed in Chapter 8.3.1. The ACSM has time resolution of 10 minutes, and can
be though of as an indicator of the accumulation and coarse mode particle composition, the efficiency
of the ACSM in the Aitken mode is low due to the low mass and a declining instrumental detection
efficiency at sizes lower than 100 nm.
Onboard the RV-Investigator the VH-TDMA was used in a different operating mode to that during
the SOAP voyage. The preselected particle diameter was alternated between 40, 100 and 150 nm.
Occasionally the selected mobility diameter was changed to 30, 100 and 150 nm, regardless the sizes
were designed to characterise the Aitken mode, the Hoppel minimum and the accumulation mode,
respectively. At each size the HGF and the VFR were measured at room temperature and at 250
°C. During the three voyages on the RV Investigator the VH-TDMA was deployed with a TSI CPC-
3787 on the H-SMPS instead of a CPC-3010, relevant VH-TDMA flow rates and operating parameters
for the VH-TDMA during the RV-Investigator voyages are set out in Table 6. The thermodenuder
temperature was set to 120 °C, not 250 °C, during the Cold Water Trial voyage. The time resolution for
a VH-TDMA scan is 3 minutes, therefore measurements were made at each preselected diameter for 6
minutes (room temp and 250 °C), and measurements of the whole suite of sizes and temperatures took
18 minutes. Similarly to the SOAP campaign the relatively short residence time of the VH-TDMA
may lead to a slight over estimation of the non-volatile fraction.
A full suite of complementary meteorological data, wind direction ,wind speed, atmospheric pressure,
solar radiation etc. are standard measurements onboard the RV Investigator. In addition, concurrent
measurements of ocean water parameters (primarily chlorophyll-a) were taken during the CAPRI-
CORN (Dawson, Strutton, and Gaube 2018; Moreau et al. 2017) and Ice-edge to Equator (Raes et al.
2018) voyages. During the CAPRICORN voyage measurements of the cloud properties were made
using a Bistatic Radar System for Atmospheric Studies (BASTA) 95-GHz cloud radar, a cloud-aerosol
Lesosphere RMAN-511 mini-Raman lidar operating at 355 nm and a Radiometrics 2-channel microwave
61
Table 6: Instrumentation deployed on the RV Investigator during the Cold Water Trial (CWT),
CAPRICORN and Ice Edge to Equator voyages
Instrument Measurement CWT CAPRI-CORN
Ice toEquator
CPC -3772/3010
CN10 Yes - Yes
CPC - 3776 CN3 - Yes -
SMPS Aerosol size distributions (14 - 700 nm) Yes Yes Yes
nano-SMPS Aerosol size distribution (4 - 400 nm) Yes Yes Yes
APS Aerosol size distributions (0.5 - 20 m) - Yes Yes
CCNc Cloud condensation nuclei number concentration Yes Yes Yes
MAAP Black carbon mass concentration (Multi Angle AbsorptionPhotometer)
Yes Yes Yes
VH-TDMA Aerosol volatility and hygroscopicity Yes Yes Yes
ACSM Non-refractory PM1 aerosol composition (AerosolChemical Speciation Monitor)
Yes - Yes
NAS TEM sample collection (Nanometer Aerosol Sampler) Yes Yes Yes
Radonmonitor
Concentrations of radon (Rn-222) Yes Yes Yes
PTR-MS VOC concentration (DMS, isoprene, monoterpenes,acetaldehyde and others)
Yes Yes Yes
Table 7: VH-TDMA operating parameters for the Southern Ocean campaigns onboard the RV In-
vestigator (Cold Water Trial, CAPRICORN and Ice-edge to Equator)
Voyage Voyagecode
Pre-selected mobilitydiameter [nm]
Thermodenudertemperature [°C]
Sample aerosol flowrate [L/min]
Sheath flowrate [L/min]
CWT IN2015_e01 40*, 100, 150 120 Pre:1.4 V:0.8 H:0.6 Pre:8 V:6 H:4
CAPRI-CORN
IN2016_v02 40*,100,150 250 Pre:1.6 V:1 H:0.6 Pre:11.8V:7.95 H:4.95
Ice toEq.
IN2016_v03 40*,100,150 250 Pre:1.6 V:1 H:0.6 Pre:11.8 V:8H:5
* 40 nm was occasionally changed to 30 nm.
62
radiometer as reported in Mace and Protat 2018b, 2018a. The microwave radiometer was also used
during the Ice-edge to Equator voyage.
8.3 Analysis and modelling
8.3.1 SOAP analysis
Nascent SSA measurements
Nascent SSA size distributions for each water sample were averaged and normalised to their maximum
value. Non-linear least square fits of up to four log-normal modes were fitted to each distribution
with a random selection of initial values for the geometric mean and standard deviation, constrained
to 10 - 320 nm and less than 2, respectively. The most appropriate fit was determined using the
bayesian information criterion, which is a measure of the error in reconstructing the measured size
distribution that applies a penalty based on the number of parameters used and therefore avoids over
fitting (Sakamoto, Ishiguro, and Kitagawa 1987).
All VH-TDMA data were inverted using the TDMAinv algorithm (Gysel, McFiggans, and Coe 2009),
and external modes were allocated based on local maxima of the resulting piecewise linear GF distri-
bution. The volatile fraction (VF) was computed using Equation 24, where d is the particle diameter,
T denotes the temperature of the thermodenuder and o denotes ambient temperature.
V F = 1− V FR = 1−(dTd0
)3
(24)
In this study SSA organic volume fractions were calculated using volatility measurements, by account-
ing for the presence of sea salt hydrates. The volatility due to hydrates was used as a proxy for the
proportion of inorganic sea salt in the natural seawater samples, which in turn provided the proportion
of organics. The hydrate fractions were computed by comparing the natural sea spray volatility profiles
and laboratory sea salt volatility profiles over the temperature range 250 - 400 °C. The organic volume
fraction was inferred from volatility measurements using the linear model outlined in Equation 25,
where VFT is the measured volatile fraction of the sea spray sample at thermodenuder temperature T
and VFT,SS is the measured volatile fraction of laboratory sea salt at temperature T. Laboratory sea
salt volatility profiles were measured using three different sea salt samples, a commercially available
sea salt (Pro Reef Sea Salt, Tropic Marin, Wartenberg, Germany), and two mixtures of laboratory
grade salts, one mimicking Sigma-Aldrich Sea Salts composition and one mimicking the Niedermeier
et al. 2008 Atlantic Ocean sea salt composition. Sea salt solutions were all made to a concentration
similar to sea water, 35 g/L and volatility profiles were within experimental error of one another. The
error in the laboratory sea salt volatile fractions were assumed to be the maximum of the standard
error in the mean across the three sea salt samples and the instrumental error (± 3%).
Measurements over the temperature range 250 - 400 °C were used in Equation 25, and it was assumed
that only hydrates evaporated for the SSA sample in this temperature range. Reported volatility
63
profiles of nascent SSA generated from natural seawater and from laboratory sea salts are consistently
parallel at 250 - 400 °C, indicating that the volatility was due to the evaporation of similar components
i.e. hydrates (Modini et al. 2010; Rasmussen et al. 2017). The proportion of volatility due to sea salt
components (hydrates) in the natural SSA sample (relative to laboratory sea salt) is given by f and
is computed from the slope of Equation 25. The proportion of the volatility due to sea salt hydrates,
f , was assumed to represent the proportion of sea salt in the natural SSA sample. The total organic
volume fraction is then given by 1− f .
V FT = f × V FT,SS +OV Fsv (25)
A further correction was applied to f for the case where the profile of inorganic, and therefore hydrates,
is different between the natural SSA sample and the laboratory sea salt sample. The correction is
represented by fio in Equation 26. Ca and Mg enrichment and associated Cl depletion can result
in an overall reduction in hydrate forming sea salt species, such as CaCl2 and MgCl2 (Salter et al.
2016; Schwier et al. 2017). The ionic composition of nascent SSA generated from natural seawater
was measured using IBA and used to compute the inorganic molecular composition, which in turn was
used to compute the volume fraction of hydrates for each sample. The same analysis was performed
based on the ionic composition of laboratory sea salts and fio was computed as the ratio of natural
seawater SSA hydrate volume fraction to laboratory sea salt hydrate volume fraction. The AIOFMAC
model was used to compute the molecular composition, using the mole fractions of Na, K, Mg, Ca,
and Cl from ion beam analyses (for natural SSA samples) and from known composition (laboratory
sea salt).
Any excess volatility in the nascent SSA generated from natural seawater was assumed to be due to the
presence of semi-volatile organics, and was represented by the intercept in Equation 25. Uncertainties
for OVFsv and VF were taken to be whichever is the maximum out of the measurement error for VF
and the standard error for the slope. The measurement error in VF is ±3%, which is due to a 1% DMA
sizing uncertainty (Johnson, Ristovski, and Morawska 2004; Modini, Harris, and Ristovski 2010). This
approach quantifies the proportion of volatility due to the presence of hydrates and due to the presence
of organics, and is therefore an improvement on previously published estimates of the SSA sv-OVF
(Mallet et al. 2016; Modini, Harris, and Ristovski 2010; Quinn et al. 2014).
OV Ftot = 1− f
fio(26)
The organic volume fractions calculated from volatility measurements were converted into PM1 mass
fractions and validated against OMFs calculated from filter samples. Volume distributions, computed
from SMPS measurements (assuming spherical particles), and assumed organic density of 1.1 g.cm−3
(Keene et al. 2007) for the volatile and non-volatile fractions, respectively, were used to convert between
organic volume and mass fractions. Note that the organic volume fractions quoted in the results are
64
computed as a fraction of the dry SSA mass, not including hydrates, to allow for a better comparison
with the measured OMF.
The HGF was computed using Equation 27, where dRH,T is the measured diameter at the RH of
the H-SMPS (90% for all except the DRH measurement) and thermodenuder temperature T, and
ddry,T is the measured dry diameter at temperature T. HGFs for sea salt were shape corrected using
the dynamic shape factor from (Zieger et al. 2017). The presence of an organic fraction has been
observed to increase the sphericity of nascent SSA (Laskin et al. 2012), therefore an organic fraction
dependent shape correction was applied (Zelenyuk et al. 2007). A single shape factor was used across
all temperatures because TEM images of laboratory sea salt showed an insignificant difference between
the apparent shape of SSA at ambient temperature and those heated to 250 °C.
HGF =dRH,T
ddry,T(27)
The organic mass fraction from SSA samples collected on filters was computed from the total organic
mass from FTIR analysis and the inorganic mass from ion beam analysis, as in Equation 28. The filter
exposed area (0.785 cm-2) was used to convert inorganic areal concentrations into total mass. The
inorganic mass was computed as the sum of Na, Mg, SO4, Cl, K, Ca, Zn, Br, Sr, and the SO4 mass
was computed by multiplying the S mass by 3 i.e. all S was assumed to be in the form of SO4. The
uncertainty in the organic mass measured using FTIR is up to 20% (Russell 2003; Russell et al. 2009),
this is taken as the uncertainty in OMF.
OMF =OMFTIR
OMFTIR + IOMIBA(28)
Enrichment factors for inorganic elements were calculated with respect to the laboratory prepared
seawater and presented with respect to Na+(Salter et al. 2016), as shown in Equation 29, where X is
the element of interest. Enrichment factors were calculated from Ion Beam Analysis of filter samples
and from TEM-EDX analysis of laboratory sea salt samples.
EF (X) =([X]/[Na+])SSA([X]/[Na+])water
(29)
The OCEANFILMS model was implemented for the surface and mixed layer nascent SSA experiments
with measured water parameters used to represent bulk seawater molecular classes. Lipids were as-
sumed to be equal to the total concentration of fatty acids, the total high molecular weight proteins
were used to represent the protein molecular class, and high molecular weight reducing sugars were
used to represent polysaccharides. Note that these measurements were not micro layer measurements,
however the seawater samples were collected via CTDs or on workboats and therefore probably don’t
represent the SML. Missing water composition data were filled using the relationships outlined in Bur-
rows et al. 2014, based on the lifetime of each molecular class, for example the bulk concentration of
proteins was assumed to be equal to one third of the polysaccharide concentration, when no other data
were available. The humic like molecular class was assumed not to be present at the surface and in
65
the mixed layer, and therefore were not included (Burrows et al. 2014). The processed molecular class
concentration was assumed to make up the remainder of the DOC after polysaccharides, proteins and
lipids have been subtracted, a minimum was applied to the concentration of processed compounds to
prevent unrealistically low or negative concentrations. The langmuir adsorption coefficients for each
molecular class was taken directly from Burrows et al. 2014. OCEANFILMS was run including the
co-adsorption of polysaccharides with all other molecular classes (Burrows et al. 2016) , and an as-
sumed bubble thickness of 0.3. The effect of the bubble thickness is to change the ratio of organics
to salt in the bubble film and therefore the SSA organic fraction, however the distribution of organic
molecular classes don’t vary. The assumed bubble thickness was not based on any measured bubble
parameters.
The ZSR approximation was used (see Chapter 7.2.5) to compute the ambient nascent SSA HGF,
defined as the average of all measurements less than 45°C, and the heated nascent SSA HGF, defined
as the average HGF for all measurements between 255 - 405 °C. The terms used in the ZSR mixture
were sea salt, non-volatile organics, semi-volatile organics and hydrates. The volume fraction of each
of these components was based on the volatility measurements (Equation 25 and Equation 26), the
hydrate fraction was computed from the difference between the VF at 255 - 405 °C and the OVFsv.
The HGFs for the hydrate component, semi-volatile organic and non-volatile organic component were
assumed to be 1 (Modini, Harris, and Ristovski 2010) and the HGF for the sea salt component was
assumed to be 2.15 at 50 nm, based on measurements of heated laboratory sea salt 2.15 ± 0.06 averaged
across 300 - 350 °C. The sea salt HGF is also consistent with the HGF of pure NaCl (Zieger et al. 2017),
which represents the HGF of the dry salt component of sea salt, i.e. without hydrates. HGFs have not
been kelvin corrected here because all measurements were performed at the same pre-selected particle
size. The hydrate fraction of the salt has been explicitly included here to account for any variation
in the proportion of hydrates between samples. Calculations using the ZSR approach were repeated
using an organic (semi-volatile and non-volatile) HGF of 1.6 as a sensitivity test for the relationship
between OVF and HGF. The HGF of 1.6 was chosen as an upper limit for organic-salt mixtures that
could possibly be present in SSA (Estillore et al. 2016, 2017). Observations have previously suggested
a primary organic HGF of 1.2 (Vaishya et al. 2013).
As a counterpoint to the ZSR assumption, which assumes the organic component is dissolved into the
bulk, the compressed film model was applied to explore the influence of partitioning organics to the
surface on the nascent SSA water uptake. The compressed film model is described in Chapter 7.2.6.
The composition of the SSA organics are unknown, therefore the compressed film model was computed
for organics with molecular areas, A0, ranging from 10 - 200 square angstroms and molecular volumes
Vorg, ranging from 0.2 to 3.6 m3mol-1. The speciation of organics into molecular classes was calculated
from the functional group concentrations as shown in Burrows et al. 2014. HGFs were computed for
three cases, assuming that just the lipids partition to the surface, that the lipids and the polysaccharides
partition to the surface and assuming that all of the organics partition to the surface. The hygroscpicity
of the bulk aerosol i.e. the component not partitioned to the surface, was computed using the ZSR
assumption as outlined in the previous paragraph, with an organic HGF of 1.
The output from the compressed film model, with all of the organics partitioned to the surface, was
66
used to calculate the CCN concentration, and was compared to the CCN concentration computed from
the ambient HGFs, assuming that the droplet surface tension is equal to the surface tension of water.
The critical supersaturation from the compressed film model, with all of the organics partitioned to
the surface, was used to compute the critical diameter using the κ-Köhler equation (Equation 10). The
CCN number concentrations were subsequently estimated assuming a nascent SSA size distribution
with number concentration 100 cm−3, mean diameter of 160 nm and geometric standard deviation of
2.6. CCN concentrations were computed by integrating the size distribution at diameters greater than
the critical diameter, which was computer (as above) for the compressed film model and assumed to
be equal to the preselected particle diameter for the full solubility (surface tension equal to surface
tension of water) case.
Ambient measurements
Ambient size distributions were compiled from SMPS (10 - 300 nm) and OPC (0.5 - 30 m) measure-
ments, and hourly averages were computed from measurements during non-polluted conditions (see
paragraph below). Log-normal modes were fitted according to the method outlined in Modini et al.
(2015). First the SSA distribution was fitted, using non-linear least squares, to diameters greater than
500 nm with the mean diameter constrained to 160 nm ± 30% and the geometric standard deviation
constrained to be between 2 and 3. The fitted log-normal SSA mode was subtracted from the data
and up to 5 log-normal modes were fitted to the residual. The log-normal modes were fitted with a
random selection of initial values for the mean diameter and geometric standard deviation, constrained
to 15-220 nm and less than 2, respectively. Similarly to nascent SSA experiments the most appropriate
fit was determined using the bayesian information criterion. All fitted modes with a mean diameter less
than 65 nm were summed and assigned to the Aitken mode, and all modes with mean diameter greater
than 65 nm (other than the SSA mode) were summed and assigned to the accumulation mode.
Air mass back trajectories were computed using the Hybrid-Single-Particle Lagrangian Integrated
Trajectory (HYSPLIT) model computed using NOAA NCEP global data at 1°resolution. A 27 member
ensemble of perturbed 3-day back trajectories were computed for each hour of the campaign. The air
mass was considered clean, not significantly influenced by terrestrial sources, if the mean number
of hours that the trajectory spent over land from all of the ensemble members was within standard
deviation of 0. Pollution periods were removed from the data using a threshold in the rolling hourly
averaged black carbon concentration of 50 ng/m3 from the Aethalometer. Overall conditions were
considered baseline when the wind direction was between 225°and 135°, the wind speed was greater
than 3 m/s, the hourly averaged black carbon concentrations were less than 50 ng/m3 and there were
no significant terrestrial influences, as defined above. It is worth noting that baseline thresholds will
not provide a perfect partition between clean marine and combustion/terrestrially influenced sources,
for example correlations between black carbon and organics below 50 ng/m3 has been observed during
marine measurements (Shank et al. 2012), and will be used here to remove the most obvious pollution
periods.
All TDMA data were inverted using the TDMAinv algorithm (Gysel, McFiggans, and Coe 2009) and
67
the inverted HGF distribution was broken up into HGF classes as shown in Table 2. The number
contribution and average HGF was computed for each HGF class. The hygroscpicity parameter and
critical diameter for cloud droplet activation were computed for each HGF class (Equation 10), and
the contributions were used to distribute the size distribution between the HGF classes. Note that
because HGFs were only measured at a single size, the aerosol composition was assumed to be uniform
across the distribution. The critical diameter was computed for the supersaturation measured by the
CCNc, 0.5%, and the surface tension was assumed to be equal to the surface tension of water. The
CCN computed from the HGF classes was also compared with CCN calculated using the average HGF
to examine the influence of mixing state, both of these were compared to the measured CCNc.
8.3.2 RV Investigator
Ambient size distributions were compiled from SMPS (10 - 700 nm) and APS (0.5 - 20 μm) measure-
ments, with the APS diameters shifted until agreement between the APS and SMPS number, area
and volume distributions was achieved according to Beddows, Dall’Osto, and Harrison 2010. This
approach was taken because there is no required a-priori knowledge of the particle density and shape
factor and the APS was not calibrated over the period that these measurements were taken. APS
measurements at diameters less than 700 nm were removed from this procedure due to known in-
strumental counting errors (Beddows, Dall’Osto, and Harrison 2010). Log-normal modes were fitted
according to the method outlined in Modini et al. (2015). First the SSA distribution was fitted, using
non-linear least squares, to diameters greater than 500 nm with the mean diameter constrained to 160
nm plus/minus 30% and the geometric standard deviation constrained to be between 2 and 3. During
the CWT voyage there were no APS measurements, therefore the SSA mode was fitted just to SMPS
data. The fitted log-normal SSA mode was subtracted from the data and up to 5 log-normal modes
were fitted to the residual. The log-normal modes were fitted with a random selection of initial values
for the mean diameter and geometric standard deviation, constrained to 15 - 220 nm and less than 2,
respectively. Similarly to nascent SSA experiments the most appropriate fit was determined using the
bayesian information criterion. All fitted modes with a mean diameter less than 65 nm were summed
and assigned to the Aitken mode, and all mode with mean diameter greater than 65 nm (other than
the SSA mode) were summed and assigned to the accumulation mode.
As outlined in Table 6 the VH-TDMA operated in a cycle of ambient and heated measurements at 3
diameters. The non-volatile fraction was determined from the ratio of the number concentration of size
selected particles when heated (120 or 250 °C), to the number concentration of size selected particles
at room temperature for each size. The method used to compute the non-volatile number fraction
(NVfrac) is shown in Equation 30, where N is the number concentration (cm-3) and d indicates that
the measurements are size selected i.e. relevant for some particular diameter. The non-volatile number
fraction is a measure of the proportion of particles that have a non-volatile component, in this context
the non-volatile component is assumed to be sea spray, therefore NVfrac is the proportion of particles
68
at mobility diameter d with an SSA fraction.
NVfrac =Nheated,d
Nunheated,d(30)
The Time of Flight Aerosol Chemical Speciation Monitor (ToF-ACSM) was operated with a vaporiser
temperature of 600 °C, which is too low to flash vaporise sea salt, therefore only a small fraction of sea
salt is detected. Peak Integration by Key Analysis (PIKA) was used to distinguish the NaCl at m/z
58 from organic contributions at similar masses, the resulting concentration was multiplied by 51 to
scale up to total sea salt concentrations (HR-SS) (Ovadnevaite et al. 2012). It should be noted that
this scaling factor was determined from the literature and ideally an instrument specific scaling factor
would be used (Mallet et al. 2016), especially since the ACSM and HR-ToF-AMS are different in other
aspects too. In addition the use of this single scaling factor on aged SSA of unknown composition may
not represent the actual SSA composition. NSS-sulfate was subsequently calculated by subtracting the
sea salt sulfate using the HR-SS and assuming a ratio of 2.71g sulfate per 35.16 g sea salt (Millero et al.
2008). The MSA concentration was computed using the fragmentation method suggested by (Langley
et al. 2010). The organic signal and the NSS-sulfate signal have had the MSA component removed
from their fragmentation calculations.
The data from all three RV Investigator voyages has been filtered for the influence of anthropogenic
pollution by flagging periods when any of the following is true:
• the wind direction, relative to the ship is between 120°and 240°
• the hourly averaged black carbon concentration (rolling average) from the MAAP greater than
30 ng/m3
• the total particle number concentration is greater than 5000 cm-3
• the date is within 5 minutes (before or after) of a pollution event as defined in the previous points
In addition air masses influenced by territorial emissions were defined by periods with Radon concen-
trations (hourly averaged) greater than 150 mBq/m3, a 5 minute window (before or after) was applied
to this procedure.
Remotely sensed Chlorophyll-a data from the Moderate Resolution Imaging Spectoradiometer (MODIS)
instrument onboard the NASA Terra and Aqua satellites were used to provide continuous measurement
of ocean biological activity (Hu, Lee, and Franz 2012). Monthly and 8-day Level 3 data were used
to provide the best spatial coverage. Remotely sensed data was compared with in-situ measurements
made during the CAPRICORN and Ice-edge to Equator voyages.
69
9 Results
Aerosol observations during the SOAP study were well complemented with detailed seawater compo-
sition measurements. The measurement results used to couple seawater composition to nascent SSA
composition are outlined in Chapter 9.1. Measurements of the impact of organic enrichment on SSA
water uptake is also outlined in Chapter 9.1. The SSA organic enrichment and subsequent impact on
water uptake are key outcomes from the SOAP voyage and directly address knowledge gaps identified
in Chapter 7.6.
