GLOBAL BIOGEOCHEMICAL CYCLES, VOL. 24, GB4014, doi:10.1029/2009GB003728, 2010
Vertical transport of anthropogenic mercury in the ocean
Sarah Strode,1,2 Lyatt Jaeglé,1 and Steven Emerson3
Received 17 November 2009; revised 28 May 2010; accepted 2 August 2010; published 18 November 2010.
[1] We investigate the vertical transport of mercury (Hg) within the ocean using a simple
box diffusion model to represent vertical water transport coupled with a particulate Hg
flux. The particulate flux assumes that the Hg content of marine particles is proportional
to the Hg concentration of surface waters via a sorption equilibrium constant, Kd. The
model is forced with the observed factor of 3 increase in atmospheric Hg deposition over
the industrial era. The modeled vertical profile of oceanic Hg shows a subsurface
maximum at ∼500 m depth due to remineralization of Hg bound to sinking organic
particles, consistent with observations. Model results indicate that surface (top 100 m)
concentrations of Hg have increased by 150% since preindustrial times. Over the past
150 years, 280 Mmol of anthropogenic Hg have accumulated in the ocean, representing
a 18% increase in the total oceanic Hg content. We find that 36% of the anthropogenic
Hg occurs in the top 400 m and only 7% occurs below 1500 m. Over the industrial
era, we find that 14% of cumulative anthropogenic emissions have accumulated in the
ocean. Our model results show that half of the accumulation of anthropogenic Hg in the
ocean is due to sinking on particulates. A sensitivity analysis indicates that the model
results are most dependent on the value of Kd.
Citation: Strode, S., L. Jaeglé, and S. Emerson (2010), Vertical transport of anthropogenic mercury in the ocean, Global
Biogeochem. Cycles, 24, GB4014, doi:10.1029/2009GB003728.
1. Introduction
[2] The ocean plays a critical role in the biogeochemical
cycling of mercury (Hg). Mason and Sheu [2002] estimate that
the ocean reservoir contains 1440 Mmol of Hg, compared to the
atmospheric reservoir of 25 Mmol, and that ocean emissions
contribute approximately one third of the current Hg source to
the atmosphere. Anthropogenic perturbation to the global Hg
cycle is significant, with historic records in lake sediments
indicating a factor of 3 increase in Hg deposition since preindustrial times [Swain et al., 1992; Lamborg et al., 2002].
[3] Oceanic Hg concentrations affect human health through
bioaccumulation in fish in the form of methylmercury. Despite
its importance to human health, the marine biogeochemistry
of Hg remains poorly understood and few observations exist.
These observations show large variability across regions and
time periods, making anthropogenic impacts difficult to detect
[Fitzgerald et al., 2007]. Instead, estimates of past changes
in oceanic Hg have relied on simple models. Using a mass
balance model separating the ocean into 2 vertical boxes,
Mason and Sheu [2002] estimated that anthropogenic activi1
Department of Atmospheric Sciences, University of Washington,
Seattle, Washington, USA.
2
Now at SAIC, NASA Goddard Space Flight Center, Greenbelt,
Maryland, USA.
3
School of Oceanography, University of Washington, Seattle,
Washington, USA.
Copyright 2010 by the American Geophysical Union.
0886‐6236/10/2009GB003728
ties resulted in a 9% increase in oceanic Hg, with most of
this increase occurring in the deep ocean, below 500 m. In
a separate study, Lamborg et al., [2002] calculated a 90%
increase in the ocean mixed layer (top 100 m), a 20% increase
in the thermocline (100–1000 m), and assumed no increase
below 1000 m. Their nonsteady state model separated the
ocean in two hemispheres and in two vertical boxes. More
recently, Sunderland and Mason [2007] examined the evolution of oceanic Hg concentrations for the surface, intermediate, and deep waters in a 14‐box model, finding a 25%
increase in the surface (0–1500 m) and an 11% increase in
the deep ocean (>1500 m). In our previous work, we have
coupled a global 3‐D atmospheric chemical transport model
of mercury (GEOS‐Chem) to a slab model of the ocean
mixed layer [Strode et al., 2007]. By conducting preindustrial and present‐day Hg simulations with GEOS‐Chem, Selin
et al. [2008] found a 180% increase in ocean mixed layer
concentrations.
[4] These studies present a somewhat conflicting picture
of how human activities have affected oceanic Hg concentrations. Estimates for the increase in mercury mass in the upper
ocean range from < 2% to 200%, and increases in the deep
ocean range from 0 to 13%. One issue is that all these studies
divide the ocean in 2–4 boxes in the vertical, each assuming
different depths.
[5] There is a large body of research on the ocean penetration of other atmospheric gases (CO2, 14C, chlorofluorocarbons, tritium) that can serve as analogs to Hg. In
particular, the ocean uptake of CO2 has been studied extensively in order to reproduce trends in atmospheric CO2
GB4014
1 of 10
GB4014
STRODE ET AL.: OCEAN MERCURY TRANSPORT
concentrations [e.g., Prentice et al., 2001, and references
therein]. Inorganic carbon measurements show that ocean
invasion of anthropogenic CO2 has been limited to the top
1500 m of the ocean [Gruber, 1998; Sabine et al., 2004].
Some of the first models used to successfully capture the
observed oceanic penetration of 14C and anthropogenic CO2
were box diffusion models [Oeschger et al., 1975]. These
models represent the ocean as a mixed layer box in contact
with the atmosphere and an eddy diffusive deep ocean with
∼10–40 vertically stacked boxes. In contrast, models with
only 2 boxes in the vertical have difficulty reproducing the
uptake of anthropogenic CO2 because the deep ocean cannot
be accurately represented as a well‐mixed box on a 100 year
timescale [Siegenthaler and Oeschger, 1978]. Presently, the
simple box diffusion models have been replaced by state‐of‐
the‐art 3‐D Ocean General Circulation Models (OGCM)
[e.g., Orr et al., 2001]. Joos et al. [1996] compared the
performance of a box diffusion model to an OGCM in
representing the penetration of a pulse of anthropogenic
CO2. After normalizing by mixed layer depth, they found
that the response of the box diffusion model was very
close to that of the OGCM. Box diffusion models thus offer
a realistic framework to examine global oceanic uptake of
anthropogenic CO2 [Joos et al., 1997].
