Upload
r
View
215
Download
1
Embed Size (px)
Citation preview
Tre
8.3 Air–Sea Exchange of Marine Trace GasesR Beale, Plymouth Marine Laboratory, Plymouth, UKMT Johnson and PS Liss, University of East Anglia, Norwich, UKPD Nightingale, Plymouth Marine Laboratory, Plymouth, UK
ã 2014 Elsevier Ltd. All rights reserved.
This article is a revision of the previous edition article by P.D. Nightingale, P.S. Liss, volume 6, pp. 1–33, © 2003, Elsevier Ltd.
8.3.1 Introduction 548.3.2 Gas Exchange Processes and Parameterizations 548.3.2.1 Gas Flux Theory 548.3.2.1.1 Mass boundary layers at the interface 548.3.2.1.2 Turbulent forcing: wind speed-based parameterizations of transfer velocity 558.3.2.1.3 Interfacial models and the Schmidt number exponent on the water side 568.3.2.1.4 Air-side transfer 578.3.2.2 Processes Driving Gas Exchange 578.3.2.2.1 The effect of wind 578.3.2.2.2 The effect of waves 578.3.2.2.3 The effect of bubbles and sea spray 588.3.2.2.4 Heat and water fluxes 598.3.2.2.5 The effect of surface films 608.3.2.2.6 Chemical enhancement 618.3.2.2.7 Biological effects 618.3.2.3 Wind Speed Parameterizations 628.3.2.3.1 Global 14C constraint 628.3.2.3.2 Local-scale natural tracer experiments 638.3.2.3.3 Deliberate tracer experiments 648.3.2.3.4 Direct flux measurements 658.3.2.3.5 NOAA/COARE 668.3.2.3.6 Reconciling observations 678.3.2.3.7 Remote sensing of kw 688.3.2.3.8 Air-side transfer velocity 698.3.2.4 Estimating Trace Gas Fluxes in Biogeochemical Studies 698.3.2.4.1 Gas-specific effects on transfer 698.3.2.4.2 Selection of wind speed parameterization 708.3.2.4.3 Averaging and interpolation/extrapolation 718.3.2.4.4 Uncertainty in estimates 728.3.2.5 Future Developments 728.3.3 The Cycling of Trace Gases Across the Air–Sea Interface 728.3.3.1 Greenhouse Gases 738.3.3.1.1 Carbon dioxide 738.3.3.1.2 Methane 748.3.3.1.3 Nitrous oxide 748.3.3.1.4 Ozone 748.3.3.1.5 Carbon monoxide 758.3.3.2 Nitrogen-Containing Gases 758.3.3.2.1 Ammonia and methylamines 758.3.3.2.2 Alkyl nitrates 758.3.3.2.3 Hydrogen cyanide and methyl cyanide 768.3.3.3 Sulfur-Containing Gases 768.3.3.3.1 Dimethyl sulfide 768.3.3.3.2 Methyl mercaptan 798.3.3.3.3 Carbonyl sulfide 798.3.3.3.4 Carbon disulfide 798.3.3.3.5 Hydrogen sulfide 798.3.3.3.6 Sulfur dioxide 798.3.3.4 Nonmethane Hydrocarbons 798.3.3.5 Oxygenated Volatile Organic Compounds 80
atise on Geochemistry 2nd Edition http://dx.doi.org/10.1016/B978-0-08-095975-7.00603-3 53
54 Air–Sea Exchange of Marine Trace Gases
8.3.3.6 Organohalogens 808.3.3.6.1 The mono-organohalogens: methyl, ethyl, and propyl iodide; methyl bromide; and methyl chloride 808.3.3.6.2 The di-organohalogens: diiodomethane, chloroiodomethane, bromoiodomethane, and dibromomethane 828.3.3.6.3 The tri-organohalogens: bromoform and chloroform 828.3.3.7 Miscellaneous Gases 838.3.3.7.1 Chlorofluorocarbons 838.3.3.7.2 Synthetic organic compounds 838.3.3.7.3 Mercury 838.3.3.7.4 Selenium and polonium 838.3.3.7.5 Hydrogen 848.3.4 Effects of Climate Change on Marine Trace Gases 848.3.4.1 Effect on Air–Sea Gas Transfer 848.3.4.2 Effect on Dissolved Gas Concentrations 848.3.4.2.1 Carbon dioxide 848.3.4.2.2 Dimethyl sulfide 848.3.4.2.3 Other gases 84References 85
8.3.1 Introduction
The ocean and atmosphere are major components of the
Earth’s surface, with reactions within and between them con-
trolling many of the properties of the Earth’s system. In this
chapter, the authors are concerned with the two-way exchange
of gases between the atmosphere and ocean, but to understand
this exchange, knowledge of processes determining gas con-
centrations in both reservoirs is also needed.
These topics are important for human affairs. Marine pro-
ductivity drives the cycling of gases such as oxygen (O2), di-
methyl sulfide (DMS), carbonmonoxide (CO), carbon dioxide
(CO2), and methyl iodide (CH3I) that are important, respec-
tively, for marine productivity, biogeochemical cycling, atmo-
spheric chemistry, climate, and human health. As one example,
the net flux of CO2 from the atmosphere to the ocean repre-
sents about a third of the annual release of anthropogenic CO2
to the atmosphere, superimposed on amuch larger natural flux
of CO2 that is cycled annually between the ocean and atmo-
sphere. As another example, about 30% of the world’s popu-
lation is thought to be at risk for iodine deficiency disorders
that impair mental development. The main source of iodine to
land is as volatile iodine compounds produced in the ocean
and transferred to the atmosphere across the air–sea interface.
Besides its impact on human health, the flux of marine iodine
species to the atmosphere is likely important for the oxidation
capacity of the troposphere, via the production of iodine oxide
radicals, and for particle formation. A third example is the
emission of volatile sulfur from the oceans in the form of
DMS. Sulfur is a bioessential element. Not only is a flux of
sulfur from the oceans needed to balance its loss from the land
via weathering, but this flux probably acts as a control on
climate via the role of sulfur compounds in the formation of
cloud condensation nuclei (CCN).
There are at least four main processes that affect the con-
centration of gases in the water column: biological production
and consumption, photochemistry, air–sea exchange, and ver-
tical mixing. The authors will not discuss the effect of vertical
mixing on gases in seawater but refer the reader instead to
Chapter 8.8. Nor will they consider the deeper oceans, as this
region is discussed in Chapters 8.4, 8.10, 8.11, and 8.18. The
authors will concentrate on the cycling of gases in the surface
ocean, including the thermocline, and the exchange of volatile
compounds across the air–sea interface. While much is known
about the cycling of gases, such as CO2 and DMS, in the water
column, frustratingly little is known about the routes of for-
mation of many of the chemical species transferred from the
oceans to the atmosphere. Equally poorly understood are the
ways in which marine emissions affect the formation and
growth of atmospheric particles, in spite of much active re-
search in this area in recent years. The first section of this
chapter deals with the theory of air–sea gas exchange and its
mechanisms and measurements, while the second concerns
the processes of production and consumption that control
the distributions of gases, including estimates of the size
of the oceanic source or sink of the compound under
discussion.
8.3.2 Gas Exchange Processes andParameterizations
8.3.2.1 Gas Flux Theory
8.3.2.1.1 Mass boundary layers at the interfaceThe flux, F, of a gas across an air–water interface is driven by a
thermodynamic concentration gradient (DC) and mediated by
the rate of transport across the interface (k), the negative sign
denoting that the flux occurs in the direction of decreasing
concentration (eqn [1]):
F ¼ �kDC [1]
As the interface is approached, turbulence is inhibited and
diffusive processes tend to dominate transfer. Mass boundary
layers on either side of the interface are defined as the domains
within which the rate of diffusion is greater than the rate of
turbulent transport (Figure 1). The thicknesses, zw and za, of
these layers are thus dependent on the diffusivity of the gas in
question and the degree of turbulence in the bulk water and
Cw
Ca
Ca,i
C
Cw,i ( = Ca,i/H)
Turb
ulen
t re
gim
eTu
rbul
ent
regi
me
Air
Wat
er
Diff
usio
n >
Tur
bul
ence
Figure 1 Interfacial concentration profiles of a gas, based on a diagramby Jahne (2009), showing mass boundary layers on either side of theinterface. Note that the gas depicted has a Henry’s law solubility (H¼Ca,i/Cw,i) of 0.25 and is undersaturated in the ocean, leading to a flux from airto sea. Figure by M.T. Johnson, shared under creative commons licenseat http://dx.doi.org/10.6084/m9.figshare.92447.
Air–Sea Exchange of Marine Trace Gases 55
atmosphere. It is assumed that all the resistance to transfer
effectively occurs within these nominal mass boundary layers,
which thus control the rate of transfer, expressed as the transfer
coefficient, k, more commonly known as the transfer velocity
(or piston velocity).
Assuming conservation of mass (i.e., no production or loss
terms), the flux through the two mass boundary layers must be
equal, and so can be written (after Liss and Slater, 1974)
F ¼ �kw Cw;i � Cw
� � ¼ �ka Ca � Ca;i
� �[2]
where kw and ka are the water- and air-side transfer velocities
(dimensions L.T�1), Cw and Ca are the bulk water- and air-
phase concentrations, and Cw,i and Ca,i are the concentrations
on either side of the interface. Cw,i and Ca,i are related by the
solubility; thus
H ¼ Ca;i
Cw;i[3]
where H is the dimensionless Henry’s law coefficient, which is
the equilibrium ratio of concentration in the gas and liquid
phases. Note that the Henry’s law solubility (H) can be pre-
sented in a number of different forms, most commonly as
liquid-over-gas form in units of mol l−1 atm−1 or dimensionless
gas-over-liquid or inverted forms. Care must be taken to apply
these different forms correctly. Sander (1999) presents a good
explanation of interconversion between forms. H in this text
has a small value when solubility is great, which is counterin-
tuitive but follows from previous works. It can then be
written that
F ¼ KwCa
H� Cw
� �¼ Ka Ca �HCwð Þ [4]
where
1
Kw¼ 1
kwþ 1
Hka[5]
and
1
Ka¼ H
kwþ 1
ka[6]
Kw and Ka are the total transfer velocities and their inverse
terms in eqns [5] and [6] represent the total resistance to
transfer across both sides of the interface, viewed from the
water and air sides, respectively. The relative magnitude of
the contributions of the air and water sides to resistance to
transfer is determined by the ratio ra/rw (Liss and Slater, 1974),
where ra (¼1/Hka) and rw (¼1/kw) are the air- and water-side
component resistance terms in eqn [5]. The diffusion coeffi-
cients and individual water- and air-side transfer velocities of
most gases do not vary much; however, their partitioning
varies greatly due to the solubility term in eqn [5], with low-
solubility gases being under water-side control and with very
soluble (or reactive – see Section 8.3.2.4) gases being limited
by an air-side resistance (Table 1).
Because molecular diffusivity is so much greater in air than
in water, even moderately soluble gases, such as CO2, will
effectively be solely under water-side control. (Under the con-
ditions used for the calculations in Table 1, only 0.3% of total
resistance to transfer is in the air phase.) Values of ra/rw of
between 0.1 and 10 correspond to the region where there is
at least a 10% contribution to total resistance to transfer from
both sides of the interface. The term ra/rw is variable with
temperature, salinity, and turbulent forcing such that under
some conditions (low temperatures and salinities and high
winds tending to increase ra/rw), gases ranging in solubility
from DMS to methanol can be subject to significant control
from both phases.
The simplifying assumption that a single phase dominates
transfer, that is, Kw¼kw (for insoluble gases) or that Ka¼ka (for
highly soluble gases), is commonly made. While this is prob-
ably reasonable, under normal environmental conditions, for
gases of solubility equal to or lower than CO2, care must be
taken when making such simplifications for more soluble
gases; dual-phase control has been recognized recently for
diiodomethane (Archer et al., 2007) and some oxygenated
volatile organics (Beale et al., 2010).
8.3.2.1.2 Turbulent forcing: wind speed-basedparameterizations of transfer velocityAs turbulent forcing increases, the mass boundary layer (i.e.,
the domain in which diffusive transport is faster than turbulent
transport) must shrink. Hence, the diffusive barrier to ex-
change is reduced and the transfer velocity increases. Wind
speed is the most commonly considered turbulent forcing
when parameterizing gas transfer velocities, although other
factors and processes can be important (Section 8.3.2.3).
A wind speed-based parameterization of transfer velocity will
generally be of the form
Table 1 Solubility (H, dimensionless gas-over-liquid form) and diffusion coefficients (D, m2 s�1) and typical transfer velocities (k, cm h�1)in air and water for a range of gases
Compound H Dwater Dair kw ka ra/rw
Nitrogen (N2) 81 0.20 1700 8.1 1720 5.8�10�5
Methane 40 0.17 1800 7.5 1730 1.1�10�4
Carbon dioxide 1.4 0.18 1400 7.7 1670 0.0032Dimethyl sulfide 0.10 0.11 990 6.2 1570 0.038Sulfur dioxide 0.038 0.16 1200 7.2 1630 0.12a
Methyl nitrate 0.025 0.13 1000 6.5 1590 0.17Diiodomethane 0.020 0.098 780 5.7 1510 0.19Acetone 0.0019 0.11 1000 6.1 1580 2.1Ammonia 0.00074 0.21 2100 8.4 1760 6.4a
Methanol 0.00022 0.16 1400 7.2 1670 20Methanal 7.6�10�6 0.18 1600 7.7 1690 590b
Hydrogen peroxide 6.9�10�7 0.20 1700 8.1 1710 6900
All values are calculated using the numerical scheme of Johnson (2010) at 25 �C, salinity of 35, and wind speed of 5 m s�1, where applicable.aValues for sulfur dioxide and ammonia are enhanced substantially above the values presented here due to hydrolysis/protonation reactions enhancing kw at seawater pH
(Section 8.3.2.4.1).bThe methanal value already accounts for enhancement due to hydrolysis as it is impossible to measure its true solubility without the effect of hydrolysis.
56 Air–Sea Exchange of Marine Trace Gases
k ¼ cþ yUxð Þ � Sc�n [7]
where c is a constant representing purely diffusive transport in
the absence of turbulent forcing (often implicitly zero in em-
pirical wind speed-based parameterizations of kw), U is the
wind speed, y is an amplitude scaling factor, and x defines the
functional form of the relationship (linear, quadratic, cubic,
etc.). The Schmidt number (Sc, the ratio of the kinematic
viscosity of the medium to the diffusivity of the gas in the
medium) relates the rate of transfer to the diffusivity of the
gas in question; the exponent n defines to what degree diffusive
processes rather than turbulent processes control transfer
through the mass boundary layers. Where the transfer velocity
for a particular gas or gases has been derived empirically,
eqn [7] also allows for the scaling to other gases (notwithstand-
ing other gas-specific effects summarized in Section 8.3.2.4.1)
through the ratio of the Schmidt numbers of the two gases:
k1k2
¼ Sc1Sc2
� ��n
[8]
Using eqn [8], transfer velocities are typically normalized to
a common Sc in order to aid comparison under different
physical conditions. Values of 600 or 660 are commonly se-
lected, the values for CO2 at 20�C in freshwater and seawater,
respectively.
8.3.2.1.3 Interfacial models and the Schmidt numberexponent on the water sideIn order to understand the process of gas exchange through the
mass boundary layers at the air–sea interface, numerous
process-based models have been proposed and tested in field
and laboratory studies. These models can tell us the degree to
which diffusive processes control gas exchange, as represented
by the Schmidt number exponent, n.
The simplest model of the interfacial mass boundary layers
is the stagnant film (or thin film) model, proposed by
Whitman (1923) and applied to the air–sea interface by Liss
and Slater (1974). This model assumes the sea surface is a flat,
solid boundary with completely stagnant mass boundary
layers, through which diffusion is the sole transport process.
Therefore, simple Fickian diffusion applies
F ¼ �DdC
dz
� �[9]
where D is the diffusivity of the compound in question.
Substituting into eqn [1], one finds that the transfer velocity
in the thin film model is directly proportional to the diffusivity
(eqn [10]):
k ¼ D
z[10]
Thus in the stagnant film model, where diffusion is the sole
limiting transfer process, n¼1. Whereas the stagnant film
models describe discontinuous layers where diffusion and tur-
bulent mixing dominate the transfer of gases to the interface,
the rigid boundary or solid wall models (e.g., Deacon, 1977;
Hasse and Liss, 1980) apply a velocity profile to either side
of the interface to describe a smooth transition between dif-
fusive and turbulent regimes. In such models, turbulence
plays a modest role in controlling the rate of transfer and is
reflected in the dependence of the predicted transfer velocity to
D2/3 (n¼2/3), demonstrating that the diffusivity of the gas
becomes less important in determining the transfer velocity
as the turbulent regime becomes more important. The rigid
boundary models are found to work well in tank and wind-
tunnel experiments studying the water-side transfer velocity
when the water surface is unruffled by waves, as might be
predicted from the rigid wall assumption, but they underesti-
mate kw when the surface ceases to be smooth (e.g., Liss and
Merlivat, 1986).
Although molecular (diffusive) processes will become
progressively more important as the interface is approached,
the existence of a stagnant film whose thickness, for a given
degree of turbulence, is invariant with time and space is clearly
physically unrealistic, particularly for nonsmooth water surfaces.
Surface renewal models of the water side of the interface im-
prove on this by periodically and instantaneously replacing the
stagnant film with material from the bulk (Danckwerts, 1951;
Air–Sea Exchange of Marine Trace Gases 57
Higbie, 1935; Ledwell, 1984). The physical processes and their
mathematical descriptions differ between the various surface
renewal models. For example, Higbie (1935) assumes a single
turbulence-dependent renewal rate, whereas Danckwerts (1951)
describes a statistical distribution of possible renewal timescales
which is modulated by turbulent forcing. All surface renewal
models, however, predict n¼0.5 and can be generalized accord-
ing to eqn [11]:
kw / D
t
� �0:5
[11]
where t is the renewal timescale. While physically more realis-
tic than the stagnant film models, surface renewal and eddy
models have until recently been used rather little in environ-
mental investigations because of the difficulty of quantifying
the renewal rate. However, developments in infrared imaging
have allowed the validation of this family of models for heat
fluxes and have broadly confirmed the predicted dependence
of kw on D0.5 for heat and momentum under wave-covered
(but not necessarily for bubbled – see Section 8.3.2.2.3) sur-
faces in the marine environment (e.g., Garbe et al., 2004; Hara
et al., 2007; Veron et al., 2011).
Similarly, the so-called large (Fortescue and Pearson, 1967)
and small eddy models (Lamont and Scott, 1970), in which
transfer in the near-surface water is described in terms of a series
of cells of rotating fluid, also lead to the conclusion that kw is
proportional to Sc�0.5. In this model, kw is related to the ¼
power of the energy dissipation rate, e (Garbe et al., 2013):
kw / enð Þ1=4Sc�1=2 [12]
where n is the kinematic viscosity of seawater. This relationship
and also the magnitude of the constant of proportionality
predicted by Lamont and Scott (1970) have been validated in
the field in various environments, including shallow waters
where bottom-driven turbulence and tidal mixing dominate
gas exchange rather than wind speed (Zappa et al., 2007).
The Schmidt number dependence n¼0.5 is currently applied
universally to most wind speed-dependent parameterizations of
kw (Section 8.3.2.3), but it is probably not strictly applicable
under all regimes and, in reality, is unlikely to be constant even
within the nonbubbled, nonsmooth regime. Asher et al. (2004)
have applied the surface penetration model of Harriott (1962),
which considers incomplete replacement of the surface by eddy
transport from the bulk to explain the apparent discrepancy
between transfer velocities of heat and mass from observations
compared with predictions from the renewal models. Their
findings imply that the diffusivity dependence of transfer veloc-
ity is not constant with turbulent forcing, that is, that there is a
continuous change in n with changing turbulent forcing as well
as step changes in the diffusivity–transfer velocity relationship
(from kw/D2/3 to kw/D1/2 at the transition between smooth
and rough surface regimes).
8.3.2.1.4 Air-side transferThere are similar models of transfer on the air side of the
interface (e.g., Duce et al., 1991; Fairall et al., 2003; Jeffery
et al., 2010), derived from the micrometeorological models
of water vapor fluxes from water surfaces, derived in turn
largely from the resistance models proposed by Wesely and
Hicks (1977) and Hicks et al. (1987). Generally, these models
find that, as in the water phase, the diffusivity dependence on
transfer is between D0.5 and D0.7 (e.g., Fairall et al., 2003;
Johnson, 2010). Analogous to the surface penetration theory,
however, some models predict no fixed diffusivity scaling (e.g.,
Jeffery et al., 2010).
8.3.2.2 Processes Driving Gas Exchange
The majority of investigations of the mechanisms driving air–
sea gas exchange and variables that influence kw have been
conducted in the laboratory, quite commonly in wind/wave
tunnels, although recent advances in direct flux measurements
and thermal imaging techniques have begun to clarify actual
mechanisms in the field (Garbe et al., 2013). The key processes
affecting gas exchange are discussed in the following sections.
8.3.2.2.1 The effect of windWind is the major driver of turbulence in the open ocean, away
from sources of tidal and bottom-driven mixing. Tangential
wind stress on the ocean surface drives viscous overturn of the
bulk surface water and reduces the diffusion-limited domain of
the mass boundary layers. However, as outlined in the follow-
ing sections, variables indirectly driven by wind speed (wave
state, age and breaking, and bubbles) and processes not di-
rectly related to wind (rain, convective effects, and surfactants)
also contribute to near-surface turbulence on the water side
and thus to kw. On the air side of the interface, surface rough-
ness is a key factor in wind-driven turbulence and is also a
function of wave conditions (Drennan et al., 2005) but is
generally not included in parameterizations of the ka–U rela-
tionship. However, as the relationship is thought to be close to
linear, such complicating factors tend to be less significant than
for kw. The rest of Section 8.3.2.2 will thus be concerned
primarily with kw.
Anomalies between different laboratory- and field-based
measurements of kw are at least in part due to the non wind
speed drivers of gas exchange, which vary between studies and
are the subject of the rest of this section. Note that there are
many complex interactions between these various processes
at numerous scales, which further complicate mechanistic
understanding and modeling of gas exchange processes. Such
interactions are beyond the scope of this work, but more
mechanistic description of the processes listed below and
their interactions can be found in Garbe et al. (2013).
8.3.2.2.2 The effect of wavesThe presence of wind-induced ripples was noted in early gas
exchange studies as being coincident with an enhancement in
kw (Kanwisher, 1963). Careful experiments with various pairs
of gases were used to identify n at different U in order to
identify which, if any, of the models discussed previously
might best represent air–water gas transfer (Jahne et al.,
1987). These experiments showed that the thin film model is
too simple and that kw varied with Sc�2/3 at low U as predicted
by boundary layer models. However, once wind-induced
waves were observed on the surface of the water, kw was
found to vary with Sc�1/2, in agreement with surface renewal
models. The exact wind speed at which this regime changes was
58 Air–Sea Exchange of Marine Trace Gases
found to vary with the facility being used but was typically about
2 m s�1 (friction velocity u*¼0.3 cm s�1) (Jahne et al., 1987).
A much tighter correlation was found between kw and the
mean square wave slope of shorter wind waves than there was
with U or u* (Jahne et al., 1987). This led to the conclusion
that a wave-related mechanism was controlling kw at interme-
diate U and that enhancements were due to increased turbu-
lence generated by waves close to the water surface. More
recently, the presence of microscale wave breaking has been
speculated to be a fundamental mechanism underlying this
increase in turbulence and hence the enhancement in kw(Zappa et al., 2001).
8.3.2.2.3 The effect of bubbles and sea sprayBubbles are thought to play an important role in gas exchange
processes under conditions that promote their formation
through entrainment of air in breaking waves: high winds,
whitecapping, and long fetch. There are at least four ways
in which bubble formation has been proposed to enhance
air–sea gas transfer rates and these are illustrated in Figure 2
(Woolf, 1997).
The initial injection of bubbles through the interface can
enhance the kinetics of air–sea gas transfer by disrupting the
sea surface microlayer or by encouraging surface renewal.
There are contradictory reports in the literature as to the mag-
nitude of this ‘kinetic’ bubble effect; however, recent tank
experiments suggest that it is probably small relative to other
bubble effects (Woolf et al., 2007), at least on ‘clean’ surfaces
with no significant microlayer.
Some bubbles will dissolve completely and inject their
constituents directly into surface water. This process is inde-
pendent of any gas-specific properties and depends only on the
air-side concentration of the gas and the volume of bubbles
dissolved. It can lead to supersaturations in surface seawater
of a few percent where, at steady state, the downward injec-
tion flux is balanced by the diffusive flux back across the
air–sea interface (Stanley et al., 2009). While important for
Solution
Advection
SurfacingEntrainment
Atmosphere
Ocean
Gasexchange
Gasexchange
Figure 2 A schematic of bubble evolution after entrainment into theupper ocean. Reproduced by permission of the Cambridge UniversityPress from The Sea Surface and Global Change (ed. P. S. Liss andR. A. Duce), 1997, 173–205.
understanding the surface concentrations of long-lived gases,
the bubble injection effect is likely to be minor for more
reactive or far from equilibrium gases (Johnson et al., 2011),
although a recent work has suggested that the injection of large
bubbles at extremely high winds may be important, as dis-
cussed in the following text.
Gases in bubbles that do not completely dissolve will tend
to equilibrate, partially or completely, with the surrounding
water. This ‘exchange effect’ also favors supersaturation due to
the effects of pressure and surface tension on the bubble below
the water surface. The effect is complicated by the fact that trace
gases may dissolve at a different rate than bulk gases (N2 and
O2) and that the pressure-driven dissolution of trace gases on
the downward trajectory may be compensated to some degree
by the resulting undersaturation of the bubble on the upward
trajectory. The capacity of the bubble to either supply or receive
gas may also be limiting in contrast to the case of transfer
across the air–sea interface where the volume of the atmo-
sphere is effectively unlimited. There is therefore a solubility
effect, as the ‘carrying capacity’ of the bubble will be smaller for
more soluble gases, and bubble-mediated gas exchange is thus
predicted to be more effective for less soluble gases (Woolf,
1997). This assertion is well supported by laboratory and
wind-tunnel experiments (e.g., Broecker and Siems, 1984; de
Leeuw et al., 2002; Merlivat and Memery, 1983; Rhee et al.,
2007; Woolf et al., 2007) but by only circumstantial evidence
in the field (Section 8.3.2.3.6). This effect is really a separate
flux term with its own transfer velocity (for transfer across the
bubble wall) and concentration gradient (between the bubble
and bulk seawater). However, as it is impossible to deconvolve
the bubble component and ocean–atmosphere interfacial
fluxes in empirical field studies of gas transfer rates, it is
commonly represented as a component of the total transfer
velocity. The bubble scheme of Woolf (1997), eqn [13], is
commonly used to model bubble fluxes, including the
National Oceanic and Atmospheric Administration/Coupled
Ocean Atmospheric Response Experiment (NOAA/COARE)
algorithm (Section 8.3.2.3.5):
F ¼ kd þ kbð Þ Ca 1þ Lð ÞH
� Cw
� �[13]
where L is the equilibrium supersaturation at which there will
be no net air–sea transfer of gas, kd is the diffusive (air–sea
interfacial) transfer velocity, and kb is the bubble-mediated
transfer velocity. Both kb and L are dependent on solubility
and D. The term kd is predicted to increase roughly linearly
with wind speed, whereas kb is predicted to be approximately
proportional to whitecap coverage (the cube of wind speed)
(Woolf, 1997).
The Woolf (1997) model has recently been extended by
McNeil and D’Asaro (2007) to include an extra term that de-
scribes the complete dissolution of larger bubbles. Previously,
it had been assumed that larger bubbles would rise to the
surface and contribute little to gas transfer other than through
turbulent disruption of the mass boundary layer/surface films,
but based on the observations of oxygen fluxes during a hurri-
cane, McNeil and D’Asaro (2007) proposed that large bubbles
might be injected deep enough during extreme wind events to
completely dissolve. These authors proposed a scheme which
Air–Sea Exchange of Marine Trace Gases 59
fully separated the bubble and diffusive fluxes and added a
further term to account for large bubble dissolution:
F ¼ kdCa
H� Cw
� �þ kb
Ca 1þLð ÞH
� Cw
� �þ Vw [14]
where V is the moles of air entrained and completely dissolved
per unit surface area per time (a function of wind speed/white-
capping, etc.) and w is the mole fraction of the gas in the
atmosphere. McNeil and D’Asaro (2007) concluded that this
extra bubble injection term (Vw) becomes significant at wind
speeds above 27 m s�1 and that it could be the dominant
mechanism for gas exchange above 37 m s�1.
Recent evidence from awind-tunnel study combinedwith an
improved bubble model, which accounts for the high void
fraction in dense bubble plumes, suggests that there may be a
further effect, whereby the flux into the bubble is ‘suffocated’ by
the small volume of water in contact with the gas in the bubbles,
that is, the flux of gas into the bubbles is limited by the decreas-
ing concentration in the water around the bubble plumes
(Woolf et al., 2007). This effect will be largest for the most
diffusive gases and may help to explain some of the apparent
inconsistencies between different field observations of gas ex-
change (Section 8.3.2.3.6).
Sea spray is the atmospheric analogue of bubbles, also vary-
ing with approximately the cube of wind speed (e.g., Bao et al.,
2011). As such, it might play an analogous role in enhancing the
air–sea flux of soluble gases (Garbe et al., 2013). This process
has not been studied, to the authors’ knowledge, and is not
included in any parameterization of gas-phase transfer velocity
to date (Johnson, 2010). However, it is recognized as an impor-
tant process for water vapor and heat transfer at very high winds
(Bao et al., 2011; DeCosmo et al., 1996; Li, 2004), so it is also
likely to be of importance for soluble trace gases and is thus
worthy of future study.
8.3.2.2.4 Heat and water fluxesThere are several ways in which heat and water transfer across
the air–sea interface might affect the exchange of trace gases.
Firstly, it has been proposed that eqn [1] gives an incomplete
description of the gas exchange process since the irreversible
thermodynamic coupling between heat and mass fluxes is not
included (Phillips, 1994, 1997). The concept has been criti-
cized by Doney (1995) on the grounds of inconsistencies in
Phillips’ treatment of the molecular boundary layer and irre-
versible effects near the air–water interface, and thematter does
not appear to have been definitively settled. It has been argued
that, at least under oceanic conditions where the temperature
gradients and thus heat exchanges across the sea surface are
quite small, any effect is very limited and can invariably be
neglected. However, Phillips’ group continues to work on the
subject (e.g., Packwood and Phillips, 2010a,b) and the gas
exchange community probably needs to revisit this body of
work.
The second and better recognized effect arises from the fact
that evaporation and other heat exchanges at the sea surface
will lead to temperature changes, for example, cooling of the
water very close to the interface if evaporation is dominant. The
magnitude of this ‘cool skin’ effect is typically 0.1–0.3 �C. Sincethe solubility of almost all gases increases with decreasing
water temperature, an error is made if the equilibrium solu-
bility is calculated at the bulk water temperature. This has
been suggested to be especially significant for gases, such as
CO2, for which the global net exchange with the oceans is a
small difference between large gross uptakes and emissions,
driven by small concentration differences and thus very sensi-
tive to the solubility term. Inclusion of the skin effect into
calculations of the net uptake of CO2 by the oceans via
eqn [5] has been estimated to lead to an enhancement by
0.7 Gt C year�1 (Robertson and Watson, 1992), which is
significant given that the global net flux is approximately
1.7 Gt C year�1. However, McGillis and Wanninkhof (2006)
suggest that these effects are actually minor due to the different
depths of the temperature and mass boundary layers, and
this assertion appears to be borne out in the field data of
Ward et al. (2004). Furthermore, in situations where an evap-
orative cool skin may exist, there is likely to be enhanced
salinity at the surface, which will approximately cancel
out the temperature effect on solubility, at least for CO2
(Takahashi et al., 2009).