The Southern Ocean aerosol size distributions, and associated log-normal fitting, are outlined in Chap-
ter 9.2. The log-normal modes were validated using observed aerosol composition, subsequently water
uptake was assigned to each log-normal mode based on the measured HGF distributions. From these
the CCN concentration was estimated and validated with the measured CCN concentrations, provid-
ing confidence in the primary and secondary contributions to CCN (as discussed in Chapter 10) and
therefore addressing the knowledge gaps outlined in Chapter 7.6.
9.1 SOAP results
The ocean water composition during the SOAP voyage is presented (Chapter 9.1.1) to provide context
on the seawater biological activity during the voyage. The nascent SSA physical properties, chemical
composition and water uptake are presented in Chapter 9.1.2, and this section outlines the data that
most directly addresses the knowledge gaps identified in Chapter 7.6. Ambient aerosol measurements
taken during the SOAP voyage are summarised in Chapter 9.1.3, these data provide valuable compar-
ison for the ambient measurements taken over the Southern Ocean in subsequent chapters.
9.1.1 Ocean water
Chlorophyll-a concentrations from water samples used to generate SSA ranged from 0.29 to 1.53 μgL−1
as shown in Figure 15, which are indicative of productive open ocean regions (O’Dowd et al. 2015), in
particular for the Southern Hemisphere. Measured chl-a concentrations are up to an order of magnitude
lower than previous SSA measurements taken in coastal waters (Frossard et al. 2014b; Quinn et al.
2014). Significant correlations were observed between chlorophyll-a and total high molecular weight
proteins and polyunsaturated fatty acids (R2 = 0.51, p-value < 0.001). The saturated faty acid
component was the largest contributor to the total fatty acid concentration, and was made up of
stearic, palmitic, myristic and lauric acid (C18 to C12, all even). Monounsaturated fatty acids were
dominated by oleic acid (C18) and polyunsaturated fatty acids were made up of docosahexaenoic and
eicosapentenoic acid (C22 and C20). Fatty acid concentrations showed significant correlation with the
alkanes (R2 = 0.75, p-value < 0.001), particularly monounsaturated fatty acids. Alkanes displayed
even carbon numbers from 16 to 28, with peak concentrations for octadecane (C18) and eicosane (C20).
It is worth noting that fatty acid and alkane measurements were not done for deep water samples,
surface and mixed layer samples only.
70
Bloom 1 displayed the highest chlorophyll-a concentrations and was dominated by dinoflagellates.
Relatively short lived aliphatic species, such as fatty acids and alkanes were elevated during bloom 1,
as were proteins, which are of intermediate lifetime. High molecular weight reducing sugars were overall
less elevated during bloom 1, and the DOC concentrations were lower than for subsequent blooms (700±100 gL-1). It is difficult to determine a chlorophyll-a peak, given that samples were taken sporadically
and at different locations inside a constantly evolving bloom, however chlorophyll-a appeared to decline
for latter bloom 1 measurements. Elevated surface concentrations were also noticeable during bloom 1,
particularly for the aliphatic species, in particular alkanes with 3.1 times higher average concentration
from surface measurements (∼0.1m) than from shallow mixed layer measurements (∼2m), and fatty
acids with 1.7 (saturated) to 4.2 (monounsaturated) times higher surface concentrations. Elevated
surface concentrations were also observed for Chl-a (1.7 times higher), carbohydrates (1.5 times higher)
and proteins (1.1 times higher). The apparent gradient in organics in the surface seawater is distinct
for bloom 1, and points to a potentially enhanced contribution from surface active species over this
bloom.
Bloom 2 was characterised as a coccolithophore bloom (Law et al. 2017b) and displayed decreasing
chlorophyll-a, fatty acid and high molecular weight proteins. The overall organic concentrations in
bloom 2 were lower than bloom 1, with average chlorophyll-a concentrations of 0.7 ± 0.4 g.L-1. The
number of measurements during bloom 3 were limited, however initially the bloom displayed similar
concentrations to bloom 2 and the final measurements of bloom 3b displayed elevated chlorophyll-a
and fatty acid concentrations.
9.1.2 Nascent SSA
Size distributions
The measured size distributions were broken up into four log-normal modes characterised by geometric
mean diameters ranging from 33 to 320 nm, as seen in Figure 16. Natural SSA size distributions were
slightly shifted towards larger diameters compared to laboratory sea salt measurements, and showed
a significant enhancement in mode 2 to the detriment of modes 1 and 4, Figure 16 and Table 8. The
size distribution of SSA generated from natural seawater samples is more narrow than laboratory sea
salt particle size distributions, which is consistent with the addition of a surfactant material (Fuentes
et al. 2010a; Modini et al. 2013) which allows the saline components of the bubble film to drain more
before bursting, producing an organically enriched particle with a more uniform distribution.
The shape of the nascent SSA size distribution was broadly similar to nascent SSA size distributions
in the literature, but shifted to slightly larger mean diameters. For example (Fuentes et al. 2010a)
fitted log-normal modes with mean mobility diameters of 20, 41, 87 and 250 nm to laboratory sea
salt generated using glass sintered filters, and modes with mean mobility diameters of 14, 48, 124
and 334 nm for plunging water generated sea salt. The use four glass filters with different pore sizes
resulted in a broader distribution than other measurements of nascent SSA using glass filters (Collins
et al. 2014; Fuentes et al. 2010a; Keene et al. 2007; Mallet et al. 2016). SSA produced from sintered
71
Figure 15: Characterisation of biological activity for water samples used to generate SSA. Note that
this is a selected subset of all water parameters. Panels show the concentration of chlorophyll-a (top),
fatty acids (2nd from top), phytoplankton carbon (2nd from bottom) and total high molecular weight
(HMW) reducing sugars and proteins (bottom). Note that total HMW proteins and reducing sugars
include particle and dissolved fractions. Shapes represent the water sample depth class.
72
glass filters does not perfectly represent real world bubble bursting from wave breaking (Collins et al.
2014; Prather et al. 2013). Observations have shown organic enrichment (King et al. 2013) and also
externally mixed organics (Collins et al. 2014) for Aitken and accumulation mode SSA using sintered
glass techniques, with slightly higher organic enrichment than that observed using plunging water or
wave breaking methods. Despite the limitations, the use of sintered glass filters allowed examination
of the enrichment of organics, in particular to identify the components of seawater that contribute to
SSA organic enrichment.
Figure 16: Nascent SSA size distributions from laboratory sea salt measurements (A) and natural
seawater measurements (B). Seawater size distributions are an average of all water samples. Lines
represent the fitted modes (dotted) and total fitted distribution (solid), black dots represent measured
values and grey bars indicate the standard deviation in dN/dlogDp.
73
Table 8: Nascent SSA log-normal parameters
Water sample Parameter Mode 1 Mode 2 Mode 3 Mode 4
Laboratory sea saltNormalised number conc. 0.34 0.12 0.40 0.14
Mean diameter 40 69 114 309
Geometric standard deviation 1.61 1.29 1.63 1.31
Natural seawater (average)Normalised number conc. 0.18 0.38 0.38 0.05
Mean diameter 34 70 120 320
Geometric standard deviation 1.47 1.45 1.58 1.44
SSA composition
Volatility measurements using the VH-TDMA indicated that the SSA volatile organic fraction made
up a relatively consistent proportion of the 50 nm SSA, with a OVFSV of 0.11 ± 0.04 (mean ± sd),
as shown in Fig. 17.The semi-volatile component due to hydrates made up 13 ± 4 %, and the SSA
compositional results are tabulated in Chapter 14. The non-volatile organic fraction, however, made up
a much larger and more variable proportion, with an average OVFNV of 0.39 ± 0.24. The 50nm OVF
was highest during bloom 1 (generally greater than 0.6), which is coincident with seawater samples
enriched in organics, during which time non-volatile organics dominated. A dominant non-volatile
organic SSA fraction has been observed for nascent SSA measurements in the North Pacific and North
Atlantic oceans (less than 15% volatilised at below 230 °C), and in the Great Barrier Reef (Bates et al.
2012; Mallet et al. 2016; Quinn et al. 2014), our observations are broadly consistent with these results.
It should be noted that a non-volatile SSA fraction is not universally observed (Modini, Harris, and
Ristovski 2010; O’Dowd et al. 2004; Ovadnevaite et al. 2011a). Organic mass fractions measured using
FTIR and IBA analysis of filter samples averaged 0.12 ± 0.6, and were elevated during bloom 1 with
surface water samples displaying OMF of approximately 0.2. The values observed here fit within the
broad range of observed OMFs for nascent SSA, for example at Mace Head summertime SSA organic
mass fractions of up to 0.8 have been observed (Facchini et al. 2008; O’Dowd et al. 2004), while
summertime OMFs of 3 - 7% have been observed for the North Atlantic and North Pacific Oceans
(Bates et al. 2012; Quinn et al. 2014).
The measured OVFs were applied to the log-normal distributions shown in Figure 16 assuming that
the first two log-normal modes contained a volatile and a non-volatile organic volume fraction, log-
normal mode 3 contained a volatile organic fraction only and mode 4 was composed of inorganic sea
salt. These compositions were applied based on the mixing state from HGF measurements (see SSA
water uptake below), which indicated an external mixture. Note that the external mixture wasn’t
observed in the volatility measurements, this is likely to be due to the relatively small observed volatile
fraction and the resulting limitation on the sensitivity of volatility measurement. The organic density
74
●● ●
● ●
●
●●
●●
●
●●
●●
●●●●●●
●●
●●
●●
●
●●
●●
●●
●●●●
●●
●●
●●●
● ●●
●
●●
●
●
● ●●
●●
●●●●●
●●●●
●●
● ●
●
●
●
● ●●
●●
●●
Feb 20 Feb 27 Mar 05
Feb 20 Feb 27 Mar 05
Feb 20 Feb 27 Mar 05
Feb 20 Feb 27 Mar 051.051.101.151.201.251.30
0.4
0.8
1.2
0.20.30.40.50.60.7
1.8
1.9
2.0
Date
● Deep
Mixed
Surface
Chl a
VF
OVF
OMF
C
HGF
OGF15
OGF30
OGF50
0.0.0.0.0.0.
●
●●
●●●● ●● ●
●●
●
●●●●●●● ●●●●●●●●●●
●
●●
●●●
●●●●●●●●●●●●●●●
Feb 20 Feb 27 Mar 050.0
0.2
0.4
0.6
Figure 17: Summary of nascent SSA properties. Chl-a time series included for context on phy-
toplankton bloom conditions (top panel). Mass fractions (MF) of Cl- and Ca2+ (multiplied by 20)
relative to total inorganic mass concentrations measured using IBA on filter samples (2nd panel from
top). VH-TDMA OVFSV, OVFNV and FTIR/IBA derived OMF from filter samples (3rd panel from
top). HGF measured using VH-TDMA (2nd panel from bottom), organic growth factor measured
using UFO-TDMA (bottom panel).
75
was assumed to be 1.1 g.cm-3(Keene et al. 2007; Modini, Harris, and Ristovski 2010) and the sea
salt density was assumed to be 2.01 g/cm-3 (Zieger et al. 2017). The organic fraction inferred using
volatility techniques and the fraction measured using FTIR/IBA of the filter samples agree reasonably
well, Figure 18. The organic mass fraction calculated using the aerosol volatility and size distributions
overestimates the organic mass fraction, Figure 18, by a factor of 1.3 on average. There are a number
of sources of uncertainty, including the assumed partitioning to log-normal modes, and the particle
densities. The presence of inorganic species, such as Ca2+, complexed with the organics are included
in organic estimates from volatility, but not from filter analyses.
Significant correlations (p-value < 0.05) of both the semi volatile organic volume fraction (OVFSV)
and the non-volatile organic volume fraction (OVFNV) with seawater high molecular weight proteins
and carbohydrates were observed, in addition the semi volatile OVF correlated with total alkanes
polyunsaturated fatty acids. Correlations with water parameters suggest that the composition of the
volatile and non-volatile OVFs were similar, but the semi-volatile OVF displayed a higher contribution
from aliphatic, lipid like species. The correlation between semi-volatile OVF and seawater alkanes was
significant for all carbon numbers for which there were alkane measurements above the measurement
detection limit, however strongest correlations were observed for the lower carbon numbers, the lowest
carbon number was C16 (R2 of 0.58, p-value < 0.001).
●●
●
●
●
●●
●
●
●
●●●●
●●●●●●●●●●
●●
●
●●●●●●●●●●●
●●●●●●●
●●
●●●●●●●●●●●●●●●●●●●
●●●●●●●●
●●●●●
●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●
●●●●
●●● ●●● ●●●●
●●
●●● ●●●●●●●●●●●●●●●●
●●
●●●●●●
●● ●●●●●●● ●●●●●●●●
●●●
●●●●● ●●●●●●●●●●●●●●●●●●●●●●
●●●●●●●●●●●●
●●●●●●●●●●●●●●● ●●●●●●● ●●●●●●●● ●●●● ●●●● ●●●●●●●●●●●●●●●●●●●●●●
●●
OVFtot slope = 3.4 + 0.6 int = −0.03 + 0.08 R2 = 0.69 P < 0.001
OVFtot slope = 3.4 + 0.6 int = −0.03 + 0.08 R2 = 0.69 P < 0.001
OVFtot slope = 3.4 + 0.6 int = −0.03 + 0.08 R2 = 0.69 P < 0.001
OVFtot slope = 3.4 + 0.6 int = −0.03 + 0.08 R2 = 0.69 P < 0.001
OVFtot slope = 3.4 + 0.6 int = −0.03 + 0.08 R2 = 0.69 P < 0.001
OVFtot slope = 3.4 + 0.6 int = −0.03 + 0.08 R2 = 0.69 P < 0.001
OVFtot slope = 3.4 + 0.6 int = −0.03 + 0.08 R2 = 0.69 P < 0.001
OVFtot slope = 3.4 + 0.6 int = −0.03 + 0.08 R2 = 0.69 P < 0.001
OVFtot slope = 3.4 + 0.6 int = −0.03 + 0.08 R2 = 0.69 P < 0.001
OVFtot slope = 3.4 + 0.6 int = −0.03 + 0.08 R2 = 0.69 P < 0.001
OVFtot slope = 3.4 + 0.6 int = −0.03 + 0.08 R2 = 0.69 P < 0.001
OVFtot slope = 3.4 + 0.6 int = −0.03 + 0.08 R2 = 0.69 P < 0.001
OVFtot slope = 3.4 + 0.6 int = −0.03 + 0.08 R2 = 0.69 P < 0.001
OVFtot slope = 3.4 + 0.6 int = −0.03 + 0.08 R2 = 0.69 P < 0.001
OVFtot slope = 3.4 + 0.6 int = −0.03 + 0.08 R2 = 0.69 P < 0.001
OVFtot slope = 3.4 + 0.6 int = −0.03 + 0.08 R2 = 0.69 P < 0.001
OVFtot slope = 3.4 + 0.6 int = −0.03 + 0.08 R2 = 0.69 P < 0.001
OVFtot slope = 3.4 + 0.6 int = −0.03 + 0.08 R2 = 0.69 P < 0.001
OVFtot slope = 3.4 + 0.6 int = −0.03 + 0.08 R2 = 0.69 P < 0.001
OVFtot slope = 3.4 + 0.6 int = −0.03 + 0.08 R2 = 0.69 P < 0.001
OVFtot slope = 3.4 + 0.6 int = −0.03 + 0.08 R2 = 0.69 P < 0.001
OVFtot slope = 3.4 + 0.6 int = −0.03 + 0.08 R2 = 0.69 P < 0.001
OVFtot slope = 3.4 + 0.6 int = −0.03 + 0.08 R2 = 0.69 P < 0.001
OVFtot slope = 3.4 + 0.6 int = −0.03 + 0.08 R2 = 0.69 P < 0.001
OVFtot slope = 3.4 + 0.6 int = −0.03 + 0.08 R2 = 0.69 P < 0.001
OVFtot slope = 3.4 + 0.6 int = −0.03 + 0.08 R2 = 0.69 P < 0.001
OVFtot slope = 3.4 + 0.6 int = −0.03 + 0.08 R2 = 0.69 P < 0.001
OVFtot slope = 3.4 + 0.6 int = −0.03 + 0.08 R2 = 0.69 P < 0.001
OVFsv slope = 0.2 + 0.1 int = 0.04 + 0.01 R2 = 0.15 P < 0.1
OVFsv slope = 0.2 + 0.1 int = 0.04 + 0.01 R2 = 0.15 P < 0.1
OVFsv slope = 0.2 + 0.1 int = 0.04 + 0.01 R2 = 0.15 P < 0.1
OVFsv slope = 0.2 + 0.1 int = 0.04 + 0.01 R2 = 0.15 P < 0.1
OVFsv slope = 0.2 + 0.1 int = 0.04 + 0.01 R2 = 0.15 P < 0.1
OVFsv slope = 0.2 + 0.1 int = 0.04 + 0.01 R2 = 0.15 P < 0.1
OVFsv slope = 0.2 + 0.1 int = 0.04 + 0.01 R2 = 0.15 P < 0.1
OVFsv slope = 0.2 + 0.1 int = 0.04 + 0.01 R2 = 0.15 P < 0.1
OVFsv slope = 0.2 + 0.1 int = 0.04 + 0.01 R2 = 0.15 P < 0.1
OVFsv slope = 0.2 + 0.1 int = 0.04 + 0.01 R2 = 0.15 P < 0.1
OVFsv slope = 0.2 + 0.1 int = 0.04 + 0.01 R2 = 0.15 P < 0.1
OVFsv slope = 0.2 + 0.1 int = 0.04 + 0.01 R2 = 0.15 P < 0.1
OVFsv slope = 0.2 + 0.1 int = 0.04 + 0.01 R2 = 0.15 P < 0.1
OVFsv slope = 0.2 + 0.1 int = 0.04 + 0.01 R2 = 0.15 P < 0.1
OVFsv slope = 0.2 + 0.1 int = 0.04 + 0.01 R2 = 0.15 P < 0.1
OVFsv slope = 0.2 + 0.1 int = 0.04 + 0.01 R2 = 0.15 P < 0.1
OVFsv slope = 0.2 + 0.1 int = 0.04 + 0.01 R2 = 0.15 P < 0.1
OVFsv slope = 0.2 + 0.1 int = 0.04 + 0.01 R2 = 0.15 P < 0.1
OVFsv slope = 0.2 + 0.1 int = 0.04 + 0.01 R2 = 0.15 P < 0.1
OVFsv slope = 0.2 + 0.1 int = 0.04 + 0.01 R2 = 0.15 P < 0.1
OVFsv slope = 0.2 + 0.1 int = 0.04 + 0.01 R2 = 0.15 P < 0.1
OVFsv slope = 0.2 + 0.1 int = 0.04 + 0.01 R2 = 0.15 P < 0.1
OVFsv slope = 0.2 + 0.1 int = 0.04 + 0.01 R2 = 0.15 P < 0.1
OVFsv slope = 0.2 + 0.1 int = 0.04 + 0.01 R2 = 0.15 P < 0.1
OVFsv slope = 0.2 + 0.1 int = 0.04 + 0.01 R2 = 0.15 P < 0.1
OVFsv slope = 0.2 + 0.1 int = 0.04 + 0.01 R2 = 0.15 P < 0.1
OVFsv slope = 0.2 + 0.1 int = 0.04 + 0.01 R2 = 0.15 P < 0.1
OVFsv slope = 0.2 + 0.1 int = 0.04 + 0.01 R2 = 0.15 P < 0.1
0.0
0.3
0.6
0.9
0.1 0.2
Org
anic
Vol
ume
Frac
tion
(VH
−TD
MA)
●
●
●
●
●
●●
●
●
●
●●●●
●
●●
●●●●●●●●●●
●●●●●●●
●●●●●●●●●●●●●●●●●●
●●●●●●●●●●
●●●●●●●
●●●●●●●●●●●●●●●●●●●
●●●●●●● ●
●●
●●●●● ●● ●● ●●●●●●●● ●●●●
●●●
●● ●●●●●●●●●●●●●●●●●●●●●●●
●●● ●●●●●●●●●●●●●● ●●●●
●●●● ●●●●●●●●●● ●●●●●● ●●●●●●●●●●●●●●●●
●●●●
●●●●●●●● ●● ●●● ●●● ●●● ●●●●●●●●●●●●●●●●●●●●●●
●●
●
OMFtot slope = 1.3 + 0.2 int = −0.02 + 0.03 R2 = 0.68 P < 0.001
OMFtot slope = 1.3 + 0.2 int = −0.02 + 0.03 R2 = 0.68 P < 0.001
OMFtot slope = 1.3 + 0.2 int = −0.02 + 0.03 R2 = 0.68 P < 0.001
OMFtot slope = 1.3 + 0.2 int = −0.02 + 0.03 R2 = 0.68 P < 0.001
OMFtot slope = 1.3 + 0.2 int = −0.02 + 0.03 R2 = 0.68 P < 0.001
OMFtot slope = 1.3 + 0.2 int = −0.02 + 0.03 R2 = 0.68 P < 0.001
OMFtot slope = 1.3 + 0.2 int = −0.02 + 0.03 R2 = 0.68 P < 0.001
OMFtot slope = 1.3 + 0.2 int = −0.02 + 0.03 R2 = 0.68 P < 0.001
OMFtot slope = 1.3 + 0.2 int = −0.02 + 0.03 R2 = 0.68 P < 0.001
OMFtot slope = 1.3 + 0.2 int = −0.02 + 0.03 R2 = 0.68 P < 0.001
OMFtot slope = 1.3 + 0.2 int = −0.02 + 0.03 R2 = 0.68 P < 0.001
OMFtot slope = 1.3 + 0.2 int = −0.02 + 0.03 R2 = 0.68 P < 0.001
OMFtot slope = 1.3 + 0.2 int = −0.02 + 0.03 R2 = 0.68 P < 0.001
OMFtot slope = 1.3 + 0.2 int = −0.02 + 0.03 R2 = 0.68 P < 0.001
OMFtot slope = 1.3 + 0.2 int = −0.02 + 0.03 R2 = 0.68 P < 0.001
OMFtot slope = 1.3 + 0.2 int = −0.02 + 0.03 R2 = 0.68 P < 0.001
OMFtot slope = 1.3 + 0.2 int = −0.02 + 0.03 R2 = 0.68 P < 0.001
OMFtot slope = 1.3 + 0.2 int = −0.02 + 0.03 R2 = 0.68 P < 0.001
OMFtot slope = 1.3 + 0.2 int = −0.02 + 0.03 R2 = 0.68 P < 0.001
OMFtot slope = 1.3 + 0.2 int = −0.02 + 0.03 R2 = 0.68 P < 0.001
OMFtot slope = 1.3 + 0.2 int = −0.02 + 0.03 R2 = 0.68 P < 0.001
OMFtot slope = 1.3 + 0.2 int = −0.02 + 0.03 R2 = 0.68 P < 0.001
OMFtot slope = 1.3 + 0.2 int = −0.02 + 0.03 R2 = 0.68 P < 0.001
OMFtot slope = 1.3 + 0.2 int = −0.02 + 0.03 R2 = 0.68 P < 0.001
OMFtot slope = 1.3 + 0.2 int = −0.02 + 0.03 R2 = 0.68 P < 0.001
OMFtot slope = 1.3 + 0.2 int = −0.02 + 0.03 R2 = 0.68 P < 0.001
OMFtot slope = 1.3 + 0.2 int = −0.02 + 0.03 R2 = 0.68 P < 0.001
OMFtot slope = 1.3 + 0.2 int = −0.02 + 0.03 R2 = 0.68 P < 0.001
OMFsv slope = 0.17 + 0.07 int = 0.02 + 0.01 R2 = 0.27 P < 0.05
OMFsv slope = 0.17 + 0.07 int = 0.02 + 0.01 R2 = 0.27 P < 0.05
OMFsv slope = 0.17 + 0.07 int = 0.02 + 0.01 R2 = 0.27 P < 0.05
OMFsv slope = 0.17 + 0.07 int = 0.02 + 0.01 R2 = 0.27 P < 0.05
OMFsv slope = 0.17 + 0.07 int = 0.02 + 0.01 R2 = 0.27 P < 0.05
OMFsv slope = 0.17 + 0.07 int = 0.02 + 0.01 R2 = 0.27 P < 0.05
OMFsv slope = 0.17 + 0.07 int = 0.02 + 0.01 R2 = 0.27 P < 0.05
OMFsv slope = 0.17 + 0.07 int = 0.02 + 0.01 R2 = 0.27 P < 0.05
OMFsv slope = 0.17 + 0.07 int = 0.02 + 0.01 R2 = 0.27 P < 0.05
OMFsv slope = 0.17 + 0.07 int = 0.02 + 0.01 R2 = 0.27 P < 0.05
OMFsv slope = 0.17 + 0.07 int = 0.02 + 0.01 R2 = 0.27 P < 0.05
OMFsv slope = 0.17 + 0.07 int = 0.02 + 0.01 R2 = 0.27 P < 0.05
OMFsv slope = 0.17 + 0.07 int = 0.02 + 0.01 R2 = 0.27 P < 0.05
OMFsv slope = 0.17 + 0.07 int = 0.02 + 0.01 R2 = 0.27 P < 0.05
OMFsv slope = 0.17 + 0.07 int = 0.02 + 0.01 R2 = 0.27 P < 0.05
OMFsv slope = 0.17 + 0.07 int = 0.02 + 0.01 R2 = 0.27 P < 0.05
OMFsv slope = 0.17 + 0.07 int = 0.02 + 0.01 R2 = 0.27 P < 0.05
OMFsv slope = 0.17 + 0.07 int = 0.02 + 0.01 R2 = 0.27 P < 0.05
OMFsv slope = 0.17 + 0.07 int = 0.02 + 0.01 R2 = 0.27 P < 0.05
OMFsv slope = 0.17 + 0.07 int = 0.02 + 0.01 R2 = 0.27 P < 0.05
OMFsv slope = 0.17 + 0.07 int = 0.02 + 0.01 R2 = 0.27 P < 0.05
OMFsv slope = 0.17 + 0.07 int = 0.02 + 0.01 R2 = 0.27 P < 0.05
OMFsv slope = 0.17 + 0.07 int = 0.02 + 0.01 R2 = 0.27 P < 0.05
OMFsv slope = 0.17 + 0.07 int = 0.02 + 0.01 R2 = 0.27 P < 0.05
OMFsv slope = 0.17 + 0.07 int = 0.02 + 0.01 R2 = 0.27 P < 0.05
OMFsv slope = 0.17 + 0.07 int = 0.02 + 0.01 R2 = 0.27 P < 0.05
OMFsv slope = 0.17 + 0.07 int = 0.02 + 0.01 R2 = 0.27 P < 0.05
OMFsv slope = 0.17 + 0.07 int = 0.02 + 0.01 R2 = 0.27 P < 0.05
0.0
0.1
0.2
0.3
0.4
0.5
0.1 0.2
Org
anic
Mas
s Fr
actio
n (V
H−T
DM
A)
Organic Mass Fraction (Filter)
Figure 18: Comparison of 50 nm organic volume fraction (left) and PM1 organic mass fraction (right)
calculated from volatility measurements (using VH-TDMA) and the OMF measured using FTIR/IBA
on filter samples. Volatile fractions shown in red, total organic fractions shown in green.