[6] In this work, we use a box diffusion model to examine
the anthropogenic perturbation to oceanic Hg over the past
150 years. The model is forced with the observed factor of
3 increase in Hg deposition over the course of the industrial
era. Compared to most of the published ocean Hg studies,
which have only 2–4 vertical levels, our model provides
greater vertical resolution and is thus more adapted to examine
vertical transport of Hg into the ocean. We describe the
model in section 2. Results are presented in section 3, where
we discuss the factors controlling the vertical distribution of
Hg in the ocean and examine the penetration of anthropogenic Hg in the ocean. We also explore the sensitivity of
our model results to a wide range of parameters to see what
values lead to reasonable simulations of the modern surface
ocean concentration. Conclusions are presented in section 4.
2. Model Description
[7] We modify the Oeschger et al. [1975] CO2 box diffusion model to represent Hg in the ocean. The Hg model
includes a single box representing the surface atmosphere
coupled to a box diffusion model of the ocean consisting
of 50 vertical layers, each 50 m deep. Vertical transport of
dissolved material in this model is parameterized entirely by
the eddy diffusion coefficient, Kz. The value of this parameter
is set by matching data from transient anthropogenic tracers
that are presently penetrating the thermocline. Oeschger
et al. [1975] demonstrated that a value of Kz = 1.3 × 10−4 m2
s−1 was able to capture the main features of the distributions
of both natural and bomb‐produced 14C, and we have used
this value here. This Kz is similar to vertical eddy diffusion
coefficients required in analytical one‐dimensional models
of the temperature and salinity distribution of the main
thermocline of the ocean [Munk, 1966] and is also within the
range of depth‐independent Kz values (1.3–2.5 × 10−4 m2
s−1) used by Joos et al. [1997] in their box diffusion model.
GB4014
These values are about ten times larger than vertical eddy
diffusion coefficients predicted from tracer release experiments and temperature microstructure distributions in the
thermocline [Ledwell et al., 1993] illustrating that the eddy
diffusion coefficients used in the one dimensional models
parameterize many different mechanisms that transport water
vertically over periods of decades. The model advances forward in time with a 6 month time step using the Crank‐
Nicholson numerical scheme for diffusion [Press et al., 1986].
[8] Given the timescales of global ocean transport, we
assume little anthropogenic change has occurred over the
past 150 years at depths below 2000 m. We thus fix the Hg
concentrations at the model’s lower boundary (2500 m) at
1.3 pM, based on deep ocean observations in the Pacific
[Sunderland and Mason, 2007]. To reduce boundary condition influence, we will only report results for the top 2000 m
of the model.
[9] Mercury exists in several forms within the ocean: dissolved gaseous mercury (Hg0aq), aqueous divalent mercury
(HgIIaq), several methylated forms (such as CH3Hg and
(CH3)2Hg), and particulate mercury [Fitzgerald et al., 2007].
Wet deposition of atmospheric divalent mercury provides a
major source of HgIIaq to the ocean. Within the ocean, HgIIaq
can become methylated, form organic or inorganic complexes,
or undergo reduction to dissolved Hg0, commonly called dissolved gaseous mercury (DGM). The aqueous reduction of
HgIIaq to Hg0aq, combined with the low solubility of Hg0,
drives an evasion flux of Hg0 from the ocean to the atmosphere. In our previous global Hg slab ocean model [Strode
et al., 2007], we found that atmospheric deposition provides
a 22.8 Mmol yr−1 source to the mixed layer. This source is
balanced by net evasion to the atmosphere (14.1 Mmol yr−1)
combined with loss to the deep ocean (8.7 Mmol yr−1) from
diffusion and particulate sinking.
[10] Our box diffusion model calculates the vertical distribution of dissolved (Hg0aq+ HgIIaq) and particulate Hg
(HgP). Mercury enters the ocean as dissolved Hg through
wet and dry deposition, and gas exchange provides a loss of
Hg0aq from the ocean to the atmosphere. The measured
percent of oceanic mercury as DGM shows large variability.
Sunderland and Mason [2007] estimate the fraction of total
Hg as Hg0 for each ocean basin, with values in the surface
ocean ranging from 6% in the Pacific to 13% in the Atlantic.
In this study, we average these two values and assume that
9.5% of the dissolved Hg in the surface ocean layer is DGM
in order to calculate gas exchange with the atmosphere. The
ocean box diffusion model uses the gas exchange formulation that we have previously described [Strode et al., 2007],
with the Henry’s Law constant from Andersson et al.
[2008]. We also use the global mean gas exchange velocity (kw = 10.7 cm h−1) determined by Strode et al. [2007].
[11] We include vertical transport of Hg via sinking of
particulate organic matter composed of dead organisms and
detritus, commonly referred to as the biological pump in the
context of the carbon cycle. Scavenging of Hg by these
organic‐rich particles can be a major sink for Hg in the open
ocean [Lamborg et al., 2002], but is poorly quantified. We
assume that dissolved and particulate mercury are in sorption
equilibrium, described with the equilibrium constant Kd =
sorbed Hg per sorbent mass/dissolved Hg per water volume
2 of 10
GB4014
STRODE ET AL.: OCEAN MERCURY TRANSPORT
[Meili, 1997; Morel et al., 1998]. In our study, we chose Kd =
1.2 × 106 l (kg particulate organic carbon)−1 in order to obtain
concentrations of dissolved Hg in the modern ocean of
0.87 pM, matching the globally averaged mixed layer concentrations we derived from Strode et al. [2007].