It is also possible that heat exchanges across an air–water
interface might lead to alterations in the value of kw. For
example, the ‘cool skin’ effect could lead to instability in the
water close to the interface, with ensuing increased turbulence
and enhancement of the transfer velocity. This idea has been
tested in a wind-tunnel study in which kw was measured as a
function of the evaporation and condensation of water mole-
cules (Liss et al., 1981). The results showed that under evapo-
rative conditions, there was no measurable enhancement in kwdue to destabilization of the near-surface water, any effect
being masked by mechanically generated mixing. However,
under condensing conditions, there was a decrease in kw of
up to 30% due to increased stability. The magnitude of this
effect in the environment is likely to be limited since evapora-
tive conditions are typically found over the oceans. However,
there are situations in which condensation is the dominant
process, as in coastal upwelling regions where cold, deep water
is brought to the surface and may then come into contact with
warm, humid air. Since such areas are often biologically rich,
the importance of lowered kw under condensing conditions
could be of some significance for the air–sea exchange of
biologically active trace gases.
Finally, a ‘significant and systematic’ enhancement in kwhas been noted for SF6 due to the influence of rainfall (Ho
et al., 1997). A first-order relationship between kw and rain rate
was observed in laboratory simulations and in simple field
experiments. More recent experiments investigating the com-
bined effects of wind and rain initially indicated that the effect
of the two drivers might be linearly additive (Ho et al., 2007),
but further experiments have shown that the enhancement
effects of rain diminish with increasing wind speeds (Garbe
et al., 2013). This may not have been observed in the earlier
experiment as wind speeds were not high enough for the re-
sults to exhibit the effect. Either way, the effect of rain on
turbulence is only likely to be highly significant at low wind
speeds, although it is not clear how rain effects may apply to
the oceans, where precipitation may also cause temperature
and salinity gradients, disrupt surfactant layers (Garbe et al.,
2013), and change, for example, carbonate chemistry for the
flux of CO2 (Turk et al., 2010).
60 Air–Sea Exchange of Marine Trace Gases
8.3.2.2.5 The effect of surface filmsA sea surface microlayer commonly exists, which is, to a variable
extent, physically, chemically, and biologically different from
bulk surface seawater. It accumulates insoluble and surface-
active organic compounds produced in the surface ocean and
can be a visible ‘oily’ film under some circumstances.
The presence of a surface film may affect gas transfer in a
number of ways. Concentrated insoluble surfactant films
(slicks) can act as a barrier to gas exchange by forming a
condensed monolayer on the sea surface (Frew, 1997) or by
providing an additional liquid phase that may enhance resis-
tance to transfer (Liss and Martinelli, 1978). These effects are
thought to be important only at low wind speeds, however, as
slicks are easily dispersed by wind and waves (Liss, 1983). The
main effect of surface-active material is believed to be due to
soluble surfactants that reduce the roughness of the sea surface
and alter its hydrodynamic properties (Frew, 1997). Indeed,
Frew has argued that soluble surfactants can reduce kw even in
the presence of breaking waves. This implies that air–sea gas
transfer is not purely a physically driven process but that sur-
face chemistry may also play an important role.
Early experiments showed that reductions in kw of up to 60%
could be observed in wind/wave tanks for a given wind speed
(Broecker et al., 1978), and contamination of the air–water
interface by surface films was a common problem, particularly
in circular tanks where there was no ‘beach’ at the end of the
tunnel for the film to collect (Jahne et al., 1987). Later experi-
ments found that the presence of surfactants influenced the
point at which small-scale waves were observed at the water
surface, coincident with a rapid increase in kw (Frew, 1997).
More recently, Lee and Saylor (2010) found that the presence
of a monolayer of oleyl alcohol greatly reduces the mass transfer
of oxygen in a wind-wave tank at low wind speeds (<4 m s�1).
There have also been observations that kw can vary with
biological activity (Goldman et al., 1988) due to the exudation
of soluble surface-active material (including carbohydrates,
0.0 0.0 1.0
Chlorophyll fluorescence
2.0
20
(a) (b)
15
10
Initi
al t
rans
fer
velo
city
, kw
(cm
h−1)
5
00.4 0.8
Surfactants (mg l−1)
Figure 3 Correlations of kw with (a) surfactant concentration, (b) in situ chlodissolved organic matter (CDOM) for seawater samples collected from MonteStates, to Bermuda (•). Reproduced by permission of the Cambridge Univer1997, 121–172.
lipids, and proteins) by phytoplankton (Frew et al., 1990).
Supporting evidence comes from laboratory experiments
using seawater collected on a transect from the United States
to Bermuda (Frew, 1997), showing that the decrease in kwcorrelates inversely with bulk water chlorophyll, dissolved or-
ganic carbon (DOC), and colored dissolved organic matter
(CDOM) (see Figure 3).
The importance of the relationship with CDOM is the possi-
bility that this parameter could be determined remotely by satel-
lites. This work implies thatUmay not be the best parameter with
which to parameterize kw in the oceans, particularly in biologi-
cally productive regions. A reasonable correlation with the total
mean square wave slope for both filmed and film-free surfaces,
particularly of shorter wind waves, suggests that this parameter
might be a more useful predictor of kw (Jahne et al., 1987);
although it is difficult to measure directly at sea mean square
wave slope can be measured by satellite (Section 8.3.2.5). This
relationship has now been shown experimentally in the field:
Frew et al. (2004) found that kw was better correlated with mean
square wave slope than with wind speed. Specifically, kw was
overpredicted by wind speed when winds were below 6 m s�1
and when CDOM levels were enhanced in the microlayer relative
to bulk water, whereas the mean square slope relationship was
not affected.
Are surfactants only important in retarding gas transfer at
very low winds? At higher winds, despite wave breaking, sur-
factants are predicted to be brought back to the sea surface via
bubble scavenging (Liss, 1975). Support for this hypothesis
comes from Asher et al. (1996), who demonstrated in a series
of laboratory experiments that soluble surfactants inhibit gas
transfer even under breaking-wave conditions. Recently, Salter
et al. (2011) reported a field experiment in which an artificial
surfactant (oleyl alcohol) was deliberately released in the
northeast Atlantic Ocean. Gas transfer rates were measured by
both the dual-tracer technique (Section 8.3.2.3.3) and direct
covariance measurements of the DMS flux (Section 8.3.2.3.4).
0.0 0.2 0.4
CDOM fluorescence
100 15050
DOC (mmol l−1)
(c) (d)
rophyll fluorescence, (c) dissolved organic carbon (DOC), and (d) coloredrey Bay, United States (▪), and on a transect from Narragansett, Unitedsity Press from The Sea Surface and Global Change (ed. Liss and Duce),
Air–Sea Exchange of Marine Trace Gases 61
Air–sea gas transfer rates were reduced by about 50% at 7 m s�1,
and kw was lowered by 25% at 11 m s�1 in the presence of the
surfactant. While the work of Salter et al. (2011) has proven that
surfactants can significantly modify air–sea gas transfer, it is not
yet clear that they do so at ambient levels. A deliberate tracer
(DT) experiment (Section 8.3.2.3.3) conducted as part of an
open ocean iron enrichment experiment in the equatorial Pacific
found that there was no measureable reduction in kw during the
development of a large algal bloom (Nightingale et al., 2000a),
even though chlorophyll-a levels increased 30-fold (Figure 4).
However, no measurements of surfactants were made, so it
is not known whether there was any change in this class of
compounds during the algal bloom. Indeed, it is possible that
the sea surface was already sufficiently covered by surfactants
that any further released by the algal bloom had no measure-
able effect. A recent study by Wurl et al. (2011) found many of
the world’s oceans are covered to a significant extent by surfac-
tants. Enrichments in the sea surface microlayer compared to
bulk seawater persisted at wind speeds of up to 10 m s�1,
consistent with the observations of Salter et al. (2011).
Asher (1997) predicted a global decrease of 20% in the esti-
mated net sea-to-air flux of CO2 by assuming that surfactant
concentrations would scale with primary productivity. Tsai and
Liu (2003) estimated that the net uptakeof CO2 could be reduced
by between 20% and 50%, the range reflecting uncertainty in
measurements of surfactant concentrations and their impact on
gas transfer. However, a simple inverse relationship between
chlorophyll a and a reduction in kw is probably unrealistic. Sea
surfacemicrolayer enrichments have been found to be greatest in
oligotrophic (low-productivity) waters rather than, asmight have
been expected, in high-productivity waters (Wurl et al., 2011).
These findings have implications for the wind speed parameter-
izations of transfer velocity derived from estimates of the global
uptake of 14C by the oceans (Section 8.3.2.3.1).
8.3.2.2.6 Chemical enhancementThe enhancement of gas transfer by chemical reaction within
the mass boundary layer(s) has been recognized as important
163.0
2.5
2.0
1.5
1.0
0.5
0.0
14
12
10
8
6
Win
d s
pee
d (m
s–1) k
w-6
60 (c
mh–1
)
Chl
orop
hyll
(mg
m3 ) p
hoto
phy
tin (m
gm
3 )
4
2
030-May 01-Jun 03-Jun
Day (1995)
05-Jun
Figure 4 Gas transfer rates during an open ocean iron enrichmentexperiment by release of deliberate tracers (DT). Estimates of kw aregiven by the horizontal black lines. The gray shaded line represents thewind speed, the dashed line represents levels of the pigment pheophytin,and the solid line represents the concentrations of chlorophyll a.Reproduced by permission of the American Geophysical Union fromGeophysical Research Letters 2000, 27, 2117–2120.
for a small number of gases, including O3 (Fairall et al., 2007),
SO2 (Liss and Slater, 1974), and possibly CO2 (Boutin and
Etcheto, 1995; Hoover and Berkshire, 1969; Wanninkhof and
Knox, 1996). Chemical reactions that are rapid relative to the
timescale of turbulent diffusive transfer across the mass bound-
ary layer serve to steepen the concentration gradient of the gas
and thus to reduce the effective depth of the mass boundary
layer, leading to enhancement of the transfer velocity (Johnson
et al., 2011). Note that while reversible reactions, such as the
hydration of CO2, will act to buffer a reaction in either
direction and thus lead to enhancement of fluxes into and
out of the ocean (albeit asymmetrically where forward and
reverse reactions occur at different rates), irreversible reactions
have the capacity to inhibit fluxes by a similar mechanism, that
is, decreasing the steepness of the concentration gradient where
either a production reaction acts against a flux into the ocean
or a breakdown reaction acts against a flux out (Johnson et al.,
2011). Here, the authors will refer only to enhancement, but
the reader should be aware that enhancement factors can be
negative in such situations.
The enhancement of CO2 exchange by hydration to car-
bonic acid and subsequent acid dissociation may be important
(Hoover and Berkshire, 1969; Wanninkhof and Knox, 1996).
Unlike the near-instantaneous hydration of SO2 (Liss and
Slater, 1974), the hydration of CO2 is relatively slow, and its
enhancement is thought to be important only when turbulent
forcing is weak (and thus the timescale of transport across the
mass boundary layer is relatively large); that is, CO2 hydration
is most likely to enhance the flux at low or zero wind speeds.
The effect on global CO2 fluxes is rather complex, however;
Boutin and Etcheto (1995) conclude that estimates of the net
global atmosphere-to-ocean CO2 flux should be reduced by
about 5% due to the bias introduced by the fact that outgassing
areas are generally associated with low average winds.
Transfer of other gases could conceivably be chemically en-
hanced by a range of different types of reactions. For instance,
hydration reactions for SO2 (Liss, 1971; Liss and Slater, 1974)
and reaction of ozone with iodide and organic matter (Fairall
et al., 2007; Martino et al., 2012) are important in controlling
their transfer rates. Photochemical reactions may also be suffi-
ciently rapid to enhance transfer (Section 8.3.2.4.1). Proton-
ation of ammonia (NH3) at seawater pH and lower leads to
chemical enhancement and thus a lowering of resistance in the
liquid phase, causing its resistance to transfer to be dominated
by the air side (Johnson et al., 2011). This is not likely to be the
case at higher pH.
Reactions between different compounds might lead to
more complex behavior than simple first-order enhancement.
For instance, air–water mass transfer of CO2 can be inhibited
or enhanced by the mass transfer of NH3 due to the reversible
and pH-dependent formation of ammonium carbamate
(Budzianowski and Koziol, 2005), although this phenomenon
has not been studied in the natural environment.
8.3.2.2.7 Biological effectsBiological activity can potentially directly or indirectly affect air–
sea gas flux in a number of ways, most of which remain largely
unstudied, and none of which are included in gas exchange
models; mainly because of the difficulty in parameterizing
such irregular phenomena. As with physicochemical processes,
62 Air–Sea Exchange of Marine Trace Gases
biological processes may enhance mass transfer by modifying
the concentration gradients in the mass boundary layers, at least
on the water side of the interface. Microbial communities at
the ocean surface tend to be different from bulk water commu-
nities, often with considerably enhanced population densities
(Cunliffe et al., 2009), which would lead to potentially rapid
processing of bioactive compounds. There is circumstantial ev-
idence for the asymmetrical biologically mediated transfer
of methane (Upstill-Goddard et al., 2003) and O2/CO2
(Garabetian, 1991; Matthews, 1999).
Some species of phytoplankton possess an enzyme (car-
bonic anhydrase) that catalyzes the hydration of CO2 inside
algal cells (Raven, 1995). Although early experiments indicated
that bovine-derived carbonic anhydrase increased the ex-
change rates of CO2 (Berger and Libby, 1969), later laboratory
investigations with natural seawater failed to find any enhance-
ment (Goldman and Dennett, 1983). However, laboratory
studies using a circular gas exchange tank showed that there
is indeed a considerable enhancement to kCO2in the presence
of various algal species (Matthews, 1999). This enhancement
was species-dependent and increased significantly as CO2 con-
centrations decreased. The lack of a response in the experi-
ments of Goldman and Dennett (1983) could well be due to
the short lifetime of the enzyme in seawater. Modeling of these
results suggested that biological enhancements might be glob-
ally significant (Matthews, 1999). However, no experiment has
yet shown that either chemical or biological enhancements by
CO2 hydration are significant at sea. Nonetheless, Calleja et al.
(2005) found that planktonic metabolism in the top few
centimeter of the ocean can control the magnitude and direc-
tion of the air–sea CO2 flux.
8.3.2.3 Wind Speed Parameterizations
A major and long-standing problem in studies of air–sea gas
exchange is the practical limitation of experimentation at the
air–sea interface. Not only is it a formidable task to try to
obtain in situ measurements close to such a dynamic region
as the sea surface, particularly in the microlayer (typically
much less than 1 mm), but it also has proved difficult to
measure the flux of gas across the air–sea interface directly.
Neither, for obvious reasons, is it straightforward to manipu-
late experimental conditions in order to probe the mechanisms
that might control gas transfer. Various ingenious techniques
have therefore been used to derive air–sea gas exchange rates
indirectly, typically by estimating air- and/or water-based bud-
gets, but also more recently and with increasing success by
direct measurement of near-instantaneous fluxes by eddy co-
variance (EC) and other micrometeorological techniques.
These approaches have largely been used to attempt to param-
eterize the kw–U relationship, and this is the focus of this
section. When reporting transfer velocity parameterizations in
the succeeding text, the authors will use k600 (i.e., kw for a
Schmidt number of 600) or k660, depending on the original
study. Gas-specific and solubility-dependent effects notwith-
standing, the transfer velocity can then be scaled to the appro-
priate Schmidt number using eqn [8].
The processes mediating gas exchange are complex,
with multiple physical (and potentially other) controls
(Section 8.3.2.2). In the ocean environment, many of these
are driven, directly or indirectly, by wind. As such, wind speed
is found to correlate well with transfer velocity in observational
field data (Ho et al., 2011). Wind speed is routinely measured
and can be derived from remotely sensed data and ground-
truthed by high-precision instruments on ships (Wanninkhof
et al., 2009); hence, it is likely to remain the key variable used
in empirical parameterizations of transfer velocity for the fore-
seeable future. Initially, the parameterizations of transfer ve-
locity fromwind speed were theoretical or based on small-scale
laboratory experiments (e.g., Liss and Slater, 1974, and refer-
ences therein). The first predictive parameterization of kw that
synthesized field experiments (Wanninkhof et al., 1985),
wind-tunnel data (Broecker et al., 1978), and theory was that
of Liss and Merlivat (1986). Although their results predict
rather low values of kw relative to subsequent field observa-
tions, they are still regularly used as a lower bound for esti-
mates of gas fluxes (e.g., Lana et al., 2011). In fact they are in
rather better agreement with recent direct measurements of
DMS flux than more recent parameterizations, although this
may be by coincidence. The parameterization of Liss and Mer-
livat (1986) assumed three discrete linear relationships to rep-
resent three different regimes known to affect transfer velocity:
the smooth regime (where the rigid boundary model holds
and n¼2/3), the rough surface regime (where turbulence be-
comes more important and n¼1/2 as predicted by renewal
models), and the breaking-wave regime (where tank experi-
ments had shown a further increase in the rate of exchange
with whitecapping):
k600 ¼ 0:17U10 U10 < 3:6 ms�1, n ¼ 2�3
� �[15]
k600 ¼ 2:85U10 � 9:65 3:6 < U10 < 13 ms�1, n ¼ 1�2
� �[16]
k600 ¼ 5:9U10 � 49:3 U10 > 13 ms�1, n ¼ 1�2
� �[17]
The following sections detail methods used to parameterize
gas exchange with wind speed in the field and the key
studies which have produced the parameterizations commonly
in use.
8.3.2.3.1 Global 14C constraintRadiocarbon (14C) is produced naturally in the atmosphere
and decays slowly in seawater with a half-life of thousands of
years and thus there must be transfer of naturally produced 14C
from the atmosphere to the ocean (Broecker and Peng, 1974).
Considering the deep-ocean 14C inventory and assuming a
preindustrial steady state, it is possible to derive a net atmo-
sphere–ocean flux for 14CO2. An estimate of the global annual
mean value for kCO2of 21�5 cm h�1 was produced by this
method (Broecker et al., 1986). Similarly, by measuring the
increased 14C present in the oceans as a result of atmospheric
nuclear weapons tests in the middle of the last century, an
independent estimate of kCO2of 22�3 cm h�1 can be derived
(Broecker et al., 1986), in very good agreement with the natural14C measurement.
This method provides a fixed point (or more realistically a
fixed region, given the uncertainties discussed in the following
text) on the kw–U curve for CO2, but it tells people nothing
about the shape or form of the relationship, how kw varies in
time and space, or how to calculate kw for other gases.
Air–Sea Exchange of Marine Trace Gases 63
Nonetheless, the parameterization of Wanninkhof (1992),
which is probably the most widely used even today, is derived
directly from the data of Broecker et al. (1986). Wanninkhof
makes the assumption that kw varies with the square of the
wind speed, in accordance with the theory that gas transfer
should scale quadratically with wind stress, as demonstrated in
eqn [11]. They fit a quadratic relationship through the global14C constraint (eqn [18]):
k660 ¼ 0:39U102 [18]
Since Wanninkhof (1992), the ocean 14C inventory, the
value of the global average wind speed, and the methods for
calculating kw based on the constraint have all been revisited
(Muller et al., 2008; Naegler et al., 2006; Sweeney et al., 2007,
and others), all revising a global average kw for CO2 downward
from the original estimates of Broecker et al. (1986) and
Wanninkhof (1992). Naegler (2009) provided a comprehen-
sive synthesis and assessment and identified reasons for dis-
agreement among the recent estimates, which include not only
complications in the inventory estimates (beyond the scope of
this summary) but also importantly errors in the normaliza-
tion of the global mean transfer velocity to a particular Schmidt
number, which is complicated by the high sensitivity of the
Schmidt number to temperature and a secondary dependence
on salinity. The current best estimate of the global mean kw for
CO2 from Naegler (2009) is 16.5 cm h�1, 23% lower than the
Broecker et al. (1986) estimate. These estimates rely on a
further assumption that is particularly important with a highly
temperature-sensitive Schmidt number applied over a wide
range of global temperature, namely, that the Schmidt number
exponent assumed in all global 14C studies is n¼0.5. Any
significant divergence from this dependence leads to large
uncertainty in the estimate.
A further large-scale estimate of kw comes fromhigh-precision
measurements of atmospheric O2/N2 ratios from baseline sites
situated around the globe (Keeling et al., 1998). The technique is
dependent on the use of CO2 data to correct for the effects of
land/atmosphere fluxes on O2/N2 ratios and on an atmospheric
transportmodel to simulate oceanic fluxes. Annual values for kO2
of 24�6 and 29�12 cm h�1 were calculated for ocean areas
north of 30�N and south of 30� S, respectively, areas with above
average winds (Keeling et al., 1998). When normalized to k600,
these values are approximately 25% higher than the estimates
predicted by Wanninkhof (1992), just within the likely error in
the estimates, but error probably cannot account for the greater
differencewithmore recent revised estimates of global average kwfrom radiocarbon data. However, a higher k600 for O2, which
is considerably less soluble than CO2, is consistent with the
influence of bubbles on apparent transfer velocities and may
explain the observed differences between Northern and South-
ern Hemisphere data and between this method and the global14C estimates.
8.3.2.3.2 Local-scale natural tracer experimentsThis section summarizes the various mass balance techniques
that have been used to derive transfer velocities from the distri-
bution and wind speed variability of natural tracers at spatial
scales of days to weeks and tens to hundreds of kilometers.
8.3.2.3.2.1 Radon-deficit technique
Radioactive decay of natural 226radium to the gas 222radon
occurs within the water column and results in a loss of radon
from the surface mixed layer to the atmosphere. A mass budget
can be made of the ‘missing’ radon by assuming a steady state
with deeper waters and hence a value for kw derived (Peng
et al., 1979). The mean value for k600 obtained using this
technique is about 14 cm h�1 and is somewhat lower than
most of the 14C-derived global mean estimates. Further mea-
surements yielded amean value for k600 of about 18 cm h�1 for
the tropical Atlantic (Smethie et al., 1985). All the radon data
showed a large amount of scatter withU, and the technique has
some well-documented shortcomings: it does not allow for the
effect of vertical mixing on the radon profile and it uses a long
averaging time imposed by the several day half-life of 222radon
(Liss, 1983) such that the condition of a steady state was rarely
fulfilled. These shortcomings were overcome by investigators
who made repeated measurements of the vertical profiles of
radon at the same station in the sub-Arctic Pacific and found a
reasonable correlation with U (Emerson et al., 1991). Again,
estimates tended to be lower than those predicted from 14C,
although they were in better agreement with revised estimates
(see Section 8.3.2.3.1). Recently, Bender et al. (2011) reanaly-
zed the data set of Peng et al. (1979) using high-quality wind
speed data and evaluated the radon-deficit method (Figure 5).
They found that it agreed rather well with other tracer-based
estimates of gas exchange, such as those of Sweeney et al.
(2007) and Nightingale et al. (2000b).
8.3.2.3.2.2 Noble gas technique
In many ways, the noble gases are ideal candidates for use in
studies of air–sea trace gas exchange because they are biologi-
cally and chemically inert and their concentrations change
only in response to physical processes. They cover a wide
range of solubility and molecular diffusivity and thus respond
differently to physical drivers. As a result of their low reactivity
and long residence times, however, the concentration gradient
across the air–sea interface is usually small, as it changes only
in response to surface warming/cooling, mixing, and bubble-
mediated transfer, with diffusive air–sea transfer acting to re-
store equilibrium (Stanley et al., 2009). Measurements there-
fore have to be made with a great degree of precision.
A monthly time series of measurements at the Bermuda
Atlantic Time Series site has been underway for several years to
constrain parameterizations of kw. The results presented by
Stanley et al. (2009) showed that the magnitude of diffusive
gas transfer was within the uncertainty of the Wanninkhof
(1992) parameterization when the effect of bubble injection,
which drives the concentration gradient counter to the diffusive
flux, was taken into account. Because of their closeness to equi-
librium, the bubble injection effect was found to be critically
important for net fluxes of the noble gases. The observations and
model presented by Stanley et al. (2009) also demonstrate a role
both completely and incompletely dissolved bubbles; with the
overall bubble effect enhancing CO2 fluxes by 5% over the
diffusive flux on average, with peaks of 30–60% in extreme
circumstances. These enhancements agree with the results from
hurricane studies and associated high-wind speed bubble
models (d’Asaro and McNeil, 2007; McNeil and d’Asaro, 2007).
Liss and Merlivat (1986)100
80
60
40
20
00 5 10
Wind speed (m s-1)
k w–6
00 (c
m h-1
)
15 20
Nightingale et al. (2000b)Sweeney et al. (2007)Wanninkhof and McGillis (1999)Wanninkhof (1992)Global 14C constraint, Naegler (2009)Best fit to all dual tracer data, Ho et al. (2011)DT, Southern Ocean, Ho et al. (2011)DT, Southern Ocean, Ho et al. (2006)DT, Southern Ocean, Wanninkhof et al. (2004)DT, equatorial Pacific, Nightingale et al. (2000b)DT, North Atlantic, McGillis et al. (2001b)DT, North Sea, Nightingale et al. (2000a)DT, Florida Shelf, Wanninkhof et al. (1997)DT, Georges Bank, Asher and Wanninkhof (1998)Radon deficiency data, Bender et al. (2011)
Figure 5 Parameterizations of kw and mass balance-based estimates, global 14C constraint (Naegler, 2009), radon deficiency data (Bender et al.,2011), and deliberate tracer (DT) data (as summarized by Ho et al., 2011). DT studies in the open ocean are shown with filled symbols, those fromcoastal and shelf seas with open symbols. Note that the two outlying DT points from Wanninkhof et al. (2004), grayed out, are omitted from the best-fitparameterization developed by Ho et al. (2011). Parameterizations denoted by red lines are constrained by the global 14C constraint. Note that the cubicrelation of Wanninkhof and McGillis (1999) appears to ‘miss’ the constraint due to the conversion from long-term averaged to short-term, steady winds(see Wanninkhof, 1992). All data have been normalized to a Schmidt number of 600, following Ho et al. (2011). Figure by M.T. Johnson, shared undercreative commons license at http://dx.doi.org/10.6084/m9.figshare.92419.
64 Air–Sea Exchange of Marine Trace Gases
8.3.2.3.2.3 Biogenic gas mass balance
These techniques involve time-series measurements of a trace
gas (typically oxygen) for which the ocean and atmosphere
system is out of equilibrium in order to determine the flux of
the gas by means of a mass budget in the water column. They
have two main difficulties. The first is to budget accurately all
the other production and/or removal processes. The second is
to evaluate dispersion and advection at the sampling site.
Initial data from oxygen budgeting in the Gulf of Maine
(Redfield, 1948) gave estimates of monthly mean gas exchange
that later proved consistent with wind/wave tank experiments
(Liss, 1983). Later results from Funka Bay in Japan showed
that kw for O2 was higher in winter than in summer and that it
increased more than linearly with wind speed (Tsunogai and
Tanaka, 1980). The use of Lagrangian techniques has helped
to overcome problems with advection and dispersion. For
example, an estimate of k600 of 19�11 cm h�1 was obtained
in a month-long budgeting study of CO2 in an SF6-labeled
patch in the North Atlantic (Feely et al., 2002) as part of the
GasEx-98 experiment. This is in reasonable agreement with
estimates from 14C techniques (Section 8.3.2.3.1) and delib-
erate tracers (Section 8.3.2.3.3) and suggests that the mass
balance technique could be more commonly used.
An alternative approach is to construct atmospheric mass
budgets. This has been done usingmeasurements of DMS in air
and water off Cape Grim, Australia, in combination with an
atmospheric model to estimate the air–sea flux of DMS (Gabric
et al., 1995). Their results are also consistent with wind speed
parameterizations derived from DT studies.
8.3.2.3.3 Deliberate tracer experimentsInert volatile tracers, particularly SF6, have been deliberately
added to water bodies in order to determine kw via water-based
mass budgeting techniques. The lack of production or removal
processes for SF6 in the water column makes the technique
simpler and far more precise than those discussed earlier. The
technique was originally utilized in enclosed lakes and a good
correlation was observed between kw and U (Upstill-Goddard
et al., 1990; Wanninkhof et al., 1985, 1987). A major improve-
ment to this technique was the corelease of 3He in lake
experiments (Watson et al., 1991). This allowed the first deter-
mination of n to be made in situ, the value of 0.51 being in
excellent agreement with the predictions of surface renewal
models.
In order to use the DT technique at sea, at least two tracers
are required in order to correct for losses by vertical and hor-
izontal mixing. Choosing two tracers with very different diffu-
sivities (normally 3He and SF6) allows the separation of
dilution effects (which will not alter the 3He/SF6 ratio) and
gas exchange. 3He is much more diffusive than SF6 and will
thus degas more rapidly, changing the ratio of 3He to SF6,
allowing the transfer velocity to be determined. One major
drawback of the DT method is that the integration period for
determination of a transfer velocity is long, typically between
Air–Sea Exchange of Marine Trace Gases 65
8 and 72 h, which means only a small number of data points
can be obtained per experiment. However, there is now quite a
large body of data from various studies, contributing to a set of
>35 data points (Figure 5). Dual-tracer determinations of kwhave been conducted in coastal seas (Asher and Wanninkhof,
1998; Nightingale et al., 2000b; Wanninkhof et al., 1997), the
open Atlantic (McGillis et al., 2001b), Pacific (Nightingale
et al., 2000a), and Southern Oceans (Ho et al., 2006, 2011;
Wanninkhof et al., 2004). Apart from the two outlying points
from the study ofWanninkhof et al. (2004) suggesting higher kwat roughly 9 and 12 m s�1 (albeit with huge associated error and
subsequently characterized as outliers by Ho et al. (2011)), the
data from all these regions show remarkably good agreement,
suggesting that the method is robust to environmental condi-
tions, although the scatter in the data does suggest that processes
other than wind speed are also likely to be important.
Asher (2009) demonstrated that up to 50% of the variabil-
ity in results from DT experiments may lie in experimental
error in the individual experiments. When used in the open
ocean, DT (and all other local-scale mass balance techniques)
requires knowledge of the change in mixed layer depth (MLD)
with time; shoaling, in particular, leads tomajor dilution of the
tracers. The various MLD estimation and modeling approaches
are thus all subject to substantial uncertainty (Ho et al., 2011).
A further complication results from the required assumption of
a Schmidt number dependence in deriving the transfer velocity
from the change in the 3He/SF6 ratio (Asher, 2009). Although
results from a lake experiment (Watson et al., 1991) and a
study in the North Sea (Nightingale et al., 2000b) have both
demonstrated Schmidt number dependence of close to n¼0.5,
250
200
150
100
50
0
0 5Wind s
k w_6
60 (c
m h
−1)
Liss and Merlivat (1986)McGillis et al. (2001)Prytherch et al. (2010a)Edson et al. (2011)Best fit to all dual tracer data, Ho et al. (2011)All dual tracer dataEC CO2, GasEx-98, Edson et al. (2011)EC CO2, SOGasEx, Edson et al. (2011)EC CO2, Knorr07,Miller et al. (2009)EC CO2, HIWASE, Prytherch et al. (2010a)EC DMS, multiple campaigns, Yang et al. (2011a)EC DMS, Knorr07, Miller et al. (2009)EC DMS, Knorr06, Marandino et al. (2009)
Figure 6 Summary of field studies of kw from eddy covariance (EC) studies.comparison. EC CO2 measurements are grayscale squares, and EC DMS measpeed binned average presented in the individual study except that of Yang eseries of DMS studies by B. Huebert (University of Hawaii) and collaborators, wThe data of Miller et al. (2009) and Marandino et al. (2009) are not includedJohnson, shared under creative commons license at http://dx.doi.org/10.608
as predicted by the surface renewal model, it is likely that this
value may not apply at higher wind speeds, which were not
covered by the aforementioned studies (e.g., Asher, 2009). This
effect is discussed in Section 8.3.2.3.6.