The mass fraction of inorganic species in SSA during SOAP was observed to vary from that of salts
in seawater, in particular an enrichment factor of 1.7 ± 0.6 relative to the composition of laboratory
sea water was observed for Ca2+ and 0.4± 0.2 for Mg2+, Figure 19. Enrichment factors observed from
TEM-EDS analysis of laboratory sea salt samples were 0.8 ± 0.3 for Ca2+ and 1.0 ± 0.1 for Mg2+
(mean ± sd), suggesting sea salt fractions similar to seawater. It should be noted that TEM-EDS
76
EF’s were based on a modest number of measurements (25 particles) and the variability in these
measurements was large. Ca2+ inorganic mass fraction and EF were observed to increase with OMF
(R2 = 0.37, P < 0.001), while the Cl- inorganic mass fraction and EF decreased with increasing OMF
(R2 = 0.45, P < 0.001) . The change in inorganic mass fraction as a function of OMF was modest for
Ca2+, however the corresponding change in enrichment factor was large, and conversely the change in
enrichment factor for Cl- was small due to the high mass fraction, as can be seen on the left hand side
of Figure 19. The mass ratio of Cl- to Na+ was 1.6 ± 0.2, which is slightly lower than the seawater
ratio of 1.8 (Seinfeld and Pandis 2006). Cl- depletion is commonly observed for ambient SSA, and
is largely attributed to atmospheric aging processes. Cl- depleted nascent SSA, as observed here,
has also previously been reported to indicate that chloride is fractionated in seawater depending on
the seawater composition or that Cl is evaporated during SSA production (Schwier et al. 2017). Wave
chamber experiments identified an externally mixed C and O containing particle type, which contained
inorganic elements such as S, Na, Mg, Ca and K, but not Cl (Ault et al. 2013b), presence of this particle
type could decrease the overall Cl contribution. Enrichment of Ca2+ is consistent with other nascent
SSA chamber experiments, a proposed explanation for this is the complexing of Ca2+ with carbonate
ions (see Chapter 7.3.3). The presence of carbonate would potentially be detected in the OVFNV from
TDMA measurements and could therefore provide an explanation for the over prediction observed in
Figure 18. Alternatively Ca2+ could be in a complex with organics, Salter et al. 2016 concluded that
if this were the case it would be with a minor amount of organic material.
Alcohol functional groups contributed 77 ± 8 % of the SSA OM, alkanes 10 ± 4 %, amines 10 ± 3 %
and carboxylic acid groups 3 ± 3 % (mean ± sd). The make up of organics across the samples was
relatively constant, as depicted by the ranges shown in Figure 19. The ratio of alkane to hydroxyl
(alcohol) functional groups indicates whether the organic fraction is aliphatic/lipid like (high ratio)
or more oxidised/carbohydrate like (low ratio). The nascent SSA generated during SOAP had alkane
to hydroxyl ratios ranging from 0.06 tot 0.25, with a mean of 0.14, these are exceptionally low values
for non-oligotrophic waters, suggesting that the SSA was heavily enriched in carbohydrates. For some
context Frossard et al. 2014b reported average ratios of 0.34 for non-productive waters to 0.93 for
productive waters. It is also worth noting that the alkane to hydroxyl ratios were lowest during blooms
1 and 3, 0.12 ± 0.04 and 0.11 ± 0.04, respectively and highest outside of phytoplankton blooms 0.2 ±0.05. Alkane to hydroxyl ratios during bloom 3 were strongly influenced by the final sample (Workboat
10) which displayed very low ratio. These results suggest that the SSA from phytoplankton blooms is
enriched in carbohydrate like organics, more so than the less biologically active regions.
Ethanol growth factors measured using the UFO-TDMA for preselected 50 nm diameter SSA were
1.2 ± 0.04 (mean ± sd) and were largely invariable for all of the water samples examined. There
were no significant correlations with the 50 nm ethanol growth factor and any of the water or particle
phase variables measured. The ethanol growth factor of volatilised SSA (for sample U7520) dropped
to within the measurement uncertainty of 1 at 200 °C, suggesting that the component contributing to
ethanol growth was volatile. The component that uptakes ethanol was more constant than the OVFSV
measured using the VH-TDMA, suggesting that it could have been a subset of the total volatile organic
component.
77
Figure 19: Inorganic mass fraction (top left) and enrichment factor (bottom) versus organic mass
fraction (bottom left) measured from IBA and FTIR analysis of filter samples. Enrichment factors
of inorganic species are with respect to laboratory prepared sea water and presented with respect to
Na+. Colour in left hand plots is displayed in bottom right plot. Left hand plots show linear trends
for species with statistically significant (P < 0.05) slope, and water sample depth (shapes). Stars
in bottom right plot represent the mean EF from TEM-EDS measurements of SSA generated from
laboratory seawater, dotted error bars show standard deviation in the mean. Contribution to organic
mass from functional groups measured by FTIR shown top right. Boxes extend from the 25th to the
75th percentile, with the line showing the median, crosses show measurements outside of the 95%
confidence interval in the median.
78
SSA was generated from a number of CTD water samples at two different depths, one in the mixed
layer and one deep water sample as shown in Table 4. Notable differences were that OVFSV and
OVFNV were over 1.4 times greater in the mixed layer than in deep water and the OMF was up to 1.8
times higher. Ca2+ and SO42- were 1.7 and 1.3 times higher in the mixed layer, respectively, Na+ was
slightly lower (a factor of 0.9) and alkanes in the aerosol phase were up to 2.4 times higher. The effect
of depth was only looked at for three CTD samples, therefore the above values should be treated with
caution, however the relative increase in SSA organic fractions, calcium and alkanes is consistent with
higher biological activity towards the ocean surface.
SSA water uptake
The HGFs observed for SSA generated from both laboratory sea salt and natural seawater samples
showed up to 3 externally mixed HGF modes. The first natural seawater SSA HGF mode averaged
1.89± 0.07 and contributed a number fraction of 0.8± 0.12 for 50 nm diameter SSA. The second mode
displayed an average HGF of 2.04±0.09 and contributed a number fraction of 0.2±0.1. The third HGF
mode was sporadically observed during SOAP measurements at 50 and 100 nm diameters (observed
during 4 samples), when present contributed number fraction of 0.01 to 0.06 and displayed an average
HGF of 2.25±0.02. The fraction of the second HGF mode at 50 nm showed significant correlation with
the proportion of log-normal mode 3, suggesting that log-normal mode 3 is made up of particles from
the second HGF mode. Similarly, the first two log-normal modes have similar hygroscopicities and are
related to the first HGF mode. The presence of externally mixed HGFs for the natural and laboratory
seawater samples suggests that the composition and/or morphology is different between log-normal
modes 1-2 and mode 3. An association between the log-normal mode and organic enrichment has
been previously observed for chamber measurements of natural seawater samples, in particular the log-
normal mode with the smallest mean diameter was associated with the greatest SSA organic fraction
(Collins et al. 2013), consistent with the results from the natural seawater samples observed here. The
presence of externally mixed HGFs for laboratory seawater samples suggests that in addition to the
differences in organic enrichment between the log-normal modes there may also be variability in the
inorganic composition, similar to the Ca2+ enrichment observed by Salter et al. (2016).
The shape corrected 50 nm ambient nascent SSA HGF averaged 1.94 ± 0.08 (mean ± sd) across all
samples, with individual samples ranging from 1.79 ± 0.05 to 2.08 ± 0.06 as shown in Figure 17. Heated
50 nm HGFs averaged 2.02 ± 0.05 across all samples, with individual samples ranging from 1.91 ± 0.06
to 2.09 ± 0.06. Particularly interesting is the distribution of HGFs throughout the voyage as shown in
Figure 20, both nascent SSA HGFs and OMFs were highest for bloom 1 on average, and decreased for
subsequent blooms. It should be noted that the relationship between HGF and OVF is similar to that
shown in Figure 20 for OMF. Conventional ZSR mixing would suggest the organic fraction and water
uptake would be inversely proportional to each other, because of the presence of a less hygroscopic
organic component. Overall, no significant relationship is observed between the HGF and the organic
fraction, Figure 21, contrary to that expected from fully soluble mixtures. The HGF decreased with
increasing OVF up to 0.4, when the OVF was between 0.4 and 0.5 the HGF sharply increased and at
OVFs greater than 0.6 the HGF is relatively stable. The SSA HGF after heating is approximately 0.1
79
Figure 20: Ambient HGF (left), heated HGF (middle) and OMF (right) by bloom. HGFs measured
using VH-TDMA, OMF measured from FTIR and IBA analysis of filter samples. Note that NA refers
to measurements taken outside of an identified bloom.
higher than that from the ambient HGF, which is a similar change in HGF as observed for laboratory
sea salt samples, and is likely to be largely due to the evaporation of hydrates.
The ZSR approach has been used to model the HGFs as shown in Figure 22, the details of this approach
are outlined in Chapter 8.3.1. The SSA HGF was computed assuming an organic HGF of 1 and 1.6,
which bounds the HGF range for organic likely to be present in SSA (Estillore et al. 2017; Fuentes
et al. 2011).
The relationship between HGF and OVF/OMF observed here is not consistent with that expected from
the ZSR assumption i.e. full solubility of organic components. Even when a HGF of 1.6 is assumed
for the organic component, the trend in HGF is not consistent between measured and modelled data.
Deviations between the ZSR model and the measured data begin to become pronounced at OVFs
greater than 0.4, at which point the measured HGF increases sharply. A threshold organic fraction,
beyond which the droplet diameter is enhanced, has previously been observed for fatty acids, and
is related to changes in the droplet surface tension (Forestieri et al. 2018; Ruehl and Wilson 2014;
Ruehl, Davies, and Wilson 2016). It should also be noted that the measured HGFs show a large
sample to sample variability. A buffered response of SSA hygroscopicity to OVF has been previously
observed (Ovadnevaite et al. 2011b; Collins et al. 2016; Forestieri et al. 2018) and is thought to
be linked to surface active organics (e.g. fatty acids). A requirement for these organics to buffer
SSA is small molecular volumes (< 10-4 m3/mol. ) and large molecular areas (> 100 Å). Given
the observed combination of a low alkane to hydroxyl ratio and an apparent non-soluble organic
component, lipopolysaccharides (LPS) could present a reasonable candidate for the composition of
the organic component. LPS have previously been identified as an important component in primary
marine aerosol (Cochran et al. 2017; Estillore et al. 2017; Facchini et al. 2008). The water uptake of
the organic component and potential partitioning to the particle surface will be examined further in
Chapter 10.
The deliquescence relative humidity was measured for the Workboat 9 seawater sample at 69% RH,
Figure 23, notably lower than that observed for NaCl/sea salt, ∼73.5 % (Zieger et al. 2017). SSA
generated from Workboat 9 seawater displayed an ambient HGF of 1.84 ± 0.06, a heated HGF of 1.94
80
●
●
●
●
●
●
1.8
1.9
2.0
2.1
2.2
0.0 0.3 0.6 0.9OVF
HG
F
Measurement●
●
Ambient
Heated
Depth● Deep
Mixed
Surface
Figure 21: HGF as a function of organic volume fraction for ambient temperature (dark blue) and
heated to 255 - 400 °C (red) measurements. Linear fit to both the heated and ambient measurements
shown with dashed lines, shaded area represent the 95% confidence interval in the linear fit.
81
● ●●
● ●
●
●
●●
●
●
●
●
●●
●●●●●●●●●●●●●
●●●●●●●●●●●●●●●●●●
●●●●
●●●●●●●●● ●●●●●●
●●●● ●
●●
●
●●●●●●●●●●●●●●●●●●●●
●●
●●
●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●
●●●●●●●●●●
●●●
●●●●●●●●●●●●●
●●●●●●●
●●
●●●●●●●
●●
●●
●●
●●
●
●●
●●●●●●●●●●●●●●●●●●●●●●●●●
1.4
1.6
1.8
2.0
0.0 0.3 0.6 0.9OVF
HG
F am
bien
t●
●●
●
●
●●
●●
● ●●●●●●●
●
●● ●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●
●●
●
●●●●●●●●●●●●
●●●●
●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●
●●●●●
●●●●●
●●●
●●●
●●
●
●●
●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●
1.50
1.75
2.00
2.25
0.0 0.3 0.6 0.9OVF
HG
F hea
ted
Figure 22: The measured HGF at ambient temperature (left) and heated to 255 - 400 °C (right) as
a function of OVF. HGF modelled using ZSR assumption shown assuming an organic HGF of 1 (blue)
and an organic HGF of 1.6 (green) shown alongside measured HGFs (ambient in dark blue, heated in
red). Linear fit to the measured and modelled data indicated by dashed line, shading represents the
95% confident interval in the linear fit.
± 0.06, an organic mass fraction of approximately 3% and an organic volume fraction of approximately
21%. The alkane to hydroxyl ratio was 0.14. The DRH observed here is consistent with observed organic
sea salt mixtures in the literature, for example a 2:1 mass ratio mixture of NaCl to glucose resulted in
a 100nm DRH of 69.2 ± 1.5 %, the mixture had a hygroscopcity (κ) of 0.8 (Estillore et al. 2017).
9.1.3 Ambient marine aerosol
Aerosol number concentrations under baseline conditions throughout the SOAP voyage averaged 580
± 340 cm-3 (mean ± sd), similar to that reported for other marine locations. The highest average
number concentrations were observed outside phytoplankton blooms, Table 9, this may have been
driven by a small number of events in particular between B2 and B3a (Figure 24). In-bloom number
concentrations were similar across B1 and B3, while B2 displayed elevated number concentrations. B1
displayed an elevated fraction of particles activated to CCN, which was relatively constant across the
other blooms and out of bloom conditions. This will be examined with regard to the size distribution
measurements. Baseline conditions were when the wind direction was between 225°and 135°, relative
to the ship, the wind speed was above 3 m/s, the hourly averaged black carbon concentration was less
than 50 ng/m3 and the number of hours the air mass spent over land (from HYSPLIT) was within
uncertainty of 0. See Chapter 8.3.1. The black carbon concentration and number of hours over land
for the SOAP voyage are shown in Figure 24.
Lognormal distributions were fitted to the number size distributions and divided up into Aitken,
82
Figure 23: Deliquescence of SSA generated from SOAP sea water sample (Workboat 9) shown in
yellow to blue colour scale. Laboratory sea salt deliquesce curve shown with red circles.
Table 9: Average baseline aerosol number concentration (CN10) and CCN concentration by bloom
for the SOAP voyage.
Bloom CN10 [cm-3] CCN concentration [cm-3]
B1 444± 240 168± 55
B2 675± 210 187± 47
B3a 470± 170 140± 99
B3b 409± 130 145± 49
None 809± 450 259± 140
83
300
1000
3000
10000
Feb 13 Feb 20 Feb 27 Mar 05
CN
[cm−3
]
30
100300
10003000
Feb 13 Feb 20 Feb 27 Mar 05
CC
N [c
m−3
]
0.00
0.25
0.50
0.75
1.00
Feb 13 Feb 20 Feb 27 Mar 05CC
N a
ctiv
atio
n ra
tio
1e−01
1e+01
1e+03
Feb 13 Feb 20 Feb 27 Mar 05
BC
[ng.
m−3
]
0
20
40
60
Feb 13 Feb 20 Feb 27 Mar 05
Hrs
ove
r la
nd
0
10
20
30
Feb 13 Feb 20 Feb 27 Mar 05Win
d sp
eed
[m/s
]
Figure 24: Time-series of aerosol properties measured during SOAP voyage. Particle number con-
centration (CN10) (top panel), CCN number concentration (2nd panel from top), CCN activation ratio
(CCN/CN) (3rd panel from top), black carbon mass concentration (2nd panel from bottom) and num-
ber of hours the air mass spent over land in the preceding 72 hours (bottom panel). The red line
shows that BC threshold of 50 ng/m3 applied here. Shading on bottom panel indicates the standard
deviation in the number of hours over land based on the results from the 27 ensemble back trajectories.
The terrestrial influence baseline has not been applied to any of the panels, the BC baseline threshold
has not been applied to the BC panel (only wd and ws thresholds) and no thresholds have been applied
to the bottom panel. Source: Reproduced from Law et al. (2017).
84
accumulation and SSA mode distribution as described Chapter 8.3.1. Aitken mode particles dominated
the number concentration, making up 68% to 89% of the total number concentration, while the SSA
mode had a small average contribution to particle number concentrations, approximately 5 - 6% for
all blooms. The relatively high number fraction of particles in the accumulation mode during bloom
1 is likely to be the reason for the elevated CCN activation ratio. The contribution of the SSA mode
observed here is consistent with previous results using this method, which show contributions of less
than 15 % from the SSA mode on average (Modini et al. 2015; Quinn et al. 2017).
Figure 25: Aerosol number size distributions from SOAP voyage by bloom. Measured data are shown
by grey points, lines show the fitted Aitken mode (green), accumulation mode (red) and SSA mode
(blue) log-normal distributions. Sum of the log-normal distributions is shown in black. Concentration
scale is on a log scale (bottom row) to show the SSA mode, linear concentration scale shown on top
row.
The ambient marine aerosol volatile component was measured on 15 occasions through the SOAP
voyage, VFR and the number fraction of the non-volatile component (NVfrac) were computed by
comparing the measurements at less than 50°C with measurements at temperatures greater than 210
°C. The NVfrac is the ratio of the heated to ambient number concentration, and it is assumed that in the
remote marine environment this variables reveals the number fraction of particle containing SSA. The
average NVfrac was 0.16 ± 0.11, and was largely independent of the preselected size, Figure 26. The
NVfrac comprises a slightly larger fraction than the number fraction of the SSA mode at these diameters,
although largely within uncertainty. This discrepancy, along with the low VFRs suggests that the aged
or cloud processed SSA could make up an important contribution or that the SSA log-normal mode
underestimates the number fraction of SSA at these sizes. The SSA fraction, observed from the
volatility measurements, also contains a large volatile volume fraction, generally greater than 50%,
which suggests that the SSA is internally mixed with a volatile, possibly secondary, component and
85
points to atmospheric ageing or cloud processing. The organic mass fraction averaged 0.5 ± 0.2 from 12
ambient filter samples collected throughout the SOAP voyage, this organic fraction was characterised
by a large hydroxyl functional group fraction, a smaller but consistent contribution from alkanes and
carboxylic acids. The organic functional groups observed here are consistent with observations in the
marine environment, which have been associated marine organics with marine saccharides from SSA
(Frossard et al. 2014b; Russell, Bahadur, and Ziemann 2011), this includes a relatively low alkane to
hydroxyl ratio, averaging 0.22. Supporting this are the inorganic fractions which were characterised
by large contributions from Na, Cl and SO4. There was an enhanced contribution from carboxylic
acids in the ambient measurements compared to the nascent SSA observations, suggesting that it
is secondary in origin, consistent with observations elsewhere (Frossard et al. 2014b). A number of
samples contained low Na fractions, particularly during B1 during which time back trajectories indicate
that the air mass could have been influenced by terrestrial sources, this is supported by intermittent
observation of Si and Al in the samples over this time.
Average HGF distributions for the baseline ambient marine periods showed distinct peaks in the mod-
erately hygroscopic mode and another in the non-hygroscopic mode. The number contribution from the
non-hygroscopic mode didn’t correlate with the concentration of black carbon or the atmospheric CO2
concentration, it did however correlate with the number of hours the air mass spent over land estimated
using the HYSPLIT model. In order to further minimise the influence from terrestrial sources, above
that already outlined in Chapter 8.3.1, the continental baseline was adjusted to remove measurements
that had spent more than 6 hours over land in the past 168 hours. The adjusted continental thresh-
old was reached by trial and error, in an attempt to reduce the influence of terrestrial air while not
removing all of the ambient HGF measurements. The tightening of the terrestrial threshold had little
influence on the overall number concentrations, all of which remained within the uncertainty presented
in Table 9, however it significantly lowered the proportion of particles in the non-hygroscopic HGF
mode. Figure 27 was produced using the data resulting from the application of the more stringent
terrestrial threshold, which resulted in virtually no data for B3a. It is worth noting that data cover-
age is an issue for the SOAP ambient HGF measurements because there were large periods when the
instrument was sampling nascent SSA from the chamber or air masses were impacted from terrestrial
emissions or ship pollution.
Baseline marine aerosol number concentrations were dominated by moderately hygroscopic particles,
with a number fraction of 0.8 ± 0.2, consistent with secondary nss-sulfates or organically enriched
SSA, in addition there was a non-hygroscopic fraction was present simultaneously to the moderately
hygroscopic mode. The non-hygroscopic mode was strongest during B1 and B2, average contributions
of 0.08 and 0.32 respectively, and a continental source for these particle cannot be ruled out given the
relationship between terrestrial air masses and the proportion of non-hygroscopic particles described
in the previous paragraph. Further evidence to support the moderately hygroscopic mode being made
up of nss-sulfates is the low number and volume fractions of refractory particles, and an observed DRH
of 76 to 79% observed for samples during B2 and B3a (Cravigan et al. 2013). The B1 nss-sulfate mode
displayed a slightly lower average HGF, 1.56 during B1 compared to 1.66 during non-bloom periods, this
relatively small decrease could be due to the enhanced precursor emissions observed during B1.