[12] Estimates of Kd based on open ocean measurements
range from 1.1 × 105 l (kg total suspended particles)−1 in the
North Pacific [Mason et al., 1998] to 106 l (kg total suspended particles)−1 in the Equatorial Pacific [Mason and
Fitzgerald, 1993]. If we assume that 52% of organic matter is carbon [Emerson and Hedges, 2008] and that organic
matter accounts for ∼20% of the suspended material mass
in surface ocean waters (top few hundred meters) [Brewer
et al., 1980], these observed values lead to a range of 1.1 × 106
to 107 l (kg particulate organic carbon)−1. Meili [1997] and
Coquery et al. [1997] report that in lakes, rivers, and estuaries Kd values span an order of magnitude: 105–106 l (kg
particulate organic matter)−1. The lower values correspond
to environments dominated by biotic particles, while the
higher values are for detrital organic matter. Converting our
assumed Kd to the same units, yields Kd = 6 × 105 l (kg
particulate organic matter)−1. Thus our assumed Kd is consistent with the range of observed values. In section 3.3, we
will examine the sensitivity of our results to different choices
of Kd.
[13] We determine the particulate Hg flux by multiplying
Kd by the concentration of dissolved Hg and by the organic
carbon particulate flux. We use the organic carbon flux
parameterization of Antia et al. [2001]: ForgC = 0.1 NPP1.77 zn,
where NPP is net primary productivity (in gC m−2 yr−1), z is
depth (in m), and n is an empirical fit to data. We assume a
global NPP value of 148 gC m−2 yr−1 (53.4 GtC yr−1) and
set n to −0.7, in between the Atlantic estimate of −0.68
[Antia et al., 2001] and the Pacific estimate of −0.734 [Pace
et al., 1987]. This yields ForgC = 28 gC−2 yr−1 (or 10 GtC
yr−1) at 100 m, consistent with observations of carbon
export [Emerson and Hedges, 2008]. The net particulate
flux of Hg into a given model layer is the difference between
the flux at the top and bottom of the layer. We assume this net
flux is remineralized within the layer and adds to the total
dissolved Hg concentration of the layer. Field observations
and laboratory experiments show that the settling velocities of
sinking particles are highly variable: 20–200 m d−1 [Alldredge
and Gotschalk, 1988; Honjo, 1996]. We assume a mean settling velocity of 100 m d−1 and calculate the particulate Hg
burden in each layer by dividing the particulate flux of Hg out
of the layer by this velocity.
[14] Both the mean surface atmospheric concentrations of
Hg0 and deposition of Hg to the ocean (in the form of dry
and wet deposition of divalent mercury and particulate mercury) are specified using results from the global GEOS‐Chem
ocean‐land‐atmosphere mercury simulation [Selin et al., 2007;
Strode et al., 2007]. Based on the simulation of Selin et al.
[2008], we assume a preindustrial deposition to the ocean
of 7.6 Mmol yr−1 and 0.5 ng m−3 for the atmospheric concentration of Hg0. For the modern‐day simulation, these
values increase by a factor of 3 to 22.8 Mmol yr−1 for
deposition and 1.5 ng m−3 for Hg0 [Selin et al., 2007]. For
simplicity, we assume that both deposition and atmospheric
GB4014
concentrations increase linearly over the 150 years between
preindustrial times and present.
[15] The box diffusion model is first run with preindustrial values for Hg deposition and atmospheric Hg0 concentration until it reaches preindustrial steady state. Deposition
and atmospheric concentration are then linearly increased
to modern levels over 150 years to give the modern, nonsteady state marine concentration profile. The Hg content
of particulate matter is proportional to the surface ocean Hg
concentration and therefore increases over the industrial era,
reflecting the greater quantity of Hg available to bind to
particles. As we will see, this biological pump represents
an effective way for transporting anthropogenic Hg to the
thermocline.
[16] We note that our approach of representing the world’s
oceans using a 1‐D box diffusion model has a number of
inherent limitations. In particular, we neglect horizontal
variations in a number of factors: atmospheric deposition of
Hg to the oceans, temperature (which affects the solubility
of Hg0), biological productivity, eddy diffusion, and the
shape of the remineralization curve. Furthermore, we do
not take into account lateral transport, which can play an
important role in controlling the vertical distribution of Hg
in the oceans [Sunderland and Mason, 2007; Sunderland
et al., 2009]. Despite these limitations, we will show that
our idealized modeling approach is a useful framework for
examining the roles of vertical diffusion and the biological
pump on the penetration of anthropogenic Hg in the ocean.
3. Results and Discussion
3.1. Evolution of the Vertical Distribution of Mercury
in the Ocean
[17] Figure 1a shows the increase in dissolved Hg concentrations for the model simulation over the past 150 years
at three levels: in the mixed layer, at 250 m, and at 1500 m
depth. The modern‐day mixed layer concentration of dissolved Hg is 0.87 pM. This value falls within the bounds of
observed mixed layer concentrations and matches by design
the globally averaged mixed layer concentrations of Strode
et al. [2007]. Model results in Figure 1a show that concentrations in the mixed layer increase by 0.55 pM (factor of
2.7 increase) between the preindustrial era and the present
as the surface ocean layer equilibrates rapidly with the
changing atmospheric input. At 250 m depth, the absolute
increase in concentrations is larger (0.75 pM) as a result of
transport of Hg on particulates. In contrast, the concentration at 1500 m has increased by only 0.2 pM, because of the
slow transport of anthropogenic Hg down to that depth.