Different wind speed parameterizations have been pro-
posed from the above dual-tracer studies, the most common
being that of Nightingale et al. (2000b) (Figure 5):
k600 ¼ 0:222u102 þ 0:333u10 [19]
In a recent synthesis of dual-tracer data on kw, Ho et al. (2011)
propose a ‘best-fit’ parameterization through all of the dual-tracer
data with a 95% confidence limit (Figure 6) (eqn [20]):
k600 ¼ 0:262� 0:022ð Þu102 [20]
This best fit overlaps with the parameterizations of
Nightingale et al. (2000b) and the ‘ecumenical’ parameteriza-
tion of Wanninkhof et al. (2009), the latter of which is
described in Section 8.3.2.3.6. The previous three parameter-
izations are plotted on Figure 5 for comparison with mass
balance estimates of k600.
8.3.2.3.4 Direct flux measurementsA number of micrometeorological approaches to direct mea-
surements of gas fluxes are possible. These were developed over
terrestrial surfaces and have been applied to the marine envi-
ronment mainly since the turn of the century (Nightingale and
Liss, 2003), although there were some successful early mea-
surements of ozone and water vapor (e.g., Lenschow et al.,
1982). The methods have been detailed by Nightingale
10peed (m s−1)
15 20
Dual-tracer data and best-fit parameterization (Ho et al., 2011) shown forsurements are pink/gray circles. Each EC data point represents a windt al. (2011a), which presents binned averages from the raw data from ahich represents the majority of published EC DMSmeasurements to date.in Yang et al. (2011a). All data are normalized to k660. Figure by M.T.4/m9.figshare.92419.
66 Air–Sea Exchange of Marine Trace Gases
(2009) and Garbe et al. (2013), so only a brief description is
given here. For parameterizations of gas exchange over the
ocean, the EC method (also known as eddy correlation or
direct covariance) is the most direct, with the fewest assump-
tions about atmospheric conditions. The resulting data are
most informative for understanding gas exchange relationships
with wind speed, which will be the focus of this section.
However, the reader should be aware that other methods
have also been applied over the ocean, such as the relaxed
eddy accumulation (e.g., Zemmelink et al., 2004) and the
gradient flux technique (e.g., Hintsa et al., 2004).
The EC technique relies on high-frequency (typically
>10 Hz) measurements of the concentration of the gas of
interest in the atmosphere and the direction and magnitude
of vertical wind velocity. Each eddy can transport mass, heat,
and momentum in three dimensions. Factoring out lateral
transport, the small differences between updraft and down-
draft concentrations of the scalar of interest (e.g., CO2) can
be used to infer the net flux at the sampling height and thus,
assuming conservation of mass, the flux at the air–sea interface.
To account for the stochastic randomness of turbulent
eddies, covariance measurements need to be averaged (from,
e.g., 10 Hz) to yield representative flux values, typically in in-
tervals of 15–60 min. Because turbulence is inherently ran-
dom, the uncertainty in hourly flux is still on the order of
20%, even for a well-resolved scalar, such as sensible heat
(Fairall et al., 2003). For gas concentration measurements,
where instrumental noise becomes significant at a high sam-
pling rate, the uncertainty in flux is even greater. Data are
commonly averaged and binned to reduce a ‘cloud’ of points
into a more useful format. While there is a healthy skepticism
from some sections of the community regarding this process,
the large number of raw data points means that the data can be
dealt with in a statistically rigorous manner and so yield mean-
ingful results (e.g., Prytherch et al., 2010a), albeit with large
error bars in some cases. Work remains to be done to tease out
the relative contributions of noise and natural variability in the
scatter of EC data. Collection and processing of EC data are
extremely complex and rely on very high-precision accelerom-
eters and sonic anemometers and considerable postsampling
computation, including correction for the ship’s motion and
flow distortion (the latter varying with the former); hence, the
technique was initially difficult to implement at sea. Further
complications include changes in air density with heat and
water fluxes (the so-called Webb effect; Webb et al., 1980),
salt contamination, and sensor corrections (e.g., Prytherch
et al., 2010b) and losses in closed path systems for CO2 sen-
sors. However, physical problems relating to motion and flow
distortion have largely been overcome, andmarine EC trace gas
flux measurements are now mainly limited by the sensors
available for collecting sufficiently high-precision, high-
frequency data. Although workers are starting to look at the
possibility of applying EC techniques to a wider range of gases
(Rowe et al., 2011), including the application of proton trans-
fer reaction/mass spectrometry (PTR/MS) to the EC measure-
ment of the oxygenated volatile organic compounds (OVOCs)
(Mingxi Yang, personal communication, 2012), successful
measurements at sea to date are limited to CO2 (e.g., Edson
et al., 2011; McGillis et al., 2001a; Miller et al., 2009; Prytherch
et al., 2010a) and DMS fluxes (e.g., Blomquist et al., 2006;
Huebert et al., 2004; Marandino et al., 2009; Yang et al.,
2011a). There are also recent studies on ozone (Bariteau
et al., 2010), SO2 (Thornton et al., 2002), and acetone
(Marandino et al., 2005), each of which reports initial results
but does not derive transfer velocity–wind speed relationships.
ECmeasurements of DMS exchange are subject to considerably
less uncertainty than CO2 because (1) DMS concentrations are
far from equilibrium between the atmosphere and ocean so
that the signal-to-noise ratio in the covariance data is much
higher; (2) there is an appreciable gas-phase resistance term for
DMS, yielding a stronger gradient in the atmosphere for a given
concentration difference compared with CO2; and (3) there are
fewer sensor problems for DMS. Figure 6 summarizes the
published EC measurements of gas exchange dependence on
wind speed using DMS and CO2. At low to moderate wind
speeds, these data appear to agree well with each other and
with mass balance estimates. At higher wind speeds, however,
data for DMS and CO2 appear to diverge, with CO2 yielding
considerably higher estimates than mass balance and some
DMS measurements yielding lower estimates. The reasons for
these discrepancies are discussed in Section 8.3.2.3.6.
EC provides a direct measurement of the flux, F, and a
measurement of the gas-phase concentration at a typical mea-
surement height of 15 m above the ocean. Measurements of
the water-phase concentration can thus be used to derive the
transfer velocity, by rearrangement of the flux equation (e.g.,
eqn [4]). Thus, the apparent transfer velocity integrates the true
physical transfer velocity and any nondiffusive effects which
may modulate the flux, such as bubbles, rainfall, and chemical
enhancement. This is, of course, also the case with mass bal-
ance approaches. Measurements of the water-phase concentra-
tion are typically conducted in bulk surface seawater (at 2–5 m
depth) and often at a substantially lower frequency than gas-
phase measurements, leading to further uncertainty in the
method (Johnson et al., 2011).
Wind speed-dependent parameterizations of transfer veloc-
ity have been derived directly from EC CO2 data sets by curve
fitting (Edson et al., 2011; McGillis et al., 2001a; Prytherch
et al., 2010a). Due to the large transfer velocities observed at
high winds and the assumed cubic dependence of whitecap-
ping and thus bubble-mediated transfer on wind speed, cubic
fits have generally been favored, and they also provide a better
fit to the global radiocarbon constraint, although whether this
is useful when comparing with a short-term EC study is debat-
able. These cubic fits to the data are shown in Figure 6.
8.3.2.3.5 NOAA/COAREThe NOAA/COARE gas flux parameterization (Fairall et al.,
2000) is a physically based model developed from the
COARE bulk flux algorithm (Fairall et al., 1996). It incorpo-
rates water column and atmospheric stability, the cool skin
effect, and the parameterizations of bubble-mediated transfer
based on the model of Woolf (1997). NOAA/COARE has two
tunable parameters when used for predicting kw from wind
speed, one relating to the diffusive resistance to transfer
through the water-side mass boundary layer and the other
relating to the magnitude of the solubility-dependent bubble
transfer term. EC measurements of CO2 transfer have been
used to tune these parameters (e.g., Hare et al., 2004), but
quite different tunings have since been proposed based on
Air–Sea Exchange of Marine Trace Gases 67
global modeling studies (e.g., Jeffery et al., 2010), leading to
some uncertainty in the optimal values. A key feature of the
tuned NOAA/COARE algorithm is that the direct diffusive gas
transfer term scales linearly with wind speed, with an addi-
tional bubble component at higher winds that scales with the
cube of the wind speed, following Woolf (1997).
8.3.2.3.6 Reconciling observationsThe earlier summary of data used to derive transfer velocity
parameterizations highlights two important discrepancies
which have not been satisfactorily resolved. As noted above,
significant variability due to processes not directly related to
wind speed is to be expected, but even so, apparent inconsis-
tencies remain.
The most troubling is the higher transfer velocities inferred
for CO2 by EC studies versus that inferred from the 3He/SF6pair by the dual-tracer method at high winds (Figures 5 and 6).
Ho et al. (2011) suggested that the two methods are measuring
empirically different properties. On the timescale of EC mea-
surements (minutes to hours), the apparent transfer velocity at
high winds may vary with the cube of wind speed because of
bubble-mediated transfer, whereas over the timescale of a
tracer study (days), the vertical mixing of the bulk water
might become limiting to the total tracer mass balance, negat-
ing the higher-order dependence. This is not unreasonable,
particularly given the high diffusivity of 3He, which will need
a greater resupply from the bulk to the mass boundary layer by
turbulent mixing than other gases.
Secondly, as noted earlier, the deliberate dual-tracer
method relies on an estimate of the Schmidt number depen-
dence of gas transfer (assumed to be n¼0.5 in all previous
studies). At higher winds, this relationship is likely to break
down, as bubbles and other processes tend to dominate over
surface renewal (Asher et al., 2004). If the diffusivity becomes
less important at higher turbulence and increased bubble
fluxes, reducing the value of n to below 0.5, this will reduce
the differential flux between 3He and SF6 relative to that pre-
dicted, leading to an underestimate of the flux and thus of kw(Asher and Wanninkhof, 1998). Jean-Baptisite and Poisson
(2000) found values of n as low as 0.2 in intermediate to
high winds in a lake study, for example. If this effect is appli-
cable to the open ocean, it could lead to potentially significant
underestimates of transfer velocity by the deliberate dual-tracer
method at high winds.
A third possible error has been identified which is also
related to bubble-mediated gas exchange. The dual-tracer
method fails to account for the solubility dependence of
bubble-mediated transfer. Bubble fluxes are predicted to be
lower for more soluble gases (Section 8.3.2.2.3), as supported
by laboratory experiment (e.g., Rhee et al., 2007; Woolf et al.,
2007). However, such effects mean that the dual-tracer
method, which is based on the exchange of 3He and SF6, two
gases that are considerably less soluble than CO2, should pre-
dict greater fluxes than those directly observed for CO2, which
is the opposite of what is found. These inconsistencies might
be reconciled by the application of an improved representation
of the bubble flux such as that outlined by Woolf et al. (2007),
which accounts for the high void fraction in dense bubble
plumes. This may lead to ‘suffocation’ of the bubble flux (i.e.,
the flux of gas into the bubbles becomes limited by its
decreasing concentration in the water around the bubble
plumes). This effect will be the largest for the most diffusive
gases, so there will likely be a differential in the effect between
the highly diffusive 3He and the rather less diffusive SF6. The
result when calculating kw from the dual-tracer data would thus
be an underestimate of kw (Woolf et al., 2007).
Such explanations may reconcile all the available data and
may also satisfy constraints from 14C and 222Rn where appro-
priate, given that these are likewise subject to significant
uncertainty. Given the uncertainties associated with EC and
dual-tracer methods and estimates of the magnitude of the
effect of bubbles on CO2 fluxes in the ocean environment, it
is impossible to determine which of the earlier explanations for
observed discrepancies between methods is correct. It is possi-
ble that under different regimes of thermal stability, wave field,
fetch, and other forcings, the observed range of kw�U relation-
ships may all be valid in particular situations. With advances in
CO2 sensors for EC measurements (Section 8.3.2.5), progress
may be made toward better understanding and resolution of
the earlier discrepancies, although novel experiments are
needed to test the Schmidt number dependence of gas
exchange at high wind speeds.
The second discrepancy, which calls into question the appli-
cability of generic transfer velocity parameterizations to different
gases, is the observed differences between apparent transfer
velocities for EC CO2 and DMS studies (Figure 6). It is clear
that the majority of DMS data predict lower transfer velocity
than the ECCO2 (or than the dual-tracer or radiocarbon-derived
parameterizations). This has been cited as a strong validation of
the importance of the solubility-dependent bubble effect (e.g.,
Yang et al., 2011a), although this conclusion is controversial
because of the large uncertainties in EC measurements. The
application of the NOAA/COARE algorithm predicts a near-
linear relationship with wind speed for DMS, as its solubility
prevents the bubble effect from significantly modifying its flux
over normal interfacial, diffusion-mediated flux (which scales to
the friction velocity, u*, in the NOAA/COARE algorithm). Thus,
the difference between the k660DMS and k660CO2predicted by
NOAA/COARE represents the magnitude of the bubble-driven
transfer for CO2 (Figure 7). The EC DMS data are strong cir-
cumstantial evidence for this inference, but other explanations
are possible. For instance, at higher wind speeds, due to the
linear nature of ka parameterizations, the air-side resistance is
predicted by the model of Johnson (2010) to increase to nearly
10% of the total resistance for DMS at 15 m s�1. Given the
uncertainty in air-side transfer velocity (Section 8.3.2.3.8), this
could explain at least some of the trend.
Yang (personal communication, 2012) has pointed out
that the effects due to the asymmetry of gas exchange may
contribute to both the discrepancy between dual-tracer and
EC CO2 measurements and between EC measurements of
DMS and CO2. The key is that 3He/SF6 and DMS are all
measured as they flux out of the ocean, whereas measurements
of CO2 under high-wind conditions tend to be made in cold
waters where the flux of CO2 is into the water. This discrepancy
has two implications: (1) any gradient on the water side be-
tween the bulk measurement used to calculate the flux and the
surface where gas exchange occurs (e.g., Jacobs et al., 2002;
Section 8.3.3) will be in opposite directions for evading and
invading gases, leading to apparently greater fluxes for gases
68 Air–Sea Exchange of Marine Trace Gases
fluxing into the ocean, and (2) any bubble effect is likely to be
greater for an invading gas due to the hydrostatic pressure effect
on bubbles, which favors invasion.
Recent DMS data collected at high winds, presented for
modeling purposes by Vlahos et al. (2011), show a flattening
off or even a decrease of kw for DMS above wind speeds of
about 13 m s�1. These data are controversial and thus have not
yet been published in an international journal. However, other
groups are now beginning to see similar effects at winds above
10 m s�1 (Tom Bell, personal communication, 2012).
Yang et al. (2011a) propose a mechanism that can explain
these observations, following Banner and Peirson (1998): as
wind increases, an increasing proportion of the total wind stress
is partitioned into wave stress and away from tangential stress.
The increase in form dragmay subsequently inhibit microbreak-
ing waves, which may decouple the dependence of diffusive
transfer on wind speed. Thus, increasing wind stress may en-
hance kw for a relatively insoluble gas, such as CO2, through
dominance of wave-related bubble processes, but at high winds,
it begins to inhibit the transfer of non bubble-mediated gases,
such as DMS. This is a compelling insight, but the underlying
theory is largely untested in the field, so the concept remains
hypothetical. An alternative explanation is that the bubble scav-
enging of (soluble) surfactants (Liss, 1975) may suppress
0 5 10Wind speed (m s−1)
15 20
k w_6
60 (c
mh−1
)
0
40
80
120
160EC CO2 dataEC DMS dataBubble k(CO2_660), Woolf (1997)Bubble k(DMS_660), Woolf (1997)kw_DMS, Woolf (199)7, Johnson (2010)
Figure 7 Demonstration of the bubble effect on DMS versus CO2
transfer. EC DMS data (pink circles) and CO2 data (gray squares) asshown in Figure 6, plotted over the predicted k660, normalized transfervelocities from the bubble model of Woolf (1997) for gases of thesolubility of DMS (dark pink solid line) and CO2 (solid black line), usingsolubility at 15 �C in the scheme of Johnson (2010). For DMS, the bubbleeffect is small, so the pink shaded area effectively represents the diffusivetransfer component for both CO2 and DMS, with the gray shaded areashowing the additional bubble-mediated transfer for CO2. Total transfervelocity, Kw, for DMS is also calculated using the scheme of Johnson(2010), applying the Woolf (1997) bubble parameterization for kw and theJohnson (2010) ka term. This demonstrates the potential error in suchcalculations by not including air-side resistance for DMS and gases ofsimilar solubility. Figure by M.T. Johnson, shared under creativecommons license at http://dx.doi.org/10.6084/m9.figshare.92419.
diffusive transfer at higher wind speeds, which would affect
DMS transfer but arguably not CO2, as its transfer may be
bubble-driven at these wind speeds.
8.3.2.3.7 Remote sensing of kwEstimation of kw in the field is technically difficult and results
are spatially and temporally limited. There is thus a role for
applying remote sensing techniques to estimate transfer veloc-
ity directly, both from the point of view of the global coverage
and the frequency of measurements. Such approaches are par-
ticularly relevant to the global estimates of air–sea CO2 ex-
change, where uncertainties in the net global flux from
applying different kw–U parameterizations are large (Table 2).
The use of remotely sensed winds from scatterometer data
for estimating transfer velocity on global and regional scales
is well established (e.g., Etcheto and Merlivat, 1988; Naegler
et al., 2006). However, there are substantial differences in
global mean wind speed from QuikSCAT wind products,
thought to be an improvement for their higher resolution,
compared with global reanalysis products such as NCEP
(Wanninkhof et al., 2009). Remotely sensed winds are used
to derive kw from an empirical kw–U parameterization or pa-
rameterizations (e.g., Boutin et al., 2002). A typical map of
global annual kCO2values derived from satellite observations is
shown in Figure 8.
A recent improvement to this approach uses multiple
remotely sensed parameters (air and water surface temperature,
humidity, wind speed, and long- and short-wave radiative
fluxes) to derive kw via the NOAA/Coupled Ocean Atmosphere
Response Experiment Gas (COAREG) algorithm (Jackson
et al., 2012). When compared with shipborne observations,
this approach was found to agree well, implying that COAREG
can be applied using satellite inputs with accuracy comparable
to that achieved with ship-based observations. The agreement
of predicted gas fluxes compared with EC measurements of
CO2 fluxes was reasonable, suggesting that this approach will
be adopted more widely in the future.
Table 2 Estimates of the annual oceanic uptake of carbon dioxide(Gt C year�1) derived from measurements of carbon dioxide and gastransfer velocity parameterizations
Parameterizationof kw
Annual oceanic uptake of carbon dioxide(Gt C year�1)
Boutin et al.(2002)
Wanninkhof andMcGillis (1999)
Takahashiet al. (2002)
Liss and Merlivat(1986)
1.2
Wanninkhof(1992)
2.2 1.4 2.0
Wanninkhof andMcGillis (1999)
2.7 2.2 3.7
Nightingale et al.(2000b)
1.7
Note that Takahashi et al. (2009) present a substantially lower estimate of the net global
ocean uptake of carbon of approximately 1.6 Tg C year�1, in line with that of Boutin
et al. (2002) (with Nightingale et al., 2000b), largely because of their use of a kwparameterization (of their own derivation) predicting lower transfer velocities than
Wanninkhof (1992) and Wanninkhof and McGillis (1999).
Figure 8 A global map of the annual average gas exchange coefficientfor carbon dioxide for the year 2001. The map was derived from acombination of wind speeds derived from the QuikSCAT satellite sensorand the W92 (Wanninkhof, 1992) parameterization of kw. Note that kwhas been corrected for the solubility of CO2 in this figure. Reproduced bykind permission of Dr. J. Boutin, Laboratoire d’OceanographieDynamique et de Climatologie (LODYC), Universite Pierre et Marie Curie,Paris, France.
Air–Sea Exchange of Marine Trace Gases 69
Another possible approach uses the laboratory finding that
kw is linearly related to the fractional area of the water surface
covered by whitecaps (Asher et al., 1998). Since it is potentially
possible to estimate whitecap coverage from the microwave
brightness temperature of the sea surface, which can be mea-
sured by satellite or airborne radiometers, this may represent a
way forward for the estimation of kw on the wide range of
scales provided by use of these various platforms. Recently,
Goddijn-Murphy et al. (2011) used satellite retrievals to better
constrain the wind speed–whitecapping relationship, which
has been a considerable source of uncertainty in applying
bubble models of gas exchange because of the huge variation
in empirical parameterizations of the said relationship from
earlier field studies.
8.3.2.3.8 Air-side transfer velocityFor more soluble gases, where the air-side transfer velocity
becomes limiting, the relationship between wind speed and
air-side transfer velocity (ka) becomes important. This is much
less well constrained than kw, with little or no field validation
of the empirical models for gases other than water vapor (e.g.,
Johnson, 2010). Typically, however, ka�U parameterizations
are close to linear (e.g., Jeffery et al., 2010; Johnson, 2010), and
physically based model predictions (e.g., Duce et al., 1991;
Jeffery et al., 2010) tend to agree reasonably with the few
wind-tunnel observations for trace gases by Mackay and Yeun
(1983). Future direct measurements of soluble trace gas fluxes
by EC methods will enable the testing of these parameteriza-
tions and the elucidation of the gas-specific relationship, in-
cluding any possible enhancement to transfer at high winds by
sea spray (Section 8.3.2.2.3).
8.3.2.4 Estimating Trace Gas Fluxes in BiogeochemicalStudies
The gas exchange community has expended a great deal of
effort to understand the mechanisms, drivers and relationships
with environmental forcings of gas exchange. As demonstrated
in Section 8.3.2.3.6, this work casts doubt on the applicability
of simple parameterizations to gases of differing solubility and
diffusivity. Where does this leave the biogeochemist wishing to
calculate the flux of a gas across an air–water interface based on
the concentration measurements of their gas of interest? Such
studies cannot be expected to tackle the big problems currently
facing the gas exchange community, but they must acknowl-
edge them and attempt to mitigate their effects on a study-by-
study basis. The following sections provide some guidance
on these issues and consider other potential pitfalls and
uncertainties, which should be mitigated where possible. Ulti-
mately, as long as the best current parameterizations are ap-
plied consistently and traceably by biogeochemical gas flux
studies, then if the concentration data remain good, the fluxes
can be revisited in the future as developments in understand-
ing and parameterizations improve. It is the concentration
gradient that is the key measurement in such studies; the
transfer velocity is a means to an end.
8.3.2.4.1 Gas-specific effects on transferWhen estimating the ocean–atmosphere flux of a particular
volatile compound, it is important to account for the gas-
specific properties, which will influence the approach taken,
or the values of parameters used. The two key properties are the
Henry’s law solubility (H) and the Schmidt number (Sc) of the
gas of interest. While these are best taken frommeasured values
in seawater, such measurements may not be available for the
gas in question, and even if they are, the quality of individual
measurements in the literature is sometimes not known. Fur-
thermore, their temperature and salinity dependence may or
may not cover the range required. For this reason, Johnson
(2010) compiled a set of formulations to calculate the air- and
water-side Schmidt numbers, employing temperature- and (for
water parameters) salinity-dependent methods from the litera-
ture. To complement this, Johnson (2010) also derived a novel
salinity dependence for the solubility of any gas, based on ion–
ion interaction and scaled particle theories, fitted to empirical
data. The scheme provides the capability to predict temperature-
and salinity-dependent Schmidt numbers and Henry’s law con-
stants for any gas given its molecular structure (or molar volume
at boiling point, where available) and its Henry’s law constant in
pure water and temperature-dependence term (available for
most compounds, e.g., Sander, 1999).
As described in Section 8.3.2.1.1, the solubility of the gas
determines the partitioning of the resistance to transfer be-
tween the air and water phases. This can be determined by
calculating the air- and water-phase resistance terms, as in, for
example, eqn [5], to determine whether one or both terms
need to be taken into account. Alternatively, a ‘belt and braces’
approach can be taken and the total transfer velocity (Kw or Ka)
can be calculated routinely, which can be done with the soft-
ware provided by Johnson (2010) for a selection of transfer
velocity parameterizations on either side of the interface. This
includes the diffusionþbubble kw model of Woolf (1997), as
used in the NOAA/COARE algorithm, which may be used to
determine the potential significance of bubble-mediated trans-
fer for the gas.
A further important consideration is whether or not the
transfer of the gas of interest might be chemically or biologi-
cally enhanced on the water side of the interface. This is a case
70 Air–Sea Exchange of Marine Trace Gases
of considering each possible reaction and comparing it to rates
of transfer across the mass boundary layer for a given wind
speed. Hoover and Berkshire (1969) presented a method for
calculating the degree of chemical enhancement of CO2 ex-
change by its hydration reaction in the mass boundary layer
(Section 8.3.2.2.6), by considering the psuedo first-order lim-
iting rate constant for the overall reaction and the capacity of
the water to take up the hydrated CO2 (a function of pH and
alkalinity). The method, which assumes the stagnant film
model, was used by Liss and Slater (1974) to estimate the
degree of chemical enhancement of SO2 transfer, and the
Hoover and Berkshire (1969) equation can be generalized to
any similar reaction (eqns [21] and [22]):
a ¼ t
t� 1ð Þ þ tanh X½ �X
[21]
where a is the chemical enhancement factor (a multiplication
factor for kw) and X is a function of the pseudo first-order rate
constant, z, the mass boundary layer thickness (z), and the
diffusivity of the gas in the medium (D) (eqn [22]):
X ¼ ztD
� �0:5
�Z [22]
The term t�1 represents the ratio of the unreacted to
reacted forms of the gas of interest, for example, NH3/NH4þ
for the protonation of ammonia (t thus being (NH3þNH4þ)/
NH4þ in this case). The function tanh(X)/X is equal to one at
small values of X, that is, where the rate of chemical reaction is
slow relative to the timescale of diffusion across the film layer.
In this case, eqn [21] simplifies to a¼1. Where the rate of
chemical reaction is sufficiently fast relative to the diffusion
timescale, the maximum attainable chemical enhancement is
a¼t /t�1), that is, the enhancement is limited by the apparent
solubility due to the reaction in question (the ‘effective’
solubility) or the buffering capacity of the system.
In an unpublished work, M.T. Johnson and P.S. Liss have
solved the Hoover and Berkshire (1969) equation ‘in reverse’ for
the rate constant, z, required for a given enhancement factor.
The results demonstrate that a rate constant of the order
10�3 s�1 or greater would be required to give a chemical en-
hancement factor to kw of 1.1 (i.e., 10% enhancement) at a wind
speed of 3 m s�1. Thus, a good guideline would be that if a
reaction happens at this rate or faster, the effects of chemical
enhancement should be further investigated, as it may be im-
portant, at least at low wind speeds. For comparison, the rates
of protonation reactions are commonly of the order of 109 s�1,
the hydrolysis of SO2 about 106 s�1 (Liss, 1971), and the pho-
tolysis of diiodomethane and hydrolysis of peroxyacetyl
nitrate (PAN) about 10�3 s�1. In the context of chemical en-
hancement rates, these correspond to enhancement factors of
the order 104–105 for NH3 and 103 for SO2 at seawater pH but
of only 1.1 for PAN and diiodomethane at 3 m s�1 and less at
higher winds.
8.3.2.4.2 Selection of wind speed parameterizationThe choice of the most appropriate parameterization for a
particular study depends on both the forcing data available
and the application to which kw is to be employed: the nature
of the problem, the spatial and temporal scales, the gas of
interest, etc. For example, when considering fluxes of CO2 at
the global scale, there is a constraint on the transfer velocity
given the bomb 14C inventory of the ocean. Notwithstanding
the uncertainties in both the inventory (Naegler, 2009) and the
global average wind speed (Wanninkhof et al., 2009), there is a
‘known point’ on the kw–U curve for global CO2 fluxes. Note
that while parameterizations derived from this value have
tended to apply a quadratic relationship, there is no evidence
from this global approach to support any particular form for
the kw–U relationship; a cubic relationship, for example, could
be fitted (e.g., Wanninkhof and McGillis, 1999). Given the
relatively well-constrained concentrations of CO2 in the atmo-
sphere and ocean, the uncertainty introduced into global flux
estimates by choice of transfer velocity is rather large; the
difference between the values of Wanninkhof (1992) versus
Wanninkhof and McGillis (1999) leads to a near doubling of
the net global ocean uptake of CO2, from 2.0 to 3.7 Gt C
year�1 (Takahashi et al., 2002). One can draw two conclusions
from this. The first is that the form of the relationship of kwwith U is of prime importance. This is due to a correlation
between wind speeds and direction of CO2 flux, that is, windy
areas tend to be sinks of CO2 and calm areas tend to be sources
of CO2. The second is that the choice of wind speed data and
the regional distribution thereof can have a large impact on the
estimate of net global CO2 flux.
While global 14C estimates of mean kw are a valuable con-
straint on long timescale global fluxes, they are not necessarily
a constraint for regional or local fluxes or over shorter
timescales. Where wind speed is the only forcing for which
data are available in a study on a smaller-than-global scale (as
is commonly the case), an empirical kw–U parameterization
must be employed to quantify the flux of a gas using wind
speeds averaged over an appropriate timescale. The most com-
monly applied parameterization in such studies is probably
still that of Wanninkhof (1992), which is inappropriate because
(1) it is based on a global constraint, which is arguably not
relevant to smaller scale studies, and (2) the value of the global14C inventory on which it is based has been demonstrated to be
an overestimate (e.g., Naegler, 2009). There is the possibility
that the dual-tracer method may underestimate CO2 flux
(Section 8.3.2.3.6), but the burden of proof currently rests
with the gas exchange community to prove that the dual-tracer
method is flawed. Until such time, the most appropriate param-
eterization to use is probably the one derived from the dual-
tracer method (e.g., Ho et al., 2011; Nightingale et al., 2000b),
with the caveat that there are probably effects which are unac-
counted for, particularly bubbles.
With sufficient environmental forcing data, probably the
best estimates of transfer velocity can be achieved by using
physically based models of boundary layer interactions, such
as NOAA/COARE. The latest incarnation, COAREG 3.0 (Fairall
et al., 2011), which includes a generalized scheme to enable
application to any gas following Johnson (2010), might be
considered the ‘state of the art’ in quantifying kw and ka (and
thus total transfer velocity). Nonetheless, it still contains some
physical parameterizations that are not current. For instance, it
applies the ‘classic’ bubble model of Woolf (1997) rather
than the dense bubble plume model (e.g., Woolf et al.,
2007), and it relies on rather uncertain empirical
Air–Sea Exchange of Marine Trace Gases 71
parameterizations such as the empirical wind speed–whitecap-
ping relationship of Monahan and O’Muircheartaigh (1980),
which is subject to significant uncertainty (Johnson et al.,
2011), although recent developments have constrained it fur-
ther (Goddijn-Murphy et al., 2011).