86
Figure 26: Ambient aerosol composition during SOAP voyage. Wind speed (top panel) provided as
reference. Volatile fraction remaining (VFR) (dark green) and the number fraction of size selected 50
or 100 nm particles with a non-volatile component (NVfrac) (dark red) are shown in the 2nd panel
from top. The OMF from FTIR and IBA analysis of filter samples is shown in third panel from top.
The mass fraction of inorganic species (MFIO) from IBA is shown in 2nd panel from bottom, and the
mass fraction of organic species (MForg) from FTIR is shown in bottom panel.
87
Figure 27: The HGF distribution averaged for each bloom (left), and box-lots of the average HGFs for
each bloom (right). All data measured with aerosol samples conditioned to 90% at room temperature.
Shaded region on the HGF distributions (left) show the standard deviation for each HGF bin. The
vertical black lines show the boundary for the HGF classes.
88
0
100
200
300
400
500
0 100 200 300 400 500
Measured CCN [cm−3 ]
Com
pute
d C
CN
[cm
−3 ]
BloomB1
B2
B3a
B3b
None
Figure 28: Measured baseline CCN concentrations at 0.5% SS compared to CCN concentrations
computed from average HGF. Baseline conditions are as defined in Chapter 8.3
The average HGFs were used to compute the critical diameter for cloud droplet activation at 0.5% SS,
hourly averaged size distributions were then integrated at diameters greater than Dcrit to compute the
CCN concentration, as shown in Figure 28. The computed and measured CCN concentrations agree
reasonably well, generally within a factor of two, which is not uncommon for such CCN closure studies
(Ovadnevaite et al. 2017). No significant biases are observed for any of the bloom periods, which
suggests that the water uptake is reasonably consistent between the sub and supersaturated regimes.
Discrepancies between computed and observed CCN are likely to be driven by the propagation of
uncertainties and errors from hourly averaging of size distributions, for example. The influence of
aerosol mixing state does not appear to be strong for these measurements, which is consistent with the
low SSA number concentrations and high contribution from nss-sulfates (Figure 27).
9.2 RV Investigator results
The ocean water chlorophyll-a concentration and key meteorological parameters during the three South-
ern Ocean voyages are presented (Chapter 9.2.1) to provide context on the observed seasonality and
zonal variability. Lognormal fits to the ambient aerosol size distribution (Chapter 9.2.2) are validated
with observed composition (Chapter 9.2.3) and water uptake data to estimate the CCN concentration
(Chapter 9.2.4). Estimated CCN concentrations are validated against the observed CCN concentration
to provide confidence in the contributions to CCN (discussed in Chapter 10).
89
Table 10: Satellite retrieved chlorophyll-a concentration averaged by voyage (representing the season)
and by latitude range.
Latitude range(°S)
CWT(January/February)
CAPRICORN(April/May)
Ice-Edge to Equator(May/June)
40 to 50 0.6± 0.2 0.4± 0.3 0.4± 0.2
>50 0.21± 0.06 0.10± 0.03 (*) 0.3± 0.1
* Chlorophyll-a concentrations retrieved from satellite at high Southern Ocean latitudes during the CAPRICORN
voyage are likely be underestimated (see text).
9.2.1 Marine biological activity and meteorology
The three Southern Ocean voyages reported here span the summer, autumn and early winter. The
seasonality in ocean biological productivity, is measured using chlorophyll-a. Data retrieved from the
MODIS instrument onboard NASA’s Aqua satellite are the primary source for chlorophyll-a data,
as shown in Figure 29, however limited in-situ measurements also provide an important compari-
son. The satellite retrieved chlorophyll-a data are the monthly average values for the closest grid cell
(4km resolution) to the ships location, monthly averages were used because higher time resolution
(e.g. 8-days) resulted increased data loss. During the Ice-Edge to Equator voyage (May/June) the
satellite retrieved and in-situ chlorophyll-a concentrations agree reasonably well, however during the
CAPRICORN voyage (March/April) in-situ chlorophyll-a concentrations are significantly higher than
the satellite retrieved data, this is likely to be an artefact from taking the monthly averages. In-situ
chlorophyll-a measurements were taken as the CAPRICORN voyage transected the oceanic subantarc-
tic front and oceanic eddies, these features are strongly associated with steep gradients in biological
productivity and are constantly moving and evolving, therefore over a month there could be significant
variability in chlorophyll-a concentrations for a given grid box.
Chlorophyll-a data were averaged by latitude, with the boundary between high and lower latitude
Southern Ocean being placed at 50 °S, which approximates the location of the sub-antarctic front. A
decrease in chlorophyll-a concentration at higher Southern Ocean latitudes was observed most promi-
nently during summertime, as shown in Table 10. In the mid-latitudes chlorophyll-a concentrations
were elevated during the summertime, however at higher latitudes photosynthetic activity slightly
increased as winter approached. North of the sub-antarctic front seasonality in chlorophyll-a was
observed, whereas south of the front the seasonality was weaker.
Wind speeds were stable across all seasons, and at all latitudes, voyage average wind speeds ranged
from 10 to 12 ± 4 m/s as shown in Figure 29. The wind speed data display an abundance of periods
when the wind driven production of SSA is expected to be strong across all seasons. As a guide wind
speeds of 4 m/s is associated with the onset of bubble formation (Gantt et al. 2012a), and whitecap
90
Figure 29: Chlorophyll-a (top panel) and meteorological parameters observed during the three South-
ern Ocean voyages , as a function of time of year. All meteorological data, wind speed (2nd panel
from top), air temperature (2nd panel form bottom) and atmospheric pressure (bottom panel) were
all measured onboard the RV Investigator. All data are coloured by latitude. Chlorophyll-a data
are the 4-km monthly averaged values from the MODIS instrument on the Aqua satellite (coloured
points) and in-situ data (black points). Grey lines show voyage average chlorophyll-a concentration.
Note that the Cold Water Trial voyage was in January/February 2015, while the CAPRICORN voyage
(March/April) and Ice-Edge to Equator voyages (May/June) were in 2016.
91
Table 11: Voyage average CN and CCN number concentrations for baseline conditions.
Voyage CN (1/cm3) CCN (1/cm3)
CWT (January/February) 514± 220 193± 66
CAPRICORN (March/April) 292± 169 89± 57
Ice-Edge to Equator (May/June) 157± 98 59± 26
coverage and SSA production has been observed to increase rapidly at wind speeds greater than 10 m/s
(Grythe et al. 2014). Air temperatures are relatively stable across all seasons in the lower latitudes,
however a sharp decrease in air temperature is observed at latitudes higher than 60 °S, which is the
location of the atmospheric polar front, and is likely to be associated with the transition from oceanic
air masses to Antarctic air masses. The meteorology of the Southern Ocean is dominated by the regular
passage of frontal systems, which were identified as sudden dips in atmospheric pressure. The passage
of fronts enhance mixing between the MBL and the free troposphere and have been associated with
injections of newly formed nss-sulfate particles into the MBL (Clarke et al. 2013; Gras 2009) .
9.2.2 Aerosol concentration and size distributions
Clean marine aerosol number concentrations displayed a seasonality, and ranged from an average of
approximately 500 cm-3 in summer to approximately 160 cm-3 in the winter, CCN concentrations
displayed a similar seasonal trend as shown in Figure 30 and Table 11. The CN and CCN number
concentrations observed here, and their seasonality, are similar to that observed from the 25-years of
data at Cape Grim (Gras 1990; Gras and Keywood 2017). The average CN and CCN concentrations
for high latitude observations are slightly lower than measurements in the mid-latitudes, for example
during the CWT voyage the average CN decreased from 550 ± 210 in the mid-latitudes to 505 ± 220
(mean ± sd) in the high latitude Southern Ocean, and CCN decreased from 230 ± 38 to 180 ± 69.
The decrease in concentrations was proportionally large for CCN than for CN.
The activation diameter is relatively variable throughout all three voyages, however a few features
are worth noting. CCN activation ratios from the summertime CWT voyage are persistently quite
high across all latitudinal zones. During the mid-season to winter-time voyages, the CCN activation
ratio increased at higher Southern Ocean latitudes, this was associated with an increase in wind speed
for the CAPRICORN voyage. The CCN activation ratio is largely governed by the shape of the size
distribution, in particular the relative contribution of the Aitken and accumulation modes.
An Aitken, accumulation and SSA mode were fitted to the measured size distributions as described
in Chapter 7.1, following the method used by Modini et al. (2015) and subsequently by Quinn et
al. (2017). The baseline aerosol size distributions averaged for each voyage are shown in Figure 31.
During periods of high biological activity the contribution from the Aitken mode was enhanced, average
92
Figure 30: Baseline aerosol CN concentration (top), CCN concentration (middle panel) and CCN
activation ratio (bottom panel) for the CWT, CAPRICORN and Ice-Edge to Equator voyages, coloured
by latitude. Black lines indicate the voyage average CN and CCN concentrations. All combustion,
terrestrial and maintenance influenced periods have been removed as described in Chapter 8.3.
93
contribution of 0.49 during the CWT compared to 0.25 during the Ice-Edge to Equator voyage. The
Aitken mode has been strongly associated with the nss-sulfates over the Southern Ocean (Gras 2009;
Gras et al. 2009), and has also been associated with entrainment from the free troposphere (Clarke
et al. 2013). The contribution from the SSA mode mirrors that of the Aitken mode, with a minimum
contribution of 0.16 during the summertime, CWT voyage, and the maximum contribution of 0.45
during the mid-season CAPRICORN voyage. The contribution from the SSA mode is larger than
previously reported, generally less than 0.15 (Modini et al. 2015; Quinn et al. 2017), the high wind
conditions observed consistently throughout all three voyages no-doubt contributed to an elevated
SSA mode. The SSA number concentrations averaged 107 cm-3 during the CAPRICORN voyage, 49
cm-3 during the CWT voyage and 42 cm-3 during the Ice-Edge to Equator voyage. To provide some
context for high wind SSA production, average SSA mode concentration of 71 cm-3 during a period
with average wind speeds of 16 m/s was observed during a voyage on the North Pacific (Modini et al.
2015), and SSA number concentrations of approximately 100 cm-3 were observed at Cape Grim and
associated with surface wind speed of over 15 m/s (Cravigan et al. 2015).
Figure 31: Baseline aerosol number size distributions averaged for the CWT (left), CAPRICORN
(middle) and Ice-Edge to Equator (right) voyages. Measured size distribution data are indicated with
grey points, lines represent the fitted Aitken mode (green), accumulation mode (red) and SSA mode
(blue) log-normal distributions. The sum of the log-normal modes is shown with a black line. All
exhaust, terrestrial and maintenance periods have been removed.
Lognormal fits were also made to measured size distribution data binned by wind speed, with ranges
of 0-10 m/s, 10-20 m/s and greater than 20 m/s as shown in Figure 32. At low wind speeds the SSA
mode number concentration was relatively low, ranging from 6-16 cm-3, which is consistent with that
observed by Modini et al. (2015) and Quinn et al. (2017) for low wind speed conditions. As might be
expected, the SSA mode contribution was much higher during high wind speed conditions, up to 178
94
cm-3 for wind speeds greater than 20 m/s during the CAPRICORN voyage, this increase was largely to
the detriment of the accumulation mode. High wind speeds also resulted in an increase in the Aitken
mode concentration, which is likely due to the concurrent low pressure systems and resulting enhanced
mixing between the free troposphere and marine boundary layer.
Figure 32: Baseline aerosol number size distributions for the CAPRICORN (left) and Ice-Edge to
Equator (right) voyages, averaged by wind speed, 0-10 m/s (top), 10-20 m/s (middle) and > 20 m/
s (bottom). Grey points represent measured data and lines represent log-normal fits. All exhaust,
terrestrial and maintenance periods have been removed.
The SSA production was examined by looking at the influence of wind speed, sea temperature, seawater
salinity and satellite retrieved ocean temperature as shown in Figure 33. Although these data are not
required for estimating CCN contributions they are briefly presented here to highlight the value of this
Southern Ocean data set. Broadly the SSA concentrations are consistent with the large number of SSA
flux parameterisations in the literature (Grythe et al. 2014; Lewis and Schwartz 2004; Ovadnevaite
et al. 2014). In addition to wind speed some SSA parameterisations include the influence of seawater
temperature, while the influence of chlorophyll-a and salinity on the flux of SSA is less commonly
95
included it is of interest. All three variables show potential skill in parametrising SSA for these
measured data. Enhanced primary marine aerosol production in the presence of marine surfactants
has been reported (Long et al. 2014; Prather et al. 2013), but not yet widely adopted. Modelling is
required to elucidate the contribution of transported SSA as well as the impact of SSA scavenging and
therefore constrain the observed SSA fluxes.
Figure 33: SSA log-normal mode number concentration as a function of wind speed coloured by
voyage (top right), sea temperature (top right), satellite retrieved chlorophyll-a concentration (bottom
left) and seawater salinity (bottom right).
9.2.3 Aerosol composition
During the CAPRICORN and Ice-Edge to Equator voyages the VH-TDMA was operated with a
heater temperature of 250°C, the rationale being that any remaining aerosol in the baseline marine
environment is SSA. The VH-TDMA pre-selected diameters were binned into those covering the Aitken
mode (< 65 nm), those covering the Hoppel minimum (70 - 110 nm) and those in the accumulation
mode (> 110 nm). Note that for the majority of the time the pre-selected diameter cycled through
40, 100 and 150 nm. The VH-TDMA was used to compute two volatility based variables, the number
fraction of non-volatile particles (NVfrac), as shown in Figure 34 and the volume fraction remaining
(VFR) as shown in Figure 35. The NVfrac is the proportion of particles that have some refractory
component, and the VFR is a measure of the proportion of this refractory component. NVfrac is a
change in concentration and measures the external particle mixing state, VFR is a change in volume
and measures the internal particle mixing state.
The NVfrac was particularly dependent on the pre-selected particle diameter, the accumulation mode
was largely made up of particles with a non-volatile SSA fraction, whereas only a small proportion of
Aitken mode particles contained a non-volatile SSA component. The NVfrac was larger during the less
biologically active Ice-Edge to Equator voyage, suggesting an increased contribution from SSA, which
96
Figure 34: Number fraction of non-volatile particles, at heater temperature of 250°C, under baseline
conditions binned by latitude. Observations were taken using the VH-TDMA during the CAPRICORN
voyage (March/April) (yellow) and the Ice-Edge to Equator voyage (May/June) (blue). Pre-Selected
diameters have been broken up into Aitken mode (top panel, < 65 nm), accumulation mode (middle
panel, 70 - 100 nm) and accumulation mode (bottom panel, > 110 nm). Exhaust, terrestrial and
maintenance influenced periods have been removed.
97
was likely driven by decreased secondary marine aerosol production. The Aitken mode NVfrac increased
at higher latitudes, particularly at latitudes greater than 55 °S during the Ice-Edge to Equator voyage.
Large high latitude NVfrac during the Ice-Edge to Equator voyage corresponded wth lower particle
number concentrations, which highlights the low wintertime secondary aerosol production in the high
latitude Southern Ocean. The chlorophyll-a concentration was relatively invariable with latitude during
the Ice-Edge to Equator voyage, which might suggests that enhanced solar radiation, and therefore
photochemistry, at lower latitudes is an important driver of nss-sulfate formation.
The VFR during periods of low biological activity was high, 0.84 ± 0.24, and consistent with the NVfrac
observed for nascent SSA at 250 °C (Mallet et al. 2016; Modini, Harris, and Ristovski 2010; Rasmussen
et al. 2017). The VFR was slightly lower during the higher biologically active CAPRICORN voyage,
0.78 ± 0.19. Lower VFRs during CAPRICORN was either a result of an enhanced volatile primary
organic component or greater condensation of gas phase products. VFRs in the accumulation mode
are slightly lower than that observed for the Aitken mode, by approximately 5 - 10 %, which suggests
that there may be a small contribution to the SSA from a secondary component from cloud processing.
The volume fraction of the semi-volatile species is lower at higher latitudes, consistent with an overall
reduced concentration of chlorophyll-a concentrations which could have decreased the contribution
from a volatile primary organic component and/or gaseous VOC emissions available to condense onto
SSA.
Overall the SSA number concentration from the volatility measurements compared reasonably well
with the number concentration of SSA from the log-normal mode fitting, as shown in Figure 36. The
number concentration of the SSA log-normal mode at 150 nm displayed significant correlation with
moderate R2 (R2 = 0.5) with the non-volatile concentration for 150 nm particles. In the Aitken mode,
SSA number concentrations were much lower, and the correlation between the volatility and log-normal
mode SSA was weaker. The non-volatile number concentration was, on average, greater than the SSA
concentration from log-normal mode fitting, particularly during the CAPRICORN voyage. The SSA
concentration estimated using volatility includes internally mixed particles with a small non-volatile
fraction i.e. low VFR, likely due to atmospheric ageing of SSA or cloud processing. The processed,
internally mixed, mode may not be entirely captured by the log-normal mode fitting procedure, as
the constraints placed on the SSA log-normal mode are based on nascent SSA (Modini et al. 2015).
The volume of SSA was calculated from the volatility measurements, by explicitly using the VFR
of the internally mixed component, and for the SSA derived from log-normal mode fitting. The
correlation between SSA volume from the two methods (R2 = 0.66) is higher than that for the number
concentration, particularly during the CAPRICORN voyage, and is likely due to explicitly accounting
for the SSA volume fraction of the internally mixed component. The log-normal mode SSA number
concentration also displayed statistically significant correlation with the mass concentration of Cl (R2
= 0.47) and sea salt (R2 = 0.32) measured using the ACSM, providing further confidence in the
log-normal mode fits. The SSA concentration measured using the volatility have stronger agreement
with the SSA mode concentrations than the mass concentrations measured using the ACSM, which is
understandable because the ACSM measures only a single component of SSA and may be biased by a
low number concentration of large particles.
98
Figure 35: Particle volume fraction remaining (VFR) after heating particles to 250 °C binned by
latitude. Observations were taken using the VH-TDMA during the CAPRICORN voyage (March/
April) (yellow) and during the Ice-Edge to Equator voyage (May/June) (blue). Pre-selected diameters
have been broken up into Aitken mode (top panel, < 65 nm), Hoppel minimum (middle panel, 70 -
110 nm), and accumulation mode (bottom panel, > 120 nm). Measurements have been filtered to only
include baseline periods.
99
Figure 36: Comparison of SSA log-normal mode with volatility (top) and composition (bottom)
measurements. The 150 nm diameter size selected non-volatile particle number concentration compared
with the SSA mode particle number concentration at 150 nm (top-left panel). Similarly the pre-selected
non-volatile volume compared with the SSA mode volume at 150 nm (top-right panel). The total SSA
number concentration is plotted against the PM1 Cl and sea salt (SS) mass concentration (bottom).
The non-volatile number fraction and VFR were determined using the VH-TDMA, concentration of
Cl and SS were computed using the ACSM, SS concentrations are based on the high resolution fit for
NaCl as described in Chapter 8.3.
100
Figure 37: PM1 mass concentrations of sulfate (top), chloride (middle) and organics (bottom) split
into 5°latitude bins. Data were measured during the CWT voyage (red) and the Ice-Edge to Equator
voyage (blue) using an ACSM. Grey lines show the detection limit for each species.
101
The non-refractory sulfate PM1 mass fractions were elevated during the CWT trial voyage relative
to the wintertime Ice-Edge to Equator voyage as shown in Figure 37. The seasonal cycle in nss
sulfates over the Southern Ocean has been well documented and therefore the difference between
summertime and late autumn sulfate concentrations could reasonably be expected. During the Ice-Edge
to Equator voyage the sulfate concentrations are lower at higher Southern Ocean latitudes (>50 °S),
which is consistent with lower biological activity and/or lower gaseous precursor emissions due to lower
seawater temperatures. The concentrations of organics was largely below the ACSM detection limit,
apart from two periods during the Ice-Edge to Equator voyage, these periods displayed moderately
high black carbon (approximately 10 - 30 ng/m3) and could therefore be due to transported and
heavily diluted anthropogenic emissions. It is worth noting that concentrations of MSA and ammonia
were almost entirely below the ACSM detection limit and are therefore not shown here. Chloride
concentrations are reasonably consistent, zonally as well as seasonally, with the notable exception of
low latitude measurements during the CWT voyage. Lower concentrations of chloride was observed in
the most Northerly latitude bin during CWT, which suggests that there could have been low primary
marine aerosol production during this time, however the averages are based on a smaller subset of
measurements and therefore should be treated with caution. The concentration of chloride decreased
across the polar front (approximately 63°S), which is consistent with a change in air mass from one
dominated by the Southern Ocean to an air mass dominated by Antarctic and sea ice marginal zone
trajectories.
9.2.4 Southern Ocean aerosol water uptake
The VH-TDMA measured aerosol water uptake at 90% RH during all three Southern Ocean voyages,
and the resulting HGF distributions are shown in Figure 38. In the Aitken mode, 40 nm preselected
particle diameter, the HGF mode at 1.5 dominated the number concentration during the CWT and
CAPRICORN voyages, this is consistent with the HGFs of a modified nss-sulfate mode. The Aitken
mode volatility showed a dominant semi-volatile component, which is also consistent with the nss-
sulfates being the major contributor. During the Ice-Edge to Equator voyage the HGF distributions
displayed a broad mode at HGFs greater than 1.75, and a small mode at approximately 1.6, consistent
with SSA externally mixed with a small nss-sulfate fraction. The nss-sulfate mode HGFs were slightly
elevated during the Ice-Edge to Equator voyage, which could suggest the presence of some organics
during summertime and early autumn voyages. An externally mixed SSA fraction is obvious for the
CAPRICORN voyage, however during the CWT the SSA HGF is difficult to distinguish from the HGF
of the nss-sulfate mode. The large overlap in HGFs between the SSA and nss-sulfate mode during the
CWT is potentially due to the presence of an enhanced SSA organic fraction or internally mixed SSA
component. It should be noted that the absence of chemical composition measurements, such as during
the CAPRICORN voyage, limits the interpretation of HGF measurements.
To compute the contribution of each component to the number of CCN, each of the log-normal modes
was assigned a composition and a HGF. The Aitken mode was assumed to made up of modified nss-
sulfates (nss-sulfates and organics), and the pre-sleected 40 nm diameter HGFs were used to calculate
the critical diameter and CCN contribution for this mode. The SSA log-normal mode was assumed
102
Figure 38: Baseline HGF distributions for pre-selected 40 nm diameter (left column), 100 nm diam-
eter (middle column) and 150 nm diameter (right column) averaged for the CWT voyage (top row),
CAPRICORN voyage (middle row) and the Ice-Edge to Equator voyage (bottom row). HGFs measured
at 90% RH and have not been corrected for the Kelvin effect.
103
to be composed of nascent SSA i.e. not cloud processed, and was assumed to a HGF equal to the
average of the HGF distribution at HGF greater than 1.75. The boundary for the SSA mode was
based on heated HGF distributions which showed a dominant mode at a HGF of approximately 2 for
the Ice-Edge to Equator voyage, 1.8 for CAPRICORN and 1.6 for the CWT voyage. The accumulation
mode was assumed to be made up of a mixture of nss-sulfates and a cloud processed component, which
is an internal mixture of nss-sulfates and SSA. The HGF of the accumulation mode was taken as the
average of the moderately hygroscopicity bin, HGF > 1.2 and HGF <1.75.
0
200
400
600
0 200 400 600
Measured CCN [cm−3]
Mod
elle
d C
CN
[cm
−3]
VoyageCAPRICORN
CWT
I2E
Figure 39: Modelled and measured CCNc from the CWT (red) CAPRICORN (orange) and Ice-Edge
to Equator (blue) voyages.