[18] Figure 1b illustrates the evolution of the Hg profile
from preindustrial to present in 25 year increments. Our
initial profile of dissolved Hg, obtained in our steady state
preindustrial simulation (blue line on Figure 1b), displays a
minimum at the surface because of Hg loss at the surface
from particle scavenging and Hg0aq evasion to the atmosphere. Organic particles, which are generated at the surface,
provide a sink for dissolved Hg. As these particles sink they
remineralize at depth providing a net source of dissolved
Hg, and thus increasing concentrations with depth. Overall,
3 of 10
GB4014
STRODE ET AL.: OCEAN MERCURY TRANSPORT
GB4014
this downward particulate flux is balanced by upward diffusion of dissolved Hg.
[19] As atmospheric deposition increase over the industrial era, oceanic Hg concentrations increase. This in turn leads
to an increase in particulate‐bound Hg flux. The modern‐
day dissolved Hg profile reaches a maximum of 1.6 pM at
500 m below the surface. This subsurface maximum is due
to the vertical penetration of anthropogenic Hg via diffusion
and increased particulate transport of Hg.
[20] Several studies have reported a subsurface maximum
in total Hg concentration in the thermocline region, which
they attributed to Hg release from particulates and/or lateral
transport of enriched intermediate waters [Cossa et al., 1992,
2004; Mason et al., 1998, 2001; Mason and Sullivan, 1999;
Laurier et al., 2004; Sunderland et al., 2009]. For example,
recent profiles of total Hg in the North Pacific Ocean reported
by Sunderland et al. [2009] display surface concentrations
of 1 pM and a maximum in intermediate waters of ∼1.6 pM
at 500–600 m depth. Figure 2 compares the present‐day modeled depth profile to selected observations from the Atlantic
[Dalziel, 1992] and Pacific [Mason and Fitzgerald, 1993;
Sunderland et al., 2009]. The model produces a subsurface
maximum similar to observations. The observed Atlantic concentrations are generally higher than the Pacific concentrations, with the modeled concentrations lying between the
two basins.
3.2. Penetration of Anthropogenic Mercury
into the Ocean
[21] Figure 1c displays the depth profile of anthropogenic
Hg concentrations, which we calculate by difference between
the time‐dependent simulation at 150 years and the preindus-
Figure 1. Evolution of Hg concentrations from preindustrial times to present in the standard box diffusion model
simulation. (a) Increase in concentration with time in the
top 50 m (mixed layer) and at 250 and 1500 m depths from
preindustrial steady state to the present. (b) Modeled concentration of total dissolved Hg with depth. Colors represent
the evolution from preindustrial steady state (initial, blue) to
present (150 years, black) in 25 year increments. (c) Present‐
day concentrations of anthropogenic dissolved Hg obtained
by difference between the 150 year profile and preindustrial
profile in Figure 1b.
Figure 2. Present‐day modeled concentration of dissolved
Hg (black line) compared with observed profiles of reactive Hg from the Atlantic [Dalziel, 1992] (blue) and Pacific
[Mason and Fitzgerald, 1993] (red). The symbols represent
the mean value at each depth, and the error bars represent the
minimum and maximum values. Also shown are mean total
Hg profiles over the Pacific from Sunderland et al. [2009]
(green).
4 of 10
GB4014
STRODE ET AL.: OCEAN MERCURY TRANSPORT
Table 1. Oceanic Mercury Mass for the Preindustrial and Modern
Time Periods
Oceanic Mercury Massa
Ocean Depth
Preindustrial
(Mmol)
Modern
(Mmol)
Anthropogenic
Increaseb
0–100 m
100–400 m
400–1500 m
Below 1500c m
Entire oceanc
14
78
432
1066
1590
35
157
590
1088
1870
150%
100%
35%
2%
18%
a
Total mercury mass (dissolved + particulate) in our standard simulation.
Anthropogenic increase defined as the increase between preindustrial
and modern burdens.
c
We estimate the burden below 2000 m by assuming a constant concentration of 1.3 pM and an average ocean depth of 3790 m (yielding a total
of 840 Mmol of mercury below 2000 m).
b
trial simulation. This vertical distribution displays a subsurface
maximum due to the downward transport of anthropogenic
mercury on particles.
[22] We calculate the mass of preindustrial and present‐
day Hg in the ocean as a function of depth by assuming a
surface area for the ocean of 360 million km2. Results for
different ocean depths are listed in Table 1. As our model
only extends to 2000 m, we estimate the burden below that
depth by assuming a constant concentration of 1.3 pM (yielding a total of 840 Mmol of mercury below 2000 m). When
we integrate over the entire ocean, we find that the oceanic
Hg burden increases by 280 Mmol between preindustrial
times and present, representing an 18% increase. Over that
time period in our simulation, anthropogenic emissions have
resulted in a cumulative anthropogenic Hg deposition to the
ocean of 1140 Mmol. Thus only 25% of the cumulative
anthropogenic mercury deposition to the ocean remains in
the ocean, while the rest is released back to the atmosphere
by air‐sea exchange. Assuming that 55% of deposition occurs
over the ocean, we estimate that 14% of total anthropogenic
deposition (to land and ocean), and hence 14% anthropogenic
emissions, have accumulated in the ocean. We can contrast
this estimate to the budget of CO2, for which ∼50% of the
total fossil fuel and cement emissions have accumulated in
the ocean [Prentice et al., 2001; Sabine et al., 2004].
[23] Our model results in Table 1 show that the mass of Hg in
the ocean has increased by 150% in the top 100 m, 100%
between 100 and 400 m, 35% between 400 and 1500 m, and
by less than 2% below 1500 m. Examining the vertical distribution of anthropogenic Hg in the ocean, we find that 17%
of the anthropogenic Hg occurs above 200 m, 36% above
400 m, and 7% below 1500 m.