Wanninkhof et al. (2009) present a parameterization that is
mechanistically consistent with people’s current understand-
ing of the gas exchange processes. It is a polynomial relation-
ship with a constant value representing diffusive flux at zero
wind, a linear term relating to the smooth surface regime, a
quadratic term for the rough surface regime, and a cubic term
for bubble-mediated transfer (eqn [23]). This is a valuable
attempt to reconcile current knowledge and arguably the
most consistent to use over a range of studies:
K660 ¼ 3þ 0:1 U10ð Þ þ 0:064 U102
� �þ 0:011 U103
� �[23]
The available parameterizations of ka are rather limited, but
those that exist are rather similar, given certain constraints,
such as the formulation of the drag coefficient (Johnson,
2010). In the transfer velocity scheme proposed by Johnson
(2010), the formulation of Jeffery et al. (2010) was adopted,
which parameterizes the NOAA/COARE scheme for water
vapor and applies it to trace gases.
The selection of an appropriate transfer velocity must be
based on the requirements of the particular study. It is impor-
tant to recognize that a given parameterization is not necessar-
ily universally applicable to gases of differing solubilities nor
to differing environmental situations. Given the wide range
of other uncertainties in quantifying air–sea gas fluxes
(Section 8.3.2.4.4), however, the uncertainty in kw parameter-
izations is often relatively small.
8.3.2.4.3 Averaging and interpolation/extrapolationTo quantify the local, regional, or global air–sea flux of a trace
gas, it is necessary to interpolate between spatial measurements
and/or extrapolate from near-instantaneous measurements in
time and then apply representative environmental parameters
to derive a flux. Furthermore, averaging of measured data is
necessary to generate preinterpolation fields for 2D estimates
from discrete measurements (e.g., Lana et al., 2011). All of
these processes introduce uncertainties, some of which are
not well quantified.
Biological production and consumption and physical mix-
ing can lead to short-term/small-scale variability in concentra-
tions, which means that all but the highest resolution
measurements are rather difficult to average. For highly bio-
reactive compounds, this is particularly problematic. For in-
stance, the standing stock of ammonia in surface waters can be
turned over on timescales of a few hours to a day in highly
productive waters, and the decoupling of production and loss
processes can lead to transient spikes in ammonium concen-
tration an order of magnitude above the ambient (Johnson
et al., 2007). Other gases, which are rapidly turned over by
biological or chemical processes, may be subject to similar
transients due to other decouplings. In evidence of this, the
frequency distributions of many data on trace gas concentra-
tions in seawater are positively skewed, with a long ‘tail’ of high
values. Combined with spatial heterogeneity due to local-scale
physical processes, this means that low spatial/temporal
resolution measurements in seawater can readily yield unre-
presentative concentrations over periods of about a day on a
research cruise. Gas-phase concentrations are more likely to be
measured continuously, but integrating over long periods can
likewise introduce uncertainties.
In order to calculate fluxes from extrapolated concentrations,
it is necessary to average temperature, salinity, and wind speeds.
Temperature is well constrained, with an average error from
satellite measurements of less than 1 �C (Wanninkhof et al.,
2009) and much less from direct measurements. Salinity, too,
is well constrained in the field and in climatological data and is a
relatively insignificant driver of gas flux variability (Johnson,
2010). Wind speed is the key uncertainty here.
In field studies quoting near-instantaneous fluxes from
concentration measurements, it is common to use 7-day aver-
aged winds as representative (e.g., Hughes et al., 2009) or,
where there is a long atmospheric integration time over a
spatial range, to apply the mean wind speed over that same
time period (e.g., Johnson et al., 2008). Either is a valid
approach, but they may yield substantially different results.
When extrapolating over wider spatial and temporal scales,
longer term-averaged and/or time-varying wind data such as
NCEP reanalysis data or satellite-retrieved winds may be used.
Reanalysis and satellite data can differ significantly due to the
lack of resolution over short time and spatial scales in the
former, which leads to a 1.3 m s�1 difference in the global
annual average wind speed between NCEP and QuikSCAT
(Naegler et al., 2006), and to variability in spatial wind distri-
butions (Wanninkhof et al., 2009). Equally important, where
variable long-term wind data are employed, it is incorrect to
use mean wind speed to drive a wind speed-dependent transfer
velocity parameterization if the k–U relationship is nonlinear.
This is because the square of the mean is not equal to the mean
of the square wind speed, a circumstance which lead
Wanninkhof (1992) to present two kw–U parameterizations,
one for short-term and one for long-term averaged wind data.
Where variables driving the flux (concentrations, wind,
temperature, and salinity) are decoupled (i.e., they are ran-
domly associated and do not covary), a probabilistic approach
can be taken to flux estimation. Probability density functions
of fluxes can be estimated based on all possible combinations
of the observed values of the controlling variables, rather than
a simple arithmetic average of spot measurements, their asso-
ciated driving variables, and calculated fluxes. Hughes et al.
(2012) use this approach to calculate the likely distribution of
fluxes of bromoform and dibromomethane from the coastal
Antarctic, to provide a robust comparison of the effect of two
differing biogeochemical regimes on bromocarbon emissions,
and to ensure that biases introduced by the chance co-
occurrence of extreme values of variables did not bias averag-
ing. After testing to ensure that the variables were decoupled
(e.g., R2<0.1), they identified the nature and numerical char-
acteristics of the distributions of all variables (normal, log-
normal, etc.) and sampled the resulting populations using
the Monte Carlo method. To ensure optimal coverage of pa-
rameter space, they employed a Latin hypercube technique to
sample 10000 combinations of controlling variables, which
they used to produce the flux distributions. The result was
average fluxes that were reasonably similar to arithmetic mean
fluxes, suggesting that the sampling regime had been sufficient
72 Air–Sea Exchange of Marine Trace Gases
to cover the parameter space. Hughes et al. (2012)were also able
to predict the likely range of fluxes and their probabilities, which
would not have been possible from normal averaging, and so
give a stronger idea of the differences between biogeochemical
regimes. Hughes et al. (2012) adopted the kw of Nightingale
et al. (2000b), after Johnson (2010), and did not investigate the
effect of different transfer velocity parameterizations. Such an
analysis, however, could have been included in their method.
Theirs is a first attempt to calculate probability density distribu-
tions of fluxes from observations, and such approaches are likely
to be refined in the future.
8.3.2.4.4 Uncertainty in estimatesThe choice and application of transfer velocity introduces con-
siderable uncertainty into gas flux calculations. In a typical
biogeochemical gas flux study, however, there are likely other
significant uncertainties relating to the concentration differ-
ence term (DC). Johnson et al. (2011) have reviewed these
uncertainties and identified key issues, from measurement
and parameterizations through to processes and experimental
design. They highlight as key uncertainties the poorly quanti-
fied gas exchange processes, including bubbles, thermal
stability, and microlayer effects, and the uncertainties in DCresulting from averaging and interpolation/extrapolation of
concentration data, which are generally sparse in space and
time for most trace gases. Extrapolating to the regional or
global scale is particularly difficult. The recently published
update to the global seawater DMS database (Lana et al.,
2011) contains approximately 50000 discrete measurements,
making DMS the secondmost measured trace gas in the surface
ocean after CO2. Nonetheless, the uncertainty introduced by
interpolation and extrapolation, as estimated by Lana et al.
(2011), to yield a regionally resolved estimate of the net global
DMS flux from the ocean, is similar to the difference in global
flux estimates derived from the concentration and wind fields
using the approaches of Liss and Merlivat (1986) versus
Wanninkhof (1992) to calculate the flux. For gases that are
measured less frequently, an estimate from the concentration
alone, without attempting to model underlying biogeochemi-
cal production and loss processes, is probably insufficient to
attempt anything other than single-point global average flux
estimates. Even in studies of a smaller spatial scope, the uncer-
tainty associated with quantifying concentrations is likely to
dominate in many cases (Johnson et al., 2011).
8.3.2.5 Future Developments
Quantifying and understanding mass transfer across air–water
interfaces is a large field of research encompassing a wide range
of disciplines, from oceanography to chemical engineering,
and approaches, including laboratory, field, and modeling
studies of EC, tracers, and heat and momentum fluxes. It is
hard to predict where the field will be in five or ten years time,
so the authors focus here on key recent developments that are
likely to further people’s understanding of trace gas exchange
in the future.
The apparent discrepancy between dual-tracer and EC mea-
surements of gas transfer may be resolved by improved sensors
for CO2 concentration. While sensors are continually being
improved (e.g., Prytherch et al., 2010b), a step change in
sensitivity is needed to reduce error in EC CO2 measurements
substantially. Various groups are working on new instrumen-
tation, including cavity ring-down techniques (Edson et al.,
2011) and photoacoustic spectrometry (Garbe et al., 2013).
Physical models of gas exchange based on parameters other
than or in addition to wind speed are likely to become more
widely used, particularly in combination with satellite-derived
variables. Satellite retrievals of whitecap coverage, a key to
current bubble models and poorly predicted from wind speed,
have recently been used to good effect (Goddijn-Murphy et al.,
2011). Mean square wave slope, a better predictor of diffusive
transfer than wind speed (Frew et al., 2004), has been retrieved
from scatterometer data (Glover et al., 2002, 2007) and from
satellite lidar (Hu et al., 2008).
Current gaps in our knowledge include the magnitude of
the water-side transfer velocity and processes controlling it at
low winds, the magnitude of surfactant effects, particularly at
high winds, and the change in scaling of the Schmidt number
with turbulent forcing. There are few empirical data from trace
gas studies to validate the physically based parameterizations
of the air-side transfer velocity and the magnitude of potential
gas-specific effects, such as chemical or biological enhance-
ment. Direct measurements of gas transfer by EC for a selection
of gases that covers a range of solubility and diffusivity, in
concert with tracer techniques and satellite observations,
would be a major step forward. Generally we are limited by
our inability to make sufficiently rapid and sensitive measure-
ments of gas concentrations for EC applications.
8.3.3 The Cycling of Trace Gases Across theAir–Sea Interface
As discussed previously, the flux of gas across the air–sea inter-
face is typically estimated from the concentration difference
driving the flux (DC) and the air-side (ka) and water-side (kw)
resistances. So far, the authors have covered ka and kw in detail;
they now discuss the DC term. Gases of environmental signif-
icance are predominantly biogenic in origin but some are also
anthropogenic. Although measurements of concentrations of
relevant compounds in seawater and air in some cases require
significant methodological development, they are often rou-
tine and so are not described here. However, as most gases of
interest are produced and/or destroyed within the ocean or
atmosphere, there is a great deal of spatial and temporal vari-
ability in their concentration fields. Some of the processes that
control this variability and so determine the observed satura-
tion anomalies are briefly discussed. Where appropriate, an
estimate of the marine contribution to the total atmospheric
flux of the gas under discussion is also included.
The key assumption of the thin film model (see
Section 8.3.2.1.1) is that the main bodies of air and water
are well mixed, that is, that the concentration of gas at the
interface between the thin film and the bulk fluid is the same
as in the bulk fluid itself, and that any production or removal
processes in the thin film are slow compared to the transport
across it. However, it is quite likely that there are near-surface
gradients (microlayer to �5 m) in the concentration of many
photochemically active gases (Plane et al., 1997). Defining and
sampling the surface microlayer remains challenging. There
Air–Sea Exchange of Marine Trace Gases 73
seems little doubt that it exists, as it can be measured in terms
of temperature difference (Section 8.3.2.2.4), a distinct bio-
logical community (Gladyshev, 1997), and its chemical prop-
erties (Hunter, 1997). It can probably best be thought of as a
series of sublayers (Hardy et al., 1997) and is usually opera-
tionally defined as the top 1 mmof the ocean surface, although
its depth depends in part on the property being studied; for
gases, its depth is probably less than 50 mm. Little research has
been published, but near-surface gradients in levels of CO,
DMS, acetone, and acetaldehyde have been reported (Law
et al., 2002; Yang et al., 2001; Zemmelink et al., 2005; Zhou
and Mopper, 1997). The key assumption in most flux calcula-
tions is that the concentrations measured at a typical sampling
depth of 4–8 m are the same as immediately below the micro-
layer. Due to variations in factors, such as bacterial abundance,
nutrient availability, DOM concentrations, and changes in
solar insolation, among others, this may often be incorrect.
8.3.3.1 Greenhouse Gases
8.3.3.1.1 Carbon dioxideAtmospheric CO2 has increased from about 280 ppm in 1800
(and several thousand years before) to almost 400 ppm in
2012 because of fossil fuel burning and deforestation, which
presently supply about 7.2�0.3 Gt C year�1 to the atmosphere
(IPCC, 2007). The observed annual increase in atmospheric
CO2 represents about 4.1�0.1 Gt C; the balance is removed
from the atmosphere and taken up by the oceans and land. The
ocean sink of this anthropogenic CO2 has been estimated to be
2.2�0.5 Gt C year�1 (IPCC, 2007).
The oceans are strongly buffered for CO2, and much of the
anthropogenic CO2 absorbed is converted to other forms of
dissolved inorganic carbon (DIC). At a typical seawater pH of
80�
80�
-108 -96 -84 -72 -60 -48 -36 -24
Net flux (g C
-12 0
70�
70�
GMT 2008 Apr 1 13:42:53
60�
60�
50�
50�
40�
40� 60� 80� 100� 120� 140� 160�180� 160
40�
30�
30�
20�
20�
40� 60� 80� 100� 120� 140� 160�180� 16020�
20�
10�
10�
0�
Figure 9 Contemporary annual air–sea CO2 fluxes for the year 2000 from aWanninkhof R, et al. (2009) Climatological mean and decadal change in surfacResearch Part II 56: 544–577, http://dx.doi.org/10.1016/j.dsr2.2008.12.009,
8, the dominant DIC species is the bicarbonate ion (HCO3�);
only 1% remains in the form of dissolved CO2. The medium-
term sink for anthropogenic CO2, over hundreds to thousands
of years, is thus thought to be the ocean, via transfer from
the atmosphere (Archer et al., 1997, and Chapter 8.10;
Watson and Orr, 2003). The major rate-limiting step in the
oceanic uptake of anthropogenic CO2 is not air–sea gas
exchange but the mixing of the surface waters with the deep
ocean (Sarmiento and Sundquist, 1992).
Large multinational research programs have determined
CO2 levels across the world’s oceans and in different seasons
in order to quantify the DC term of eqn [1] and so to estimate
the global air–sea flux from these measurements, about
3 million of which were compiled by Takahashi et al. (2009)
(Figure 9).
The transfer velocity term is derived from a combination of
U-based parameterizations and global maps of U (see
Section 8.3.2.1.2 and Figure 8). Estimates of the oceanic up-
take of anthropogenic CO2 by these techniques can vary sig-
nificantly depending on the parameterization of kw and the
wind speed distribution used. This approach has tended to give
lower estimates of uptake than those derived from global
circulation models and atmospheric isotopic measurements
(Battle et al., 2000; Keeling et al., 1996; Sarmiento et al.,
2000). The use of the recent, much larger DpCO2 data sets
(Takahashi et al., 2009) has produced much better agreement,
although the tendency for the climatological DpCO2 approach
to give a smaller flux remains: 2.2�0.3 Gt C year�1 from
inverse modeling and 1.9�0.7 Gt C year�1 from pCO2 clima-
tology (Gruber et al., 2009).
There is also a huge natural flux of CO2 between the ocean
and the atmosphere of almost 90 Gt C year�1 that was believed
to be almost in balance prior to 1800. There was probably a net
m-2 year-1)
12 24 36 48 60 72 84 96 108
80�
70�
�140�120�100� 80� 60� 40� 20� 20�0�
�140�120�100� 80� 60� 40 20� 20�0�
60�
70�
80�
60�
50�
50�
40�
40�
30�
30�
20�
20�
10�
10�
0�
pCO2 climatology. Reproduced from Takahashi T, Sutherland SC,e ocean pCO2, and net sea-air CO2 flux over the global oceans. Deep-Seaby permission of Elsevier.
74 Air–Sea Exchange of Marine Trace Gases
flux from the ocean to the atmosphere of about 0.6 Gt C year�1
that balanced the supply of DIC to the oceans via rivers
(Sarmiento and Sundquist, 1992). This huge influx and efflux
is due to a combination of marine productivity (the biological
pump) and ocean circulation (the solubility pump). The biolog-
ical pump is discussed in detail in Chapters 8.4, 8.10, and 8.18.
Ocean circulation also results in air–sea exchange of CO2, as the
solubility of CO2 is temperature-dependent: warming decreases
solubility and there is a net transfer from the oceans to the
atmosphere, whereas the opposite occurs on cooling. Almost
all of the anthropogenic CO2 is thought to be taken up by the
solubility pump (Sarmiento and Gruber, 2006), as CO2 avail-
ability does not normally limit biological productivity in the
world’s oceans. Indeed, the inclusion of biological processes
in the oceanic global circulation models increased estimates
of the oceanic sink of anthropogenic CO2 by only 4.9% (Orr
et al., 2001).
That the net uptake of anthropogenic CO2 is only about 2%
of the CO2 cycled annually across the air–sea interface should
be of major concern, as it suggests that the net flux is extremely
sensitive to changes in the cycling of CO2. Any changes in the
ocean circulation or the biogeochemistry of the mixed layer
could have a major impact on the magnitude or even the sign
of the present net sink of CO2 and hence on the Earth’s climate.
The flux of anthropogenic CO2 into the oceans, for example,
will reduce the pH of surface waters in a process termed ocean
acidification (Chapter 8.19). Changes in pH have been shown
to affect biogenic calcification (e.g., Riebesell et al., 2000) and
could affect the community structure within the water column,
thus altering the biological pump. The subject of ocean acidi-
fication is currently the topic of intensive research; a summary
of this work can be found in Chapter 8.19 and in the book
edited by Gattuso and Hansson (2011).
8.3.3.1.2 MethaneMethane (CH4) is second only to CO2 in its ability to influence
global warming as a greenhouse gas. It has an atmospheric
lifetime of 12 years (IPCC, 2007) and is present at concentra-
tions in the range of 1720–1860 ppb (Frankenberg et al., 2011;
Manning et al., 2011). It is a dominant sink pathway for the
hydroxyl (OH) radical, thereby influencing the oxidizing ca-
pacity of the atmosphere (Bates et al., 1996). The distribution
of CH4 in the ocean is controlled largely by biological pro-
cesses, including methanogenic and methanotrophic bacteria.
Methanogenesis is a form of bacterial respiration that occurs in
anaerobic conditions and leads to the production of CH4.
Thus, the sediment-rich estuaries, the seafloor, and the oxy-
gen-depleted zones, such as the base of the mixed layer, are ‘hot
spots’ for CH4 production (Kelley and Jeffrey, 2002; Schmale
et al., 2010; Upstill-Goddard et al., 2000; Zhou et al., 2009).
Methanogenesis also occurs in anoxic microzones present in
the water column, such as detritus and the guts of zooplankton
(Marty, 1993; Oremland, 1979). Methanotrophic bacteria can
utilize CH4 for energy and as a source of carbon for assimila-
tion into cellular material (Murrell and McDonald, 2000),
which provides an oceanic sink. Overall, the ocean represents
a CH4 source to the atmosphere, releasing about 4–15 Tg CH4
year�1 (IPCC, 2007, and references therein). This is small
(maximum of 2.6%) compared with the overall global source
of CH4 of about 582 Tg year�1 (IPCC, 2007).
8.3.3.1.3 Nitrous oxideNitrous oxide (N2O) is a trace gas with a global warming
potential 298 times that of CO2 over a 100-year horizon
(IPCC, 2007). N2O is long-lived in the atmosphere (114
years) (IPCC, 2007) and can therefore be transported long
distances to the stratosphere, where it represents the major
precursor to the ozone-depleting nitric oxide (NO) radical
(Crutzen, 1970). The marine environment is a net producer
of N2O via microbial nitrification and denitrification. The
oxygen status of the ocean largely determines which areas
will be significant sources of N2O to the atmosphere, with
the yield highest under hypoxic and suboxic conditions
(Codispoti, 2010). In addition to sediments, the pycnocline
and oxygen minimum zones, estuaries, and upwelling regions
are also supersaturated with N2O due to the enhanced eutro-
phication and a high abundance of suspended particles (de
Wilde and de Bie, 2000; Nevison et al., 2004; Rees et al., 2011).
Coastal regions also exhibit elevated N2O fluxes to the atmo-
sphere, particularly those which are subject to a high riverine
input (Ferron et al., 2010), because N2O has a strong negative
relationship with salinity; N2O dissolved in river water ‘salts
out’ as it mixes with seawater (Barnes and Owens, 1998). This
can be compared to the open ocean, which is typically close to
air–sea equilibrium (Charpentier et al., 2010; Forster et al.,
2009) and represents about 40% of the annual oceanic emis-
sions to the atmosphere (Bange et al., 1996). Bange (2006)
revised the oceanic emission of N2O to the atmosphere and
concluded it may be as high as 11 Tg N year�1 and thus
represents a source that was previously underestimated in
global budgets.
8.3.3.1.4 OzoneOzone (O3) is the primary precursor for hydroxyl radical (OH)
formation, and so is a fundamental reactant species in the
atmosphere. It also is an air pollutant and a greenhouse gas.
The oceans are an important reservoir for atmospheric O3 be-
cause of the high reactivity of the gas with components in the
surface water. Dominant reactions appear to be with iodide ions
and organic matter, with possibly smaller contributions from
DMS and alkenes (ethene and propene) (Ganzeveld et al., 2009;
Garland et al., 1980; Martino et al., 2012). Its resistance in the
water phase, combined with a paucity of direct O3 fluxes mea-
sured over the remote oceans, creates uncertainty in the reported
deposition rates (Fairall et al., 2007). The preferred technique
for such measurements is the micrometeorological technique of
eddy correlation (see Section 8.3.2.3.4), which provides a
means of sampling at a high temporal resolution. Using this
technique, O3 deposition velocities to the ocean of between
0.008 and 0.260 cm s�1 have been reported from a ship-based
experiment (Bariteau et al., 2010). This result is comparable to
that of Whitehead et al. (2009), who report a mean deposition
flux of 0.096 cm s�1 from a fixed jetty situated at Roscoff, a
coastal site on the French coast. Whitehead et al. (2009) report
a significant difference between the O3 flux at low and high tide
with mean values of 0.128 and 0.0302 cm s�1, respectively. The
rate at low tide is slower during the night, reportedly due to O3
deposition tomacroalgal surfaces. Lenschow et al. (1982) report
deposition velocities of 0.057 cm s�1 over the remote Pacific
Ocean using aircraft-based eddy correlation. A modeling study
Air–Sea Exchange of Marine Trace Gases 75
by Ganzeveld et al. (2009) estimated an O3 flux to the global
marine environment of 300 Tg year�1, including biogeochemi-
cal, water-side turbulence, and atmospheric dependencies,
which appear to play a role in both the temporal and spatial
distribution of O3 deposition velocities.
8.3.3.1.5 Carbon monoxideCarbon monoxide (CO) provides a significant sink for the
hydroxyl radical (OH), thereby limiting the oxidative ability
of the atmosphere. The photodissociation of CDOM provides
the largest in situ source of oceanic CO (Valentine and Zepp,
1993), producing 50–60 Tg CO–C year�1 (Xie and Zafiriou,
2009; Zafiriou et al., 2003). The largest oceanic sink is thought
to be microbial consumption, accounting for approximately
84% of the total sink processes, the remainder being attributed
to gas exchange (Zafiriou et al., 2003). The highest CO levels
are observed in surface seawater, with a stratified vertical pro-
file caused by the decrease in light penetration with depth
(Johnson and Bates, 1996). This creates a state of super-
saturation at the ocean interface, resulting in net gas transfer
from sea to air. The combination of photodegradation, which
is reliant on UV light, CDOM concentrations, and rapid mi-
crobial consumption drives the in situ concentration of CO and
produces large diurnal variations (e.g., Kitidis et al., 2011; Xie
et al., 2009; Zafiriou et al., 2008). Coastal areas and estuaries
where DOC levels are enhanced due to terrestrial input are
zones of increased CO production (Stubbins et al., 2011).
Upwelling regions also exhibit higher CO fluxes due to the
elevated phytoplankton biomass and thus enhanced DOM
available for phototransformation (Day and Faloona, 2009).
More recently, research has been conducted on the dark pro-
duction of CO from DOM below the photic zone with varying
pH, temperature, and ionic strength of the seawater. Dark
production may represent a substantial source at 5–16 Tg
CO–C year�1, estimated from global extrapolations (Zhang
et al., 2008, and references therein), although this remains
uncertain. The overall oceanic source of CO to the marine
boundary layer is estimated to be 1–6 Tg CO–C year�1
(Stubbins et al., 2006).
8.3.3.2 Nitrogen-Containing Gases
Nitrogen is a biolimiting element such that its abundance, or
lack of it, directly influences the biological productivity of the
ocean. Marine nitrogen exists in multiple oxidation states,
which results in a complex biogeochemical cycle that is largely
controlled by phytoplankton and bacteria. Thus, nitrogen
availability and subsequent saturation anomalies of the surface
ocean often exhibit both spatial and temporal variations.
8.3.3.2.1 Ammonia and methylaminesAmmonia (NH3) and its organic analogues, the amines (gen-
eral formula RxNHy), are important in the atmosphere as the
only basic gas-phase compounds sufficiently abundant to neu-
tralize fine-mode aerosol acidity (Johnson and Bell, 2008).
Ammonia is readily converted to ammonium (NH4þ) in the
atmosphere, which is responsible for neutralizing approxi-
mately 50% of the non sea-salt sulfate acidity in aerosol
in marine air (Johnson and Bell, 2008; Savioe et al., 1993).
Ammonium has also been shown to significantly enhance the
rate of particle nucleation over a binary system of sulfuric acid
and water alone (e.g., Korhonen et al., 1999). Amines are
produced both anthropogenically via industrial processes and
also biogenically through agriculture and decomposition,
which leads to more than 150 species being detected in the
atmosphere (Ge et al., 2011). Some amines have been found to
enhance particle nucleation more strongly than ammonia at
ambient concentrations (Kurten et al., 2008). They are also a
major constituent of secondary organic aerosol in marine air
(Facchini et al., 2008).
The ocean–atmosphere exchange of ammonia is thought to
be bidirectional, the major control on flux direction over
global scales being temperature (Johnson et al., 2008), with
sinks in cold, high latitude oceans and sources in warmer
water. As a soluble gas, the magnitude and direction of
ocean–atmosphere NH3 exchange are particularly sensitive to
the gas-phase concentration term Ca over the range of normal
seawater and atmospheric concentrations. Thus, coastal waters
tend to be a sink for ammonia in populated areas where
agricultural, domestic, and industrial sources combine to
raise Ca, sometimes multiple orders of magnitude above am-
bient (Dentener et al., 2006; Johnson et al., 2008).
Seawater concentrations of NH3 have been observed to be
10–100 times greater than those of the methylamines, of which
monomethylamine was the most abundant. While inshore wa-
ters acted as sources and sinks for methylamines, offshore waters
were a consistent sink (Gibb et al., 1999). As soluble gases, the
transfer of ammonia and amines tends to be under gas-phase
(ka) control, particularly as kw will typically be influenced by
rapid partitioning of these gases into their protonated forms in
seawater, such as ammonium and dimethylammonium. Overall,
the ocean is thought to be a source of ammonia to the atmo-
sphere, with net emissions estimated at 5 Tg NH3–N year�1
(Galloway et al., 2004), however such estimates are subject to
large uncertainties and based on sparse marine data, particularly
gas phase ammonia measurements over the ocean.
8.3.3.2.2 Alkyl nitratesAlkyl (methyl, ethyl, and propyl) nitrates are important com-
pounds in the oxidant chemistry of the atmosphere. In partic-
ular, they are a significant component of its reactive nitrogen
pool (NOy), which plays a critical role in regulating the levels
of tropospheric ozone. They exhibit a lower reactivity than
other NOy species, which allows long-range transport of nitro-
gen oxides; thus, they are of particular importance in remote
marine environments (Simpson et al., 2003). Tropospheric
measurements made over the Pacific Ocean inferred the pres-
ence of an oceanic source of alkyl nitrates (Atlas et al., 1993).
Oceanic methyl and ethyl nitrate in the surface Atlantic Ocean
have since been shown to be supersaturated with respect to the
overlying atmosphere, particularly in regions of enhanced bi-
ological productivity, such as the equatorial upwelling, where
800% supersaturation was reported (Chuck et al., 2002). The
production mechanism is still relatively uncertain, despite
the inference that concentrations may coincide with elevated
chlorophyll-a levels (Chuck et al., 2002). Dahl et al. (2003)
suggested that the reaction of alkyl peroxy radicals with nitric
oxide in seawater may represent an important in situ photo-
chemical production pathway for alkyl nitrates. In addition to
this, bacterial reactions, free radical processes, or other
76 Air–Sea Exchange of Marine Trace Gases
mechanisms involving dissolved organic matter (DOM) at
depth could also increase the oceanic yield, as implied by the
detection of alkyl nitrates below the photic zone (Dahl et al.,
2007; Hughes et al., 2010).
8.3.3.2.3 Hydrogen cyanide and methyl cyanideHydrogen cyanide (HCN, also known as methanenitrile) and
methyl cyanide (CH3CN, also known as acetonitrile or etha-
nenitrile) are trace gases in the atmosphere occurring at the
100–200 ppt (parts per trillion) level (Murphy et al., 2010;
Williams et al., 2004). Their main source is biomass burning,
and measurements made in air masses affected by such regions
show enhanced concentrations up to nearly threefold that in
clean air (Salisbury et al., 2003; Singh et al., 2003). Concen-
tration measurements over the oceans show lower amounts in
the marine boundary layer of about 30 ppt, and this has been
attributed to uptake by the oceans (Jost et al., 2003; Li et al.,
2000; Singh et al., 2003). It is proposed that oceanic deposi-
tion is a major sink for these species from the atmosphere and
that approximately 0.4–1.4 Tg of nitrogen is deposited to the
oceans annually by this mechanism (Jost et al., 2003; Singh
et al., 2003). This contradicts direct CH3CN flux estimations
made by Williams et al. (2004) in the Tropical Atlantic, which
indicated a flux from sea to air. The paucity of seawater mea-
surements limits the understanding of production and loss
processes for HCN and CH3CN in seawater. Williams et al.
(2004) assumed a near-surface source (biological or atmo-
spheric deposition) and from this inferred that CH3CN has a
slow degradation in seawater, explaining its detection at depth
even though there is no currently known subsurface produc-
tion mechanism.
8.3.3.3 Sulfur-Containing Gases
Sulfur is an element that is essential to life. It had been specu-
lated for many years that there must be a major source of
volatile sulfur from the oceans to the land via the atmosphere
in order to balance the loss from the land via weathering
(Eriksson, 1959). This cycle is important in supplying the
element sulfur to the terrestrial environment where it is essen-
tial for plant growth (e.g., Zhao et al., 1999).