Overall the modelled CCN concentration agree reasonably well with the observations, as show in
Figure 39. There is a small positive bias in the modelled CCN concentrations, in particular during the
CAPRICORN voyage. The CCN closure provides some confidence in the assumed composition and
HGFs for the log-normal modes. In addition, agreement between sub saturated and supersaturated
water uptake indicates that the surface tension effects aren’t likely to make significant contribution to
water uptake for these measurements. It should be noted that it is possible that compensating errors
and assumptions could contribute to agreement between measured and modelled CCN. The reduction
in heated HGFs between the Ice-Edge to Equator, CAPRICORN and CWT voyages is consistent with
an increasing organic volume fraction and solubility of the organic component. The contribution of
SSA to the CCN number concentrations is discussed further in Chapter 10.3.
104
10 Discussion and implications
The relationship between the organic composition in sea water and the nascent SSA organic fraction
is discussed in Chapter 10.1, and the impact of SSA organics on SSA water uptake is discussed in
Chapter 10.2. Modelled SSA enrichment and SSA water uptake are compared to that measured during
the SOAP voyage, which provides a direct validation for modelling and addresses the knowledge gaps
outlined in Chapter 7.6. the results relating to nascent SSA organic enrichment and nascent SSA
organic surface partitioning are based on chamber measurements taken during the SOAP voyage. The
seasonal and zonal contribution of SSA to CCN concentrations over the Southern Ocean are discussed
in Chapter 10.3, and directly compared with observations from remote sensing providing important
validation for these measurements over the Southern Ocean.
10.1 SSA organic enrichment
The observations and modelling discussed in this section are based on the nascent SSA chamber
measurements taken during the SOAP voyage. The organic enrichment of SSA was examined using the
chlorophyll-a based emissions scheme suggested by Gantt et al. 2011, the OCEANFILMS-1 emissions
scheme (Burrows et al. 2014) and the OCEANFILMS-2 emissions scheme which allows for the co-
adsoprtion of polysaccharides (Burrows et al. 2016), as shown in Figure 40. Measured OMF values
displayed a relatively poor relationship with chlorophyll-a, as has been observed in previous studies.
Unsurprisingly the correlation between chlorophyll-a and OMF is strongest for the emission scheme
that directly relates these two variables. The relationship between measured OMFs and chlorophyll-a
displays much greater scatter, which is also reflected in the OCEANFILMS models. As described in
Chapter 7.5.1, the enrichment of SSA is not just a product of phytoplankton biomass, which is largely
what is measured by chlorophyll-a, but more likely due to the demise of phytoplankton communities
and the resulting release of a range of organic material, some of which is available for transfer into
the aerosol phase. OCEANFILMS captures the broad molecular classes in seawater and therefore the
dependence between chlorophyll-a and SSA organics is more obfuscated.
The correlation between SSA OMF estimated using chlorophyll-a (Gantt et al. 2011) and the measured
SSA OMF is not very strong, as shown in Figure 41. The nature of the sampling method used in this
study, 23 spot samples taken over an 18 day period, was not favourable for the use of chlorophyll-a as
a marker for SSA organics. Chlorophyll-a is best used as marker over much longer timescales, of the
order of weeks to months, for example to describes the seasonality in SSA organic enrichment. The
OCEANFILMS-1 model improves on the representation of primary marine aerosol compared to the
chlorophyll-a model, however the magnitude of the modelled OMF is low, which is likely due to the
under-representation of more soluble DOC, such as polysaccharides. OCEANFILMS-2 includes the
co-adsorption of polysaccharides and reproduces the SSA OMF reasonably well. It is worth noting
that both OCEANFILMS models over predict when the organic fraction is low, OMF < 0.05.
The organic macromolecular classes associated with the SSA were determined from the functional group
composition using the conversions outlined in Burrows et al. 2014, which relates the concentration of
105
Figure 40: Filter organic mass fraction as a function of seawater chlorophyll-a concentration.
Measured OMF (red) is from FTIR and IBA analysis of filter samples, modelled OMFs are from
chlorophyll-a based emission scheme (yellow) (Gantt et al. 2011), OCEANFILMS-1 (light blue) and
OCEANFILMS-2 (dark blue). OCEANFILMS models assumed a bubble thickness of 0.3 μm.
Figure 41: OMF modelled using parameterisation from Gantt et al. (2011) (left), OCEANFILMS-1
(middle) and OCEANFILMS-2 (right) compared to that measured in this study. Lines show the linear
fit between Modelled and measured OMF, shading shows the 95% confidence interval in the linear
model, the R2 is also printed in each panel.
106
each molecular class to a weighted sum of the functional group concentrations. The conversion between
functional groups and macromolecular classes is based on the properties of characteristic molecules,
for example cholesterol and simple sugars, and is therefore not a perfect representation of the marine
environment. As expected OCEANFILMS-1 underestimated the proportion of polysaccharides and
over estimated the proportion of lipids, as shown in Figure 42. OCEANFILMS-2 displayed an improved
representation of polysaccharides, however the proportion of lipids was still overestimated.
The proportion of the processed molecular class was extremely underrepresented for both OCEAN-
FILMS models. It is worth noting that the functional groups that make up the processed class are
assumed to be the same as those for polysaccharides, which effectively splits this high hydroxyl com-
ponent across the two classes. The polysaccharide and processed classes should be considered as the
same (or very similar) components in the macromolecular class distributions from measured data. The
polysaccharide, or hydroxyl dominated class, is underestimated by OCEANFILMS-2 and in partic-
ular OCEANFILSM-1. According to the OCEANFILMS model the processed class are assigned a
much lower langmuir adsorption coefficient than the polysaccharide class, which results in lower en-
richment into the aerosol phase. During the SOAP voyage seawater concentrations of the processed
class, which is the DOC component not assigned to any other macromolecular class, were of similar
scale to the polysaccharide component, ranging from 0.1 - 2.2 times the polysaccharide concentration.
The under-representation of the processed component characterises the difference between the mea-
sured molecular classes and those from OCEANFILMS-2. The large polysaccharide fraction observed
in this work, along with the observed surface tension effects (as discussed in Chapter 10.2) indicate
the possibility of a strong contribution from LPS to the SSA organic fraction.
Over prediction of alkane to hydroxyl ratios for particularly clean marine measurements is a known
issue for OCEANFILMS-2, and broadening the model to different saccharides with varying molecular
weights has been identified for future research (Burrows et al. 2016). Particularly of interest here is
research into the interaction of surfactants with divalent cations, which have been observed to impact
the orientation of surfactant head groups, and thus the surface pressure, for modelled surfactant salt
systems (Adams, Casper, and Allen; Casillas-Ituarte et al. 2010; Casper et al. 2016). In particular Ca2+
has been observed to form particularly stable complexes, bridging neighbouring surfactant molecules,
and having a condensation effect on the monolayer (Casper et al. 2016). The enhanced enrichment
of Ca2+ with higher OMF observed herein may therefore be associated with the complexation with
surfactants, in which case the enrichment of Ca2+ could influence the organic enrichment and the SSA
water uptake.
OCEANFILMS does improve the prediction of organic enrichment from seawater parameters, relative
to chlorophyll-a based models. Limitations remain in the implementation of OCEANFILMS, which
requires the availability of surface water concentrations for the five macromolecular classes, which are
generally generated using biogeochemical modelling. In addition there are remaining uncertainties as
to the global applicability of the model organics applied in OCEANFILMS, in regions with different
phytoplankton populations for example. Further broadening and/or refining the organics of interest
in OCEANFILMS is likely to hinder its application. Similar issues of global applicability are also
present for chlorophyll-a based estimates of OMF. Chlorophyll-a is, however, globally observed at a
107
Figure 42: Measured organic composition inferred from functional groups (right) and modelled or-
ganic composition from OCEANFILMS (OCEANFILMS-1 left panel, OCEANFILMS-2 middle panel).
daily timescale via satellite and is extremely important for large scale simulations of SSA organic
enrichment. Measurements over the sparsely observed Southern and South Pacific Oceans, such as
those reported herein, are important to validate emission schemes developed in other parts of the
world, particualrly given the importance of SSA in this region.
10.2 SSA organic partitioning
The compressed film model was used to represent the nascent SSA water uptake from chamber mea-
surements taken during the SOAP voyage. A range of organics, characterised by different molecular
areas and molar volumes, was used to test the sensitivity of HGF to the organic composition. The
critical molecular surface areas that minimised the error in the modelled HGF were most commonly
35 - 45 square angstroms for both ambient and heated samples, which is consistent with laboratory
observations of marine salts mixed with a phospholipid found in the SML (DPPC) (Casper et al. 2016).
Molar volumes of less than 10 cm3/mol were typically observed. This is well below the threshold ob-
served by Forestieri et al. (2018) to induce surface tension effects significantly different than those
when a surface tension of water is assumed. It should be noted that significant sample to sample
variability was observed in the fitted molecular volume and surface area.
The proportion of SSA organics at the particle surface were tested assuming that partitioning occurs
on the basis of the organic molecular classes as computed from the distribution of functional groups
or from the OCEANFILMS model. Four cases were tested assuming that the lipids partitioned to
the particle surface, that the lipids and polysaccharides, that the lipids polysaccharides and processed
organics and assuming that all of the organics partitioned to the particle surface, as shown in Figure 43
and Figure 44. The most notable feature is that for both ambient (Figure 43) and heated measurements
(Figure 44) the error in the modelled HGF is significantly reduced at high OVFs when all of the organics
are partitioned to the surface. At low OVFs the sensitivity to partitioning of organics is low. Note
that Figure 43 and Figure 44 display the error in the HGF as a function of OMF, very similar results
are achieved when the error is reported as a function of OVF. The results presented here suggest
108
All Surface
Lipid + Poly + Proc Surface
Lipid + Poly Surface
Lipid Surface
0.05 0.10 0.15 0.20
0.0
0.2
0.4
0.6
0.8
0.0
0.2
0.4
0.6
0.8
0.0
0.2
0.4
0.6
0.8
0.0
0.2
0.4
0.6
0.8
OMF
HG
Fmea
sure
d −
HG
Fmod
elle
d Bloom1
2
3a
3b
None
OrganicSpeciationFunctional groups
OceanFILMS
Ambient
Figure 43: Compressed film model HGF error (absolute value of measured minus modelled HGF)
measured at ambient temperature for lipids partitioned to the surface (top), lipids and polysaccharides
partitioned to the surface (2nd panel from top), lipids polysaccharides and processed organics parti-
tioned to the surface (2nd panel from bottom) and all organics partitioned to the surface (bottom).
Organic speciation derived from FTIR measurements (circles) and from OCEANFILMS model (trian-
gles). Compressed film model A0 and Vorg values were determined by those that resulted in the best
fit between observed and modelled HGF for each sample.
109
that at high OVFs all of the organics, including the proteins, partition to the surface of the SSA.
A number of possible scenarios describe the partitioning of organics, for example all of the organics
might partition to the surface for all samples (low and high OVF), alternatively at low OVFs only
some fraction of the organics might partition to the surface e.g. lipids and polysaccharides (or even
no surface partitioning). At low OVFs a surface monolayer is not necessarily formed, even with all
of the organics partitioned to the surface. Regardless, at high OVFs the protein component is likely
to be surface active. The measurements at high OVFs are all during B1, and therefore the apparent
enhanced surface activity of the protein component could be driven by a change in the proteinaceous
composition between phytoplankton blooms.
Directly comparing the HGFs modelled using the compressed film model and those modelled using the
ZSR assumption, as in Figure 45, reveals the potential for a small fraction of organic partitioned to the
surface when the OVF is low, and a high fraction partitioned to the surface when the OVF is high. The
compressed film model output also highlights the contribution of surface tension to the observed SSA
HGF. The surface tension effects on HGF observed at high organic volume fractions are approximately
equivalent to the reduction in HGF predicted by the ZSR assumption, i.e. by Raoult’s Law. The role
of surface tension is important for a number of reasons, one is that OVFs inferred in studies using
water uptake techniques, using the ZSR assumption, have often been lower than those measured using
more direct analyses of SSA chemical composition. For example, water uptake measurements made by
Fuentes et al. (2011) and Modini et al. (2010) yielded Aitken mode SSA organic volume fraction of up
to 40%, while measurements with similar seawater chlorophyll-a and DOC concentrations measured
OVFs of up to 80% (Keene et al. 2007; Facchini et al. 2008; Prather et al. 2013). A contribution to
this apparent discrepancy could be that estimating the OVF from water uptake measurements is done
using the ZSR assumption, therefore the presence of any surface active organics, and the resulting
depression of surface tension, would likely result in an under-prediction of OVF. The contribution of
surface tension to the discrepancy between reported observations is unknown as there is also likely
to be a contribution from variability in the marine organics and resulting SSA enrichment. As was
outlined in the previous section chlorophyll-a (and DOC) is an imperfect variable for the description
of the SSA organic fraction.
The impact of the reduction in SSA surface tension should be measured by what influence it has on the
overall CCN concentrations, and to test this the CCN concentrations predicted from the compressed
film model, with all organics partitioned to the surface, were compared with those computing using
the ZSR assumption, when both models were applied to a theoretical SSA size distribution (Fig. 46).
The compressed film model was applied assuming that the critical diameter was equal to the VH-
TDMA pre-selected particle diameter (50 nm), and the κ-Köhler equation was used to calculate CCN
concentrations from the ZSR modelled HGFs using the same critical SS as in the compressed film model.
Setting up the calculation as described above gives a comparison of the ZSR CCN concentration, which
varied with OVF, to the compressed film CCN, which was stable across all samples. In addition the
surface tension required to obtain the compressed film CCN was also calculated. The decrease in the
surface tension was up to 30 mN.m−1, which is a large reduction in surface tension, similar to that
modelled for binary SSA proxies by Forestieri et al. 2018. The increase in CCN associated with the
110
All Surface
Lipid + Poly + Proc Surface
Lipid + Poly Surface
Lipid Surface
0.05 0.10 0.15 0.20
0.0
0.2
0.4
0.6
0.8
0.0
0.2
0.4
0.6
0.8
0.0
0.2
0.4
0.6
0.8
0.0
0.2
0.4
0.6
0.8
OMF
HG
Fmea
sure
d −
HG
Fmod
elle
d Bloom1
2
3a
3b
None
OrganicSpeciationFunctional groups
OceanFILMS
Heated
Figure 44: Compressed film model HGF error (absolute value of measured minus modelled HGF)
measured at 255 - 400 °C for lipids partitioned to the surface (top), lipids and polysaccharides parti-
tioned to the surface (2nd panel from top), lipids polysaccharides and processed organics partitioned
to the surface (2nd panel from bottom) and all organics partitioned to the surface (bottom). Organic
speciation derived from FTIR measurements (circles) and from OCEANFILMS model (triangles).
Compressed film model A0 and Vorg values were determined by those that resulted in the best fit
between observed and modelled HGF for each sample.
111
●
●
●
●●
●
●
●
●
●●●●●
●●
●
●
●
●●
●●
●●
●●●●
●●●
●
●
●
1.4
1.6
1.8
2.0
0.25 0.50 0.75 1.00OVF
HG
F am
bien
t
Bloom● 1
2
3a
3b
None
Data●●
●●
●●
Compressed Film Model
Observations
ZSR Model
Figure 45: Ambient HGF modelled using compressed film model (green) as a function of OVF,
modelled using the ZSR assumption (light blue) and observed (dark blue). Compressed film model
output is for case where all organics are partitioned to the particle surface. Compressed film model
A0 and Vorg values were determined by those that reduced the overall error between observed and
modelled HGF for all of the samples.
112
surface tension modification was up to 10 cm−3, which represents a 17% decrease in CCN concentrations
from the ZSR approach. Translated into CDNC, a 17% decrease would have a potentially large impact
on the cloud fraction, cloud liquid water content and cloud radiative forcing (Rosenfeld et al. 2019).
The results presented here suggest that the inclusion of surface tension effects for SSA could improve
the representation of SSA CCN in atmospheric modelling.
●
●
●
●
●
●●
●●●
●
●
●●
●
70
75
80
85
0.00 0.25 0.50 0.75 1.00OVF
CC
Nc
[cm
−3]
●
●
●
●
●
●
●
●●●
●
●
●
●
●
0.04
0.05
0.06
0.07
0.00 0.25 0.50 0.75 1.00OVF
Surfa
ce T
ensi
on [N
/m]
Figure 46: Modelled CCNc concentrations (left) and surface tension (right) as a function of OVF.
The CCN concentrations are modelled assuming an SSA size distribution with number concentration
of 100 cm-3, a mean diameter of 160 nm and a geometric standard deviation of 2.6. Blue line shows
the CCN concentration predicted using the compressed film model with all organics partitioned to
the surface and with an assumed Dcrit of 50 nm, red points show the predicted CCN concentrations
using the ZSR assumption. The surface tension required to obtain a Dcrit of 50 nm is represented
with yellow circles (right), and the blue line shows the surface tension of water. The dashed lines are
a LOESS fit to the data, and the shaded regions are a 95% confidence interval in the fit.
During the Southern Ocean voyages the voyage average heated HGF was observed to decrease from
the Ice-Edge to Equator voyage, from approximately 2 to approximately 1.8, and then subsequently
to approximately 1.6 for the CWT voyage. The seasonal reduction in SSA water uptake broadly
follows that expected from the ZSR assumption for internally mixed sea salt and organics, and that
modelled using chlorophyll-a based SSA parameterisations (Gantt et al. 2012a; Vignati et al. 2010).
The absence of apparent surface tension effects may be due to the differences in the macromolecular
profile over the broader Southern Ocean to that observed during SOAP, in particular during B1. In
addition the biological activity is unlikely to be as strong, which could result in insufficient organic
fractions for the formation of a monolayer. It is worth noting that Collins et al. (2016) observed water
uptake inconsistent with the ZSR assumption, however this was for intense phytoplankton blooms
(chlorophyll-a > 1 mg/m3) and combined with the results reported here, it appears unlikely to be
applicable across large oceanic regions.
113
10.3 SSA contribution to CCN
The contribution to CCNc from SSA has been previously observed from in-situ measurements over the
Southern Ocean (Fossum et al. 2018; Quinn et al. 2017), however these studies have largely reported
on the results from single voyages and do not capture the seasonality in the aerosol composition. In
addition studies have examined the seasonality in the contribution of SSA to CCN over the Southern
Ocean, however these have generally been modelling or remote sensing studies (McCoy et al. 2015).
The in-situ measured SSA log-normal mode contribution to CCN number concentrations for the CWT,
CAPRICORN and Ice-Edge to Equator voyages are shown in Figure 47 for 0.5% SS, and in Figure 48
for 0.2% SS. During the summertime fractional contributions from SSA are the lowest, 0.48 ± 0.17
at 0.2% SS, and increase to the maximum observed value in the late-autumn 0.79 ± 0.20. It should
be noted that although CCN concentrations were measured at 0.5% SS, this is likely to exceed the
in-cloud SS for marine stratocumulus, therefore values for 0.2% SS are also reported.
The seasonal trend observed here is similar to that from remote sensing reported by McCoy et al.
(2015), where a summertime minimum of approximately 0.55 and a wintertime maximum of approxi-
mately 0.85 were reported. The SSA contributions reported by McCoy et al. (2015) are slightly higher
than those from computed for the SSA log-normal mode (nascent), however it should be noted that
there is an additional SSA contribution from the accumulation/cloud processed mode. The in-situ
results presented here are consistent with those presented by Fossum et al. (2018), which reported
summertime SSA contributions to CCN over the Southern Ocean of 8-51% at wind speed less than 16
m/s and up to 100% at higher wind speeds. Quinn et al. ( 2017) reported Southern Ocean SSA con-
tributions to CCN of the order of 20 - 40% at 0.2% SS, from largely summertime measurements.
Although the fractional contributions to CCN were higher during the winter months, the overall num-
ber concentration of SSA, and the number of SSA CCN at 0.2% SS increased slightly during the
summertime (though not significant). SSA CCN was 53 ± 30 cm-3 in February and 40 ± 27 cm-3
in May. It is worth noting that average SSA concentrations in April were very high (approximately
77 cm-3), however this is driven by a single high wind speed period. A potential increase in the SSA
number concentration during the biologically active summer months is consistent with previous obser-
vations of enhanced SSA production due to the presence of surface active marine organics (Long et al.
2014; Fuentes et al. 2010b). The precise mechanism for increased production of SSA in the presence
of certain organic molecular classes is still unclear. McCoy et al. (2015) also reported a positive cor-
relation between SSA OMF and the cloud droplet number concentration over the Southern Ocean, in
particular over the high latitude Southern Ocean, where contributions from SSA are generally higher
(as seen in Figure 49). The authors postulated that it could be due to surface tension effects leading
to enhanced SSA CCN activation or enhanced SSA production. No surface tension effects on SSA
activation to CCN were observed for the ambient Southern Ocean measurements reported here.
The potential for increasing SSA flux with increased ocean organic activity could require modifications
to existing SSA organic parameterisations, which largely apply an organic mass to existing wind speed
and temperature dependent fluxes i.e. assume that the emission rate is independent of the composition.
Overall these existing parameterisations will lead to lower HGFs, through the ZSR assumption, and
114
Figure 47: Seasonality in fraction of CCN (0.5% SS) from SSA mode for the CWT (red), CAPRI-
CORN (orange) and Ice-Edge to Equator (blue) voyages.
Figure 48: Seasonality in fraction of CCN (0.2% SS) from SSA mode for the CWT (red), CAPRI-
CORN (orange) and Ice-Edge to Equator (blue) voyages.
115
lower CCN concentrations, which may in turn exacerbate cloud radiative biases over the Southern
Ocean. Even though the observed changes in CCN are modest, small increases in the CDNC can have
a large influence on cloud radiative properties, particularly over pristine environments with low CDNC
concentrations (Rosenfeld et al. 2019).
Figure 49: Fraction of CCN (0.5% SS) from SSA mode for the CWT, CAPRICORN and Ice-Edge
to Equator voyages by latitude bins, 40 - 50 °S (red) and 50 - 65 °S (blue).
116
11 Conclusions
Measurements of the composition and water uptake properties of the marine aerosol over the southern
hemisphere are sorely needed for the validation of atmospheric models. The Southern Ocean, in
particular, is scarcely observed and has a high impact on modelled radiative forcing because of the
under-representation of low level cloud, most notably at the high latitudes during the summer months.
This work reports on four research voyages which measured the marine aerosol water uptake, size
resolved concentration and composition over the Southern and Southern Pacific Oceans across a number
of seasons. These measurements are key to establishing the aerosol cloud interactions that drive cloud
radiative forcing in the region.
Chamber measurements of primary marine aerosol were generated using 23 water samples collected
from differing water depths and across 3 phytoplankton blooms tracked during the SOAP voyage,
revealing that SSA was an internal mixture of sea salt and organics. Volatility measurements at a
preselected particle mobility diameter of 50 nm indicated that the SSA OVF ranged from 0.05 to 0.87,
and was largely made up of a refractory component. Filter measurements of PM for diameters less than
approximately 1 μm were analysed for the concentration of organic functional groups using FTIR, and
the concentration of inorganic species was determined using IBA. The organic mass fractions ranged
from 0.03 to 0.23, and had a large proportion of hydroxyl functions groups consistent with an SSA
organics dominated by polysaccharides. The alkane to hydroxyl ratio was very low for these SSA, which
also suggests a polysaccharide rich, less aliphatic organic species. Ca2+ was observed to be 1.7 times
higher in the aerosol phase than in seawater, which is consistent with other primary marine aerosol
studies. A possible explanation for this is that Ca2+ complexes with organics in the SML, which is
supported by the correlation of the Ca2+ enrichment factor with the OMF. There is an associated
depletion of Cl-, potentially due to the absence of a CaCl2 component. Comparisons of the SSA
generated using mixed layer SSA and using deep water samples revealed that SSA generated from
surface mixed layer water were enriched in organics, calcium, sulfate and alkanes.