3.3. Sensitivity Studies
[24] The computational efficiency of the diffusion model
makes it ideal for exploring parameter space. Below we
examine the sensitivity of our results to assumptions about:
particulate sinking, modern and cumulative deposition, the
gas exchange coefficient, the particulate‐dissolved Hg partitioning coefficient, and the eddy diffusion coefficient.
3.3.1. Particulate Sinking
[25] The standard simulation (described in section 2)
includes the effects of both vertical diffusion and particulate
GB4014
remineralization. In order to separate the roles of these
two transport modes, we conduct a sensitivity simulation
in which the particulate concentration of Hg (and thus the
amount of particulate Hg remineralizing at depth) is held
fixed at its preindustrial levels. Thus, in this sensitivity simulation vertical transport of anthropogenic Hg is only controlled by diffusion.
[26] Figure 3a shows the resulting vertical profiles for this
fixed particulate Hg concentration simulation. The modern‐
day profile (150 years) shows a gradual increase in concentration with depth, contrary to the pronounced subsurface
maximum in the standard simulation (Figure 1b). Surface
concentrations are somewhat higher (0.95 pM) compared
to the standard simulation (0.87 pM) because less of the
anthropogenic Hg deposited from the atmosphere is removed
by particle sinking. At depth, Hg concentrations do not increase
as much, reaching 1.2 pM at 500 m compared to 1.6 pM in
the standard simulation.
[27] This simulation shows the concentration of anthropogenic Hg decreasing with depth (Figure 3b), as transport
of anthropogenic Hg is solely controlled by diffusion. In
Figure 3. (a) Concentration of dissolved Hg with depth for
the box diffusion model run with Hg particulate concentration
fixed to preindustrial levels. Colors represent the evolution
from preindustrial steady state (blue) to present (black) in
25 year increments. (b) Concentration of dissolved anthropogenic Hg after 150 years.
5 of 10
GB4014
STRODE ET AL.: OCEAN MERCURY TRANSPORT
GB4014
Figure 4a. Sensitivity of modern surface concentrations of Hg to modern and cumulative deposition.
Line colors represent simulations with modern deposition in Mmol yr−1 given in the legend. For each
value of modern deposition, surface concentrations are calculated for four different cumulative depositions
(using functions described in section 3.3.2) over the last 150 years. The black solid circle indicates the
parameters used in the standard simulation.
contrast, we find a subsurface maximum in anthropogenic
Hg for our standard simulation (Figure 1c). For the fixed
particulate Hg concentration simulation, we find a 135 Mmol
increase in the ocean Hg burden, a factor of 2 smaller than
in the standard simulation. Thus the increasing uptake of Hg
in the biological pump over the industrial era, as in our standard simulation, accounts for half of the modeled anthropogenic increase, and represents an effective way of increasing
transport of anthropogenic Hg in the ocean interior where it
will not be lost by gas exchange.
3.3.2. Modern and Cumulative Deposition
[28] The standard model run assumes that deposition to the
ocean increased linearly by a factor of 3 over the last 150 years
from 7.6 to 22.8 Mmol yr−1. We conduct sensitivity simulations with a range of modern Hg depositions (15–30 Mmol
yr−1) and a range in the time course of change since the preindustrial period. In an idealized set of simulations, we are
changing the shape of the time‐dependent deposition function to be linear, quadratic, cubic, or square root. The square
root assumption results in larger cumulative deposition since
deposition increases earlier in the industrial era, while the
cubic function increases later, resulting in lower cumulative
deposition. For a modern deposition value of 22.8 Mmol
yr−1, the square root, linear, quadratic and cubic curves lead
to cumulative deposition values of 2660, 2280, 1900, and
1710 Mmol, respectively. This corresponds to cumulative
anthropogenic deposition values of 1520, 1140, 760, and
570 Mmol.
[29] Figure 4a shows the concentration of Hg in the surface
ocean plotted against cumulative anthropogenic deposition
for different modern deposition rates. Figure 4a illustrates
that surface ocean concentrations are highly sensitive to the
rate of modern deposition (concentrations increase by ∼60%
for a doubling of modern deposition), but have only a small
sensitivity to the cumulative deposition. For a modern deposition value of 22.8 Mmol yr−1, surface concentration increases
by only 5% for a 55% increase in cumulative deposition
(cyan line). This is consistent with the fast response time of
dissolved Hg in the surface ocean, which has a residence
time of 7 months [Strode et al., 2007].
[30] Figure 4b shows the sensitivity of ocean Hg concentrations at 500 m depth to cumulative and modern deposition, generated from the same runs as Figure 4a. The lines
for different modern deposition values are overlapping in
Figure 4b. This indicates that concentrations at 500 m depth
are mostly sensitive to the time course of deposition and
less sensitive to the modern rate of deposition, because of
the time required for Hg to reach this depth. For a doubling in cumulative deposition, Hg concentrations at 500 m
depth increase by almost 50%. For a doubling in modern
6 of 10
GB4014
STRODE ET AL.: OCEAN MERCURY TRANSPORT
GB4014
Figure 4b. Mercury concentration at 500 m depth as a function of time‐integrated deposition for different
values of modern deposition. The model runs are the same as in Figure 4a.
deposition with the same cumulative deposition, concentrations increase by less than 5%, as shown by the nearly overlapping lines.
3.3.3. Gas Exchange Velocity and Modern Deposition
[31] Since both the gas exchange velocity, kw, and modern
atmospheric deposition to the ocean are uncertain, we use the
box diffusion model to explore whether other combinations
of kw and deposition give realistic surface Hg concentrations. Figure 5 (left) shows the surface concentration of
dissolved Hg as a function of the modern deposition rate and
a scaling factor applied to kw (a scaling factor of 1 corresponds to the standard simulation where kw = 10.7 cm h−1).