8.3.3.3.1 Dimethyl sulfideAlthough the vector of the flux of sulfur from ocean to land via
the atmosphere was originally believed to be hydrogen sulfide,
oceanic measurements subsequently indicated that DMS was the
dominant volatile organic sulfur species in the oceans (Lovelock
et al., 1972). Indeed, Lovelock had predicted that this compound
was the missing link in the biogeochemical cycle of sulfur, and
he has used the existence of DMS and other methylated com-
pounds produced bymarine organisms as evidence in support of
the Gaia hypothesis (Lovelock, 1979). In addition, DMS is of
interest since one of its oxidation products in the troposphere is
sulfur dioxide (Section 8.3.3.3.6), which can itself be further
oxidized to sulfuric acid and then form aerosol sulfate (Plane,
1989). DMS is therefore thought to be a major source of atmo-
spheric acidity, particularly in remote areas away from anthro-
pogenic influence (Keene et al., 1998), and may also act as a
source of cloud condensation nucleii (CCN) (Andreae et al.,
1995). This link led to the CLAW hypothesis (named after the
four authors of Charlson et al., 1987), the idea that phytoplank-
ton may influence global climate. The hypothesis has stimulated
much research but remains controversial and has yet to
be successfully tested. In a recent paper, Quinn and Bates
(2011) critically examined the CLAW hypothesis and concluded
that “. . .. a dimethyl sulphide biological control over cloud
condensation nuclei probably does not exist and sources of
these nuclei to the marine boundary layer and the response of
clouds to changes in aerosol are much more complex than was
recognized twenty years ago.” Although it has proved difficult to
establish the link between DMS and CCN proposed in CLAW,
the idea still has considerable appeal and explains why it remains
a topic of considerable discussion and research even a quarter of
a century after it was first proposed. Knowing what people now
know about the complexity of factors controlling both the DMS
production in the ocean and, especially in this context, the
mechanisms of CCN formation in the atmosphere, it was per-
haps simplistic to expect a single substance to lead directly to
CCN. People now appreciate much more of the complexity of
the atmospheric system containing many reactive gases and
a wide range of particles of different sizes and sources all poten-
tially reacting together. So, it is not so much that CLAW needs
to be ‘retired’ (Quinn and Bates, 2011) but that people need
to replace it with something better, bearing in mind CLAW’s
central idea that marine biology can play an important role
in controlling the properties of the atmosphere (Liss and
Lovelock, 2007).
DMS is produced from dimethylsulfoniopropionate
(DMSP), a compound that is thought to regulate osmotic
pressure in the cells of some species of phytoplankton (Kirst
et al., 1991). There may be an antigrazing function for DMSP
(Kirst et al., 1991) as the cleavage of DMSP to DMS also
produces acrylic acid, a compound believed to have anti-
microbial properties (Sieburth, 1960). This cleavage is known
to be catalyzed by a DMS lyase enzyme (Steinke et al., 1998).
DMSP may also play a role in polar regions as a cryoprotectant
(Kirst et al., 1991). Sunda et al. (2002) showed from laboratory
experiments that DMS can act as a scavenger of hydroxyl and
other reactive oxygen species and so potentially act as a protec-
tant against oxidative stress.
The presence of DMSP in phytoplankton is limited to
specific groups of algae, particularly prymnesiophytes (Keller
et al., 1989). DMS production is enhanced by microzooplank-
ton grazing (Dacey and Wakeham, 1986), viral infection (Malin
et al., 1998), senescence (Nguyen et al., 1988), and bacterial
conversion (Kiene and Bates, 1990). Detailed process experi-
ments have revealed that DMSP andDMS are both rapidly cycled
in the water column by a complex interaction between phyto-
plankton, microzooplankton, bacteria, and viruses (Archer et al.,
2002b; Kiene et al., 2000; Stefels et al., 2007). Only about 10%
of the total DMS produced (and less than 1.3% of the DMSP
synthesized) appears to be transferred to the atmosphere, with
the bulk of the DMS being recycled in the water column or
photooxidized (Archer et al., 2002a).
In order to unravel some of the complexities of DMS cycling
in seawater, mainly by bacteria, molecular techniques have
been used to identify the genes involved in particular pro-
cesses. Howard et al. (2006) showed that the bacterial gene
dmdA, isolated from a Roseobacter strain, catalyzed the first
Air–Sea Exchange of Marine Trace Gases 77
vital step in the demethylation of DMSP. Todd et al. (2007)
found in addition that the dddD gene was able to cleave DMSP
to DMS.
Besides the several functions of DMS mentioned earlier, it
has been proposed that DMS can act as a signaling compound
between organisms. Nevitt et al. (1995) showed that the for-
aging behavior of some sea birds was affected by the presence
of DMS artificially added to air at sea. They argued that since
DMS is known to be released during zooplankton grazing of
DMSP-containing phytoplankton, the ability to sense DMS
would confer an advantage in the birds’ search for zooplankton
food sources. At a much smaller scale, Steinke et al. (2006)
reported laboratory experiments in which they used video
microscopy to show that a tethered copepod was able to detect
the presence of micromolar concentrations of DMS injected
80January
April M
AugJuly
October Nove
Febr
40
0
−40
−80
80
40
0
−40
−80
80
40
0
−40
−80
80
40
0
−40
−80−160 −80 0 80 160 −160
0 5
−80 0
Figure 10 Monthly climatology (L10) of DMS concentrations (nM). Note thaonly a few specific regions exceed 15 nM DMS concentration. Reproduced frsurface dimethylsulfide concentrations and emission fluxes in the global oceaGB003850.
into the water, as adjudged by the frequency of its ‘tail flapping.’
At an even smaller scale, Johnston et al. (2008) proposed that
bacteria may make DMS to attract other organisms.
Not unexpectedly, given its biological source and the com-
plexity of its production and destruction processes in seawater,
measured concentrations of DMS in marine waters are highly
variable in both space and time, so it is particularly difficult to
characterize global concentration fields. With this in mind,
Kettle et al. (1999) assembled a database of all available sea-
water DMS concentrations, about 15000 individual measure-
ments at that time. Lana et al. (2011) increased this database to
more than 47000 observations. Their results for the monthly
climatology are shown in Figure 10.
In general, DMS concentrations are larger at high latitudes,
with a trend toward higher concentrations in summer. The
ay June
Septemberust
mber December
uary March
10 15<
80 160 −160 −80 0 80 160
t the scale is capped at 15 nM to ensure readability of the plots, althoughom Lana A, Bell TG, Simo R, et al. (2011) An updated climatology ofn. Global Biogeochemical Cycles 25: GB1994, http://dx.doi.org/10.1029/
78 Air–Sea Exchange of Marine Trace Gases
global marine flux of DMS to the atmosphere is estimated
by Lana et al. (2011) at 28.1 (17.6–34.4) Tg S year�1, which
is about 17% greater than the equivalent flux calculated using
the smaller database of Kettle et al. (1999).
Once emitted from the oceans across the air–sea interface,
DMS is unstable and subject to oxidation in the marine atmo-
sphere. Major oxidants are the free radicals hydroxyl (OH) and
nitrate (NO3), which yield a number of gaseous and particulate
products (Vogt and Liss, 2009; von Glasow and Crutzen,
2004), as shown in Figure 11.
In a modeling paper, von Glasow et al. (2004) argued for
the importance of bromine oxide (BrO) as an additional oxi-
dant. Their study indicates that typical levels of BrO in the
marine atmosphere may lead to a significant additional path-
way for DMS oxidation, which could reduce column concen-
trations by 60% and reduce particle production/growth and
thus the potential of DMS to affect climate. As shown in
Figure 11, most of the DMS oxidation routes do not appear
to lead to new particle formation, with any nongaseous prod-
ucts being taken up onto existing particle surfaces and thus
enabling them to grow. Whether such a scheme really describes
the complexity of the actual system in the marine atmosphere
is open to question.
Addition
Uptake
Uptake?
O
OH (lower T),BrO
CH3SCH3
CH3SCH3
CH3SCH2OO CH3S CH3
O
CH3SOH
O
O
CH3SCH3
Dim
DimethylsulfoxideDMSO
DimethylsulfideDMS
OH
<0.1
>0.9
NO NO2
O3
Methylsulfinicacid, MSIA
OH (higher T).NO3, CI
Abstraction
Other products, slowCH3SCHO
SO2,SO
3
Other pathway
Figure 11 Atmospheric oxidation of DMS by both addition and abstraction ppreexisting particles. Reproduced from von Glasow R and Crutzen PJ (2004)Atmospheric Chemistry and Physics 4: 589–608.
The study of the DMS cycle using field measurements in the
atmosphere is difficult in the Northern Hemisphere due to
overprinting of the natural sulfur signal by anthropogenic in-
puts of sulfur dioxide and resulting sulfate particles, which are
hard to distinguish from the products of DMS oxidation. In
addition, particle numbers are significantly higher in the atmo-
sphere of the Northern Hemisphere than in the Southern
Hemisphere because of greater pollution and larger amounts
of terrestrial dust from the larger exposed land area. As shown
by Twomey (1991) and Lohmann and Feichter (2005), the
sensitivity of climate, in particular cloud formation, to the
number density of particles in the size range of CCN is non-
linear, with the effect being much stronger at low particle
numbers. For both these reasons, it is easier and more relevant
to study the sulfur cycle in the Southern Hemisphere. Vallina
et al. (2006) used a 3-year time series of satellite observations
of chlorophyll a and CCN, rainfall amount, wind speed, and
model-derived OH to examine the seasonality of CCN over the
Southern Ocean at 40–60� S. They estimated that the biogenic
(DMS) contribution to CCN numbers is 35% in winter and
80% in summer, confirming the central role of marine bio-
genic emissions in controlling both the number density and
seasonality of CCN over oceanic areas remote from land.
Uptake
Uptake
Uptake
UptakeUptake
NewCCN
SO CH3SO2 CH3SO3
O
O
CH3SOH
O
O
CH3SOH
O3?
OH ?
ethylsulfoneDMSO2
Only important if noclouds present
NO2
O3
NO2 HO2
O3
Methylsulfonicacid, MSA
s
Decomp Unimol.
OH
MSA
SO2 H2SO4
athways. For most reactions, DMS oxidation products are taken up ontoModel study of multi-phase DMS oxidation with a focus on halogens.
Air–Sea Exchange of Marine Trace Gases 79
8.3.3.3.2 Methyl mercaptanMethyl mercaptan (CH3SH), also known as methanethiol, is
also produced from DMSP in a pathway that competes with
DMS production and may even be the dominant product
(Kiene et al., 2000). However, CH3SH is thought to be rapidly
removed from the water column due to assimilation into pro-
teins by marine bacteria (Kiene, 1996) and reaction with DOM
(Kiene et al., 2000). CH3SHmay also have a small photochem-
ical sink in seawater (Flock and Andreae, 1996). Seawater
concentrations are assumed to be considerably lower than for
DMS, although few measurements have been made (Kiene,
1996). What data there are indicate that levels of CH3SH
in the Atlantic Ocean range from 150 to 1500 pM and are
typically about 10% those of DMS (Kettle et al., 2001). Open
ocean values were close to 300 pM but increased dramatically
in upwelling and coastal regions such that the mean concen-
tration was as high as 20% than that of DMS. In some areas, the
ratio of CH3SH to DMS was unity. This would make CH3SH
the second most abundant volatile sulfur compound in seawa-
ter and suggests that CH3SH should receive more attention in
future studies of DMS production or DMSP degradation and in
estimates of the flux of biogenic sulfur from the oceans.
8.3.3.3.3 Carbonyl sulfideCarbonyl sulfide (COS) is produced photochemically from the
interaction of UV light with CDOM (Uher and Andreae, 1997)
and its principal loss mechanism in the water column is hydro-
lysis (Andreae and Ferek, 1992). Typical atmospheric concentra-
tions are about 500 ppt(v), while seawater concentrations vary
from 1 to 100 pM (Kettle et al., 2001). A diurnal cycle for COS
has been reported in surface waters, with the ocean acting as a
sink late at night and in early morning but as a source to the
atmosphere for the remainder of the day (Kettle et al., 2001). The
presence of significant undersaturations suggests that in situ deg-
radation rates are high. The net flux from the oceans to the air is
thought to be about 0.1 Tg COS year�1 (Watts, 2000), although
there is a considerable uncertainty in this number. A modeling
exercise suggests that the flux could be between �0.1 and
þ0.2 Tg COS year�1 (Kettle et al., 2002). As CDOM levels are
higher in coastal areas, fluxes from the coastal seas could be up to
0.2 Tg COS year�1, based on the sparse data set of Watts (2000).
A total flux of 0.3 Tg COS year�1 represents about 40% of the
total atmospheric source.
8.3.3.3.4 Carbon disulfideCarbon disulfide (CS2) is also known to have a photochemical
source from CDOM (Xie et al., 1998), although a biological
source has also been reported (Xie et al., 1999). CS2 has no
significant sink in the water column other than transfer to the
atmosphere, although a small diurnal signal has been observed
(Kettle et al., 2001). There are very few published oceanic
measurements. Seawater concentrations are typically about
5 pM S, although higher levels (up to 150 pM) have been
reported from the upwelling areas (Kettle et al., 2001). The
global oceanic flux to the atmosphere has been estimated at
0.13–0.24 Tg CS2 year�1 (Xie and Moore, 1999), about
20–35% of the total atmospheric source (Watts, 2000).
8.3.3.3.5 Hydrogen sulfideHydrogen sulfide (H2S) was originally thought to be the
dominant volatile sulfur compound in the oceans but is now
known to make a rather minor contribution to the total marine
flux of sulfur to the atmosphere (Watts, 2000). It is produced
mainly as a product of the hydrolysis of COS (Elliott, 1989)
and from a particulate organic material, although there is some
evidence for a direct algal source (Andreae et al., 1991). H2S is
rapidly oxidized in surface waters (2–50 h) and exhibits a
diurnal cycle with a maximum before dawn (Andreae, 1990).
The oceanic source strength has been estimated at 1.8 Tg year�1,
approximately 25% of the total source to the atmosphere
(Watts, 2000).
8.3.3.3.6 Sulfur dioxideThere are large terrestrial sources of sulfur dioxide (SO2) to the
atmosphere whereby it is oxidized to sulfuric acid (H2SO4) and
subsequently affects the Earth’s radiative balance via the forma-
tion of non sea-salt sulfate aerosol. Marine boundary layer SO2
concentrations are typically around 25 ppt above the Pacific
Ocean (Yang et al., 2011b), although concentrations increase
with proximity to coastal regions due to terrestrial influences
(Thornton et al., 1999; Yang et al., 2011b). Like ozone, SO2 is
subject to deposition into the oceans, with no reemission. This
arises from the high reactivity of the gas in seawater, which
ensures its rapid destruction and effective zero surface water
concentration driving the one-way flux (Liss, 1971). Dry deposi-
tion rates to the Pacific Ocean for SO2 are estimated at
0.6 mmol m�2 day�1 and are shown to be minor in the overall
SO2 budget (Yang et al., 2011b). Exchange of SO2 across the
air–sea interface is controlled by resistance in the gas phase (ka)
due to the high solubility and reactivity of SO2 in seawater.
8.3.3.4 Nonmethane Hydrocarbons
Nonmethane hydrocarbons (NMHC, such as ethane, ethene,
propane, propene, and isoprene) are trace atmospheric con-
stituents that play important roles both in providing a sink for
hydroxyl radicals and in controlling ozone concentrations
(Donahue and Prinn, 1990). The dominant route of oceanic
production is the phototransformation of DOM in surface
seawater, which is responsible for the highest observed con-
centration in alkenes, particularly ethene (Kansal, 2009;
Riemer et al., 2000). Production may also be biological; for
example, isoprene is emitted by phytoplankton and seaweed
(Broadgate et al., 2004). Thus, isoprene exhibits a significant
correlation with chlorophyll-a abundance so that its emission
from seawater is strongly seasonal (Broadgate et al., 1997).
Furthermore, ethane and propane have been reported as
products emitted from the autolysis of diatom cells and
n-butane from phytoplankton species, such as cryptophytes
and dinoflagellates (Kameyama et al., 2009; McKay et al.,
1996). Marine algae may also use NMHC as signaling com-
pounds (Steinke et al., 2002). The lifecycles of marine NMHC
are still largely unconstrained, as highlighted by modeling
results, suggesting that a larger oceanic source of butanes and
pentanes is required to match atmospheric measurements
made in some coastal regions (Pozzer et al., 2010). A global
estimate for C2–C4 oceanic emissions is 2.1 Tg year�1, with
80 Air–Sea Exchange of Marine Trace Gases
85% attributed to alkenes, of which ethene contributes 40% to
the total emission (Plass-Dulmer et al., 1995). If total NMHC
oceanic emissions are considered, the value increases to
between 5 and 10 Tg C year�1 (Kansal, 2009).
8.3.3.5 Oxygenated Volatile Organic Compounds
Oxygenated volatile organic compounds (OVOCs) include low
molecular weight alcohols, aldehydes, ketones, and peroxides.
Both the photolysis and atmospheric oxidation of OVOCs pro-
duce other stable trace gases (e.g., peroxyacetyl nitrate (PAN))
and influence the levels of hydroxyl radicals (HOx) and the
global ozone budget (Jacob et al., 2005; Singh et al., 1995).
Data on OVOC concentrations in the ocean are sparse, and
thus, it is highly uncertain which processes drive the production
and destruction of these compounds in seawater. Carbonyl
compounds are known to be formed via phototransformation
of DOM in the surface ocean, of which formaldehyde, acetal-
dehyde, propionaldehyde, acetone, glyoxal, and methylglyoxal
are key entities (Kieber et al., 1990; Miller and Moran, 1997;
Mopper et al., 1991; Zhou andMopper, 1997). Rapidmicrobial
acetaldehyde consumption has also been demonstrated by
Mopper and Stahovec (1986), suggesting an active in situ
sink. Acetaldehyde concentrations in Atlantic seawater have
been observed up to 27 nM (Beale et al., 2011) but are less
than 4 nM in the North Pacific (Kameyama et al., 2010). This
compound is thought to be released from the ocean to the
atmosphere (57 Tg year�1), with no current evidence to suggest
an oceanic reservoir (Millet et al., 2010). Both source and sink
global oceanic budgets are notable for acetone, with 0–15 Tg -
year�1 (Singh et al., 2004) and 2 Tg year�1 (Fischer et al., 2012)
estimated, respectively. Concentrations of in situ methanol are
sufficiently higher than for other OVOCs reported, with a max-
imumvalue of 429 nM (average of 99 nM) observed during two
research cruises in the Atlantic Ocean (Beale et al., 2011),
compared with 159 nM in the North Pacific (Kameyama et al.,
2010). Dixon et al. (2010, 2011) reported uptake rates, as low
as 1 day, of methanol in Atlantic water by methylotrophic
bacteria, highlighting that this may represent a rapid oceanic
sink for this reactive OVOC. There is currently only one data set
reporting oceanic concentrations of ethanol and 1-, 2-propanol
(Beale et al., 2010). Naik et al. (2010) reviewed the global
budget of ethanol and concluded that there appears to be a
missing source in explaining ethanol concentrations measured
over remote oceanic regions. Beale et al. (2010) suggested the
ocean may be this missing source and predicted that the
magnitude of the flux is likely to change depending on location
and time of day. Global budget estimates for oceanic methanol
include 85 Tg year�1 (source) and 101 Tg year�1 (sink) (Millet
et al., 2008). Ethanol is suspected to be released to
the atmosphere at a rate between 3 and 5 Tg year�1 and
1-propanol between 2 and 3 Tg year�1, whereas 2-propanol is
reported to change direction depending on location, yielding a
global budget estimate between 1 Tg year�1 (oceanic sink) and
5 Tg year-�1 (oceanic source) (Beale et al., 2010).
8.3.3.6 Organohalogens
The oceans are the main source of reactive halogens to the
atmosphere, where they photodissociate to form radical
species, which catalytically destroy both tropospheric and
stratospheric ozone (Butler et al., 2007; Richter and Wallace,
2004). (The atmospheric chemistry of halogenated com-
pounds is discussed in detail by Glasow and Crutzen in
Chapter 5.2) Whether they are mono-, di-, or trihalogenated
often depends on their route of production, which is predom-
inantly via macro- and microalgae (Carpenter, 2003), summa-
rized in Figure 12. However, the magnitude of the marine
halocarbon source is likely to be influenced by multiple
factors, such as location, community speciation, solar insola-
tion, nutrient levels, and water temperature (Chuck et al.,
2005; Wang et al., 2009), so that significant uncertainty sur-
rounds their global emission estimates.
8.3.3.6.1 The mono-organohalogens: methyl, ethyl, andpropyl iodide; methyl bromide; and methyl chlorideOf the mono-iodocarbons, methyl iodide (CH3I, also known
as iodomethane) is considered dominant. High levels in the
coastal seas and in the Southern Ocean have been associated
with blooms of the algae Phaeocystis (Nightingale, 1991), and
evidence from laboratory culture experiments also suggests
that some diatoms and marine cyanobacteria release CH3I
(Brownell et al., 2010; Hughes et al., 2011; Manley and de la
Cuesta, 1997; Smythe-Wright et al., 2006; Tokarczyk and
Moore, 1994). Bacteria are also able to produce CH3I via
methylation of iodide present in seawater (Amachi et al.,
2001), and marine aggregates may represent an additional
source of volatile mono-iodocarbons (Hughes et al., 2008;
Figure 12). However, the presence of supersaturated levels of
CH3I in the tropical Atlantic Ocean, an area of low biological
activity, suggests an abiotic source as well (Happell and
Wallace, 1996), probably by reaction between photochemi-
cally produced methyl and iodine radicals in surface seawater
(Moore and Zafiriou, 1994). Known sinks for marine CH3I are
summarized in Figure 12 and include nucleophilic attack by
the chloride ion (Cl�), hydrolysis (Zafiriou, 1975), direct
photolysis, and reactions involving natural photosensitizers
(Moore, 2006; Zika et al., 1984). Methyl iodide appears highly
supersaturated in almost all surface waters with an average of
3820% in the Atlantic and 2900% in the Pacific Ocean relative
to the overlying atmosphere (Butler et al., 2007). However,
regions closer to equilibrium have been reported in colder,
polar waters (Chuck et al., 2005) and undersaturations have
been reported in the Greenland and Norwegian seas during fall
(Happell and Wallace, 1996). The global ocean-to-atmosphere
flux of CH3I has been estimated at 0.2–0.3 Tg year�1 (Bell
et al., 2002; Jones et al., 2010). Ethyl- (C2H5I) and propyl
iodide (C3H7I) are likely to exhibit similar routes of oceanic
production and destruction to that of CH3I, despite exhibiting
a lower abundance, so that less emphasis has been placed on
their gas exchange rates (Archer et al., 2007; Jones et al., 2010;
Klick and Abrahamsson, 1992). The best estimates for their
global sea-to-air flux are 0.02 and <0.01 Tg I year�1 for C2H5I
and C3H7I, respectively (Jones et al., 2010), fairly minor con-
tributions to the total marine iodocarbon pool.
Methyl bromide (CH3Br, also known as bromomethane) is
a toxic compound commonly used in agriculture as a fumi-
gant, the use of which is currently limited by international
agreement due to its role on stratospheric ozone destruction
(Butler, 1995). The ocean represents the dominant source and
Atmosphere
I, IO
LandOcean CDOM + I- CH3I, CH2I2,
CH2ClI
CCNformation
Albedo Ozonedestruction
Transfer of I to terrestrialecosystems via rain and aerosolsAerosol
Sea–airgas exchange
Macroalgae
Phytoplankton
BacteriaDOM/POM
Naturalphotosensitizers
Nucleophilicattack
Hydrolysis
Directphotolysis
hvhv
hv
O3
O3
O2
Cl–
Figure 12 Summary of the oceanic in situ sources and sinks of volatile iodine and their impact on atmospheric processes. Iodinated species shown arefor example only, as emission of other compounds (e.g., I2) is possible. Oceanic sources are represented by bold full lines, and sinks by bold dashedlines with italic text.
Air–Sea Exchange of Marine Trace Gases 81
a significant sink of CH3Br to the atmosphere, although the
overall budget estimate remains unbalanced (Hu et al., 2010).
Uncertainty also surrounds the exact nature of in situ oceanic
sources, but a correlation has been established between CH3Br
and pigment indicators of prymnesiophytes, a class of phyto-
plankton (Baker et al., 1999). Laboratory incubations have
shown that some other species of phytoplankton (e.g., Prochlor-
ococcus marinus) also release CH3Br but at rates that were too low
to support the observed oceanic concentrations (Brownell et al.,
2010; Scarratt and Moore, 1996). CH3Br is known to have
significant loss mechanisms in the water column via reaction
with Cl�, hydrolysis, and biological uptake (Goodwin et al.,
1998; King and Saltzman, 1997). The ocean is suspected to act
as a net sink for atmospheric CH3Br, although fluxes are reported
in both directions depending on location and season (Tokarczyk
and Moore, 2006). It appears that polar and tropical waters are
largely undersaturated (King et al., 2000; Lobert et al., 1997), but
temperate and coastal regions are harder to characterize because
of seasonal variation (Tokarczyk and Moore, 2006). Sea surface
temperature can be used as a proxy to estimate CH3Br saturation
(Groszko and Moore, 1998). This method has been refined to
include seasonal data, which improves the relationship, particu-
larly in more temperate waters (King et al., 2002). The global net
oceanic sink for CH3Br is estimated at 10–18 Gg year�1 (King
et al., 2002). For comparison, a global emission rate for CH3Br
from coastal regions alone is estimated at 0.5–3.6 Gg year�1 (Hu
et al., 2010).
Early measurements of methyl chloride (CH3Cl, also
known as chloromethane) suggested that the ocean was a
major source of this compound to the atmosphere (Lovelock,
1975; Singh et al., 1983). The exact production mechanism in
seawater is still unconstrained. The formation routes may be
biological via phytoplankton emissions, although some labo-
ratory culture studies report release rates too low to sustain the
sea-to-air flux (Scarratt and Moore, 1998; Tait and Moore,
1995). There is also evidence that photochemical production
may contribute to oceanic CH3Cl (Ooki et al., 2010), most
likely through reaction with CDOM in subsurface waters
(Moore, 2008). Reaction between Cl� with CH3I has also
been proposed as an in situ production mechanism (Zafiriou,
1975). Apart from loss to the atmosphere, microbial uptake is
also suspected as an oceanic sink, which may be similar in
magnitude to that of gas exchange (Tokarczyk et al., 2003).
The flux of CH3Cl is both into and out of the ocean. The
saturation anomalies appear to be strongly temperature-
dependent with undersaturations in cooler waters (<12 �C).Thus, there is often a marked seasonality in emissions
(MacDonald and Moore, 2007) and a strong latitudinal trend.
Measurements in the Labrador Sea were at or below saturation,
while warmer waters south of the Gulf Stream and all Pacific
samples were consistently supersaturated (Moore et al., 1996a).
This result agrees with that of MacDonald and Moore (2007),
who report a consistent oceanic sink in cooler waters at higher
latitudes in the Atlantic Ocean. The ocean acts as a net source of
82 Air–Sea Exchange of Marine Trace Gases
CH3Cl with an estimated 0.2–0.4 Tg year�1 released to the
atmosphere (Keene et al., 1999). Net coastal emissions are
estimated at approximately 19–98 Gg year�1 (Hu et al., 2010).
8.3.3.6.2 The di-organohalogens: diiodomethane,chloroiodomethane, bromoiodomethane, anddibromomethaneThe polyhalocarbons differ from the monohalogenated gases
in their routes of oceanic production. Whereas the latter tend
to be formed via the methylation of organic halogens, di- and
trihalogenated species originate from the halogenation of or-
ganic precursors catalyzed by haloperoxidase enzymes pro-
duced within algal cells (Moore et al., 1996b).
Diiodomethane (CH2I2) is rarely detected in the marine
boundary layer (Carpenter, 2003; Kurihara et al., 2010;Mahajan
et al., 2010) or is present at very low concentrations (Jones et al.,
2010), but it represents a dominant component of the marine
volatile iodocarbon pool (Archer et al., 2007; Hopkins et al.,
2009). Macroalgae are prolific producers of iodocarbons, and
brown algae in particular are dominant emitters of CH2I2(Carpenter et al., 2000). In addition to this, microalgae, such
as diatoms and some species of prasinophytes, have been shown
to release this compound (also CH2ClI) to the surrounding
seawater (Kurihara et al., 2010; Moore et al., 1996b; Figure 12).
Archer et al. (2007) report that CH2I2 contributed 17% to the
total annual flux of iodine atoms from the shelf seas of
the Western English Channel with a daily average flux to the
atmosphere of 4 nmol m�2 day�1. This can be compared to
2 nmol m�2 day�1 in the western North Pacific (Kurihara et al.,
2010). Such calculations are uncertain, however, when the
air-side concentration (Ca) is difficult to determine.
In the surface ocean, CH2I2 is able to absorb UV radiation
with a wavelength >290 nm, which results in a short lifetime
(�12 min) due to direct photolysis (Figure 12) and thereby
limits transfer across the air–sea interface (Martino et al.,
2005). The product of CH2I2 photodegradation is chloroiodo-
methane (CH2ClI), with a yield of 25–30% (Martino et al.,
2005). This abiotic source means that fluxes of CH2ClI are not
restricted to coastal areas where macroalgal emissions also
dominate. Supersaturations of this species have been noted in
temperate and tropical waters with levels closer to equilibrium
in cooler, polar waters (Chuck et al., 2005). The photolytic
lifetime of this species is longer than its precursor but remains
short, at�13 h; nevertheless, it still provides a greater potential
for flux to the atmosphere (Martino et al., 2005). This is
corroborated by the flux estimates which are reported at 19
and 4 nmol m�2 day�1 by Archer et al. (2007) and Kurihara
et al. (2010), respectively.
Bromoiodomethane (CH2BrI) may be formed via biotic
(macro- and microalgae) and abiotic (nucleophilic substitu-
tion) pathways, but its concentration is low in seawater com-
pared to CH2I2. It is also vulnerable to fast photodegradation
and represents a small proportion of the overall iodocarbon
pool (4%) (Archer et al., 2007). When the annual global fluxes
of these three dihalogenated species (CH2I2, CH2ClI, and
CH2BrI) are combined, they are likely to provide a flux com-
parable to that of methyl iodide alone (�0.3 Tg I year�1)
(Jones et al., 2010).
Dibromomethane (CH2Br2) is a short-lived dihalogen
and, together with bromoform (a trihalogenated species;
Section 8.3.3.6.3), represents a significant contributor to
stratospheric bromine. Its production in seawater is largely
macroalgal, and therefore, concentrations are highest in coastal
waters (Moore and Tokarczyk, 1993) and areas of upwelling
(Butler et al., 2007), which exhibit a high abundance of micro-
algae. Active sink processes are similar to the other dihalogens:
exchange with the atmosphere and nucleophilic attack by the
Cl� ion, perhaps forming CH2BrCl (Moore and Tokarczyk,
1993). Bacterial degradation has also been observed in kelp
beds but may represent removal of less than 1% of that pro-
duced by macroalgae (Goodwin et al., 1997). Supersaturations
of this species were observed at nearly all times in the open
ocean, with periods of undersaturation in the Southern Ocean
assumed to be caused by mixing with CH2Br2-depleted subsur-
face waters rather than by biological consumption (Butler
et al., 2007). Modeling by Liang et al. (2010) indicated that a
global oceanic CH2Br2 emission of 57 Gg year�1 is required to
reproduce the levels of bromine observed in the troposphere.