The SSA organic fraction displayed a weak correlation with chlorophyll-a, which is consistent with
previous studies showing that chlorophyll-a doesn’t sufficiently describe the molecular complexity of
marine organics. Chlorophyll-a is best used to correlate SSA organic enrichment over larger spatial
scales and at temporal scales of the order of months. The OCEANFILMS model provides an improved
representation of the SSA organic fraction observed in this study, in particular when the co-adsorption
of polysaccharides is included i.e. OCEANFILMS-2 is applied. Using the distribution of the organic
functional groups, the composition of the SSA organic fraction was framed in terms of the molecular
classes in OCEANFILMS. The observed SSA was dominated by a polysaccharide component, with a
relative small lipid fraction, OCEANFILMS-2 couldn’t reproduce the observed polysaccharide fraction
indicating that further work on representing the adsorption of polysaccharide species is required.
Water uptake measurements revealed that the SSA hygroscopicity was largely invariable with the
organic mass fraction, and deviated from the regularly used water uptake mixing rule, the ZSR as-
sumption. Deviations between the observed HGF and the ZSR rule were particularly noteworthy at
organic volume fractions above 0.4, and particularly during B1. It should be noted that the ZSR rule
117
is commonly used in atmospheric models to represent the hygroscopicity of internally mixed aerosol.
The representation of hygroscopicity was drastically improved when the compressed film model was
applied. The observation of surface tension effects could be one reason for the discrepancies that have
been observed between organic fractions estimated using water uptake and those estimated using other
methods. The compressed film model partitions organics to the surface of an aqueous aerosol, which
lowers the particle surface tension and increases water uptake. In the case when all of the organics were
partitioned to the surface, the error in the modelled HGF droped to almost to within the uncertainty
of the measured HGFs.
The combination of very low alkane to hydroxyl ratios, and the surface tension effects observed from
hygroscicity measurements indicate that the organic component could be made up of lipopolysaccha-
rides, which have previously been identified as an important component in primary marine aerosols.
Ca2+ has also been identified to complex with surface active species and alter their orientation and
therefore their surface active properties. Further work should focus on determining the specific species
which drive the surfactant behaviour of SSA and identify the regions and conditions under which these
species become important contributors to SSA. It is apparent that when these species are present in
large enough proportions the SSA CCN concentration estimated using the ZSR approach is estimated
to be under predicted by up to 17%, which is certainly enough to have an impact on cloud radiative
forcing in clean marine conditions.
The aerosol size distributions, volatility, composition and water uptake properties were also measured
during three voyages in the Southern Ocean spanning the summer, autumn and early winter. The
contribution of the nascent SSA mode to CCN was lowest during the summer, 48% of the CCN at
0.2% SS in January, and increased to 79% in May. The number concentration of nascent SSA CCN
decreased slightly from January to May, indicating that the seasonal change in contribution to CCN
is driven by a seasonally varying secondary component. Seasonality in CCN contributions over the
Southern Ocean have previously been observed from remote sensing measurements (McCoy et al.
2015). The data reported here display good agreement with the seasonality observed from remote
sensing measurements, and therefore provide valuable in-situ validation. Across all three Southern
Ocean voyages the CCN contribution from SSA was greater at higher Southern Ocean latitudes, in
particular beyond 50 °S. The zonal trends observed here are also consistent with previous observations
from remote sensing.
No significant surface tension effects were observed in the water uptake of the SSA mode, which suggests
that high organic fractions associated with intense phytoplankton blooms could be required to observed
the surface tensions effects observed during the SSA chamber experiments during the SOAP voyage.
The surface tension effects may not be important on large spatial and seasonal scales but could be
of importance for phytoplankton bloom regions, and for ocean basins with lower wind speeds which
are likely to have a more intact SML, increasing the chance of surfactant enrichment to SSA. The
results presented here provide an excellent data set to examine the aerosol sources over the Southern
Ocean with an eye to quantifying the impact of seasonal and zonal aerosol variability on aerosol cloud
interactions.
118
12 References
Adams, Ellen M, Clayton B Casper, and Heather C Allen. “Effect of cation enrichment on dipalmi-
toylphosphatidylcholine (DPPC) monolayers at the air-water interface”. Journal of Colloid and In-
terface Science 478 IS -:353–364. http : / / www . sciencedirect . com / science / article / pii /
S0021979716303757.
Allan, J D, et al. 2009. “Composition and properties of atmospheric particles in the eastern Atlantic
and impacts on gas phase uptake rates”. Atmospheric Chemistry and Physics 9, no. 2323 (): 9299–
9314. doi:10.5194/acp-9-9299-2009. http://www.atmos-chem-phys.net/9/9299/2009/.
Alroe, J, et al. 2018. “Determining the link between hygroscopicity and composition for semi-volatile
aerosol species”. Atmospheric Measurement Techniques 11, no. 77 (): 4361–4372. doi:10.5194/amt-
11-4361-2018. https://www.atmos-meas-tech.net/11/4361/2018/.
Andreae, M O, and D Rosenfeld. 2008. “Aerosol–cloud–precipitation interactions. Part 1. The nature
and sources of cloud-active aerosols”. Earth-Science Reviews 89, no. 11 (): 13–41. http://www.
sciencedirect.com/science/article/pii/S0012825208000317.
Ault, Andrew P, et al. 2013a. “Raman microspectroscopy and vibrational sum frequency generation
spectroscopy as probes of the bulk and surface compositions of size-resolved sea spray aerosol parti-
cles”. Physical Chemistry Chemical Physics 15, no. 1717 (): 6206. doi:10.1039/c3cp43899f. http:
//xlink.rsc.org/?DOI=c3cp43899f.
Ault, Andrew P, et al. 2013b. “Size-Dependent Changes in Sea Spray Aerosol Composition and Proper-
ties with Different Seawater Conditions”. Environ. Sci. Technol. 47, no. 1111 (): 5603–5612. doi:doi:
10.1021/es400416g. http://dx.doi.org/10.1021/es400416g.
Ayers, G P, and J L Gras. 1991. “Seasonal relationship between cloud condensation nuclei and aerosol
methanesulphonate in marine air”. Nature 353, no. 63476347 (): 834–835. doi:10.1038/353834a0.
http://www.nature.com/doifinder/10.1038/353834a0.
Ayers, G P, et al. 1995. “Dimethylsulfide in marine air at Cape Grim, 41°S”. J. Geophys. Res 100 ():
21013–21021. doi:10.1029/95JD02144. http://dx.doi.org/10.1029/95JD02144.
Bates, T S, et al. 2012. “Measurements of ocean derived aerosol off the coast of California”. J. Geophys.
Res 117 (): n/a. doi:10.1029/2012JD017588. http://dx.doi.org/10.1029/2012JD017588.
Beddows, David C S, Manuel Dall’Osto, and Roy M Harrison. 2010. “An Enhanced Procedure for the
Merging of Atmospheric Particle Size Distribution Data Measured Using Electrical Mobility and
Time-of-Flight Analysers”. Aerosol Science and Technology 44, no. 1111 (): 930–938. doi:10.1080/
02786826.2010.502159. http://www.tandfonline.com/doi/abs/10.1080/02786826.2010.
502159.
Berg, Olle H, Erik Swietlicki, and Radovan Krejci. 1998. “Hygroscopic growth of aerosol particles in
the marine boundary layer over the Pacific and Southern Oceans during the First Aerosol Charac-
terization Experiment (ACE 1)”. J. Geophys. Res 103 (): 16535–16545. http://dx.doi.org/10.
1029/97JD02851.
Bialek, J, et al. 2014. “Hygroscopic and chemical characterisation of Po Valley aerosol”. Atmospheric
Chemistry and Physics 14, no. 33 (): 1557–1570. doi:10.5194/acp-14-1557-2014. http://www.
atmos-chem-phys.net/14/1557/2014/.
119
Bialek, Jakub, et al. 2012. “On the contribution of organics to the North East Atlantic aerosol num-
ber concentration”. Environmental Research Letters 7, no. 44 (): 044013. doi:10 . 1088 / 1748 -
9326/7/4/044013. http://stacks.iop.org/1748- 9326/7/i=4/a=044013?key=crossref.
13097dfb3ed934088d2da4c012d77078.
Bigg, E K. 2007. “Sources, nature and influence on climate of marine airborne particles”. Environmental
Chemistry 4, no. 33 (): 155–161. doi:10.1071/EN07001. http://dx.doi.org/10.1071/EN07001.
Bigg, E Keith, and Caroline Leck. 2008. “The composition of fragments of bubbles bursting at the
ocean surface”. Journal of geophysical research 113 (): D11209. doi:10.1029/2007JD009078. http:
//www.agu.org/pubs/crossref/2008/2007JD009078.shtml.
Biskos, G, et al. 2006a. “Nanosize Effect on the Deliquescence and the Efflorescence of Sodium Chloride
Particles”. Aerosol Science and Technology 40, no. 22 (): 97–106. doi:10.1080/02786820500484396.
http://www.tandfonline.com/doi/abs/10.1080/02786820500484396.
Biskos, G, et al. 2006b. “Nanosize effect on the hygroscopic growth factor of aerosol particles”. Geo-
physical Research Letters 33, no. 77 (): n/a. doi:10.1029/2005GL025199. http://dx.doi.org/10.
1029/2005GL025199.
Blanchard, Duncan C. 1989. “The Ejection of Drops from the Sea and Their Enrichment with Bacteria
and Other Materials: A Review”. Estuaries 12, no. 33 (): 127–137. http://www.jstor.org/stable/
1351816.
Blot, R, et al. 2013. “Ultrafine sea spray aerosol over the southeastern Pacific: open-ocean contributions
to marine boundary layer CCN”. Atmospheric Chemistry and Physics 13, no. 1414 (): 7263–7278.
doi:10.5194/acp-13-7263-2013. http://www.atmos-chem-phys.net/13/7263/2013/.
Bodas-Salcedo, A, et al. 2012. “The surface downwelling solar radiation surplus over the southern ocean
in the met office model: The role of midlatitude cyclone clouds”. Journal of Climate 25, no. 2121
(): 7467–7486. http://www.scopus.com/inward/record.url?eid=2-s2.0-84870009577&
partnerID=40&md5=225d6ed5d95939dbed0a048e2e0a1a5d.
Burrows, S M, et al. 2014. “A physically based framework for modeling the organic fractionation
of sea spray aerosol from bubble film Langmuir equilibria”. Atmospheric Chemistry and Physics
14, no. 2424 (): 13601–13629. doi:10.5194/acp- 14- 13601- 2014. http://www.atmos- chem-
phys.net/14/13601/2014/.
Burrows, S M, et al. 2018. “OCEANFILMS sea-spray organic aerosol emissions – Part 1: implementation
and impacts on clouds”. Atmospheric Chemistry and Physics Discussions 2018 (): 1–27. doi:10.5194/
acp-2018-70. https://www.atmos-chem-phys-discuss.net/acp-2018-70/.
Burrows, Susannah M, et al. 2016. “OCEANFILMS-2: Representing coadsorption of saccharides in
marine films and potential impacts on modeled marine aerosol chemistry”. Geophysical Research
Letters 43, no. 1515 (): 8306–8313. doi:10.1002/2016GL069070. http://dx.doi.org/10.1002/
2016GL069070.
Carslaw, K S, et al. 2013. “Large contribution of natural aerosols to uncertainty in indirect forcing”.
Nature 503, no. 74747474 (): 67–71. doi:Article. http://dx.doi.org/10.1038/nature12674.
Casillas-Ituarte, Nadia N, et al. 2010. “Surface organization of aqueous MgCllt;subgt;2lt;/subgt; and
application to atmospheric marine aerosol chemistry”. Proc Natl Acad Sci USA 107, no. 1515 ():
6616. http://www.pnas.org/content/107/15/6616.abstract.
120
Casper, Clayton B, et al. 2016. “Surface Potential of DPPC Monolayers on Concentrated Aqueous
Salt Solutions”. J. Phys. Chem. B 120, no. 88 (): 2043–2052. doi:10.1021/acs.jpcb.5b10483.
http://pubs.acs.org/doi/abs/10.1021/acs.jpcb.5b10483.
Cavalli, F, et al. 2004. “Advances in characterization of size-resolved organic matter in marine aerosol
over the North Atlantic”. J. Geophys. Res 109 (): n/a. doi:10.1029/2004JD005137. http://dx.doi.
org/10.1029/2004JD005137.
Ceburnis, Darius, et al. 2008. “Marine aerosol chemistry gradients: Elucidating primary and secondary
processes and fluxes”. Geophysical Research Letters 35, no. 77 (): n/a. doi:10.1029/2008GL033462.
http://dx.doi.org/10.1029/2008GL033462.
Chan, Man Nin, and Chak K Chan. 2003. “Hygroscopic Properties of Two Model Humic-like Substances
and Their Mixtures with Inorganics of Atmospheric Importance”. Environmental science & technology
37, no. 22 (): 5109–5115. doi:10.1021/es034272o.
Charlson, R J, et al. 1987. “Oceanic phytoplankton, atmospheric sulphur, cloud”. Nature 3526 (): 16.
http://www.atmos.washington.edu/~dipierro/classes/Aerosol_chemistry/CLAW_326655a0.
pdf.
Chen, H, et al. 1973. “A general method of predicting the water activity of ternary aqueous solutions
from binary data”. The Canadian Journal of Chemical Engineering 51, no. 22 (): 234–241. doi:10.
1002/cjce.5450510214. http://dx.doi.org/10.1002/cjce.5450510214.
Claeys, Magda, et al. 2010. “Chemical characterisation of marine aerosol at Amsterdam Island during
the austral summer of 2006–2007”. Journal of Aerosol Science 41, no. 11 (): 13–22. http://www.
sciencedirect.com/science/article/pii/S0021850209001426.
Clarke, A D, et al. 2013. “Free troposphere as the dominant source of CCN in the Equatorial Pacific
boundary layer: long-range transport and teleconnections”. Atmospheric Chemistry and Physics Dis-
cussions 13, no. 11 (): 1279–1326. doi:10.5194/acpd-13-1279-2013. http://www.atmos-chem-
phys-discuss.net/13/1279/2013/.
Clegg, Simon L, Peter Brimblecombe, and Anthony S Wexler. 1998. “Thermodynamic Model of the
System H +−NH 4+−Na +−SO 42-−NO 3-−Cl -−H 2O at 298.15 K”. The Journal of Physical
Chemistry A 102, no. 12 (): 2155–2171. doi:10.1021/jp973043j.
Cochran, Richard E, et al. 2016. “Analysis of Organic Anionic Surfactants in Fine and Coarse Fractions
of Freshly Emitted Sea Spray Aerosol”. Environ. Sci. Technol. 50, no. 55 (): 2477–2486. doi:10.1021/
acs.est.5b04053. http://pubs.acs.org/doi/abs/10.1021/acs.est.5b04053.
Cochran, Richard E, et al. 2017. “Molecular Diversity of Sea Spray Aerosol Particles: Impact of Ocean
Biology on Particle Composition and Hygroscopicity”. Chem 2, no. 55 (): 655–667. doi:10.1016/j.
chempr.2017.03.007. http://linkinghub.elsevier.com/retrieve/pii/S2451929417301201.
Cohen, David D, et al. 2004. “IBA methods for characterisation of fine particulate atmospheric pol-
lution: a local, regional and global research problem”. The 13th International Conference on Par-
ticle Induced X-ray Emission (PIXE 2013) 219–220 IS - (): 145–152. //www.sciencedirect.com/
science/article/pii/S0168583X04000710.
Collins, D B, et al. 2014. “Direct aerosol chemical composition measurements to evaluate the physico-
chemical differences between controlled sea spray aerosol generation schemes”. Atmospheric Measure-
121
ment Techniques 7, no. 1111 (): 3667–3683. doi:10.5194/amtd-7-6457-2014. http://www.atmos-
meas-tech.net/7/3667/2014/amt-7-3667-2014.pdf.
Collins, Douglas B, et al. 2013. “Impact of marine biogeochemistry on the chemical mixing state and
cloud forming ability of nascent sea spray aerosol”. Journal of Geophysical Research: Atmospheres
(): n/a. doi:10.1002/jgrd.50598. http://dx.doi.org/10.1002/jgrd.50598.
Collins, Douglas B, et al. 2016. “Phytoplankton blooms weakly influence the cloud forming ability of sea
spray aerosol”. Geophysical Research Letters 43, no. 1818 (): 9975–9983. doi:10.1002/2016GL069922.
http://onlinelibrary.wiley.com/doi/10.1002/2016GL069922/full.
Covert, David S, et al. 1998. “Comparison of directly measured CCN with CCN modeled from the
number-size distribution in the marine boundary layer during ACE 1 at Cape Grim, Tasmania”.
J. Geophys. Res 103 (): 16597–16608. doi:10.1029/98JD01093. http://dx.doi.org/10.1029/
98JD01093.
Cravigan, Luke, et al. 2013. “Marine aerosol hygroscopicity and volatility, measured on the Chatham
Rise (New Zealand)”. AIP Conference Proceedings 1527, no. 11 (): 547–550. doi:10.1063/1.4803329.
http://scitation.aip.org/content/aip/proceeding/aipcp/10.1063/1.4803329.
Cravigan, Luke T. 2015. “Sources and composition of sub-200 nm sea spray aerosol inferred from
volatility and hygroscopicity.” ().
Cravigan, Luke T, et al. 2015. “Observation of sea-salt fraction in sub-100 nm diameter particles
at Cape Grim”. Journal of Geophysical Research: Atmospheres (): 2014JD022601. doi:10.1002/
2014JD022601. http://dx.doi.org/10.1002/2014JD022601.
Dall’Osto, M, et al. 2010. “Aerosol properties associated with air masses arriving into the North East
Atlantic during the 2008 Mace Head EUCAARI intensive observing period: an overview”. Atmo-
spheric Chemistry and Physics 10, no. 1717 (): 8413–8435. http://www.atmos-chem-phys.net/10/
8413/2010/acp-10-8413-2010.pdf.
Dawson, H R S, P G Strutton, and P Gaube. 2018. “The Unusual Surface Chlorophyll Signatures
of Southern Ocean Eddies”. J. Geophys. Res.: Oceans 123, no. 99 (): 6053–6069. doi:10.1029/
2017JC013628. http://doi.wiley.com/10.1029/2017JC013628.
Deane, Grant B, and M Dale Stokes. 2002. “Scale dependence of bubble creation mechanisms in break-
ing waves”. Nature 418, no. 69006900 (): 839–844. doi:10.1038/nature00967. http://www.nature.
com/doifinder/10.1038/nature00967.
DeCarlo, Peter F, et al. 2004. “Particle Morphology and Density Characterization by Combined Mo-
bility and Aerodynamic Diameter Measurements. Part 1: Theory”. Aerosol Science and Technology
38, no. 12 (): 1185–1205. doi:10.1080/027868290903907.
Elliott, Scott, et al. 2018. “Does Marine Surface Tension Have Global Biogeography? Addition for the
OCEANFILMS Package”. Atmosphere 9, no. 66 ().
Estillore, Armando D, et al. 2017. “Linking hygroscopicity and the surface microstructure of model
inorganic salts, simple and complex carbohydrates, and authentic sea spray aerosol particles”. Phys-
ical Chemistry Chemical Physics 19, no. 3131 (): 21101–21111. doi:10.1039/C7CP04051B. http:
//xlink.rsc.org/?DOI=C7CP04051B.
122
Estillore, Armando D, et al. 2016. “Water Uptake and Hygroscopic Growth of Organosulfate Aerosol”.
Environ. Sci. Technol. 50, no. 88 (): 4259–4268. doi:10.1021/acs.est.5b05014. http://pubs.acs.
org/doi/abs/10.1021/acs.est.5b05014.
Facchini, Maria Cristina, et al. 1999. “Cloud albedo enhancement by surface-active organic solutes in
growing droplets”. Nature 401, no. 67506750 (): 257–259. doi:10.1038/45758. http://www.nature.
com/doifinder/10.1038/45758.
Facchini, Maria Cristina, et al. 2008. “Primary submicron marine aerosol dominated by insoluble
organic colloids and aggregates”. Geophysical Research Letters 35, no. 1717 (): n/a. http://dx.doi.
org/10.1029/2008GL034210.
Facchini, Maria Cristina, et al. 2000. “Surface tension of atmospheric wet aerosol and cloud/fog droplets
in relation to their organic carbon content and chemical composition”. Atmospheric Environment VL
- 34, no. 2828 (): 4853–4857. doi:http://dx.doi.org/10.1016/S1352-2310(00)00237-5. http:
//www.sciencedirect.com/science/article/pii/S1352231000002375.
Fanourgakis, G S, et al. 2019. “Evaluation of global simulations of aerosol particle and cloud conden-
sation nuclei number, with implications for cloud droplet formation”. Atmospheric Chemistry and
Physics 19 (13): 8591–8617.
Fitzgerald, James W. 1991. “Marine aerosols: A review”. Atmospheric Environment. Part A. Gen-
eral Topics 25, no. 33 (): 533–545. http://www.sciencedirect.com/science/article/pii/
096016869190050H.
Flato, G, et al. 2013. “Evaluation of climate models”. Climate Change 2013: The Physical Science
Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental
Panel on Climate Change (): 741–882.
Fletcher, Catherine A, et al. 2007. “Hygroscopic and volatile properties of marine aerosol observed
at Cape Grim during the P2P campaign”. Environmental Chemistry 4, no. 33 (): 162–171. doi:10.
1071/EN07011. http://www.publish.csiro.au/view/journals/dsp_journal_fulltext.cfm?
nid=188&f=EN07011.
Forestieri, S D, et al. 2018. “Establishing the impact of model surfactants on cloud condensation nuclei
activity of sea spray aerosol mimics”. Atmospheric Chemistry and Physics 18, no. 1515 (): 10985–
11005. doi:10.5194/acp-18-10985-2018. https://www.atmos-chem-phys.net/18/10985/2018/.
Fossum, Kirsten N, et al. 2018. “Summertime Primary and Secondary Contributions to Southern Ocean
Cloud Condensation Nuclei”. Scientific Reports 8, no. 11 (): 13844. doi:10.1038/s41598-018-32047-
4. https://www-nature-com/articles/s41598-018-32047-4.
Frossard, Amanda A, and Lynn M Russell. 2012. “Removal of Sea Salt Hydrate Water from Seawater-
Derived Samples by Dehydration”. Environmental science amp; technology 46, no. 2424 (): 13326–
13333. doi:10.1021/es3032083. http://pubs.acs.org/doi/abs/10.1021/es3032083.
Frossard, Amanda A, et al. 2014a. “Side-by-Side Comparison of Four Techniques Explains the Apparent
Differences in the Organic Composition of Generated and Ambient Marine Aerosol Particles”. Aerosol
Science and Technology 48, no. 33 (): v. doi:10.1080/02786826.2013.879979. http://www.
tandfonline.com/doi/abs/10.1080/02786826.2013.879979.
123
Frossard, Amanda A, et al. 2014b. “Sources and composition of submicron organic mass in marine
aerosol particles”. Journal of Geophysical Research: Atmospheres 119, no. 2222 (): 2014JD021913.
doi:10.1002/2014JD021913. http://dx.doi.org/10.1002/2014JD021913.
Fuentes, E, et al. 2010a. “Laboratory-generated primary marine aerosol via bubble-bursting and atom-
ization”. Atmospheric Measurement Techniques 3, no. 11 (): 141–162. doi:10.5194/amt-3-141-2010.
http://www.atmos-meas-tech.net/3/141/2010/.
– . 2010b. “On the impacts of phytoplankton-derived organic matter on the properties of the primary
marine aerosol – Part 1: Source fluxes”. Atmospheric Chemistry and Physics 10, no. 1919 (): 9295–
9317. doi:10.5194/acp-10-9295-2010. http://www.atmos-chem-phys.net/10/9295/2010/.
Fuentes, E, et al. 2011. “On the impacts of phytoplankton-derived organic matter on the properties of
the primary marine aerosol – Part 2: Composition, hygroscopicity and cloud condensation activity”.