Figure 5. Sensitivity of ocean surface dissolved Hg concentrations (in pM) to (left) deposition and kw,
the gas exchange velocity and (right) to Kd, the dissolved particulate Hg partitioning coefficient and Kz,
the eddy diffusion coefficient. The kw scale factor represents the scaling applied to the gas exchange
velocity used in the standard run. Colors and contours represent total dissolved Hg concentration
in pM. The black solid circle marks the parameters used in the standard simulation.
7 of 10
GB4014
STRODE ET AL.: OCEAN MERCURY TRANSPORT
The scaling of kw can represent either an adjustment of the
gas exchange velocity or an altered fraction of total Hg as
dissolved gaseous Hg, since the model cannot distinguish
these two possibilities.
[32] Figure 5 (left) demonstrates that the surface concentration is sensitive to both deposition and kw. It is possible to
maintain a surface concentration of 0.87 pM dissolved Hg
with an increase or decrease in kw provided deposition is
increased or decreased concurrently. A smaller value of kw
would provide slower evasion of Hg from the ocean to the
atmosphere, but this smaller loss could be balanced by a
smaller source from deposition while maintaining realistic
surface concentrations.
3.3.4. Eddy Diffusion Coefficient and Kd
[33] Figure 5 (right) shows the sensitivity of the surface
ocean Hg concentration to the assumed eddy diffusion coefficient Kz and the dissolved particulate partitioning coefficient Kd. Increasing Kd leads to faster removal of Hg via
particulate sinking from the surface layer, reducing the surface concentration of dissolved Hg. Because of the vertical
gradient induced by particulates, diffusion transports dissolved Hg upward from the subsurface maximum to the surface. Thus increasing Kz leads to higher surface concentrations.
Increasing Kz also smoothes out the vertical concentration
gradient, leading to a lower sensitivity to Kd when Kz is
large.
[34] Higher surface concentrations result in a larger loss
of Hg from the ocean via gas exchange with the atmosphere. Consequently, the dependence of the ocean burden
of anthropogenic Hg on Kd and Kz is opposite that of the
surface concentration. Increasing Kd increases the oceanic
burden of anthropogenic Hg because Hg is transported more
quickly into the ocean interior where it will not be lost by
gas exchange. A doubling of Kd leads to a 40% increase
in the anthropogenic Hg burden in the ocean (from 280 to
400 Mmol), while a factor of 2 decrease in Kd results in a
20% decrease in the anthropogenic Hg burden.
3.4. Comparison to Previous Studies
[35] Estimates from previous modeling studies for the total
Hg oceanic burden increase range from 120 to 270 Mmol
[Mason and Sheu, 2002; Lamborg et al., 2002; Sunderland
and Mason, 2007; Selin et al., 2008]. Our result (280 Mmol)
is thus at the upper end of these studies. Expressed in terms
of percentage increase in the oceanic Hg burden our 18%
increase is also at the upper end of previous estimates (9–
18%).
[36] In the top 100 m of the ocean, we find a 150%
increase in the burden of Hg. This is within the range given
by previous studies (90–200%). This factor of 2–3 increase
is consistent with the limited historical archives of Hg in
museum seabird feathers [Monteiro and Furness, 1997].
[37] Our mixed layer burden of total Hg (36 Mmol in the
top 100 m) is lower than previous studies (54–64 Mmol).
This is likely due to a smaller particulate Hg concentration in our simulation. Our particulate Hg concentration is
sensitive to the choice of settling velocity, which varies
depending on the size of the particles, so an overestimate in
settling velocity could lead to an underestimate in particulate
Hg mass in the surface ocean.
GB4014
[38] There are substantial differences between models
in estimates of the distribution of anthropogenic Hg below
the ocean mixed layer. The Mason and Sheu [2002] model
assumes that less than 5% of the anthropogenic Hg is
present at depths shallower than 500 m, with the remaining
95% occurring below 500 m (we find values of 45% and
55%, respectively). Sunderland and Mason [2007] find that
50% of the anthropogenic Hg is above 1500 m and 50%
below that depth (we find values of 92% and 8%, respectively). Results from our simulation indicate a much slower
penetration of anthropogenic Hg compared to these studies
(see Figure 1c and Table 1). However, our results are consistent with the Hg model of Lamborg et al. [2002], who
assume that 100% of the anthropogenic Hg is confined to
the top 1000 m of the ocean (we find that 76% of anthropogenic Hg is in the top 1000 m).
4. Conclusions
[39] In this paper, we have described the development of a
box diffusion model for ocean Hg. The model includes air‐
sea exchange, atmospheric deposition, and vertical transport
of Hg within the ocean due to diffusion and particulate
sinking. This simple model framework allows us to investigate the role of vertical transport processes on oceanic Hg
distribution in greater detail than previous models. We use
this model to explore how the Hg content of the ocean has
changed over the past 150 years of the industrial era.
[40] Vertical transport of Hg via sinking particulate organic
matter is critical for determining the shape of the Hg profile
and the penetration of anthropogenic Hg in the box diffusion
model. In our standard simulation, we assume that dissolved
and particulate Hg are in sorption equilibrium described
with a partitioning constant Kd. The amount of Hg bound to
particles thus increases over the industrial era, following the
increase in dissolved Hg concentration in the mixed layer.
Particulate Hg remineralizes at depth, leading to a subsurface concentration maximum at 500 m in the modern ocean.
[41] Overall, we find an oceanic uptake of anthropogenic
Hg emissions of 280 Mmol for the past 150 years in our
standard simulation. By conducting a sensitivity simulation
where the flux of particulate‐bound Hg is kept fixed at
preindustrial levels, we find half of the increase in Hg
burden over the industrial era is associated with particulate
sinking while the other half is due to diffusion. Particulate
sinking enhances the ocean uptake of anthropogenic Hg by
transporting Hg away from the surface, thus reducing loss to
the atmosphere. Our standard model implies that 14% of the
anthropogenic Hg emissions have accumulated in the ocean
over the industrial era, while the remaining anthropogenic
deposition is recycled back to the atmosphere via air‐sea
exchange.