8.3.3.6.3 The tri-organohalogens: bromoformand chloroformBromoform (CHBr3, also known as tribromomethane) is a
short-lived trace gas, which represents a significant source of
organic bromine to the atmosphere (Quack and Wallace,
2003). Early measurements from the Arctic Ocean suggested
that the main source of this compound was coastal (Dyrssen
and Fogelqvist, 1981; Fogelqvist, 1985) and subsequent re-
search highlighted as well those regions subject to upwelling
(Butler et al., 2007; Moore and Tokarczyk, 1993; Nightingale,
1991). Macroalgae are prolific in their release of polybromi-
nated compounds and are estimated to produce approximately
70% of the global CHBr3 emissions to the atmosphere, with
60% of this due to brown algae emissions (Carpenter and Liss,
2000). In addition, some diatoms (Moore et al., 1996b) and
cyanobacteria (Karlsson et al., 2008) have also been linked to
CHBr3 production. In situ oceanic sinks are reported as hydro-
lysis, reductive dehalogenation, halogen substitution, photolysis,
and, most importantly, sea-to-air transfer (Quack and Wallace,
2003). Microbial consumption is not thought to be an active
removal process (Goodwin et al., 1997). Bromoform is still
suspected to be the primary marine source of reactive bromine
(Chuck et al., 2005) with observations of supersaturation in
most surface waters in the open ocean (Butler et al., 2007),
implying that the net flux is to the overlying atmosphere. How-
ever, regions linked to equilibrium or undersaturation have also
been reported (Chuck et al., 2005; Nightingale, 1991), indicating
that saturation anomalies of CHBr3 vary greatly on regional and
temporal scales, highlighting the importance of simultaneous air
(Ca)- and water (Cw)-side measurements (Chuck et al., 2005). A
global sea–air flux of 10 Gmol Br year�1 is estimated for CHBr3(Quack and Wallace, 2003).
Chloroform (CHCl3, also known as trichloromethane) is
the second most abundant chlorine species in the atmosphere
behind methyl chloride (Rhew et al., 2008). Its presence in the
atmosphere is dominated by natural emissions (Keene et al.,
1999), with production by macroalgae a likely in situ oceanic
source (Baker et al., 2001; Nightingale et al., 1995). CHCl3production in the ocean has also been linked to carbon tetra-
chloride degradation. Seawater data are sparse but mea-
surements made in the Southern Ocean ranged between
Air–Sea Exchange of Marine Trace Gases 83
1.5 and 2.9 pM (Abrahamsson et al., 2004) and Nightingale
(1991) noted that its distribution often correlates with that of
dichlorobromomethane. Estimates of the global annual emis-
sion of CHCl3 from the ocean lie between 0.1 Tg year�1
(Nightingale, 1991) and 0.4 Tg year�1 (McCulloch, 2003). If
coastal waters are excluded, the flux decreases to 0.09 Tg year�1
(McCulloch, 2003).
8.3.3.7 Miscellaneous Gases
8.3.3.7.1 ChlorofluorocarbonsThese are man-made volatile compounds containing F and/or
Cl atoms. They have a variety of uses, most of which lead
ultimately to their emission to the atmosphere and the poten-
tial for uptake by the oceans across the sea surface.
Measurements ofmethyl chloroformor1,1,1,-trichloroethane
(CH3CCl3) showed that this compound was significantly under-
saturated in the equatorial PacificOcean (Butler et al., 1991). Loss
rates were supported roughly by known hydrolysis rates, and the
authors calculated that about 6% of atmospheric CH3CCl3 is
removed by consumption in the oceans. With this exception,
most of the chlorofluorocarbons were originally thought to be
stable in the water column.
The first evidence that this assumption was incorrect came
from the observations of carbon tetrachloride (CCl4) removal in
the Baltic Sea under anoxic conditions (Krysell et al., 1994). A
later investigation in the Black Sea found that reductions in CCl4,
CHCl3, CH3CCl3, dibromomethane, dibromochloromethane,
and bromodichloromethane were related to oxygen/hydrogen
sulfide concentrations (Tanhua et al., 1996). Most of the CCl4was transformed to CHCl3 as an intermediate product. Subse-
quent work in a fjord in Norway showed that CFC-11 was also
removed in anoxic waters (Shapiro et al., 1997). Loss rates of
both CCl4 and CFC-11 in anoxic waters are probably due to
biological rather than chemical removal (Lee et al., 1999). It
also seems likely that some of the chlorofluorocarbons are re-
moved in fully oxygenated surface waters. Observations show
there is a deficit of CCl4 in the Antarctic surface and bottom
waters (Meredith et al., 1996). Finally, fluorinated compounds,
such as CFC-113, are degraded in warm surface waters of the
temperate North Atlantic, the tropical western Pacific, the Eastern
Mediterranean, and even the Weddell Sea (Roether et al., 2001).
CFC-113 depletions were of the order of 3% year�1, with possi-
bly accelerated rates in the mixed layer or near the surface.
Selected dechlorination of chlorinated compounds by soil
bacteria has long been recognized (Vogel et al., 1987). It seems
likely that there is a biological transformation of these com-
pounds by marine bacteria, particularly as they can transform
CH3Br. Not only are these compounds likely to be removed
from oceanic and coastal waters under anoxic and suboxic
conditions, but given that compounds such as CH4 and N2O
are thought to be produced in suboxic microenvironments
within the water column (see Sections 8.3.3.1.2 and
8.3.3.1.3, respectively), it seems reasonable that the same
sites might also be areas of chlorofluorocarbon removal.
8.3.3.7.2 Synthetic organic compoundsSynthetic organic compounds are a vast group of man-made
chemicals; here, the authors consider only a small subset,
including the polychlorinated biphenyls (PCBs) and various
chlorinated organic pesticides (e.g., DDT, chlordane, and
dieldrin), for which there are particular environmental con-
cerns. These are emitted to the atmosphere during use or
disposal and are found dispersed throughout the environment.
Deposition to the oceans occurs by both wet and dry
processes in varying proportions for the different compounds,
although gaseous deposition is always a significant route
(25–85% of the total) (Duce et al., 1991). Reemission is also
possible where the concentration gradient changes sign, be-
cause of either reduced air concentrations or elevated water
concentration or both. Many of these compounds are interest-
ing with respect to gas exchange because both gas (ka)- and
liquid (kw)-phase resistances are significant for air–sea ex-
change, contrary to the usual situation for gases in which one
or the other resistance is dominant.
8.3.3.7.3 MercuryOceanic volatile forms of mercury (Hg) include Hg metal atoms
(Hgo) and, to a lesser extent, the fully dimethylated form
(Me2Hg). There is evidence that these species are formed biolog-
ically (Fantozzia et al., 2012; Fitzgerald, 1989) and, in the case of
Hgo, by photochemical processes in seawater (Costa, 1999).
Their emission frommarine waters constitutes a significant com-
ponent of the total global cycling of Hg (Fitzgerald, 1989; Lam-
borg et al., 2002). This is important both biogeochemically and
for pollution, since volatilization from seawater not only de-
creases the lifetime of Hg in the water column but also brings
about its wider geographical dispersion to other environments.
The consequence is that Hg from pollution sources transported
to the oceans by rivers and through the atmosphere can be
recycled back to terrestrial environments. This conclusion is
supported by a model of the global mercury cycle, which in-
dicates that almost 90% of the emission of Hg from the ocean is
recycled from previously depositedmaterial (Strode et al., 2007).
8.3.3.7.4 Selenium and poloniumThere are considerable similarities between the oceanic (bio)geo-
chemical behavior of the elements sulfur (see Section 8.3.3.3)
and selenium(Se). This is not surprising in viewof their proximity
as adjacent members of Group VIb of the periodic table. For
example, both elements require a significant source of a volatile
form(s) to be emitted from the oceans to the atmosphere in order
for their geochemical budgets to balance (Mosher and Duce,
1987). In the case of sulfur, the carrier molecule is DMS, but
until a decade ago, the equivalent volatile form of Se was un-
known. Measurements of volatile Se species in samples from the
North Atlantic have since been reported (Amouroux et al., 2001).
The dominant volatile forms were found to be dimethyl selenide
(DMSe – the direct analogue of DMS) and the mixed S/Se com-
pounddimethyl selenyl sulfide (DMSeS). Amouroux et al. (2001)
calculate that the flux of volatile Se from the global oceans to the
atmosphere is more than enough to balance the geochemical
budget of this element. These volatile Se species are assumed to
form by routes analogous to those discussed earlier for DMS
(Section 8.3.3.3.1), an assumption supported by the positive
correlation found by Amouroux et al. (2001) between the seawa-
ter concentrations of DMSe and DMS. Once emitted to the atmo-
sphere, DMSe (and DMSeS) is likely subject to similar
atmospheric transformations as described previously for DMS,
although the four orders of magnitude smaller size of the Se
84 Air–Sea Exchange of Marine Trace Gases
sea-to-air flux compared with that of S means that their impor-
tance for atmospheric properties will be insignificant.
On emission to the atmosphere, the volatile Se compounds
are oxidized and incorporated into particulate phases, a frac-
tion of which will be dry and wet deposited to land. Measure-
ments of the Se content of mosses in Norway showed
decreased concentrations with distance from the sea, indicat-
ing a marine source (E. Steinnes, 2003, and personal commu-
nication, 2012). This may have important implications as
Rayman (2000) has shown that Se is vital for human health
but that people in many European countries do not reach the
recommended intake level. Thus, in both their biogeochemical
cycling and impact on health, there is a considerable similarity
in the behavior of selenium and iodine (Section 8.3.3.6).
Extension of the Group VIb similarity to later members of
the group (Te and Po) is problematic. Polonium isotopes have
been measured in aerosols at a coastal site with both onshore
and offshore winds (Kim et al., 2000). Air masses which have
traveled over the oceans have higher ‘excess’ 210Po, attributed
to volatilization of Po from the sea surface. This process may be
analogous to that already described for S and Se, although the
evidence is only circumstantial. In the case of tellurium, there
are no known marine measurements of volatile forms due to
its extremely low concentration in seawater, although it may
experience a similar cycle.
8.3.3.7.5 HydrogenMolecular hydrogen (H2) is formed in surface ocean waters by
microbial and photochemical processes (Conrad and Seiler,
1986; Moore et al., 2009; Punshon and Moore, 2008; Wilson
et al., 2010) so that surface seawaters are generally at or
well above supersaturation with respect to the concentrations
of the gas in the overlying atmosphere. These supersaturations
imply a net global flux of H2 from sea to air in the range
3–6 Tg year�1 (Conrad and Seiler, 1986; Novelli et al., 1999;
Rhee et al., 2005; Schmidt, 1974), with an uncertainty on each
estimate of 50% or more. This flux accounts for only a few
percent of the total global H2 production rate, most of which is
from photochemical oxidation of a variety of volatile organic
compounds in the atmosphere (Rhee et al., 2005).
8.3.4 Effects of Climate Change on MarineTrace Gases
In this brief and somewhat speculative section, the authors re-
view some potential effects of climate and other global changes
on the exchange of trace gases between the atmosphere and
oceans. First, the effects of climate change on Kw are discussed,
followed by some examples of how trace gas concentration fields
and ocean/atmosphere fluxes may be affected.
8.3.4.1 Effect on Air–Sea Gas Transfer
The factors controlling kw have been discussed in detail in
Section 8.3.2.2. Wind is clearly the main driver of kw, either
directly or indirectly via its role in the formation of waves and
bubbles. Thus, in order to predict how change in climate may
alter Kw, knowledge of how the wind field may vary is a
prerequisite. Models of a warmer world show greater warming
at high (particularly northern) latitudes than at lower latitudes
(IPCC, 2007), and since it is the equator–pole temperature
gradients which ultimately drive the wind systems, warming
should result in reduced windiness and thence lower Kw. This
may be counteracted to some extent by the possibility of
greater storminess (cyclones and hurricanes) in the tropics
(IPCC, 2007, in particular Section 10.3.6.3, pp. 786–788).
However, a better prediction of Kw in the future (or the past)
is arguably more dependent on improved knowledge of winds
and storminess than it is on the current uncertainties in the
relationship between Kw and wind (Section 8.3.2.4). A second
factor that could influence transfer velocity is the effect of
temperature change on the physical terms Sc (the Schmidt
number) and H (Henry’s law solubility) (Section 8.3.2). The
temperature (and salinity) dependence of these terms is rea-
sonably well understood and can be straightforwardly incor-
porated into predictive models as required (Johnson, 2010).
The effects are likely to be quite small in comparison with other
changes and uncertainties. Prediction of the role of surface films
in affecting kw is difficult not only because of the lack of under-
standing of their importance in the field (Section 8.3.2.2.5) but
also because any temporal change is likely to be driven by the
availability of suitable surface-active organic material, in turn,
dependent on marine biological activity, which is itself hard to
predict (see next section).
8.3.4.2 Effect on Dissolved Gas Concentrations
8.3.4.2.1 Carbon dioxideThe attempts to quantify changes in gas fluxes under altered
climates are often undertaken using coupled atmosphere–
ocean models. These generally start with a purely physical
approach. The ability to incorporate biological processes and,
even more, their changes with time is rudimentary. However,
physical changes, particularly in vertical mixing, are likely to
dominate for future (and past and present) uptake of man-
made CO2 because production and hence CO2 uptake by
marine biota are generally not thought to be carbon limited.
The situation is more complex for uptake of natural CO2 where
future changes in biological activity may lead to significant
alterations in the oceans’ ability to take up CO2 (IPCC, 2007;
Sarmiento and Gruber, 2006).
8.3.4.2.2 Dimethyl sulfideIn the case of DMS, several coupled models to estimate future
oceanic emissions, incorporating biological schemes with dif-
ferent degrees of complexity, generally conclude that any
global changes will be quite small under climate changes
expected this century (Bopp et al., 2003; Gabric, et al., 2004;
Kloster et al., 2007; Vallina et al., 2007). However, all the
models agree that there is large spatial inhomogeneity in any
effects, with increases in DMS in parts of the high latitudes and
decreases at low latitudes (see Vogt and Liss, 2009, for a more
detailed discussion).
8.3.4.2.3 Other gasesEven the most sophisticated current models do not incorporate
many important factors, themselves subject to climatic and
other global changes, which could cause major alterations in
trace gas formation and destruction in the oceans. For example,
Air–Sea Exchange of Marine Trace Gases 85
altered solar radiation entering the oceans will potentially
affect the concentrations of those trace gases (such as COS –
Section 8.3.3.3.3) for which photochemistry is a significant
mode of formation. Elevated CO2 uptake by the oceans will
itself lead to a significant decrease in the pH of surface waters,
which is likely to adversely affect phytoplankton, particularly
those with calcium carbonate structures (Riebesell et al., 2000).
The effect of pH changes on trace gas production is, as yet, little
studied but initial results from mesocosm experiments of
DMS/DMSP and various organohalogens indicate that signifi-
cant changes are possible (Avgoustidi et al., 2012; Hopkins
et al., 2009). Inputs of iron and nitrogen from the atmosphere
or from deeper waters affect biological activity in the surface
oceans and may affect trace gas fluxes. Fe additions, for exam-
ple, can increase DMS (Turner et al., 2004), and N additions
can enhance the production of N2O and the drawdown of
CO2 (Duce et al., 2008). Again, the impact of such changes
on trace gas formation and destruction is largely unknown but
is receiving increasing attention, for example, in the SOLAS
(Surface Ocean – Lower Atmosphere Study) project (http://
www.solas-int.org).
References
Abrahamsson K, Bertilsson S, Chierici M, et al. (2004) Variations of biochemicalparameters along a transect in the Southern Ocean, with special emphasis onvolatile halogenated organic compounds. Deep Sea Research II 51: 2745–2756.
Amachi S, Kamagata Y, Kanagawa T, and Muramatsu Y (2001) Bacteria mediatemethylation of iodine in marine and terrestrial environments. Applied andEnvironmental Microbiology 67: 2718–2722.
Amouroux D, Liss PS, Tessier E, Hamren-Larsson M, and Donard OFX (2001) Role ofoceans as biogenic sources of selenium. Earth and Planetary Science Letters189: 277–283.
Andreae MO (1990) Ocean–atmosphere interactions in the global biogeochemicalsulfur cycle. Marine Chemistry 30: 1–29.
Andreae TW, Cutter GA, Hussain N, Radfordknoery J, and Andreae MO (1991)Hydrogen-sulfide and radon in and over the western North-Atlantic Ocean. Journalof Geophysical Research 96: 18753–18760.
Andreae MO, Elbert W, and Demora SJ (1995) Biogenic sulfur emissions and aerosolsover the tropical south-Atlantic. 3. Atmospheric dimethylsulfide, aerosols and cloudcondensation nuclei. Journal of Geophysical Research 100: 11335–11356.
Andreae MO and Ferek RJ (1992) Photochemical production of carbonyl sulfide inseawater and its emission to the atmosphere. Global Biogeochemical Cycles6: 175–183.
Archer D, Kheshgi H, and MaierReimer E (1997) Multiple timescales for neutralization offossil fuel CO2. Geophysical Research Letters 24: 405–408.
Archer SD, Gilbert FJ, Nightingale PD, et al. (2002) Transformation ofdimethylsulphoniopropionate to dimethyl sulphide during summer in the North Seawith an examination of key processes via a modelling approach. Deep Sea ResearchII 49: 3067–3101.
Archer SD, Goldson LE, Liddicoat MI, Cummings DG, and Nightingale PD (2007)Marked seasonality in the concentrations and sea-to-air flux of volatile iodocarboncompounds in the western English Channel. Journal of Geophysical Research112: C08009.
Archer SD, Smith GC, Nightingale PD, et al. (2002) Dynamics of particulatedimethylsulphoniopropionate during a Lagrangian experiment in the northern NorthSea. Deep Sea Research II 49: 2979–2999.
Asher W (1997) The sea-surface microlayer and its effect on global air–sea gas transfer.In: Liss PS and Duce RA (eds.) The Sea Surface and Global Change, pp. 251–285.Cambridge: Cambridge University Press.
Asher WE (2009) The effects of experimental uncertainty in parameterizing air–sea gasexchange using tracer experiment data. Atmospheric Chemistry and Physics9: 131–139.
Asher WE, Jessup AT, and Atmane MA (2004) On the use of the active controlled fluxtechnique for in situ measurement of the air–sea transfer velocity of heat and gas.Journal of Geophysical Research 09: C08S14.
Asher W, Karle L, Higgins B, and Farley P (1996) The influence of bubble plumes onair–seawater gas transfer velocities. Journal of Geophysical Research101: 12,027–12,041.
Asher W, Wang Q, Monahan EC, and Smith PM (1998) Estimation of air–sea gastransfer velocities from apparent microwave brightness temperature. MarineTechnology Society Journal 32: 32–40.
Asher WE and Wanninkhof R (1998) The effect of bubble-mediated gas transfer onpurposeful dual-gaseous tracer experiments. Journal of Geophysical Research103: 10555–10560.
Atlas E, Pollock W, Greenberg J, Heidt L, and Thompson AM (1993) Alkyl nitrates,nonmethane hydrocarbons, and halocarbon gases over the equatorial Pacific-Oceanduring Saga-3. Journal of Geophysical Research 98: 16933–16947.
Avgoustidi V, Nightingale PD, Joint I, Steinke M, Turner SM, and Liss PS (2012)Decreased marine dimethyl sulphide production under elevated CO2 in mesocosmand in vitro studies. Environmental Chemistry 9: 399–401. http://dx.doi.org/10.1071/EN11125.
Baker JM, Reeves CE, Nightingale PD, Penkett SA, Gibb SW, and Hatton AD (1999)Biological production of methyl bromide in the coastal waters of the North Sea andopen ocean of the northeast Atlantic. Marine Chemistry 64: 267–285.
Baker JM, Sturges WT, Sugier J, et al. (2001) Emissions of CH3Br, organochlorines andorganoiodines from temperate macroalgae. Chemosphere: Global Change Science3: 93–106.
Bange HW (2006) New directions: The importance of oceanic nitrous oxide emissions.Atmospheric Environment 40: 198–199.
Bange HW, Rapsomanikis S, and Andreae MO (1996) Nitrous-oxide in coastal waters.Global Biogeochemical Cycles 10: 197–207.
Banner M and Peirson W (1998) Tangential stress beneath wind-driven air–waterinterfaces. Journal of Fluid Mechanics 364: 115–145.
Bao J-W, Fairall CW, Michelson SA, and Bianco L (2011) Parameterizations of sea-spray impact on the air–sea momentum and heat fluxes. Monthly Weather Review139: 3781–3797.
Bariteau L, Helmig D, Fairall CW, Hare JE, Hueber J, and Lang EK (2010) Determinationof oceanic ozone deposition by ship-borne eddy covariance flux measurements.Atmospheric Measurement Techniques 3: 441–455.
Barnes J and Owens NJP (1998) Denitrification and nitrous oxide concentrations in theHumber Estuary, UK, and adjacent coastal zones. Marine Pollution Bulletin37: 247–260.
Bates TS, Kelly KC, Johnson JE, and Gammon RH (1996) A reevaluation of the openocean source of methane to the atmosphere. Journal of Geophysical Research101: 6953–6961.
Battle M, Bender ML, Tans PP, et al. (2000) Global carbon sinks and their variabilityinferred from atmospheric O2 and delta C13. Science 287: 2467–2470.
Beale R, Liss PS, Dixon JL, and Nightingale PD (2011) Quantification of oxygenatedvolatile organic compounds in seawater by membrane inlet—Proton transferreaction/mass spectrometry. Analytica Chimica Acta 706: 128–134.
Beale R, Liss PS, and Nightingale PD (2010) First oceanic measurements of ethanol andpropanol. Geophysical Research Letters 37: L24607.
Bell N, Hsu L, Jacob DJ, et al. (2002) Methyl iodide: Atmospheric budget and use as atracer of marine convection in global models. Journal of Geophysical Research 107:art. no.-4340.
Bender ML, Kinter S, Cassar N, and Wanninkhof R (2011) Evaluating gas transfervelocity parameterizations using upper ocean radon distributions. Journal ofGeophysical Research 116: C02010.
Berger R and Libby WF (1969) Equilibration of atmospheric carbon dioxide withseawater: Possible enzymatic control of the rate. Science 164: 1395–1397.
Blomquist BW, Fairall CW, Huebert B, and Kleiber DJ (2006) DMS sea–airtransfer velocity: Direct measurements by eddy covariance and parameterizationbased on the NOAA/COARE Gas Transfer Model. Geophysical Research Letters33: L07601.
Bopp L, Aumont O, Belviso S, and Monfray P (2003) Potential impact of climate changeon marine dimethyl sulfide emissions. Tellus 55B: 11–22.
Boutin J and Etcheto J (1995) Estimating the chemical enhancement effect on the air–sea CO2exchange using the ERS-1 scatterometer wind speeds. In: Jahne B and Monahan EC(eds.) Air–Water Gas Transfer, pp. 827–841. Hanau: AEON Verlag and Studio.
Boutin J, Etcheto J, Merlivat L, and Rangama Y (2002) Influence of gas exchangecoefficient parameterisation on seasonal and regional variability of CO2 air–seafluxes. Geophysical Research Letters 29: art. no.-1182.
Broadgate WJ, Liss PS, and Penkett SA (1997) Seasonal emissions of isoprene andother reactive hydrocarbon gases from the ocean. Geophysical Research Letters24: 2675–2678.
Broadgate WJ, Malin G, Kupper FC, Thompson A, and Liss PS (2004) Isoprene andother non-methane hydrocarbons from seaweeds: A source of reactivehydrocarbons to the atmosphere. Marine Chemistry 88: 61–73.
86 Air–Sea Exchange of Marine Trace Gases
Broecker HC, Petermann J, and Siems W (1978) The influence of wind on CO2exchange in a wind wave tunnel, including the effects of monolayers. Journal ofMarine Research 36: 595–610.
Broecker HC and Siems W (1984) The role of bubbles for gas transfer from water to airat higher wind speeds. Experiments in the wind-wave facility in Hamburg.In: Brutsaert W and Jirka GH (eds.) Gas Transfer at Water Surfaces, pp. 229–236.Dordrecht: Reidel.
Broecker W and Peng T-H (1974) Gas exchange rates between air and sea. Tellus26: 21–35.
Broecker WS, Ledwell JR, Takahashi T, et al. (1986) Isotopic versus micrometeorologicocean CO2 fluxes—A serious conflict. Journal of Geophysical Research91: 517–527.
Brownell DK, Moore RM, and Cullen JJ (2010) Production of methyl halidesby Prochlorococcus and Synechococcus. Global Biogeochemical Cycles24: GB2002.
Budzianowski W and Koziol A (2005) Stripping of ammonia from aqueoussolutions in the presence of carbon dioxide: Effect of negative enhancement of masstransfer. Trans ICHemE A, Chemical Engineering Research and Design83: 196–204.
Butler JH (1995) Ozone depletion—Methyl bromide under scrutiny. Nature376: 469–470.
Butler JH, Elkins JW, Thompson TM, Hall BD, Swanson TH, and Koropalov V (1991)Oceanic consumption of Ch3CCl3—Implications for tropospheric OH. Journal ofGeophysical Research 96: 22347–22355.
Butler JH, King DB, Lobert JM, et al. (2007) Oceanic distributions and emissions ofshort-lived halocarbons. Global Biogeochemical Cycles 21: GB1023.
Calleja MLI, Duarte CM, Navarro N, and Agustı S (2005) Control of air–sea CO2disequilibria in the subtropical NE Atlantic by planktonic metabolism under theocean skin. Geophysical Research Letters 32: L08606.
Carpenter LJ (2003) Iodine in the marine boundary layer. Chemical Reviews103: 4953–4962.
Carpenter LJ and Liss PS (2000) On temperate sources of bromoform and otherreactive organic bromine gases. Journal of Geophysical Research105: 20539–20547.
Carpenter LJ, Malin G, Liss PS, and Kupper FC (2000) Novel biogenic iodine-containing trihalomethanes and other short-lived halocarbons in the coastal EastAtlantic. Global Biogeochemical Cycles 14: 1191–1204.
Charlson RJ, Lovelock JE, Andreae MO, and Warren SG (1987) Oceanic phytoplankton,atmospheric sulfur, cloud albedo and climate. Nature 326: 655–661.
Charpentier J, Farıas L, and Pizarro O (2010) Nitrous oxide fluxes in the central andeastern South Pacific. Global Biogeochemical Cycles 24: GB3011.
Chuck AL, Turner SM, and Liss PS (2002) Direct evidence for a marine source of C1 andC2 alkyl nitrates. Science 297: 1151–1154.
Chuck AL, Turner SM, and Liss PS (2005) Oceanic distributions and air–sea fluxes ofbiogenic halocarbons in the open ocean. Journal of Geophysical Research110: C10022.
Codispoti LA (2010) Interesting times for marine N2O. Science 327: 1339–1340.Conrad R and Seiler W (1986) Exchange of CO and H2 between ocean and atmosphere.
In: Buat-Menard P (ed.) The Role of Air–Sea Exchange in Geochemical Cycling,pp. 269–282. Dordrecht: Reidel.
Costa MLPS (1999) Photoreduction of mercury in sea water and its possibleimplications for Hg-0 air–sea fluxes. Marine Chemistry 68: 87–95.
Crutzen PJ (1970) The influence of nitrogen oxides on the atmospheric ozone content.Journal of the Royal Meteorological Society 96: 320–325.
Cunliffe M, Whitely A, Schafer H, and Newbold L (2009) Comparison ofbacterioneuston and bacterioplankton dynamics during a phytoplanktonbloom in a fjord microcosm. Applied and Environmental Microbiology75: 7173–7181.
D’Asaro E and McNeil C (2007) Air–sea gas exchange at extreme wind speedsmeasured by autonomous oceanographic floats. Journal of Marine Systems66: 92–109.
Dacey JWH and Wakeham SG (1986) Oceanic dimethylsulfide—Production duringzooplankton grazing on phytoplankton. Science 233: 1314–1316.
Dahl EE, Saltzman ES, and De Bruyn WJ (2003) The aqueous phase yield of alkylnitrates from ROOþNO: Implications for photochemical production in seawater.Geophysical Research Letters 30: 1271–1273.
Dahl EE, Yvon-Lewis SA, and Saltzman ES (2007) Alkyl nitrate (C1–C3) depth profiles inthe tropical Pacific Ocean. Journal of Geophysical Research 112: C01012.
Danckwerts PV (1951) Significance of liquid-film coefficients in gas absorption.Industrial and Engineering Chemistry 43: 1460–1467.
Day DA and Faloona I (2009) Carbon monoxide and chromophoric dissolved organicmatter cycles in the shelf waters of the northern California upwelling system. Journalof Geophysical Research 114: C01006.
de Leeuw G, Kunz GJ, Caulliez G, et al. (2002) LUMINY—An overview. In: Donelan MA,Drennan WM, Saltzman ES, and Wanninkhof R (eds.) Gas Transferat Water Surfaces: Geophysical Monograph, vol. 127, pp. 291–295. Washington,DC: AGU.
de Wilde HPJ and de Bie MJM (2000) Nitrous oxide in the Schelde estuary: Productionby nitrification and emission to the atmosphere. Marine Chemistry 69: 203–216.
Deacon EL (1977) Gas transfer to and across an air–water interface. Tellus29: 363–374.
DeCosmo J, Katsaros KB, Smith SD, et al. (1996) Air–sea exchange of water vapor andsensible heat: The Humidity Exchange Over the Sea (HEXOS) results. Journal ofGeophysical Research 101: 12001–12016.
Dentener FJ, Drevet J, Lamarque JF, et al. (2006) Nitrogen and sulfur deposition onregional and global scales: A multimodel evaluation. Global Biogeochemical Cycles20: GB4003.
Dixon JL, Beale R, and Nightingale PD (2010) Microbial methanol uptake in northeastAtlantic waters. The ISME Journal (2011) 5: 704–716.
Dixon JL, Beale R, and Nightingale PD (2011) Rapid biological oxidation of methanol inthe tropical Atlantic: Significance as a microbial carbon source. Biogeosciences8: 2707–2716.
Donahue NM and Prinn RG (1990) Nonmethane hydrocarbon chemistry inthe remote marine boundary-layer. Journal of Geophysical Research95: 18387–18411.
Doney SC (1995) Comment on “Experimental demonstration of coupling of heat andmatter fluxes at a gas water interface” by Leon Phillips. Journal of GeophysicalResearch 100: 14347–14350.
Drennan W, Taylor P, and Yelland M (2005) Parameterizing the sea surface roughness.Journal of Physical Oceanography 35: 835–848.
Duce RA, LaRoche J, Altieri K, et al. (2008) Impacts of atmospheric anthropogenicnitrogen on the open ocean. Science 320: 893–897.
Duce RA, Liss PS, Merrill JT, et al. (1991) The atmospheric input of trace species to theworld ocean. Global Biogeochemical Cycles 5: 193–259.
Dyrssen D and Fogelqvist E (1981) Bromoform concentrations of the Arctic Ocean in theSvalbard Area. Oceanologica Acta 4: 313–317.
Edson JB, Fairall CW, Bariteau L, et al. (2011) Direct covariance measurement of CO2gas transfer velocity during the 2008 Southern Ocean Gas Exchange Experiment:Wind speed dependency. Journal of Geophysical Research 116: C00F10. http://dx.doi.org/10.1029/2011JC007022 [printed 117(C4), 2012].
Elliott S (1989) The effect of hydrogen peroxide on the alkaline hydrolysis of carbondisulfide. Environmental Science and Technology 24: 264–267.
Emerson S, Quay P, Stump C, Wilbur DO, and Knox M (1991) O2, Ar, N2, and222Rn in
surface waters of the subarctic ocean: Net biological O2 production. GlobalBiogeochemical Cycles 5: 49–70.
Eriksson E (1959) The yearly circulation of chloride and sulfur in nature;meteorological, geochemical and pedological implications. Tellus 11: 375–403.
Etcheto J and Merlivat L (1988) Satellite determination of the carbon dioxide exchangecoefficient at the ocean–atmosphere interface—A 1st step. Journal of GeophysicalResearch 93: 15669.