Atmospheric Chemistry and Physics 11, no. 66 (): 2585–2602. doi:10.5194/acp-11-2585-2011.
http://www.atmos-chem-phys.net/11/2585/2011/.
Gantt, B, and N Meskhidze. 2013. “The physical and chemical characteristics of marine primary organic
aerosol: a review”. Atmospheric Chemistry and Physics 13, no. 88 (): 3979–3996. doi:10.5194/acpd-
12-21779-2012. http://www.atmos-chem-phys.net/13/3979/2013/acp-13-3979-2013.pdf.
Gantt, B, et al. 2012a. “Global distribution and climate forcing of marine organic aerosol – Part 2:
Effects on cloud properties and radiative forcing”. Atmospheric Chemistry and Physics 12, no. 1414
(): 6555–6563. doi:10.5194/acp-12-6555-2012. http://www.atmos-chem-phys.net/12/6555/
2012/acp-12-6555-2012.pdf.
Gantt, B, et al. 2012b. “Model evaluation of marine primary organic aerosol emission schemes”. At-
mospheric Chemistry and Physics 12, no. 1818 (): 8553–8566. doi:10.5194/acp-12-8553-2012.
http://www.atmos-chem-phys.net/12/8553/2012/.
Gantt, B, et al. 2011. “Wind speed dependent size-resolved parameterization for the organic mass
fraction of sea spray aerosol”. Atmospheric Chemistry and Physics 11, no. 1616 (): 8777–8790. http:
//www.atmos-chem-phys.net/11/8777/2011/acp-11-8777-2011.pdf.
Gao, Qiuju. 2012. Marine biogenic polysaccharides as a potential source of aerosol in the high Arctic :
Towards a link between marine biology and cloud formation. http://urn.kb.se/resolve?urn=urn:
nbn:se:su:diva-72433.
Glassmeier, F, et al. 2019. “An emulator approach to stratocumulus susceptibility”. Atmospheric Chem-
istry and Physics Discussions 2019 (): 1–20. doi:10.5194/acp-2018-1342. https://www.atmos-
chem-phys-discuss.net/acp-2018-1342/.
Good, N, et al. 2010. “Consistency between parameterisations of aerosol hygroscopicity and CCN
activity during the RHaMBLe discovery cruise”. Atmospheric Chemistry and Physics 10, no. 77 ():
3189–3203. http://www.atmos-chem-phys.net/10/3189/2010/acp-10-3189-2010.pdf.
Gras, J L. 1990. “Baseline atmospheric condensation nuclei at Cape Grim 1977–1987”. Journal of
Atmospheric Chemistry 11, no. 11 (): 89–106. doi:10.1007/BF00053669. http://dx.doi.org/10.
1007/BF00053669.
– . 1995. “CN, CCN and particle size in Southern Ocean air at Cape Grim”. Atmospheric Research 35,
no. 22 (): 233–251. http://www.sciencedirect.com/science/article/pii/0169809594000215.
124
– . 2009. “Postfrontal nanoparticles at Cape Grim: impact on cloud nuclei concentrations”. Environ-
mental Chemistry 6, no. 66 (): 515–523. doi:10.1071/EN09076. http://dx.doi.org/10.1071/
EN09076.
Gras, J L, and G P Ayers. 1983. “Marine aerosol at southern mid-latitudes”. J. Geophys. Res 88 ():
10661–10666. doi:10.1029/JC088iC15p10661. http://dx.doi.org/10.1029/JC088iC15p10661.
Gras, J L, and M Keywood. 2016. “Cloud condensation Nuclei over the Southern Ocean: wind depen-
dence and seasonal cycles”. Atmospheric Chemistry and Physics Discussions 2016 (): 1–26. doi:10.
5194/acp-2016-998. http://www.atmos-chem-phys-discuss.net/acp-2016-998/.
– . 2017. “Cloud condensation nuclei over the Southern Ocean: wind dependence and seasonal cycles”.
Atmospheric Chemistry and Physics 17, no. 77 (): 4419–4432. doi:10.5194/acp-17-4419-2017.
http://www.atmos-chem-phys.net/17/4419/2017/.
Gras, J L, et al. 2009. “Postfrontal nanoparticles at Cape Grim: observations”. Environmental Chemistry
6, no. 66 (): 508–514. doi:10.1071/EN09075. http://dx.doi.org/10.1071/EN09075.
Grythe, H, et al. 2014. “A review of sea-spray aerosol source functions using a large global set of sea salt
aerosol concentration measurements”. Atmospheric Chemistry and Physics 14, no. 33 (): 1277–1297.
doi:10.5194/acp-14-1277-2014. http://www.atmos-chem-phys.net/14/1277/2014/.
Gysel, M, G B McFiggans, and H Coe. 2009. “Inversion of tandem differential mobility analyser
(TDMA) measurements”. Journal of Aerosol Science 40, no. 22 (): 134–151. doi:10 . 1016 / j .
jaerosci.2008.07.013. http://linkinghub.elsevier.com/retrieve/pii/S0021850208001778.
Hartz, Kara E Huff, et al. 2006. “Cloud condensation nuclei activation of limited solubility organic
aerosol”. Atmospheric Environment 40 (4): 605–617.
Hawkins, L N, et al. 2010. “Carboxylic acids, sulfates, and organosulfates in processed continental
organic aerosol over the southeast Pacific Ocean during VOCALS-REx 2008”. Journal of geophysical
research 115 (): 8350. doi:10.1029/2009JD013276. http://doi.wiley.com/10.1029/2009JD013276.
Heintzenberg, J, D C Covert, and R Van Dingenen. 2000. “Size distribution and chemical composition
of marine aerosols: a compilation and review”. Tellus B 52, no. 44 (). http://www.tellusb.net/
index.php/tellusb/article/view/17090.
Hersey, S P, et al. 2009. “Aerosol hygroscopicity in the marine atmosphere: a closure study using high-
time-resolution, multiple-RH DASH-SP and size-resolved C-ToF-AMS data”. Atmospheric Chemistry
and Physics 9, no. 77 (): 2543–2554. doi:10.5194/acp-9-2543-2009. http://www.atmos-chem-
phys.net/9/2543/2009/.
Hoppel, W A, et al. 1990. “Aerosol size distributions and optical properties found in the marine bound-
ary layer over the Atlantic Ocean”. J. Geophys. Res 95 (): 3659–3686. doi:10.1029/JD095iD04p03659.
http://dx.doi.org/10.1029/JD095iD04p03659.
Hori, Masahiro, et al. 2003. “Activation capability of water soluble organic substances as CCN”. Journal
of Aerosol Science 34 (4): 419–448.
Hu, Chuanmin, Zhongping Lee, and Bryan Franz. 2012. “Chlorophyll aalgorithms for oligotrophic
oceans: A novel approach based on three-band reflectance difference”. J. Geophys. Res 117 (): 5545.
doi:10.1029/2011JC007395. http://doi.wiley.com/10.1029/2011JC007395.
125
Huang, Yi, et al. 2014. “An Evaluation of WRF Simulations of Clouds over the Southern Ocean with
A-Train Observations”. Monthly Weather Review 142, no. 22 (): 647–667. doi:10.1175/MWR-D-13-
00128.1. http://journals.ametsoc.org/doi/abs/10.1175/MWR-D-13-00128.1.
Hwang, Yen-Ting, and Dargan M W Frierson. 2013. “Link between the double-Intertropical Conver-
gence Zone problem and cloud biases over the Southern Ocean”. Proceedings of the National Academy
of Sciences 110, no. 1313 (): 4935–4940. doi:10.1073/pnas.1213302110. http://www.pnas.org/
content/110/13/4935.full.
Hyder, Patrick, et al. 2018. “Critical Southern Ocean climate model biases traced to atmospheric model
cloud errors”. Nature Communications 9, no. 1 (): 3625.
Jimi, S I, et al. 2007. “A short climatology of nanoparticles at the Cape Grim Baseline Air Pollution
Station, Tasmania”. Environmental Chemistry 4, no. 55 (): 301–309. doi:10.1071/EN07038. http:
//www.publish.csiro.au/?paper=EN07038.
Johnson, G R, Z Ristovski, and L Morawska. 2004. “Method for measuring the hygroscopic behaviour
of lower volatility fractions in an internally mixed aerosol”. Journal of Aerosol Science 35, no. 44 ():
443–455. doi:10.1016/j.jaerosci.2003.10.008. http://linkinghub.elsevier.com/retrieve/
pii/S0021850203004464.
Johnson, G R, et al. 2008. “A robust, portable H-TDMA for field use”. Journal of Aerosol Science 39,
no. 1010 (): 850–861. doi:10.1016/j.jaerosci.2008.05.005. http://linkinghub.elsevier.com/
retrieve/pii/S0021850208000992.
Kay, Jennifer E, et al. 2016. “Global Climate Impacts of Fixing the Southern Ocean Shortwave Ra-
diation Bias in the Community Earth System Model (CESM)”. Journal of Climate 29, no. 12 ():
4617–4636.
Keene, William C, et al. 2007. “Chemical and physical characteristics of nascent aerosols produced
by bursting bubbles at a model airsea interface”. J. Geophys. Res 112 (): D21202. doi:10.1029/
2007JD008464. http://onlinelibrary.wiley.com/doi/10.1029/2007JD008464/full.
King, Stephanie M, et al. 2013. “Investigating Primary Marine Aerosol Properties: CCN Activity of Sea
Salt and Mixed Inorganic–Organic Particles”. Environ. Sci. Technol. 46, no. 1919 (): 10405–10412.
doi:doi:10.1021/es300574u. http://dx.doi.org/10.1021/es300574u.
Koehler, K A, et al. 2006. “Water activity and activation diameters from hygroscopicity data - Part II:
Application to organic species”. Atmospheric Chemistry and Physics 6 (3): 795–809.
Koponen, Ismo K, et al. 2002. “Number size distributions and concentrations of marine aerosols: Ob-
servations during a cruise between the English Channel and the coast of Antarctica”. J. Geophys.
Res 107 (): 4753. doi:10.1029/2002JD002533. http://dx.doi.org/10.1029/2002JD002533.
Langley, L, et al. 2010. “Contributions from DMS and ship emissions to CCN observed over the
summertime North Pacific”. Atmospheric Chemistry and Physics 10, no. 33 (): 1287–1314. doi:10.
5194/acp-10-1287-2010. https://www.atmos-chem-phys.net/10/1287/2010/.
Laskin, Alexander, et al. 2012. “Tropospheric chemistry of internally mixed sea salt and organic par-
ticles: Surprising reactivity of NaCl with weak organic acids”. J. Geophys. Res 117 (): D15302.
doi:10.1029/2012JD017743. http://dx.doi.org/10.1029/2012JD017743.
126
Law, C S, et al. 2017a. “An Overview of the Surface Ocean Aerosol Production (SOAP) campaign”.
Atmospheric Chemistry and Physics Discussions 2017 (): 1–42. doi:10.5194/acp-2017-535. https:
//www.atmos-chem-phys-discuss.net/acp-2017-535/.
– . 2017b. “Overview and preliminary results of the Surface Ocean Aerosol Production (SOAP) cam-
paign”. Atmospheric Chemistry and Physics 17, no. 2222 (): 13645–13667. doi:10.5194/acp-17-
13645-2017. https://www.atmos-chem-phys.net/17/13645/2017/.
Lawler, M J, et al. 2014. “Composition of 15–85 nm particles in marine air”. Atmospheric Chemistry
and Physics 14, no. 2121 (): 11557–11569. doi:10.5194/acp-14-11557-2014. http://www.atmos-
chem-phys.net/14/11557/2014/.
Leck, C, and E K Bigg. 2010. “New particle formation of marine biological origin”. Aerosol Science and
Technology 44, no. 77 (): 570–577. http://www.tandfonline.com/doi/abs/10.1080/02786826.
2010.481222.
Leck, C, and E Svensson. 2015. “Importance of aerosol composition and mixing state for cloud droplet
activation over the Arctic pack ice in summer”. Atmospheric Chemistry and Physics 15, no. 55 ():
2545–2568. doi:10.5194/acp-15-2545-2015. http://www.atmos-chem-phys.net/15/2545/2015/.
Leck, Caroline, and E Keith Bigg. 2005a. “Biogenic particles in the surface microlayer and overlaying
atmosphere in the central Arctic Ocean during summer”. Tellus B 57, no. 44 (): 305–316. doi:10.
1111/j.1600-0889.2005.00148.x. http://dx.doi.org/10.1111/j.1600-0889.2005.00148.x.
– . 2005b. “Source and evolution of the marine aerosol—A new perspective”. Geophysical Research
Letters 32, no. 1919 (): n/a. http://dx.doi.org/10.1029/2005GL023651.
Lee, Christopher, et al. 2015. “Advancing Model Systems for Fundamental Laboratory Studies of Sea
Spray Aerosol Using the Microbial Loop”. The Journal of Physical Chemistry A 119, no. 3333 ():
8860–8870. doi:10.1021/acs.jpca.5b03488. http://pubs.acs.org/doi/abs/10.1021/acs.jpca.
5b03488.
Lee, L A, et al. 2012. “Mapping the uncertainty in global CCN using emulation”. Atmospheric Chemistry
and Physics 12, no. 2020 (): 9739–9751. doi:10.5194/acp-12-9739-2012. http://www.atmos-
chem-phys.net/12/9739/2012/.
Leeuw, Gerrit de, et al. 2011. “Production flux of sea spray aerosol”. 49, no. 22 (): n/a. doi:10.1029/
2010RG000349. http://dx.doi.org/10.1029/2010RG000349.
Lewis, E R, and S E Schwartz. 2004. “Sea salt aerosol production: mechanisms, methods, measure-
ments and models: a critical review”. 152 (). http://books.google.com/books?hl=en&
lr=&id=AEIo8ebgw00C&oi=fnd&pg=PR9&dq=Lewis+Sea+Salt+Aerosol+
ProductionMechanisms+Methods+Measurements+andModelsA+Critical+Review&ots=eM-
s1ofOc4&sig=oSyDraKXpBvDiBA8EMuKKDw-GGI.
Lightstone, James M, et al. 2000. “Deliquescence, Efflorescence, and Water Activity in Ammonium Ni-
trate and Mixed Ammonium Nitrate/Succinic Acid Microparticles”. The Journal of Physical Chem-
istry A 104, no. 41 (): 9337–9346. doi:10.1021/jp002137h.
Limpert, Eckhard, Werner A Stahel, and Markus Abbt. 2001. “Log-normal Distributions across the
Sciences: Keys and Clues: On the charms of statistics, and how mechanical models resembling gam-
bling machines offer a link to a handy way to characterize log-normal distributions, which can
127
provide deeper insight into variability and probability—normal or log-normal: That is the question”.
BioScience 51, no. 5 (): 341–352.
Long, M S, et al. 2011. “A sea-state based source function for size- and composition-resolved marine
aerosol production”. Atmospheric Chemistry and Physics 11, no. 33 (): 1203–1216. doi:10.5194/acp-
11-1203-2011. http://www.atmos-chem-phys.net/11/1203/2011/.
Long, M S, et al. 2014. “Light-enhanced primary marine aerosol production from biologically productive
seawater”. Geophysical Research Letters 41, no. 77 (): 2661–2670. doi:10.1002/2014GL059436. http:
//dx.doi.org/10.1002/2014GL059436.
Mace, Gerald G, and Alain Protat. 2018a. “Clouds over the Southern Ocean as Observed from the
R/V Investigatorduring CAPRICORN. Part I: Cloud Occurrence and Phase Partitioning”. J. Appl.
Meteor. Climatol. 57, no. 88 (): 1783–1803. doi:10.1175/JAMC-D-17-0194.1. http://journals.
ametsoc.org/doi/10.1175/JAMC-D-17-0194.1.
– . 2018b. “Clouds over the Southern Ocean as Observed from the R/V Investigatorduring CAPRI-
CORN. Part II: The Properties of Nonprecipitating Stratocumulus”. J. Appl. Meteor. Climatol. 57,
no. 88 (): 1805–1823. doi:10.1175/JAMC-D-17-0195.1. http://journals.ametsoc.org/doi/10.
1175/JAMC-D-17-0195.1.
Mallet, Marc, et al. 2016. “Sea spray aerosol in the Great Barrier Reef and the presence of non-
volatile organics”. Journal of Geophysical Research: Atmospheres (): 2016JD024966. doi:10.1002/
2016JD024966. http://onlinelibrary.wiley.com/doi/10.1002/2016JD024966/full.
Mann, G W, et al. 2010. “Description and evaluation of GLOMAP-mode: a modal global aerosol
microphysics model for the UKCA composition-climate model”. Geoscientific Model Development 3,
no. 22 (): 519–551. doi:10.5194/gmd-3-519-2010. http://www.geosci-model-dev.net/3/519/
2010/.
Maria, S F, et al. 2003. “Source signatures of carbon monoxide and organic functional groups in
Asian Pacific Regional Aerosol Characterization Experiment (ACEAsia) submicron aerosol types”.
J. Geophys. Res 108 (): n/a. doi:10.1029/2003JD003703. http://onlinelibrary.wiley.com/doi/
10.1029/2003JD003703/full.
Marshall, Julia, et al. 2005. “Optical Properties of Aerosol Particles over the Northeast Pacific”. J.
Appl. Meteor. 44, no. 88 (): 1206–1220. doi:10.1175/JAM2267.1. http://journals.ametsoc.org/
doi/abs/10.1175/JAM2267.1.
Maßling, A, et al. 2007. “Hygroscopic growth of sub-micrometer and one-micrometer aerosol particles
measured during ACE-Asia”. Atmospheric Chemistry and Physics 7, no. 1212 (): 3249–3259. http:
//www.atmos-chem-phys.net/7/3249/2007/acp-7-3249-2007.pdf.
McCoy, D T, et al. 2015. “Natural aerosols explain seasonal and spatial patterns of Southern Ocean
cloud albedo”. Science Advances 1, no. 66 (): e1500157. doi:10 . 1126 / sciadv . 1500157. http :
//advances.sciencemag.org/cgi/doi/10.1126/sciadv.1500157.
McFiggans, G, et al. 2006. “The effect of physical and chemical aerosol properties on warm cloud
droplet activation”. Atmospheric Chemistry and Physics 6, no. 99 (): 2593–2649. doi:10.5194/acp-
6-2593-2006. http://www.atmos-chem-phys.net/6/2593/2006/.
McMurry, Peter H, and Mark R Stolzenburg. 1989. “On the sensitivity of particle size to relative hu-
midity for Los Angeles aerosols”. Atmospheric Environment (1967) 23, no. 22 (): 497–507. doi:http:
128
//dx.doi.org/10.1016/0004-6981(89)90593-3. http://www.sciencedirect.com/science/
article/pii/0004698189905933.
Merikanto, J, et al. 2009. “Impact of nucleation on global CCN”. Atmospheric Chemistry and Physics
9, no. 2121 (): 8601–8616. doi:10.5194/acp-9-8601-2009. http://www.atmos-chem-phys.net/9/
8601/2009/.
Meskhidze, Nicholas, et al. 2013. “Production mechanisms, number concentration, size distribution,
chemical composition, and optical properties of sea spray aerosols”. Atmospheric Science Letters 14,
no. 44 (): 207–213. doi:10.1002/asl2.441. http://dx.doi.org/10.1002/asl2.441.
Mifflin, A L, M L Smith, and S T Martin. 2009. “Morphology hypothesized to influence aerosol particle
deliquescence”. Physical Chemistry Chemical Physics 11, no. 4343 (): 10095–10107. doi:10.1039/
b910432a. http://dx.doi.org/10.1039/B910432A.
Millero, Frank J, et al. 2008. “The composition of Standard Seawater and the definition of the Reference-
Composition Salinity Scale”. Deep Sea Research Part I: Oceanographic Research Papers 55, no. 11
(): 50–72. http://www.sciencedirect.com/science/article/pii/S0967063707002282.
Miyazaki, Yuzo, Kimitaka Kawamura, and Maki Sawano. 2010. “Size distributions and chemical char-
acterization of water-soluble organic aerosols over the western North Pacific in summer”. J. Geophys.
Res 115 (): n/a. http://dx.doi.org/10.1029/2010JD014439.
Modini, R L, B Harris, and Z Ristovski. 2010. “The organic fraction of bubble-generated, accumulation
mode Sea Spray Aerosol (SSA)”. Atmospheric Chemistry and Physics 10, no. 66 (): 2867–2877. http:
//eprints.qut.edu.au/38660/.
Modini, R L, et al. 2013. “Effect of soluble surfactant on bubble persistence and bubble-produced
aerosol particles”. Journal of Geophysical Research: Atmospheres 118, no. 33 (): 1388–1400. doi:10.
1002/jgrd.50186. http://dx.doi.org/10.1002/jgrd.50186.
Modini, R L, et al. 2015. “Primary marine aerosolcloud interactions off the coast of California”. Journal
of Geophysical Research: Atmospheres 120, no. 99 (): 4282–4303. doi:10.1002/2014JD022963. http:
//onlinelibrary.wiley.com/doi/10.1002/2014JD022963/full.
Modini, Robin L, et al. 2010. “Observation of the suppression of water uptake by marine particles”.
Atmospheric Research 98, no. 22 (): 219–228. http://www.sciencedirect.com/science/article/
pii/S0169809510000748.
Modini, Robin Lewis. 2010. Investigation of the effect of organics on the water uptake of marine
aerosols. https://eprints.qut.edu.au/46884/.
Moreau, Sébastien, et al. 2017. “Eddy-induced carbon transport across the Antarctic Circumpolar
Current”. Global Biogeochemical Cycles 31, no. 99 (): 1368–1386. doi:10.1002/2017GB005669. http:
//doi.wiley.com/10.1002/2017GB005669.
Myhre, G, et al. 2013. “Anthropogenic and natural radiative forcing” (). doi:10.1002/2013GL059099/
full. http://scholar.google.com/scholar?q=related:hBk07QcI_TYJ:scholar.google.com/
&hl=en&num=20&as_sdt=0,5&as_ylo=2013&as_yhi=2013.
Niedermeier, D, et al. 2008. “LACIS-measurements and parameterization of sea-salt particle hygro-
scopic growth and activation”. Atmospheric Chemistry and Physics 8, no. 33 (): 579–590. doi:10.
5194/acp-8-579-2008. http://www.atmos-chem-phys.net/8/579/2008/.
129
Nozière, Barbara, Christine Baduel, and Jean-Luc Jaffrezo. 2014. “The dynamic surface tension of
atmospheric aerosol surfactants reveals new aspects of cloud activation”. 5 SP - (). doi:10.1038/
ncomms4335. http://dx.doi.org/10.1038/ncomms4335.
O’Dowd, Colin D, Jason A Lowe, and Michael H Smith. 1999. “Observations and modelling of aerosol
growth in marine stratocumulus—case study”. Atmospheric Environment VL - 33, no. 18 (): 3053–
3062.
Oppo, C, et al. 1999. “Surfactant components of marine organic matter as agents for biogeochemical
fractionation and pollutant transport via marine aerosols”. Marine Chemistry 63, no. 33 (): 235–253.
http://www.sciencedirect.com/science/article/pii/S0304420398000656.
Orellana, Mónica V, et al. 2011. “Marine microgels as a source of cloud condensation nuclei in the
high Arctic”. Proceedings of the National Academy of Sciences 108, no. 3333 (): 13612–13617. http:
//www.pnas.org/content/108/33/13612.abstract.
Ott, Wayne R. 2012. “A Physical Explanation of the Lognormality of Pollutant Concentrations”. Jour-
nal of the Air & Waste Management Association 40, no. 10 (): 1378–1383.
Ovadnevaite, J, et al. 2014. “A sea spray aerosol flux parameterization encapsulating wave state”.
Atmospheric Chemistry and Physics 14, no. 44 (): 1837–1852. doi:10.5194/acp-14-1837-2014.
http://www.atmos-chem-phys.net/14/1837/2014/.