[42] The box diffusion model predicts that 36% of the
anthropogenic Hg occurs in the top 400 m and 93% in the
top 1500 m. Our results indicate that mixed layer concentrations of Hg (top 100 m) have increased by 150% since preindustrial times, in the middle of previous estimates (90–200%).
[43] Surface Hg concentrations are sensitive to assumptions about the rate of modern deposition, the gas exchange
velocity, the Kd partitioning constant, and the value of the
8 of 10
GB4014
STRODE ET AL.: OCEAN MERCURY TRANSPORT
eddy diffusion coefficient used in the model. Since the surface ocean layer equilibrates rapidly with the atmosphere,
it is relatively insensitive to historic deposition rates. In
contrast, the thermocline region responds more slowly to
changing atmospheric input, and is more sensitive to cumulative deposition over the industrial era.
[44] At present the distribution of Hg in the ocean is
poorly constrained by a limited number of observations,
which show large variability in time and space [e.g., Laurier
et al., 2004; Fitzgerald et al., 2007]. There is a clear need
to obtain more systematic observations of speciated vertical
Hg profiles for different regions of the ocean. Observations
to constrain the partitioning of Hg between dissolved and
particulate phases in open ocean environments are particularly important, given the high sensitivity of penetration of
anthropogenic Hg to particulate sinking.
[45] Acknowledgment. This work was supported by funding from
the National Science Foundation under grant ATM 0238530.
References
Alldredge, A. L., and C. Gotschalk (1988), In situ settling of marine snow,
Limnol. Oceanogr., 33, 339–351, doi:10.4319/lo.1988.33.3.0339.
Andersson, M. E., K. Gårdfeldt, I. Wängberg, and D. Strömberg (2008),
Determination of Henry’s law constant for elemental mercury, Chemosphere, 73(4), 587–592, doi:10.1016/j.chemosphere.2008.05.067.
Antia, A. N., et al. (2001), Basin‐wide particulate carbon flux in the
Atlantic Ocean: Regional export patterns and potential for atmospheric
CO 2 sequestration, Global Biogeochem. Cycles, 15(4), 845–862,
doi:10.1029/2000GB001376.
Brewer, P. G., Y. Nozaki, D. W. Spencer, and A. P. Fleer (1980), Sediment
trap experiments in the deep North Atlantic: Isotopic and elemental
fluxes, J. Mar. Res., 38, 703–728.
Coquery, M., D. Cossa, and J. Sanjuan (1997), Speciation and sorption of
mercury in two macro‐tidal estuaries, Mar. Chem., 58, 213–227,
doi:10.1016/S0304-4203(97)00036-4.
Cossa, D., P. Michel, J. Noel, and D. Auger (1992), Vertical mercury profile in relation to arsenic, cadmium and copper at the eastern North Atlantic ICES reference station, Oceanol. Acta, 15(6), 603–608.
Cossa, D., M.‐H. Cotté‐Krief, R. P. Mason, and J. Bretaudeau‐Sanjuan
(2004), Total mercury in the water column near the shelf edge of the
European continental margin, Mar. Chem., 90, 21–29, doi:10.1016/j.
marchem.2004.02.019.
Dalziel, J. A. (1992), Reactive mercury on the Scotian Shelf and in the
adjacent northwest Atlantic Ocean, Mar. Chem., 37, 171–178,
doi:10.1016/0304-4203(92)90076-M.
Emerson, S. R., and J. I. Hedges (2008), Chemical Oceanography and the
Marine Carbon Cycle, 468 pp., Cambridge Univ. Press, Cambridge,
U. K, doi:10.1017/CBO9780511793202.
Fitzgerald, W. F., C. H. Lamborg, and C. R. Hammerschmidt (2007),
Marine biogeochemical cycling of mercury, Chem. Rev., 107, 641–662,
doi:10.1021/cr050353m.
Gruber, N. (1998), Anthropogenic CO2 in the Atlantic Ocean, Global Biogeochem. Cycles, 12(1), 165–191, doi:10.1029/97GB03658.
Honjo, S. (1996), Fluxes of particles to the interior of the open oceans, in
Particle Flux in the Ocean, SCOPE Report, vol. 57, edited by V. Ittekkot
et al., pp. 91–154, John Wiley, New York.
Joos, F., M. Bruno, R. Fink, U. Siegenthaler, T. F. Stocker, C. Le Quéré,
and J. Sarmiento (1996), An efficient and accurate representation of
complex oceanic and biospheric models of anthropogenic carbon
intake, Tellus, Ser. B, 48(3), 397–417, doi:10.1034/j.1600-0889.1996.
t01-2-00006.x.
Joos, F., J. C. Orr, and U. Siegenthaler (1997), Ocean carbon transport in
a box diffusion versus a general circulation model, J. Geophys. Res.,
102(C6), 12,367–12,388, doi:10.1029/97JC00470.
Lamborg, C. H., W. F. Fitzgerald, J. O’Donnell, and T. Torgersen (2002),
A non‐steady state compartmental model of global‐scale mercury biogeochemistry with interhemispheric atmospheric gradients, Geochim.
Cosmochim. Acta, 66, 1105–1118, doi:10.1016/S0016-7037(01)00841-9.
GB4014
Laurier, F. J. G., R. P. Mason, G. A. Gill, and L. Whalin (2004), Mercury
distributions in the North Pacific Ocean—20 years of observations, Mar.
Chem., 90, 3–19, doi:10.1016/j.marchem.2004.02.025.