Facchini MC, Decesari S, Rinaldi M, et al. (2008) Important source of marine secondaryorganic aerosol from biogenic amines. Environmental Science and Technology42: 9116–9121.
Fairall C, Bradley E, Hare J, Grachev A, and Edson J (2003) Bulk parameterization ofair–sea fluxes: Updates and verification for the coare algorithm. Journal of Climate16: 571–591.
Fairall C, Bradley E, Rogers D, Edson J, and Young G (1996) Bulk parameterizationof air–sea fluxes for TOGA COARE. Journal of Geophysical Research101: 3747–3764.
Fairall C, Hare J, Edson J, and McGillis W (2000) Parameterization andmicrometeorological measurements of air–sea gas transfer. Boundary-LayerMeteorology 96: 63–105.
Fairall CW, Helmig D, Ganzeveld L, and Hare J (2007) Water-side turbulenceenhancement of ozone deposition to the ocean. Atmospheric Chemistry and Physics7: 443–451.
Fairall C, Yang M, Bariteau L, et al. (2011) Implementation of the coupledocean–atmosphere response experiment flux algorithm with CO2, dimethyl sulfide,and O3. Journal of Geophysical Research 116: C00F09.
Fantozzia L, Mancaa G, Ammoscatoa I, et al. (2012) The cycling and sea–air exchangeof mercury in the waters of the Eastern Mediterranean during the 2010 MED-OCEANOR cruise campaign. Science of the Total Environment. http://dx.doi.org/10.1016/j.scitotenv.2012.09.062.
Feely RA, Wanninkhof R, Hansell DA, Lamb MF, Greeley D, and Lee K (2002) Watercolumn CO2 measurements during the GasEx-98 Expedition. In: Donelan MA,Drennan WM, Saltzman ES, and Wanninkhof R (eds.) Gas Transfer at WaterSurfaces: Geophysical Monograph, vol. 127, pp. 173–180. Washington, DC: AGU.
Air–Sea Exchange of Marine Trace Gases 87
Ferron S, Ortega T, and Forja JM (2010) Nitrous oxide distribution in thenorth-eastern shelf of the Gulf of Cadiz (SW Iberian Peninsula). Marine Chemistry119: 22–32.
Fischer EV, Jacob DJ, Millet DB, Yantosca RM, and Mao J (2012) The role of the oceanin the global atmospheric budget of acetone. Geophysical Research Letters39: L01807.
Fitzgerald WF (1989) Atmospheric and oceanic cycling of mercury. In: Riley JP andChester R (eds.) Chemical Oceanography, vol. 10, pp. 151–186. London: Academic.
Flock OR and Andreae MO (1996) Photochemical and non-photochemical formationand destruction of carbonyl sulfide and methyl mercaptan in ocean waters. MarineChemistry 54: 11–26.
Fogelqvist E (1985) Carbon-tetrachloride, tetrachloroethylene, 1,1,1-trichloroethane andbromoform in arctic seawater. Journal of Geophysical Research 90: 9181–9193.
Forster G, Upstill-Goddard RC, Gist N, Robinson C, Uher G, and Woodward EMS(2009) Nitrous oxide and methane in the Atlantic Ocean between 50�N and 52�S:Latitudinal distribution and sea-to-air flux. Deep Sea Research II 56: 964–976.
Fortescue GE and Pearson JRA (1967) On gas absorption into a turbulent liquid.Chemical Engineering Science 22: 1163–1176.
Frankenberg C, Aben I, Bergamaschi P, et al. (2011) Global column-averaged methanemixing ratios from 2003 to 2009 as derived from SCIAMACHY: Trends andvariability. Journal of Geophysical Research 116: D04302.
Frew NM (1997) The role of organic films in air–sea gas exchange. In: Liss PS andDuce RA (eds.) The Sea Surface and Global Change, pp. 121–172. Cambridge:Cambridge University Press.
Frew N, Bock E, Schimpf U, et al. (2004) Air-sea gas transfer: Its dependence on windstress small-scale roughness and surface films. Journal of Geophysical Research109: C08S17.
Frew NM, Goldman JC, Dennett MR, and Johnson AS (1990) Impact ofphytoplankton-generated surfactants on air–sea gas-exchange. Journal ofGeophysical Research 95: 3337–3352.
Gabric AJ, Ayers GP, and Sander GC (1995) Independent marine and atmosphericmodel estimates of the sea–air flux of dimethylsulfide in the Southern Ocean.Geophysical Research Letters 22: 3521–3524.
Gabric AJ, Simo R, Cropp RA, Hirst AC, and Dachs J (2004) Modelling estimates of theglobal emission of dimethylsulfide under enhanced greenhouse conditions. GlobalBiogeochemical Cycles 18: GB3016.
Galloway JN, Dentener FJ, Capone DG, et al. (2004) Nitrogen cycles: Past, present, andfuture. Biogeochemistry 70: 153–226.
Ganzeveld L, Helmig D, Fairall CW, Hare J, and Pozzer A (2009) Atmosphere–ocean ozone exchange: A global modeling study of biogeochemical, atmospheric,and waterside turbulence dependencies. Global Biogeochemical Cycles23: GB4021.
Garabetian F (1991) 14C-glucose uptake and 14C–CO2 production in surface microlayerand surface water samples: Influence of UV and visible radiation. Marine EcologyProgress Series 77: 21–26.
Garbe CS, Rutgersson A, Delille B, et al. (2013) Transfer across the air–sea interface.In: Liss PS and Johnson MT (eds.) Ocean–Atmosphere Interactions of Gases andParticles. Berlin: Springer.
Garbe CS, Schimpf U, and Jahne B (2004) A surface renewal model to analyze infraredimage sequences of the ocean surface for the study of air–sea heat and gasexchange. Journal of Geophysical Research 109: C08S15.
Garland JA, Elzerman AW, and Penkett SA (1980) The mechanism for dry deposition ofozone to seawater surfaces. Journal of Geophysical Research 85: 7488–7492.
Gattuso J-P and Hansson L (2011) Ocean Acidification, p. 326. Oxford: OxfordUniversity Press.
Ge X, Wexler AS, and Clegg SL (2011) Atmospheric amines—Part I. A review.Atmospheric Environment 45: 524–546.
Gibb SW, Mantoura RFC, Liss PS, and Barlow RG (1999) Distributions andbiogeochemistries of methylamines and ammonium in the Arabian Sea. Deep SeaResearch II 46: 593–615.
Gladyshev M (1997) Biophysics of the surface film of aquatic ecosystems. In: Liss LPSand Duce RA (eds.) The Sea Surface and Global Change, pp. 321–338. Cambridge:Cambridge University Press.
Glover DM, Frew NM, and McCue SJ (2007) Air–sea gas transfer velocity estimatesfrom the Jason-1 and TOPEX altimeters: Prospects for a long-term global timeseries. Journal of Marine Systems 66: 173–181.
Glover DM, Frew NM, McCue SJ, and Bock EJ (2002) A multi-year time series of globalgas transfer velocity from the TOPEX dual frequency, normalised radar backscatteralgorithm. In: Donelan MA, Drennan WM, Saltzman ES, and Wanninkhof R (eds.)Gas Transfer at Water Surfaces: Geophysical Monographs, vol. 127, pp. 325–333.Washington, DC: AGU.
Goddijn-Murphy L, Woolf D, and Callaghan A (2011) Parameterizations and algorithmsfor oceanic whitecap coverage. Journal of Physical Oceanography 41: 742–756.
Goldman JC and Dennett MR (1983) Carbon-dioxide exchange between air andseawater—No evidence for rate catalysis. Science 220: 199–201.
Goldman JC, Dennett MR, and Frew NM (1988) Surfactant effects on air seagas-exchange under turbulent conditions. Deep Sea Research 35: 1953–1970.
Goodwin KD, Lidstrom ME, and Oremland RS (1997) Marine bacterial degradation ofbrominated methanes. Environmental Science and Technology 31: 3188–3192.
Goodwin KD, Schaefer JK, and Oremland RR (1998) Bacterial oxidation ofdibromomethane and methyl bromide in natural waters and enrichment cultures.Applied and Environmental Microbiology 64: 4629–4636.
Groszko W and Moore RM (1998) Ocean–atmosphere exchange of methyl bromide: NWAtlantic and Pacific Ocean studies. Journal of Geophysical Research103: 16737–16741.
Gruber N, Gloor M, Mikaleff Fletcher SE, et al. (2009) Oceanic sources, sinks, andtransport of atmospheric CO2. Global Biogeochemical Cycles 23: GB1005. http://dx.doi.org/10.1029/2008GB003349.
Happell JD and Wallace DWR (1996) Methyl iodide in the Greenland/Norwegian Seasand the tropical Atlantic Ocean: Evidence for photochemical production.Geophysical Research Letters 23: 2105–2108.
Hara T, VanInwegen E, Wendelbo J, et al. (2007) Estimation of air–sea gas and heatfluxes from infrared imagery based on near surface turbulence models.In: Garbe CS, Handler RA, and Jahne B (eds.) Transport at the Air SeaInterface—Measurements, Models and Parameterizations. Berlin: Springer.
Hardy JT, Hunter KA, Calmet D, et al. (1997) Report Group 2—Biological effects ofchemical and radiative change in the sea surface. In: Liss LPS and Duce RA (eds.)The Sea Surface and Global Change, pp. 35–70. Cambridge: Cambridge UniversityPress.
Hare J, Fairall C, McGillis W, Edson J, Ward B, and Wanninkhof R (2004) Evaluation ofthe NOAA/COARE air–sea gas transfer parameterization using GasEx data. Journalof Geophysical Research 109: C08S02.
Harriott P (1962) A random eddy modification of the penetration theory. ChemicalEngineering Science 17: 149–154.
Hasse L and Liss P (1980) Gas exchange across the air–sea interface. Tellus32: 470–481.
Hicks BB, Baldocchi DD, Meyers TP, Hosker RP, and Matt DR (1987) A preliminarymultiple resistance routine for deriving dry deposition velocities from measuredquantities. Water, Air, and Soil Pollution 36: 311–330.
Higbie R (1935) The rate of absorption of a pure gas into a still liquid during shortperiods of exposure. American Institute of Chemical Engineers 35: 365–389.
Hintsa EJ, Dacey JWH, McGillis WR, Edson JB, Zappa CJ, and Zemmelink HJ (2004)Sea-to-air fluxes from measurements of the atmospheric gradient of dimethylsulfideand comparison with simultaneous relaxed eddy accumulation measurements.Journal of Geophysical Research 109: C01026.
Ho DT, Bliven LF, Wanninkhof R, and Schlosser P (1997) The effect of rain on air–watergas exchange. Tellus 49B: 149–158.
Ho DT, Law CS, Smith MJ, Schlosser P, Harville M, and Hill P (2006) Measurementsof air–sea gas exchange at high wind speeds in the southern ocean:Implications for global parameterizations. Geophysical Research Letters33: 16,611–16,616.
Ho DT, Veron F, Harrison E, Bliven LF, Scott N, and McGillis WR (2007) The combinedeffect of rain and wind on air–water gas exchange: A feasibility study. Journal of MarineSystems 66(1–4): 150–160. http://dx.doi.org/10.1016/j.jmarsys.2006.02.012.
Ho DT, Wanninkhof R, Schlosser P, Ullman DS, Hebert D, and Sullivan KF (2011)Toward a universal relationship between wind speed and gas exchange: Gas transfervelocities measured with 3He/SF6 during the southern ocean gas exchangeexperiment. Journal of Geophysical Research 116: C00F04.
Hoover TE and Berkshire DC (1969) Effects of hydration in carbon dioxideexchange across an air–water interface. Journal of Geophysical Research74: 456–464.
Hopkins FE, Turner SM, Nightingale PD, Steinke M, Bakker D, and Liss PS (2009)Ocean acidification and marine trace gas emissions. PNAS. http://dx.doi.org/10.1073/pnas.0907163107.
Howard EC, Henriksen JR, Buchan A, et al. (2006) Bacterial taxa that limit sulfur fluxfrom the ocean. Science 314: 649–651.
Hu Y, Stamnes K, Vaughan M, et al. (2008) Sea surface wind speed estimation fromspace-based lidar measurements. Atmospheric Chemistry and Physics Discussions8: 2771–2793.
Hu L, Yvon-Lewis SA, Liu Y, Salisbury JE, and O’Hern JE (2010) Coastalemissions of methyl bromide and methyl chloride along the eastern Gulf ofMexico and the east coast of the United States. Global Biogeochemical Cycles24: GB1007.
Huebert BJ, Blomquist BW, Hare JE, Fairall CW, Johnson JE, and Bates TS (2004)Measurement of the sea–air DMS flux and transfer velocity using eddy correlation.Geophysical Research Letters 31: L23113.
88 Air–Sea Exchange of Marine Trace Gases
Hughes C, Chuck AL, Rossetti H, et al. (2009) Seasonal cycle of seawater bromoformand dibromomethane concentrations in a coastal bay on the western AntarcticPeninsula. Global Biogeochemical Cycles 23: GB2024.
Hughes C, Franklin DJ, and Malin G (2011) Iodomethane production by two importantmarine cyanobacteria: Prochlorococcus marinus (CCMP 2389) and Synechococcussp. (CCMP 2370). Marine Chemistry 125: 19–25.
Hughes C, Johnson M, von Glasow R, et al. (2012) Climate-induced change in biogenicbromine emissions from the Antarctic marine biosphere. Global BiogeochemicalCycles 26: GB3019. http://dx.doi.org/10.1029/2012GB004295.
Hughes C, Kettle AJ, Unazi GA, Weston K, Jones MR, and Johnson MT (2010) Seasonalvariations in the concentrations of methyl and ethyl nitrate in a shallow freshwaterlake. Limnology and Oceanography 55: 305–314.
Hughes C, Malin G, Turley CM, Keely BJ, and Nightingale PD (2008) The production ofvolatile iodocarbons by biogenic marine aggregates. Limnology and Oceanography53: 867–872.
Hunter KA (1997) Chemistry of the sea-surface microlayer. In: Liss LPS and Duce RA(eds.) The Sea Surface and Global Change. Cambridge: Cambridge UniversityPress.
IPCC (2007) Climate change 2007: The physical science basis. In: Solomon S, Qin D,and Manning M, et al. (eds.) Contribution of Working Group 1 to the FourthAssessment Report of the Intergovernmental Panel on Climate Change, p. 996.Cambridge: Cambridge University Press.
Jackson DL, Wick GA, and Hare JE (2012) A comparison of satellite-derived carbondioxide transfer velocities from a physically based model with GasEx cruiseobservations. Journal of Geophysical Research 117: C00F13.
Jacob DJ, Field BD, Li Q, et al. (2005) Global budget of methanol: Constraints fromatmospheric observations. Journal of Geophysical Research 110: D08303.
Jacobs C, Kjeld JF, Nightingale P, et al. (2002) Possible errors in CO2 measurementsdue to near-surface concentration gradients. Journal of Geophysical Research107: 3128. http://dx.doi.org/10.1029/2001JC000983.
Jahne B (2009) Air–sea gas exchange. In: Steele JH, Turekian KK, and Thorpe SA (eds.)Encyclopedia Ocean Sciences, pp. 3434–3444. Oxford: Elsevier.
Jahne B, Munnich KO, Bosinger R, Dutzi A, Huber W, and Libner P (1987) On theparameters influencing air–water gas exchange. Journal of Geophysical Research92: 1937–1949.
Jean-Baptisite P and Poisson A (2000) Gas transfer experiment on a lake(Kerguelen Islands) using 3He and SF6. Journal of Geophysical Research105: 1177–1186.
Jeffery C, Robinson I, and Woolf D (2010) Tuning a physically-based model of theair–sea gas transfer velocity. Ocean Modelling 31: 28–35.
Johnson JE and Bates TS (1996) Sources and sinks of carbon monoxide in the mixedlayer of the tropical South Pacific Ocean. Global Biogeochemical Cycles10: 347–359.
Johnson M (2010) A numerical scheme to calculate temperature and salinity dependentair–water transfer velocities for any gas. Ocean Science 6: 913–932.
Johnson M, Hughes C, Bell T, and Liss P (2011) A rumsfeldian analysis of uncertaintyin air–sea gas exchange. In: Gas Transfer at Water Surface 2010, pp. 464–484.Kyoto: Kyoto University Press.
Johnson M, Sanders R, Avgoustidi V, et al. (2007) Ammonium accumulation during asilicate-limited diatom bloom indicates the potential for ammonia emission events.Marine Chemistry 106: 63–75.
Johnson MT and Bell TG (2008) Coupling between dimethylsufide emissionsand the ocean–atmosphere exchange of ammonia. Environmental Chemistry5: 259–267.
Johnson MT, Liss PS, Bell TG, et al. (2008) Field observations of theocean–atmosphere exchange of ammonia: Fundamental importance of temperatureas revealed by a comparison of high and low latitudes. Global BiogeochemicalCycles 22: GB1019.
Johnston AWB, Todd JD, Sun L, Nikolaidou-Katsaridou MN, Curson ARJ, and Rogers R(2008) Molecular diversity of bacterial production of the climate-changing gas,dimethyl sulphide, a molecule that impinges on local and global symbioses. Journalof Experimental Botany 59: 1059–1067.
Jones CE, Hornsby KE, Sommariva R, et al. (2010) Quantifying the contribution ofmarine organic gases to atmospheric iodine. Geophysical Research Letters37: L18804.
Jost C, Trentmann J, Sprung D, and Andreae MO (2003) Deposition of acetonitrile tothe Atlantic Ocean off Namibia and Angola and its implications for the atmosphericbudget of acetonitrile. Geophysical Research Letters 30. http://dx.doi.org/10.1029/2003GL017347.
Kameyama S, Tanimoto H, Inomata S, et al. (2010) High-resolution measurement ofmultiple volatile organic compounds dissolved in seawater using equilibrator inlet-proton transfer reaction-mass spectrometry (EI-PTR-MS). Marine Chemistry122: 59–73. http://dx.doi.org/10.1016/j.marchem.2010.08.003.
Kameyama S, Tsunogai U, Nakagawa F, et al. (2009) Enrichment of alkanes within aphytoplankton bloom during an in situ iron enrichment experiment in the westernsubarctic Pacific. Marine Chemistry 115: 92–101.
Kansal A (2009) Sources and reactivity of NMHCs and VOCs in the atmosphere: Areview. Journal of Hazardous Materials 166: 17–26.
Kanwisher J (1963) On the exchange of gases between the atmosphere and the sea.Deep Sea Research 10: 195–207.
Karlsson A, Auer N, Schulz-Bull D, and Abrahamsson K (2008) Cyanobacterialblooms in the Baltic—A source of halocarbons. Marine Chemistry 110:129–139.
Keeling RF, Piper SC, and Heimann M (1996) Global and hemispheric CO2 sinksdeduced from changes in atmospheric O2 concentration. Nature 381: 218–221.
Keeling RF, Stephens BB, Najjar RG, Doney SC, Archer D, and Heimann M (1998)Seasonal variations in the atmospheric O2/N2 ratio in relation to the kinetics ofair–sea gas exchange. Global Biogeochemical Cycles 12: 141–163.
Keene WC, Khalil MAK, Erickson DJ, et al. (1999) Composite global emissions ofreactive chlorine from anthropogenic and natural sources: Reactive ChlorineEmissions Inventory. Journal of Geophysical Research 104: 8429–8440.
Keene WC, Sander R, Pszenny AAP, Vogt R, Crutzen PJ, and Galloway JN (1998)Aerosol pH in the marine boundary layer: A review and model evaluation. Journal ofAerosol Science 29: 339–356.
Keller MD, Bellows WK, and Guillard RRL (1989) Dimethyl sulfide production inmarine-phytoplankton. ACS Symposium Series 393: 167–182.
Kelley CA and Jeffrey WH (2002) Dissolved methane concentration profiles and air–seafluxes from 41�S to 27�N. Global Biogeochemical Cycles 16. http://dx.doi.org/10.1029/2001GB001809.
Kettle AJ, Andreae MO, Amouroux D, et al. (1999) A global database of sea surfacedimethylsulfide (DMS) measurements and a procedure to predict sea surface DMSas a function of latitude, longitude, and month. Global Biogeochemical Cycles13: 399–444.
Kettle AJ, Kuhn U, von Hobe M, Kesselmeier J, Liss PS, and Andreae MO (2002)Comparing forward and inverse models to estimate the seasonal variation ofhemisphere-integrated fluxes of carbonyl sulfide. Atmospheric Chemistry andPhysics 2: 343–361.
Kettle AJ, Rhee TS, von Hobe M, Poulton A, Aiken J, and Andreae MO (2001) Assessingthe flux of different volatile sulfur gases from the ocean to the atmosphere. Journalof Geophysical Research 106: 12193–12209.
Kieber RJ, Zhou X, and Mopper K (1990) Formation of carbonyl compounds fromUV-induced photodegradation of humic substances in natural waters: Fate ofriverine carbon in the sea. Limnology and Oceanography 35: 1503–1515.
Kiene RP (1996) Production of methanethiol from dimethylsulfoniopropionate in marinesurface waters. Marine Chemistry 54: 69–83.
Kiene RP and Bates TS (1990) Biological removal of dimethyl sulfide from sea–water.Nature 345: 702–705.
Kiene RP, Linn LJ, and Bruton JA (2000) New and important roles for DMSP in marinemicrobial communities. Journal of Sea Research 43: 209–224.
Kim G, Hussain N, and Church TM (2000) Excess Po-210 in the coastal atmosphere.Tellus 52B: 74–80.
King DB, Butler JH, Montzka SA, Yvon-Lewis SA, and Elkins JW (2000) Implications ofmethyl bromide supersaturations in the temperate North Atlantic Ocean. Journal ofGeophysical Research 105: 19763–19769.
King DB, Butler JH, Yvon-Lewis SA, and Cotton SA (2002) Predicting oceanic methylbromide saturation from SST. Geophysical Research Letters 29. http://dx.doi.org/10.1029/2002GL016091.
King DB and Saltzman ES (1997) Removal of methyl bromide in coastal seawater:Chemical and biological rates. Journal of Geophysical Research102: 18715–18721.
Kirst GO, Thiel C, Wolff H, Nothnagel J, Wanzek M, and Ulmke R (1991)Dimethylsulfoniopropionate (DMSP) in ice-algae and its possible biological role.Marine Chemistry 35: 381–388.
Kitidis V, Tilstone GH, Smyth TJ, Torres R, and Law CS (2011) Carbon monoxideemission from a Mauritanian upwelling filament. Marine Chemistry 127: 123–133.
Klick S and Abrahamsson K (1992) Biogenic volatile iodated hydrocarbons in theocean. Journal of Geophysical Research 97: 12683–12687.
Kloster S, Six KD, Feichter J, et al. (2007) Response of dimethylsulfide (DMS) in theocean and atmosphere to global warming. Journal of Geophysical Research112: G03005.
Korhonen P, Kulmala M, Laaksonen A, Viisanen Y, McGraw R, and Seinfled JH (1999)Ternary nucleation of H2SO4, NH3, and H2O in the atmosphere. Journal ofGeophysical Research 104: 26349–26354.
Krysell M, Fogelqvist E, and Tanhua T (1994) Apparent removal of the transient tracercarbon-tetrachloride from anoxic seawater. Geophysical Research Letters21: 2511–2514.
Air–Sea Exchange of Marine Trace Gases 89
Kurihara MK, Kimura M, Iwamoto Y, et al. (2010) Distributions of short-livediodocarbons and biogenic trace gases in the open ocean and atmosphere in thewestern North Pacific. Marine Chemistry 118: 156–170.
Kurten T, Loukonen V, Vehkamaki H, and Kulmala M (2008) Amines are likely toenhance neutral and ion-induced sulfuric acid–water nucleation in the atmospheremore effectively than ammonia. Atmospheric Chemistry and Physics 8: 4095–4103.
Lamborg CH, Fitzgerald WF, O’Donnell J, and Torgersen T (2002) A non-steady statebox model of global-scale mercury biogeochemistry with interhemisphericatmospheric gradients. Abstracts of Papers of the American Chemical Society223: 072-ENVR.
Lamont JC and Scott DS (1970) An eddy cell model of mass transfer into the surface ofa turbulent liquid. AIChE Journal 16: 513–519.
Lana A, Bell TG, Simo R, et al. (2011) An updated climatology of surface dimethylsulfideconcentrations and emission fluxes in the global ocean. Global BiogeochemicalCycles 25: GB1994. http://dx.doi.org/10.1029/GB003850.
Law CS, Sjoberg TN, and Ling RD (2002) Atmospheric emission and cycling of carbonmonoxide in the Scheldt Estuary. Biogeochemistry 59: 69–94.
Ledwell JR (1984) The variation of the gas transfer coefficient with molecular diffusivity.In: Brutsaert W and Jirka GH (eds.) Gas Transfer at Water Surfaces, pp. 293–302.Dordrecht: Reidel.
Lee BS, Bullister JL, and Whitney FA (1999) Chlorofluorocarbon CFC-11 and carbontetrachloride removal in Saanich Inlet, an intermittently anoxic basin. MarineChemistry 66: 171–185.
Lee RJ and Saylor JR (2010) The effect of a surfactant monolayer on oxygen transferacross an air–water interface during mixed convection. International Journal of Heatand Mass Transfer 53: 3405–3413.
Lenschow DH, Pearson R, and Stankov BB (1982) Measurements of ozone vertical fluxto ocean and forest. Journal of Geophysical Research 87: 8833–8837.
Li Q, Jacob DJ, Bey I, et al. (2000) Atmospheric hydrogen cyanide (HCN): Biomassburning source, ocean sink? Geophysical Research Letters 27: 357–360.
Li W (2004) Modelling air–sea fluxes during a western Pacific typhoon: Role of seaspray. Advances in Atmospheric Sciences 21: 269–276.
Liang Q, Stolarski RS, Kawa SR, et al. (2010) Finding the missing stratospheric Bry: Aglobal modeling study of CHBr3 and CH2Br2. Atmospheric Chemistry and Physics10: 2269–2286.
Liss PS (1971) Exchange of SO2 between the atmosphere and natural waters. Nature233: 327–329.
Liss P (1975) Chemistry of the sea surface microlayer. In: Riley J and Skirrow G (eds.)Chemical Oceanography, vol. 2, pp. 192–244. London: Academic.
Liss PS (1983) Gas transfer: Experiments and geochemical implications. In: Liss PS andSlinn WGN (eds.) Air–Sea Exchange of Gases and Particles, pp. 241–298.Dordrecht: Reidel.
Liss PS, Balls PW, Martinelli FN, and Coantic M (1981) The effect of evaporation andcondensation on gas transfer across an air–water-interface. Oceanologica Acta4: 129–138.
Liss PS and Lovelock JE (2007) Climate change: The effect of DMS emissions.Environmental Chemistry 4: 377–378.
Liss PS and Martinelli FN (1978) The effect of oil films on the transfer of oxygen andwater vapour across an air–water interface. Thalassia Jugoslavica 14: 215–220.
Liss PS and Merlivat L (1986) Air–sea gas exchange rates: Introduction and synthesis.In: Buat-Menard P (ed.) The Role of Air–Sea Gas Exchange in Geochemical Cycling,pp. 113–127. Dordrecht: Reidel.
Liss P and Slater P (1974) Flux of gases across the air–sea interface. Nature247: 181–184.
Lobert JM, Yvon-Lewis SA, Butler JH, Montzka SA, and Myers RC (1997)Undersaturation of CH3Br in the Southern Ocean. Geophysical Research Letters24: 171–172.
Lohmann U and Feichter J (2005) Global indirect aerosol effects: A review. AtmosphericChemistry and Physics 5: 715–737.
Lovelock JE (1975) Natural halocarbons in the air and seawater. Nature 256: 193–194.Lovelock JE (1979) Gaia: A new look at life on Earth. Oxford: Oxford University Press.Lovelock JE, Maggs RJ, and Rasmussen RA (1972) Atmospheric dimethyl sulfide and
the natural sulfur cycle. Nature 237: 452–453.MacDonald S and Moore RM (2007) Seasonal and spatial variations in methyl chloride
in NW Atlantic waters. Journal of Geophysical Research 112: C05028.Mackay D and Yeun ATK (1983) Mass transfer coefficient correlations for
volatilization of organic solutes from water. Environmental Science and Technology17: 211–217.
Mahajan AS, Shaw M, Oetjen H, et al. (2010) Evidence of reactive iodine chemistry inthe Arctic boundary layer. Journal of Geophysical Research 115: D20303.
Malin G, Wilson WH, Bratbak G, Liss PS, and Mann NH (1998) Elevated production ofdimethylsulfide resulting from viral infection of cultures of Phaeocystis pouchetii.Limnology and Oceanography 43: 1389–1393.
Manley SL and de la Cuesta JL (1997) Methyl iodide production from marinephytoplankton cultures. Limnology and Oceanography 42: 142–147.
Manning AJ, O’Doherty S, Jones AR, Simmonds PG, and Derwent RG (2011)Estimating UK methane and nitrous oxide emissions from 1990 to 2007using an inversion modeling approach. Journal of Geophysical Research116: D02305.
Marandino CA, De Bruyn WJ, Miller SD, Prather MJ, and Saltzman ES (2005) Oceanicuptake and the global atmospheric acetone budget. Geophysical Research Letters32: L15806.
Marandino CA, De Bruyn WJ, Miller SD, and Saltzman ES (2009) Open ocean DMS air/sea fluxes over the eastern South Pacific Ocean. Atmospheric Chemistry andPhysics 9: 345–356.
Martino M, Leze B, Baker AR, and Liss PS (2012) Chemical controls on ozonedeposition to water. Geophysical Research Letters 39: L05809.
Martino M, Liss PS, and Plane JMC (2005) The photolysis of dihalomethanes in surfaceseawater. Environmental Science and Technology 39: 7097–7101.
Marty DG (1993) Methanogenic bacteria in seawater. Limnology and Oceanography38: 452–456.
Matthews B (1999) The Rate of Air–Sea CO2 Exchange: Chemical Enhancement andCatalysis by Marine Microalgae. PhD Thesis, University of East Anglia, UK.
McCulloch A (2003) Chloroform in the environment: Occurrence, sources, sinks andeffects. Chemosphere 50: 1291–1308.
McGillis WR, Edson JB, Hare JE, and Fairall CW (2001) Direct covariance air–sea CO2fluxes. Journal of Geophysical Research 106: 16729–16745.
McGillis WR, Edson JB, Ware JD, et al. (2001) Carbon dioxide flux techniquesperformed during GasEx-98. Marine Chemistry 75: 267–280.
McGillis WR and Wanninkhof R (2006) Aqueous CO2 gradients for air–sea fluxestimates. Marine Chemistry 98: 100–108.
McKay WA, Turner MF, Jones BMR, and Halliwell CM (1996) Emissions ofhydrocarbons from marine phytoplankton—Some results from controlledlaboratory experiments. Atmospheric Environment 30: 2583–2593.
McNeil C and D’Asaro E (2007) Parameterization of air sea gas fluxes at extremewindspeeds. Journal of Marine Systems 66: 110–121.
Meredith MP, VanScoy KA, Watson AJ, and Locarnini RA (1996) On the use of carbontetrachloride as a transient tracer of Weddell Sea deep and bottom waters.Geophysical Research Letters 23: 2943–2946.
Merlivat L and Memery L (1983) Gas exchange across an air–water interface:Experimental results and modeling of bubble contribution to transfer. Journal ofGeophysical Research 88: 707–724.
Miller S, Marandino C, de Bruyn W, and Saltzman ES (2009) Air-sea gas exchange ofCO2 and DMS in the North Atlantic by eddy covariance. Geophysical ResearchLetters 36. http://dx.doi.org/10.1029/2009gl038907.