Ovadnevaite, Jurgita, et al. 2011a. “Detecting high contributions of primary organic matter to marine
aerosol: A case study”. Geophysical Research Letters 38, no. 22 (): n/a. http://dx.doi.org/10.
1029/2010GL046083.
Ovadnevaite, Jurgita, et al. 2012. “On the effect of wind speed on submicron sea salt mass con-
centrations and source fluxes”. J. Geophys. Res 117 (): n/a. doi:10.1029/2011JD017379. http:
//dx.doi.org/10.1029/2011JD017379.
Ovadnevaite, Jurgita, et al. 2011b. “Primary marine organic aerosol: A dichotomy of low hygroscop-
icity and high CCN activity”. Geophysical Research Letters 38, no. 2121 (): L21806. doi:10.1029/
2011GL048869. http://dx.doi.org/10.1029/2011GL048869.
Ovadnevaite, Jurgita, et al. 2017. “Surface tension prevails over solute effect in organic-influenced
cloud droplet activation”. Nature 546, no. 76607660 (): 637–641. doi:10.1038/nature22806. http:
//www.nature.com/doifinder/10.1038/nature22806.
O’Dowd, Colin, et al. 2015. “Connecting marine productivity to sea-spray via nanoscale biological
processes: Phytoplankton Dance or Death Disco?” Scientific Reports 5 (): 14883. doi:10.1038/
srep14883. http://www.nature.com/articles/srep14883.
O’Dowd, Colin D, et al. 2004. “Biogenically driven organic contribution to marine aerosol”. Nature
431, no. 70097009 (): 676–680. doi:10.1038/nature02959. http://www.nature.com/doifinder/
10.1038/nature02959.
Peng, Changgeng, Man Nin Chan, and Chak K Chan. 2001. “The Hygroscopic Properties of Dicar-
boxylic and Multifunctional Acids: Measurements and UNIFAC Predictions”. Environmental science
& technology 35, no. 22 (): 4495–4501. doi:10.1021/es0107531.
Petters, M D, and S M Kreidenweis. 2007. “A single parameter representation of hygroscopic growth
and cloud condensation nucleus activity”. Atmospheric Chemistry and Physics 7, no. 88 (): 1961–
1971. http://hal-insu.archives-ouvertes.fr/hal-00296196/.
130
– . 2013. “A single parameter representation of hygroscopic growth and cloud condensation nucleus
activity amp;ndash; Part 3: Including surfactant partitioning”. Atmospheric Chemistry and Physics
13, no. 22 (): 1081–1091. doi:10.5194/acp-13-1081-2013. http://www.atmos-chem-phys.net/
13/1081/2013/.
Petters, Sarah Suda, and Markus Dirk Petters. 2016. “Surfactant effect on cloud condensation nuclei
for two-component internally mixed aerosols”. Journal of Geophysical Research: Atmospheres 121,
no. 44 (): 1878–1895. doi:10.1002/2015JD024090. http://doi.wiley.com/10.1002/2015JD024090.
Pradeep Kumar, P, K Broekhuizen, and J P D Abbatt. 2003. “Organic acids as cloud condensation nu-
clei: Laboratory studies of highly soluble and insoluble species”. Atmospheric Chemistry and Physics
3 (3): 509–520.
Prather, Kimberly A, et al. 2013. “Bringing the ocean into the laboratory to probe the chemical
complexity of sea spray aerosol”. Proceedings of the National Academy of Sciences 110, no. 1919 ():
7550–7555. http://www.pnas.org/content/110/19/7550.abstract.
Prenni, Anthony J, Paul J DeMott, and Sonia M Kreidenweis. 2003. “Water uptake of internally mixed
particles containing ammonium sulfate and dicarboxylic acids”. Atmospheric Environment VL - 37
(30): 4243–4251.
Prenni, Anthony J, et al. 2007. “Cloud droplet activation of secondary organic aerosol”. J. Geophys.
Res 112 (): D10223. doi:10.1029/2006JD007963. http://dx.doi.org/10.1029/2006JD007963.
Prisle, N L, et al. 2010. “Surfactants in cloud droplet activation: mixed organic-inorganic particles”.
Atmospheric Chemistry and Physics 10, no. 1212 (): 5663–5683. doi:10.5194/acp-10-5663-2010.
https://www.atmos-chem-phys.net/10/5663/2010/.
Protat, Alain, et al. 2017. “Shipborne observations of the radiative effect of Southern Ocean clouds”.
Journal of Geophysical Research: Atmospheres 122 (1): 318–328. doi:10.1002/2016JD026061. https:
//agupubs.onlinelibrary.wiley.com/doi/abs/10.1002/2016JD026061.
Putaud, J P, et al. 2000. “Chemical mass closure and assessment of the origin of the submicron aerosol
in the marine boundary layer and the free troposphere at Tenerife during ACE-2”. Tellus B 52, no.
22 (): 141–168. doi:10.1034/j.1600-0889.2000.00056.x. http://dx.doi.org/10.1034/j.1600-
0889.2000.00056.x.
Quinn, P K, and T S Bates. 2012. “The case against climate regulation via oceanic phytoplankton
sulphur emissions”. Nature 480, no. 73757375 (): 51–56. doi:10.1038/nature10580. http://dx.
doi.org/10.1038/nature10580.
Quinn, P K, and D J Coffman. 1998. “Local closure during the First Aerosol Characterization Exper-
iment (ACE 1): Aerosol mass concentration and scattering and backscattering coefficients”. J. Geo-
phys. Res 103 (): 16575–16596. doi:10.1029/97JD03757. http://dx.doi.org/10.1029/97JD03757.
Quinn, P K, et al. 2017. “Small fraction of marine cloud condensation nuclei made up of sea spray
aerosol”. Nature Geoscience 30 (): 869. doi:10 . 1038 / ngeo3003. http : / / www . nature . com /
doifinder/10.1038/ngeo3003.
Quinn, Patricia K, et al. 2015. “Chemistry and Related Properties of Freshly Emitted Sea Spray
Aerosol”. Chemical Reviews (): 150406123611007. doi:10.1021/cr500713g. http://pubs.acs.org/
doi/abs/10.1021/cr500713g.
131
Quinn, Patricia K, et al. 2014. “Contribution of sea surface carbon pool to organic matter enrichment
in sea spray aerosol”. Nature Geoscience 7, no. 33 (): 228–232. doi:10 . 1038 / ngeo2092. http :
//dx.doi.org/10.1038/ngeo2092.
Rader, D J, and P H Mcmurry. 1986. “Application of the tandem differential mobility analyzer to
studies of droplet growth or evaporation”. Journal of Aerosol Science 17, no. 55 (): 771–787. doi:http:
//dx.doi.org/10.1016/0021-8502(86)90031-5. http://www.sciencedirect.com/science/
article/pii/0021850286900315.
Raes, Eric J, et al. 2018. “Oceanographic boundaries constrain microbial diversity gradients in the
South Pacific Ocean”. Proceedings of the National Academy of Sciences 2 (): 201719335. doi:10.
1073/pnas.1719335115. http://www.pnas.org/lookup/doi/10.1073/pnas.1719335115.
Raina, Jean-Baptiste, et al. 2013. “DMSP biosynthesis by an animal and its role in coral thermal stress
response”. Nature 502, no. 74737473 (): 677–680. doi:10.1038/nature12677. http://dx.doi.org/
10.1038/nature12677.
Rasmussen, Berit Brøndum, et al. 2017. “What controls volatility of sea spray aerosol? Results from
laboratory studies using artificial and real seawater samples”. Journal of Aerosol Science 107 ():
134–141. doi:10.1016/j.jaerosci.2017.02.002. http://linkinghub.elsevier.com/retrieve/
pii/S0021850216303810.
Rinaldi, M, et al. 2009. “On the representativeness of coastal aerosol studies to open ocean studies:
Mace Head – a case study”. Atmospheric Chemistry and Physics 9, no. 2424 (): 9635–9646. doi:10.
5194/acp-9-9635-2009. http://www.atmos-chem-phys.net/9/9635/2009/.
Rinaldi, Matteo, et al. 2013. “Is chlorophyll-a the best surrogate for organic matter enrichment in
submicron primary marine aerosol?” Journal of Geophysical Research: Atmospheres 118, no. 1010
(): 4964–4973. doi:10.1002/jgrd.50417. http://dx.doi.org/10.1002/jgrd.50417.
Rinaldi, Matteo, et al. 2010. “Primary and Secondary Organic Marine Aerosol and Oceanic Biological
Activity: Recent Results and New Perspectives for Future Studies”. Advances in Meteorology 2010
(). doi:10.1155/2010/310682. http://dx.doi.org/10.1155/2010/310682.
Roberts, G C, and A Nenes. 2005. “A Continuous-Flow Streamwise Thermal-Gradient CCN Chamber
for Atmospheric Measurements”. Aerosol Science and Technology 39, no. 33 (): 206–221. doi:10.
1080/027868290913988. http://www.tandfonline.com/doi/abs/10.1080/027868290913988.
Rose, C, et al. 2015. “Airborne measurements of new particle formation in the free troposphere above
the Mediterranean Sea during the HYMEX campaign”. Atmospheric Chemistry and Physics 15, no.
1717 (): 10203–10218. doi:10.5194/acp-15-10203-2015. http://www.atmos-chem-phys.net/15/
10203/2015/.
Rose, C, et al. 2017. “CCN production by new particle formation in the free troposphere”. Atmospheric
Chemistry and Physics 17, no. 22 (): 1529–1541. doi:10.5194/acp-17-1529-2017. http://www.
atmos-chem-phys.net/17/1529/2017/.
Rosenfeld, Daniel, et al. 2019. “Aerosol-driven droplet concentrations dominate coverage and water of
oceanic low-level clouds”. Science 363, no. 64276427 (): eaav0566. http://science.sciencemag.
org/content/363/6427/eaav0566.abstract.
132
Ruehl, C R, J F Davies, and K R Wilson. 2016. “An interfacial mechanism for cloud droplet formation
on organic aerosols”. Science 351, no. 62806280 (): 1447–1450. doi:10.1126/science.aad4889.
http://www.sciencemag.org/cgi/doi/10.1126/science.aad4889.
Ruehl, Christopher R, and Kevin R Wilson. 2014. “Surface Organic Monolayers Control the Hygro-
scopic Growth of Submicrometer Particles at High Relative Humidity”. The Journal of Physical
Chemistry A 118, no. 2222 (): 3952–3966. doi:10.1021/jp502844g. http://pubs.acs.org/doi/10.
1021/jp502844g.
Russell, Lynn M. 2003. “Aerosol Organic-Mass-to-Organic-Carbon Ratio Measurements”. Environmen-
tal science amp; technology 37, no. 1313 (): 2982–2987. doi:10.1021/es026123w. http://pubs.acs.
org/doi/abs/10.1021/es026123w.
Russell, Lynn M, Ranjit Bahadur, and Paul J Ziemann. 2011. “Identifying organic aerosol sources
by comparing functional group composition in chamber and atmospheric particles”. Proceedings of
the National Academy of Sciences (). http://www.pnas.org/content/early/2011/02/09/
1006461108.abstract.
Russell, Lynn M, et al. 2009. “Carbohydrate-like composition of submicron atmospheric particles and
their production from ocean bubble bursting”. Proceedings of the National Academy of Sciences 107,
no. 1515 (): 6652. http://www.pnas.org/content/early/2009/12/23/0908905107.abstract.
Russell, Lynn M, et al. 2013. “Eastern Pacific Emitted Aerosol Cloud Experiment”. Bull. Amer. Meteor.
Soc. 94, no. 55 (): 709–729. doi:10.1175/BAMS-D-12-00015.1. http://journals.ametsoc.org/
doi/abs/10.1175/BAMS-D-12-00015.1.
Sakamoto, Y, M Ishiguro, and G Kitagawa. 1987. “Akaike information criterion statistics.” Mathematics
and Computers in Simulation 29, no. 55 (): 452.
Salter, M E, et al. 2016. “Calcium enrichment in sea spray aerosol particles”. Geophysical Research
Letters 43, no. 1515 (): 8277–8285. doi:10.1002/2016GL070275. http://doi.wiley.com/10.1002/
2016GL070275.
Sanchez, K J, et al. 2017. “Top-down and bottom-up aerosol–cloud closure: towards understand-
ing sources of uncertainty in deriving cloud shortwave radiative flux”. Atmospheric Chemistry and
Physics 17, no. 1616 (): 9797–9814. doi:10.5194/acp-17-9797-2017. https://www.atmos-chem-
phys.net/17/9797/2017/.
Sanchez, Kevin J, et al. 2018. “Substantial Seasonal Contribution of Observed Biogenic Sulfate Particles
to Cloud Condensation Nuclei”. Scientific Reports 8, no. 11 (): 3235. doi:10.1038/s41598-018-
21590-9. https://www-nature-com/articles/s41598-018-21590-9.
Schmitt-Kopplin, P., et al. 2012. “Dissolved organic matter in sea spray: a transfer study from marine
surface water to aerosols”. Biogeosciences 9 (4): 1571–1582. doi:10.5194/bg-9-1571-2012. https:
//www.biogeosciences.net/9/1571/2012/.
Schwier, A N, et al. 2015. “Primary marine aerosol emissions from the Mediterranean Sea during pre-
bloom and oligotrophic conditions: correlations to seawater chlorophyll a from a mesocosm study”.
Atmospheric Chemistry and Physics 15, no. 1414 (): 7961–7976. doi:10.5194/acp-15-7961-2015.
http://www.atmos-chem-phys.net/15/7961/2015/.
Schwier, A N, et al. 2017. “Primary marine aerosol physical flux and chemical composition during a
nutrient enrichment experiment in mesocosms in the Mediterranean Sea”. Atmospheric Chemistry
133
and Physics 17, no. 2323 (): 14645–14660. doi:10.5194/acp-17-14645-2017. https://www.atmos-
chem-phys.net/17/14645/2017/.
Sciare, J, et al. 2009. “Long-term observations of carbonaceous aerosols in the Austral Ocean atmo-
sphere: Evidence of a biogenic marine organic source”. J. Geophys. Res 114 (): n/a. http://dx.doi.
org/10.1029/2009JD011998.
Seinfeld, John H, and Spyros N Pandis. 2006. “Atmospheric Chemistry and Physics : From Air Pollution
to Climate Change” ().
Sellegri, K, et al. 2008. “Role of the volatile fraction of submicron marine aerosol on its hygroscopic prop-
erties”. Atmospheric Research 90, no. 22 (): 272–277. http://www.sciencedirect.com/science/
article/pii/S0169809508000896.
Sellegri, K, et al. 2006. “Surfactants and submicron sea spray generation”. J. Geophys. Res 111 (): n/a.
http://dx.doi.org/10.1029/2005JD006658.
Shank, L M, et al. 2012. “Organic matter and non-refractory aerosol over the remote Southeast Pacific:
oceanic and combustion sources”. Atmospheric Chemistry and Physics 12, no. 11 (): 557–576. doi:10.
5194/acp-12-557-2012. http://www.atmos-chem-phys.net/12/557/2012/.
Sievering, H, et al. 2004. “Aerosol non-sea-salt sulfate in the remote marine boundary layer under clear-
sky and normal cloudiness conditions: Ocean-derived biogenic alkalinity enhances sea-salt sulfate
production by ozone oxidation”. J. Geophys. Res 109 (): D19317. doi:10.1029/2003JD004315. http:
//dx.doi.org/10.1029/2003JD004315.
Simpson, Rebecca M C, et al. 2014. “Dimethyl sulfide: Less important than long-range transport as
a source of sulfate to the remote tropical Pacific marine boundary layer”. Journal of Geophysical
Research: Atmospheres (): n/a. doi:10.1002/2014JD021643. http://dx.doi.org/10.1002/
2014JD021643.
Spracklen, D V, et al. 2005. “A global off-line model of size-resolved aerosol microphysics: I. Model
development and prediction of aerosol properties”. Atmospheric Chemistry and Physics 5, no. 88 ():
2227–2252. doi:10.5194/acp-5-2227-2005. http://www.atmos-chem-phys.net/5/2227/2005/.
Spracklen, Dominick V, et al. 2008. “Contribution of particle formation to global cloud condensation
nuclei concentrations”. 35, no. 66 (): L06808. doi:10.1029/2007GL033038. http://onlinelibrary.
wiley.com/doi/10.1029/2007GL033038/full.
Stokes, M D, et al. 2013. “A Marine Aerosol Reference Tank system as a breaking wave analogue for
the production of foam and sea-spray aerosols”. Atmospheric Measurement Techniques 6, no. 44 ():
1085–1094. doi:10.5194/amt-6-1085-2013. http://www.atmos-meas-tech.net/6/1085/2013/.
Stokes, R H, and R A Robinson. 1966. “Interactions in Aqueous Nonelectrolyte Solutions. I. Solute-
Solvent Equilibria”. The Journal of Physical Chemistry 70, no. 77 (): 2126–2131. doi:10.1021/
j100879a010. http://pubs.acs.org/doi/abs/10.1021/j100879a010.
Swietlicki, E, et al. 1999. “A closure study of sub-micrometer aerosol particle hygroscopic behaviour”.
Atmospheric Research 50, no. 33 (): 205–240. http://www.sciencedirect.com/science/article/
pii/S0169809598001057.
Swietlicki, E, et al. 2000. “Hygroscopic properties of aerosol particles in the northeastern Atlantic during
ACE2”. Tellus B 52, no. 22 (): 201–227. http://onlinelibrary.wiley.com/doi/10.1034/j.1600-
0889.2000.00036.x/abstract.
134
Swietlicki, E, et al. 2008. “Hygroscopic properties of submicrometer atmospheric aerosol particles mea-
sured with H-TDMA instruments in various environments - A review”. Tellus, Series B: Chemical
and Physical Meteorology 60 B, no. 33 (): 432–469. http://www.scopus.com/inward/record.url?
eid=2-s2.0-45449110434&partnerID=40&md5=14d9b1616331ab8e9b71f5174692fd5e.
Tang, I N, and H R Munkelwitz. 1994. “Water activities, densities, and refractive indices of aqueous
sulfates and sodium nitrate droplets of atmospheric importance”. J. Geophys. Res 99, no. D9 ():
18801–18808.
Twohy, C H, et al. 2013. “Impacts of aerosol particles on the microphysical and radiative properties
of stratocumulus clouds over the southeast Pacific Ocean”. Atmospheric Chemistry and Physics 13,
no. 55 (): 2541–2562. doi:10.5194/acp-13-2541-2013. http://www.atmos-chem-phys.net/13/
2541/2013/.
Vaattovaara, P, et al. 2006. “The composition of nucleation and Aitken modes particles during coastal
nucleation events: evidence for marine secondary organic contribution”. Atmospheric Chemistry and
Physics 6, no. 1212 (): 4601–4616. doi:10.5194/acp-6-4601-2006. http://www.atmos-chem-
phys.net/6/4601/2006/.
Vaishya, Aditya, et al. 2013. “Bistable effect of organic enrichment on sea spray radiative properties”.
Geophysical Research Letters (): 2013GL058452.
VanReken, Timothy M, et al. 2005. “Cloud condensation nucleus activation properties of biogenic
secondary organic aerosol”. J. Geophys. Res 110, no. D7 (): D07206.
Vignati, E, et al. 2010. “Global scale emission and distribution of sea-spray aerosol: Sea-salt and organic
enrichment”. Atmospheric Environment VL - 44, no. 55 (): 670–677. http://www.sciencedirect.
com/science/article/pii/S1352231009009571.
Wang, Xiaofei, et al. 2017. “The role of jet and film drops in controlling the mixing state of submicron
sea spray aerosol particles”. Proceedings of the National Academy of Sciences 114, no. 2727 (): 6978–
6983. doi:10.1073/pnas.1702420114. http://www.pnas.org/content/114/27/6978.full.
Webb, M, et al. 2001. “Combining ERBE and ISCCP data to assess clouds in the Hadley Centre,
ECMWF and LMD atmospheric climate models”. Climate Dynamics 17, no. 1212 (): 905–922. doi:10.
1007/s003820100157. http://link.springer.com/10.1007/s003820100157.
Wex, H, et al. 2010. “Influence of the external mixing state of atmospheric aerosol on derived CCN num-
ber concentrations”. Geophysical Research Letters 37, no. 1010 (): L10805. doi:10.1029/2010GL043337.
http://dx.doi.org/10.1029/2010GL043337.
Wex, H, et al. 2009. “Towards closing the gap between hygroscopic growth and activation for secondary
organic aerosol: Part 1 – Evidence from measurements”. Atmospheric Chemistry and Physics 9, no.
1212 (): 3987–3997. doi:10.5194/acp-9-3987-2009. http://www.atmos-chem-phys.net/9/3987/
2009/.
Williams, K D, and M J Webb. 2008. “A quantitative performance assessment of cloud regimes in
climate models”. Climate Dynamics 33, no. 11 (): 141–157. doi:10.1007/s00382- 008- 0443- 1.
http://link.springer.com/article/10.1007/s00382-008-0443-1/fulltext.html.
Woodhouse, M T, et al. 2013. “Sensitivity of cloud condensation nuclei to regional changes in dimethyl-
sulphide emissions”. Atmospheric Chemistry and Physics 13, no. 55 (): 2723–2733. doi:10.5194/acp-
13-2723-2013. http://www.atmos-chem-phys.net/13/2723/2013/.
135
Yoon, Y J, et al. 2007. “Seasonal characteristics of the physicochemical properties of North Atlantic
marine atmospheric aerosols”. J. Geophys. Res 112 (): n/a. http : / / dx . doi . org / 10 . 1029 /
2005JD007044.
Zahorowski, W, S D Chambers, and A Henderson-Sellers. 2004. “Ground based radon-222 observations
and their application to atmospheric studies”. Journal of Environmental Radioactivity 76, no. 11 ():
3–33. http://www.sciencedirect.com/science/article/pii/S0265931X04001183.
Zelenyuk, Alla, et al. 2007. “Measurements and interpretation of the effect of a soluble organic sur-
factant on the density, shape and water uptake of hygroscopic particles”. Journal of Aerosol Science
38, no. 99 (): 903–923. doi:10.1016/j.jaerosci.2007.06.006. http://www.sciencedirect.com/
science/article/pii/S0021850207001036.
Zheng, G, et al. 2018. “Marine boundary layer aerosol in the eastern North Atlantic: seasonal variations
and key controlling processes”. Atmospheric Chemistry and Physics 18, no. 2323 (): 17615–17635.
doi:10.5194/acp-18-17615-2018. https://www.atmos-chem-phys.net/18/17615/2018/.
Zhou, Jingchuan, et al. 2001. “Hygroscopic properties of aerosol particles over the central Arctic
Ocean during summer”. J. Geophys. Res 106 (): 32111–32123. http://dx.doi.org/10.1029/
2000JD900426.
Zieger, P, et al. 2017. “Revising the hygroscopicity of inorganic sea salt particles”. Nature Commu-
nications 8 (): ncomms15883. doi:10.1038/ncomms15883. http://www.nature.com/articles/
ncomms15883.
136
13 Appendix A
Summary of studies using laboratory generated SSA. Source:(Cravigan 2015). Abbreviations for
measurement techniques are:
• Ion chromatography (IC)
• Fourier transform infra-red spectroscopy (FTIR)
• Evolved gas analysis (EGA)
• Total organi carbon (TOC)- used on filter extractions, usually to obtain the WSOC concentration
• Proton nuclear magnetic resonance (HNMR)
• Differential aerosol sizing and hygroscopicity spectrometer probe (DASH-SP)
• Thermal optical analysis (TOA)
• Transmission electron microscopy with energy dispersive X-ray (TEM-EDX)
• Aerosol mass spectrometry (AMS)
• Cloud condensation nuclei counter (CCNc)
• Volatility hygroscopicity tandem differential mobility analyser (VH-TDMA)
137
14 Appendix B
Summary of nascent SSA composition observed from chamber measurements during SOAP voyage.
Note that the organic volume fractions are computed as a fraction of the dry SSA mass, not including
hydrates.
141