Miller WL and Moran MA (1997) Interaction of photochemical and microbial processesin the degradation of refractory dissolved organic matter from a coastal marineenvironment. Limnology and Oceanography 42: 1317–1324.
Millet DB, Guenther A, Siegel DA, et al. (2010) Global atmospheric budget ofacetaldehyde: 3-D model analysis and constraints from in-situ and satelliteobservations. Atmospheric Chemistry and Physics 10: 3405–3425.
Millet DB, Jacob DJ, Custer TG, et al. (2008) New constraints on terrestrial and oceanicsources of methanol. Atmospheric Chemistry and Physics Discussions8: 7609–7655.
Monahan E and O’Muircheartaigh I (1980) Optimal power-law description of oceanicwhitecap coverage dependence on wind speed. Journal of Physical Oceanography10: 2094–2099.
Moore RM (2006) Methyl halide production and loss rates in sea water from fieldincubation experiments. Marine Chemistry 101: 213–219.
Moore RM (2008) A photochemical source of methyl chloride in saline waters.Environmental Science and Technology 42: 1933–1937.
Moore RM, Groszko W, and Niven SJ (1996) Ocean–atmosphere exchange of methylchloride: Results from NW Atlantic and Pacific Ocean studies. Journal ofGeophysical Research 101: 28529–28538.
Moore RM, Punshon S, Mahaffey C, and Karl D (2009) The relationship betweendissolved hydrogen and nitrogen fixation in ocean waters. Deep Sea Research I56: 1449–1458.
Moore RM and Tokarczyk R (1993) Volatile biogenic halocarbons in the NorthwestAtlantic. Global Biogeochemical Cycles 7: 195–210.
Moore RM, Webb M, Tokarczyk R, and Wever R (1996) Bromoperoxidase andiodoperoxidase enzymes and production of halogenated methanes in marine diatomcultures. Journal of Geophysical Research 101: 20899–20908.
Moore RM and Zafiriou OC (1994) Photochemical production of methyl iodide inseawater. Journal of Geophysical Research 99: 16415–16420.
Mopper K and Stahovec WL (1986) Sources and sinks of low molecular weight organiccarbonyl compounds in seawater. Marine Chemistry 19: 305–321.
90 Air–Sea Exchange of Marine Trace Gases
Mopper K, Zhou X, Kieber RJ, Kieber DJ, Sikorski RJ, and Jones RD (1991)Photochemical degradation of dissolved organic carbon and its impact on theoceanic carbon cycle. Nature 353: 60–62.
Mosher BW and Duce RA (1987) A global atmospheric selenium budget. Journal ofGeophysical Research 92: 13289–13298.
Muller SA, Joos F, Plattner GK, Edwards NR, and Stocker TF (2008) Modeled naturaland excess radiocarbon: Sensitivities to the gas exchange formulation and oceantransport strength. Global Biogeochemical Cycles 22: GB3011.
Murphy JG, Oram DE, and Reeves CE (2010) Measurements of volatile organiccompounds over West Africa. Atmospheric Chemistry and Physics 10:5281–5294.
Murrell JC and McDonald IR (2000) Methylotrophy. In: Lederberg J (ed.) Encyclopediaof Microbiology, 2nd edn., vol. 3. New York, NY: Academic.
Naegler T (2009) Reconciliation of excess 14C-constrained global CO2 piston velocityestimates. Tellus B 61: 372–384.
Naegler T, Ciais P, Rodgers K, and Levin I (2006) Excess radiocarbon constraints onair–sea gas exchange and the uptake of CO2 by the oceans. Geophysical ResearchLetters 33: L11802.
Naik V, Fiore AM, Horowitz LW, et al. (2010) Observational constraints on the globalatmospheric budget of ethanol. Atmospheric Chemistry and Physics10: 5361–5370.
Nevison CD, Lueker TJ, and Weiss RF (2004) Quantifying the nitrous oxide source fromcoastal upwelling. Global Biogeochemical Cycles 18: GB1018.
Nevitt GA, Veit RR, and Kareiva P (1995) Dimethyl sulphide as a foraging cue forAntarctic Procellariiform seabirds. Nature 376: 680–682. http://dx.doi.org/10.1038/376680ao.
Nguyen BC, Belviso S, Mihalopoulos N, Gostan J, and Nival P (1988) Dimethylsulfide production during natural phytoplanktonic blooms. Marine Chemistry24: 133–141.
Nightingale PD (1991) Low Molecular Weight Halocarbons in Seawater. PhD Thesis,University of East Anglia, UK
Nightingale PD (2009) Air–sea gas exchange. In: Le Quere C and Saltmann ES (eds.)Surface Ocean-Lower Atmosphere Processes, pp. 69–97. Washington, DC:American Geophysical Union (Geophysical Monograph 187).
Nightingale PD and Liss PS (2003) Gases in seawater. In: Holland HD and Turekian KK(eds.) Treatise on Geochemistry, vol. 6, pp. 49–81. Oxford: Elsevier.
Nightingale PD, Liss PS, and Schlosser P (2000a) Measurements of air–sea gastransfer during an open ocean algal bloom. Geophysical Research Letters27: 2117–2120.
Nightingale PD, Malin G, Law CS, et al. (2000b) In situ evaluation of air–sea gasexchange parameterizations using novel conservative and volatile tracers. GlobalBiogeochemical Cycles 14: 373–387.
Nightingale PD, Malin G, and Liss PS (1995) Production of chloroform and otherlow-molecular-weight halocarbons by some species of macroalgae. Limnology andOceanography 40: 680–689.
Novelli PC, Lang PM, Masarie KA, Hurst DM, Myers R, and Elkins JW (1999) Molecularhydrogen in the troposphere: Global distribution and budget. Journal ofGeophysical Research 104: 30,427–30,444.
Ooki A, Tsuda A, Kameyama S, et al. (2010) Methyl halides in surface seawater andmarine boundary layer of the northwest Pacific. Journal of Geophysical Research115: C10013.
Oremland RS (1979) Methanogenic activity in plankton samples and in fish intestines: Amechanism for in-situ methanogenesis in oceanic waters. Limnology andOceanography 24: 1136–1141.
Orr JC, Maier-Reimer E, Mikolajewicz U, et al. (2001) Estimates of anthropogeniccarbon uptake from four three-dimensional global ocean models. GlobalBiogeochemical Cycles 15: 43–60.
Packwood DM and Phillips LF (2010a) Irreversible thermodynamics of a gas–liquidinterface. Advances in Geosciences 19: 499–509. http://dx.doi.org/10.1142/9789812838162_0037.
Packwood DM and Phillips LF (2010b) Non-equilibrium thermodynamics of the gas–liquid interface: Measurement of the Onsager heat of transport for nitrous oxide atthe surface of water. Journal of Non-Equilibrium Thermodynamics 35(1): 75–84.http://dx.doi.org/10.1515/JNETDY.2010.005.
Peng TH, Broecker WS, Mathieu G, Li YH, and Bainbridge AE (1979) Radon evasionrates in the Atlantic and Pacific Oceans as determined during the GEOSECSProgram. Journal of Geophysical Research 84: 2471–2486.
Phillips LF (1994) Experimental demonstration of coupling of heat and matterfluxes at a gas–water interface. Journal of Geophysical Research99: 18577–18584.
Phillips LF (1997) The physical chemistry of air–sea gas exchange. In: Liss PS andDuce RA (eds.) The Sea Surface and Global Change, pp. 207–250. Cambridge:Cambridge University Press.
Plane JMC (1989) Gas-phase atmospheric oxidation of biogenic sulfur-compounds—A review. ACS Symposium Series 393: 404–423.
Plane JMC, Blough NV, Ehrhardt MG, Waters K, Zepp RG, and Zika RG (1997) ReportGroup 3—Photochemistry in the sea-surface microlayer. In: Liss PS and Duce RA (eds.)The Sea Surface and Global Change, pp. 71–92. Cambridge: Cambridge UniversityPress.
Plass-Dulmer C, Koppmann R, Ratte M, and Rudolph J (1995) Lightnonmethane hydrocarbons in seawater. Global Biogeochemical Cycles 9:79–100.
Pozzer A, Pollmann J, Taraborrelli D, et al. (2010) Observed and simulated globaldistribution and budget of atmospheric C2–C5 alkanes. Atmospheric Chemistry andPhysics 10: 4403–4422.
Prytherch J, Yelland M, Pascal R, Moat B, Skjelvan I, and Neill C (2010a) Directmeasurements of the CO2 flux over the ocean: Development of a novel method.Geophysical Research Letters 37: L03607.
Prytherch J, Yelland M, Pascal R, Moat B, Skjelvan I, and Srokosz M (2010b) Openocean gas transfer velocity derived from long-term direct measurements of the CO2flux. Geophysical Research Letters 37: L23607.
Punshon S and Moore RM (2008) Photochemical production of molecular hydrogen inlake water and coastal seawater. Marine Chemistry 108: 215–220.
Quack B and Wallace DWR (2003) Air-sea flux of bromoform: Controls, rates, andimplications. Global Biogeochemical Cycles 17. http://dx.doi.org/10.1029/2002GB001890.
Quinn PK and Bates TS (2011) The case against climate regulation via oceanicphytoplankton sulphur emissions. Nature 380: 51–56. http://dx.doi.org/10.1038/nature10580.
Raven JA (1995) Phycological reviews. 15. Photosynthetic and nonphotosynthetic rolesof carbonic-anhydrase in algae and cyanobacteria. Phycologia 34: 93–101.
Rayman MP (2000) The importance of selenium to human health. The Lancet356: 233–241.
Redfield AC (1948) The exchange of oxygen across the sea surface. Journal of MarineResearch 7: 347–361.
Rees AP, Brown IJ, Clark DR, and Torres R (2011) The Lagrangian progression ofnitrous oxide within filaments formed in the Mauritanian upwelling. GeophysicalResearch Letters 38: L21606.
Rhee T, Nightingale P, Woolf D, Caulliez G, Bowyer P, and Andreae M (2007) Influenceof energetic wind and waves on gas transfer in a large wind-wave tunnel facility.Journal of Geophysical Research 112: 5027.
Rhee TS, Brenninkmeijer CAM, and Rockmann T (2005) The overwhelming role of soilsin the global atmospheric hydrogen cycle. Atmospheric Chemistry and PhysicsDiscussions 5: 11,215–11,248.
Rhew RC, Arn The Y, Abel T, Atwood A, and Mazeas O (2008) Chloroform emissionsfrom the Alaskan Arctic tundra. Geophysical Research Letters 35: L21811.
Richter U and Wallace DWR (2004) Production of methyl iodide in the tropical AtlanticOcean. Geophysical Research Letters 31: L23S03.
Riebesell U, Zondervan I, Rost B, Tortell PD, Zeebe RE, and Morel FMM (2000)Reduced calcification of marine plankton in response to increased atmospheric CO2.Nature 407: 364–367.
Riemer DD, Milne PJ, Zika RG, and Pos WH (2000) Photoproduction of nonmethanehydrocarbons (NMHCs) in seawater. Marine Chemistry 71: 177–198.
Robertson JE and Watson AJ (1992) Thermal skin effect of the surface ocean and itsimplications for CO2 uptake. Nature 358: 738–740.
Roether W, Klein B, and Bulsiewicz K (2001) Apparent loss of CFC-113 in the upperocean. Journal of Geophysical Research 106: 2679–2688.
Rowe M, Fairall C, and Perlinger J (2011) Chemical sensor resolution requirements fornear-surface measurements of turbulent fluxes. Atmospheric Chemistry and Physics11: 5263–5275.
Salisbury G, Williams J, Holzinger R, et al. (2003) Ground-based PTR-MSmeasurements of reactive organic compounds during the MINOS campaign inCrete, July–August 2001. Atmospheric Chemistry and Physics 3: 925–940.
Salter M, Upstill-Goddard R, Nightingale P, et al. (2011) Impact of an artificial surfactantrelease on air sea gas fluxes during Deep Ocean Gas Exchange Experiment II.Journal of Geophysical Research 116: C11016.
Sander R (1999) Compilation of Henry’s Law constants for inorganic and organicspecies of potential importance in environmental chemistry (Version 3). http://www.henrys-law.org
Sarmiento JL and Gruber N (2006) Ocean Biogeochemical Dynamics, pp. 503.Princeton, NJ: Princeton University Press.
Sarmiento JL, Monfray P, Maier-Reimer E, Aumont O, Murnane RJ, and Orr JC (2000)Sea-air CO2 fluxes and carbon transport: A comparison of three ocean generalcirculation models. Global Biogeochemical Cycles 14: 1267–1281.
Sarmiento JL and Sundquist ET (1992) Revised budget for the oceanic uptake ofanthropogenic carbon-dioxide. Nature 356: 589–593.
Air–Sea Exchange of Marine Trace Gases 91
Savioe DI, Prospero JM, Larsen RJ, et al. (1993) Nitrogen and sulfur species inAntarctic aerosols at Mawson, Palmer Station, and Marsh (King George Island).Journal of Atmospheric Chemistry 17: 95–122.
Scarratt MG and Moore RM (1996) Production of methyl chloride and methyl bromidein laboratory cultures of marine phytoplankton. Marine Chemistry 54: 263–272.
Scarratt MG and Moore RM (1998) Production of methyl bromide and methyl chloridein laboratory cultures of marine phytoplankton II. Marine Chemistry 59: 311–320.
Schmale O, Schneider von Deimling J, Gulzow W, Nausch G, Waniek JJ, and Rehder G(2010) Distribution of methane in the water column of the Baltic Sea. GeophysicalResearch Letters 37: L12604.
Schmidt U (1974) Molecular hydrogen in the atmosphere. Tellus 26: 78–90.Shapiro SD, Schlosser P, Smethie WM, and Stute M (1997) The use of H-3 and
tritiogenic He-3 to determine CFC degradation and vertical mixing rates inFramvaren Fjord, Norway. Marine Chemistry 59: 141–157.
Sieburth JM (1960) Acrylic acid, an “antibiotic” principle in Phaeocystis blooms inAntarctic waters. Science 132: 676–677.
Simpson IJ, Blake NJ, Blake DR, et al. (2003) Photochemical production and evolutionof selected C2–C5 alkyl nitrates in tropospheric air influenced by Asian outflow.Journal of Geophysical Research 108. http://dx.doi.org/10.1029/2002JD002830.
Singh HB, Kanakidou M, Crutzen PJ, and Jacob DJ (1995) High concentrations andphotochemical fate of oxygenated hydrocarbons in the global troposphere. Nature378: 50–54.
Singh HB, Salas LJ, Chatfield RB, et al. (2004) Analysis of the atmosphericdistribution, sources and sinks of oxygenated volatile organic chemicals based onmeasurements over the Pacific during TRACE-P. Journal of Geophysical Research109: D15S07.
Singh HB, Salas L, Herlth D, et al. (2003) In situ measurements of HCN and CH3CN overthe Pacific Ocean: Sources, sinks, and budgets. Journal of Geophysical Research108. http://dx.doi.org/10.1029/2002JD003006.
Singh HB, Salas LJ, and Stiles RE (1983) Methyl halides in and over the EasternPacific (40-degrees-N–32-degrees-S). Journal of Geophysical Research88: 3684–3690.
Smethie WM, Takahashi T, and Chipman DW (1985) Gas exchange and CO2 flux in thetropical Atlantic Ocean determined from Rn-222 and pCO2 measurements. Journalof Geophysical Research 90: 7005–7022.
Smythe-Wright D, Boswell SM, Breithaupt P, Davidson RD, Dimmer CH, and EirasDiaz LB (2006) Methyl iodide production in the ocean: Implications for climatechange. Global Biogeochemical Cycles 20: GB3003.
Stanley RHR, Baschek B, Lott DE III, and Jenkins WJ (2009) A new automated methodfor measuring noble gases and their isotopic ratios in water samples. Geochemistry,Geophysics, Geosystems 10: Q05,008.
Stefels J, Steinke M, Turner S, Malin G, and Belviso S (2007) Environmental constraintson the production and removal of the climatically active gas dimethylsulphide(DMS) and implications for ecosystem modeling. Biogeochemistry 83: 245–275.
Steinke M, Malin G, and Liss PS (2002) Trophic interactions in the sea: An ecologicalrole for climate relevant volatiles? Journal of Phycology 38: 630–638.
Steinke M, Stefels J, and Stamhuis E (2006) Dimethyl sulfide triggers search behaviorin copepods. Limnology and Oceanography 51: 1925–1930.
Steinke M, Wolfe GV, and Kirst GO (1998) Partial characterisation ofdimethylsulfoniopropionate (DMSP) lyase isozymes in 6 strains of Emilianiahuxleyi. Marine Ecology Progress Series 175: 215–225.
Strode SA, Jaegle L, Selin NE, et al. (2007) Air–sea exchange in the global mercurycycle. Global Biogeochemical Cycles 21: GB1017. http://dx.doi.org/10.1029/2006GB002766.
Stubbins A, Law CS, Uher G, and Upstill-Goddard RC (2011) Carbon monoxideapparent quantum yields and photoproduction in the Tyne estuary. Biogeosciences8: 703–713.
Stubbins A, Uher G, Kitidis V, Law CS, Upstill-Goddard RC, and Woodward EMS(2006) The open-ocean source of atmospheric carbon monoxide. Deep SeaResearch II 53: 1685–1694.
Sunda W, Kieber DJ, Kiene RP, and Huntsman S (2002) An oxidant function for DMSPand DMS in marine algae. Nature 418: 317–320.
Sweeney C, Gloor E, Jacobson A, et al. (2007) Constraining global air–sea gasexchange for CO2 with recent bomb
14C measurements. Global BiogeochemicalCycles 21: GB2015.
Tait VK and Moore RM (1995) Methyl chloride (CH3Cl) production in phytoplanktoncultures. Limnology and Oceanography 40: 189–195.
Takahashi T, Sutherland SC, Sweeney C, et al. (2002) Global sea–air CO2 flux based onclimatological surface ocean pCO(2), and seasonal biological and temperatureeffects. Deep Sea Research II 49: 1601–1622.
Takahashi T, Sutherland SC, Wanninkhof R, et al. (2009) Climatological mean and decadalchange in surface ocean pCO2, and net sea–air CO2 flux over the global oceans. Deep-Sea Research Pt II 56: 544–577. http://dx.doi.org/10.1016/j.dsr2.2008.12.009.
Tanhua T, Fogelqvist E, and Basturk O (1996) Reduction of volatile halocarbons inanoxic seawater, results from a study in the Black Sea. Marine Chemistry54: 159–170.
Thornton DC, Bandy AR, Blomquist BW, Driedger AR, and Wade TP (1999) Sulfurdioxide distribution over the Pacific Ocean 1991–1996. Journal of GeophysicalResearch 104: 5845–5854.
Thornton DC, Bandy AR, Tu FH, et al. (2002) Fast airborne sulfur dioxide measurementsby atmospheric pressure ionization mass spectrometry apims. Journal ofGeophysical Research-Atmospheres 107: (D22)12.
Todd JD, Rogers R, Li YG, et al. (2007) Structural and regulatory genes required tomake the gas dimethyl sulfide in bacteria. Science 315: 666–669.
Tokarczyk R and Moore RM (1994) Production of volatile organohalogens byphytoplankton cultures. Geophysical Research Letters 21: 285–288.
Tokarczyk R and Moore RM (2006) A seasonal study of methyl bromideconcentrations in the North Atlantic (35�–60�N). Journal of Geophysical Research111: D08304.
Tokarczyk R, Saltzman ES, Moore RM, and Yvon-Lewis SA (2003) Biologicaldegradation of methyl chloride in coastal seawater. Global Biogeochemical Cycles17. http://dx.doi.org/10.1029/2002GB001949.
Tsai W and Liu K (2003) An assessment of the effect of sea-surface surfactant on globalatmosphere–ocean CO2 flux. Journal of Geophysical Research 108: 3127.
Tsunogai S and Tanaka N (1980) Flux of oxygen across the air–sea interface asdetermined by the analysis of dissolved components in sea-water. GeochemicalJournal 14: 227–234.
Turk D, Zappa CJ, Meinen CS, et al. (2010) Rain impacts on CO2 exchange in thewestern equatorial pacific ocean. Geophysical Research Letters 7: L23610.
Turner SM, Harvey MJ, Law CS, Nightingale PD, and Liss PS (2004) Iron-inducedchanges in oceanic sulfur biogeochemistry. Geophysical Research Letters31: L14307.
Twomey S (1991) Aerosols, clouds and radiation. Atmospheric Environment25A: 2435–2442. http://dx.doi.org/10.1016/0960-1686(91)90159-5.
Uher G and Andreae MO (1997) Photochemical production of carbonyl sulfide in NorthSea water: A process study. Limnology and Oceanography 42: 432–442.
Upstill-Goddard RC, Barnes J, Frost T, Punshon S, and Owens NJP (2000) Methane inthe southern North Sea: Low-salinity inputs, estuarine removal, and atmosphericflux. Global Biogeochemical Cycles 14: 1205–1217.
Upstill-Goddard R, Frost T, Henry G, Franklin M, Murrell J, and Owens N (2003)Bacterioneuston control of air-water methane exchange determined with a laboratorygas exchange tank. Global Biogeochemical Cycles 17: 1108.
Upstill-Goddard RC, Watson AJ, Liss PS, and Liddicoat MI (1990) Gas transfer in lakesmeasured with SF6. Tellus 42B: 364–377.
Valentine RL and Zepp RG (1993) Formation of carbon-monoxide from thephotodegradation of terrestrial dissolved organic-carbon in natural-waters.Environmental Science and Technology 27: 409–412.
Vallina SM, Simo R, and Gasso S (2006) What controls CCN seasonality in theSouthern Ocean? A statistical analysis based on satellite-derived chlorophyll andCCN and model-estimated OH radical and rainfall. Global Biogeochemical Cycles20: GB101. http://dx.doi.org/10.1029/2005GB002597.
Vallina SM, Simo R, and Manizza M (2007) Weak response of oceanicdimethylsulfide to upper mixing shoaling induced by global warming. Proceedingsof the National Academy of Sciences of the United States of America104: 16,004–16,009.
Veron F, Melville WK, and Lenain L (2011) The effects of small-scale turbulence onair–sea heat flux. Journal of Physical Oceanography 41: 205–220.
Vlahos P, Monahan EC, Huebert BJ, and Edson JB (2011) Wind-dependence of DMStransfer velocity: Comparison of model with recent southern ocean observations.Kyoto: Kyoto University Press.
Vogel TM, Criddle CS, and McCarty PL (1987) Transformations of halogenatedaliphatic-compounds. Environmental Science and Technology 21: 722–736.
Vogt M and Liss PS (2009) Dimethylsulfide and climate. SOLAS Geophysical ResearchSeries 187. http://dx.doi.org/10.1029/2008GM000790.
von Glasow R and Crutzen PJ (2004) Model study of multi-phase DMS oxidation with afocus on halogens. Atmospheric Chemistry and Physics 4: 589–608.
von Glasow R, von Kuhlmann R, Lawrence MG, Platt U, and Crutzen PJ (2004) Impact ofreactive bromine chemistry in the trophosphere. Atmospheric Chemistry andPhysics 4: 2481–2497.
Wang L, Moore RM, and Cullen JJ (2009) Methyl iodide in the NW Atlantic: Spatial andseasonal variation. Journal of Geophysical Research 114: C07007.
Wanninkhof R (1992) Relationship between wind speed and gas exchange over theocean. Journal of Geophysical Research 97: 7373–7382.
Wanninkhof R, Asher WE, Ho DT, Sweeney C, and McGillis WR (2009) Advances inquantifying air–sea gas exchange and environmental forcing. Annual Review ofMarine Science 1: 213–244.
92 Air–Sea Exchange of Marine Trace Gases
Wanninkhof R, Hitchcock G, Wiseman WJ, et al. (1997) Gas exchange, dispersion, andbiological productivity on the west Florida shelf: Results from a Lagrangian tracerstudy. Geophysical Research Letters 24: 1767–1770.
Wanninkhof R and Knox M (1996) Chemical enhancement of CO2 exchange in naturalwaters. Limnology and Oceanography 41: 689–697.
Wanninkhof R, Ledwell JR, and Broecker WS (1985) Gas exchange wind speed relationmeasured with sulfur hexafluoride on a lake. Science 227: 1224–1226.
Wanninkhof R, Ledwell JR, Broecker WS, and Hamilton M (1987) Gas-exchange onMono lake and Crowley lake, California. Journal of Geophysical Research92: 14567–14580.
Wanninkhof R and McGillis WR (1999) A cubic relationship between air–sea CO2exchange and wind speed. Geophysical Research Letters 26: 1889–1892.
Wanninkhof R, Sullivan KF, and Top Z (2004) Air–sea gas transfer in the southernocean. Journal of Geophysical Research 109: C08S19.
Ward B, Wanninkhof R, McGillis WR, et al. (2004) Biases in the air–sea flux of CO2resulting from ocean surface temperature gradients. Journal of GeophysicalResearch 109: C08S08.
Watson AJ and Orr JC (2003) Carbon dioxide fluxes in the global ocean.In: Fasham MJR (ed.) Ocean Biogeochemistry: A JGOFS Synthesis. Global Change,IGBP Series, pp. 123–143. Berlin: Springer.
Watson AJ, Upstill-Goddard RC, and Liss PS (1991) Air sea gas exchange inrough and stormy seas measured by a dual tracer technique. Nature349: 145–147.
Watts SF (2000) The mass budgets of carbonyl sulfide, dimethyl sulfide, carbondisulfide and hydrogen sulfide. Atmospheric Environment 34: 761–779.
Webb E, Pearman G, and Leuning R (1980) Correction of the flux measurements fordensity effects due to hear and water vapor transfer. Quarterly Journal of RoyalMeteorological Society 106: 85–100.
Wesely M and Hicks B (1977) Some factors that affect the deposition rates of sulfurdioxide and similar gases on vegetation. APCA 27: 1110–1116.
Whitehead JD, McFiggans GB, Gallagher MW, and Flynn MJ (2009) Directlinkage between tidally driven coastal ozone deposition fluxes, particleemission fluxes, and subsequent CCN formation. Geophysical Research Letters36: L04806.
Whitman WG (1923) The two-film theory of gas absorption. Chemical and MetallurgicalEngineering 29: 146–148.
Williams J, Holzinger R, Gros V, Xu X, Atlas E, and Wallace DWR (2004) Measurementsof organic species in air and seawater from the tropical Atlantic. GeophysicalResearch Letters 31: L23S06.
Wilson ST, Foster RA, Zehr JP, and Karl DM (2010) Hydrogen production byTrichodesmium erythraeum Cyanothece sp. and Crocosphaera watsonii. AquaticMicrobial Ecology 59: 197–206.
Woolf D (1997) Bubbles and their role in air–sea gas exchange. In: Liss PS andDuce RA (eds.) The Sea Surface and Global Change, pp. 173–205. Cambridge:Cambridge University Press.
Woolf D, Leifer I, Nightingale P, et al. (2007) Modelling of bubble-mediated gas transfer;fundamental principles and a laboratory test. Journal of Marine Systems 66: 71–91.
Wurl O, Wurl E, Miller L, Johnson K, and Vagle S (2011) Formation and globaldistribution of sea-surface microlayers. Biogeosciences 8: 121–135.
Xie H, Belanger S, Demers S, Vincent WF, and Papakyriakou TN (2009)Photobiogeochemical cycling of carbon monoxide in the southeastern Beaufort Seain spring and autumn. Limnology and Oceanography 54: 234–249.
Xie HX and Moore RM (1999) Carbon disulfide in the North Atlantic and Pacific Oceans.Journal of Geophysical Research 104: 5393–5402.
Xie HX, Moore RM, and Miller WL (1998) Photochemical production ofcarbon disulphide in seawater. Journal of Geophysical Research 103:5635–5644.
Xie HX, Scarratt MG, and Moore RM (1999) Carbon disulphide production inlaboratory cultures of marine phytoplankton. Atmospheric Environment33: 3445–3453.
Xie H and Zafiriou OC (2009) Evidence for significant photochemical production ofcarbon monoxide by particles in coastal and oligotrophic marine waters.Geophysical Research Letters 36: L23606.
Yang M, Blomquist BW, Fairall CW, et al. (2011a) Air–sea exchange of dimethylsulfidein the Southern Ocean: Measurements from SO GasEx compared to temperate andtropical regions. Journal of Geophysical Research 116: C00F05.
Yang M, Huebert BJ, Blomquist BW, et al. (2011b) Atmospheric sulfur cycling in thesoutheastern Pacific—Longitudinal distribution, vertical profile, and diel variabilityobserved during VOCALS-REx. Atmospheric Chemistry and Physics 11: 5079–5097.
Yang GP, Watanabe S, and Tsunogai S (2001) Distribution and cycling ofdimethylsulfide in surface microlayer and subsurface seawater. Marine Chemistry76: 137–153.
Zafiriou OC (1975) Reaction of methyl halides with seawater and marine aerosols.Journal of Marine Research 33: 75–81.
Zafiriou OC, Andrews SS, and Wang W (2003) Concordant estimates of oceanic carbonmonoxide source and sink processes in the Pacific yield a balanced global“blue-water” CO budget. Global Biogeochemical Cycles 17. http://dx.doi.org/10.1029/2001GB001638.
Zafiriou OC, Xie H, Nelson NB, Najjar RG, and Wang W (2008) Diel carbon monoxidecycling in the upper Sargasso Sea near Bermuda at the onset of spring and inmidsummer. Limnology and Oceanography 53: 835–850.
Zappa CJ, Asher WE, and Jessup AT (2001) Microscale wave breaking and air–watergas transfer. Journal of Geophysical Research 106: 9385–9391.
Zappa CJ, McGillis WR, Raymond PA, et al. (2007) Environmental turbulent mixingcontrols on the air–water gas exchange in marine and aquatic systems. GeophysicalResearch Letters 34: L10601.
Zemmelink HJ, Gieskes WWC, Klaassen W, et al. (2004) Relaxed eddy accumulationmeasurements of the sea-to-air transfer of dimethylsulfide over the northeasternPacific. Journal of Geophysical Research 109: C01025.
Zemmelink HJ, Houghton L, Sievert SM, Frew NM, and Dacey JWH (2005) Gradients indimethylsulfide, dimethylsulfoniopropionate, dimethylsulfoxide, and bacteria nearthe sea surface. Marine Ecology Progress Series 295: 33–42.
Zhang Y, Xie H, Fichot CG, and Chen G (2008) Dark production of carbon monoxide(CO) from dissolved organic matter in the St. Lawrence estuarine system:Implications for the global coastal and blue water CO budgets. Journal ofGeophysical Research 113: C12020.
Zhao FJ, Hawkesford MH, and McGrath SP (1999) Sulphur assimilation and effects onyield and quality of wheat. Journal of Cereal Science 30: 1–17. http://dx.doi.org/10.1006/jcrs.1998.0241.
Zhou XL and Mopper K (1997) Photochemical production of low-molecular-weightcarbonyl compounds in seawater and surface microlayer and their air–seaexchange. Marine Chemistry 56: 201–213.
Zhou H, Yin X, Yang Q, Wang H, Wu Z, and Bao S (2009) Distribution, source andflux of methane in the western Pearl River Estuary and northern South China Sea.Marine Chemistry 117: 21–31.
Zika RG, Gidel LT, and Davis DD (1984) A comparison of photolysis and substitutiondecomposition rates of methyl iodide in the ocean. Geophysical Research Letters11: 353–356.