Ledwell, J. R., A. J. Watson, and C. S. Law (1993), Evidence for slow mixing across the pycnocline from an open‐ocean tracer‐release experiment,
Nature, 364(6439), 701–703, doi:10.1038/364701a0.
Mason, R. P., and W. F. Fitzgerald (1993), The distribution and biogeochemical cycling of mercury in the equatorial Pacific Ocean, Deep Sea
Res., Part I, 40, 1897–1924, doi:10.1016/0967-0637(93)90037-4.
Mason, R. P., and G.‐R. Sheu (2002), Role of the ocean in the global mercury cycle, Global Biogeochem. Cycles, 16(4), 1093, doi:10.1029/
2001GB001440.
Mason, R. P., and K. A. Sullivan (1999), The distribution and speciation of
mercury in the South and equatorial Atlantic, Deep Sea Res., Part II, 46,
937–956, doi:10.1016/S0967-0645(99)00010-7.
Mason, R. P., K. R. Rolfhus, and W. F. Fitzgerald (1998), Mercury in the
North Atlantic, Mar. Chem., 61, 37–53, doi:10.1016/S0304-4203(98)
00006-1.
Mason, R. P., N. M. Lawson, and G.‐R. Sheu (2001), Mercury in the
Atlantic Ocean: Factors controlling air‐sea exchange of mercury and
its distribution in the upper waters, Deep Sea Res., Part II, 48(13),
2829–2853, doi:10.1016/S09670645(01)00020-0.
Meili, M. (1997), Mercury in lakes and rivers, Met. Ions Biol. Syst., 34,
21–51.
Monteiro, L. R., and R. W. Furness (1997), Accelerated increase in mercury contamination in North Atlantic mesopelagic food chains as indicated by time series of seabird feathers, Environ. Toxicol. Chem., 16,
2489–2493, doi:10.1002/etc.5620161208.
Morel, F. M. M., A. M. L. Kraepiel, and M. Amyot (1998), The chemical cycle and bioaccumulation of mercury, Annu. Rev. Ecol. Syst., 29,
543–566, doi:10.1146/annurev.ecolsys.29.1.543.
Munk, W. H. (1966), Abyssal recipes, Deep Sea Res. Oceanogr. Abstr., 13,
707–730.
Oeschger, H., U. Siegenthaler, U. Schotterer, and A. Gugelmann (1975), A
box diffusion model to study carbon dioxide exchange in nature, Tellus,
27, 168–192, doi:10.1111/j.2153-3490.1975.tb01671.x.
Orr, J. C., et al. (2001), Estimates of anthropogenic carbon uptake
from four three‐dimensional global ocean models, Global Biogeochem.
Cycles, 15(1), 43–60, doi:10.1029/2000GB001273.
Pace, M. L., G. A. Knauer, D. M. Karl, and J. H. Martin (1987), Primary
production, new production and vertical flux in the eastern Pacific Ocean,
Nature, 325(6107), 803–804, doi:10.1038/325803a0.
Prentice, I. C., G. D. Farquhar, M. J. R. Fasham, M. L. Goulden, M. Heimann,
V. J. Jaramillo, H. S. Kheshgi, C. Le Quéré, R. J. Scholes, and D. W. R.
Wallace (2001), The carbon cycle and atmospheric carbon dioxide, in
Climate Change 2001: The Scientific Basis: Contribution of Working
Group I to the Third Assessment Report of the Intergovernmental Panel
and Climate Change, edited by J. T. Houghton et al., pp. 183–237,
Cambridge Univ. Press, New York.
Press, W. H., B. P. Flannery, S. A. Teukolsky, and W. T. Vetterling (1986),
Numerical Recipes: The Art of Scientific Computing, Cambridge Univ.
Press, New York.
Sabine, C. L., et al. (2004), The oceanic sink for anthropogenic CO2,
Science, 305(5682), 367–371, doi:10.1126/science.1097403.
Selin, N. E., D. J. Jacob, R. J. Park, R. M. Yantosca, S. Strode, L. Jaeglé,
and D. Jaffe (2007), Chemical cycling and deposition of atmospheric
mercury: Global constraints from observations, J. Geophys. Res., 112,
D02308, doi:10.1029/2006JD007450.
Selin, N. E., D. J. Jacob, R. M. Yantosca, S. Strode, L. Jaeglé, and E. M.
Sunderland (2008), Global 3‐D land‐ocean‐atmosphere model for
mercury: Present‐day versus preindustrial cycles and anthropogenic
enrichment factors for deposition, Global Biogeochem. Cycles, 22,
GB2011, doi:10.1029/2007GB003040.
Siegenthaler, U., and H. Oeschger (1978), Predicting future atmospheric
carbon dioxide levels, Science, 199(4327), 388–395, doi:10.1126/science.
199.4327.388.
Strode, S. A., L. Jaeglé, N. E. Selin, D. J. Jacob, R. J. Park, R. M. Yantosca,
R. P. Mason, and F. Slemr (2007), Air‐sea exchange in the global
mercury cycle, Global Biogeochem. Cycles, 21, GB1017, doi:10.1029/
2006GB002766.
Sunderland, E. M., and R. P. Mason (2007), Human impacts on open ocean
mercury concentrations, Global Biogeochem. Cycles, 21, GB4022,
doi:10.1029/2006GB002876.
Sunderland, E. M., D. P. Krabbenhoft, J. W. Moreau, S. A. Strode, and
W. M. Landing (2009), Mercury sources, distribution, and bioavailability in the North Pacific Ocean: Insights from data and models, Global
Biogeochem. Cycles, 23, GB2010, doi:10.1029/2008GB003425.
9 of 10
GB4014
STRODE ET AL.: OCEAN MERCURY TRANSPORT
Swain, E. B., D. R. Engstrom, M. E. Brigham, T. A. Henning, and P. L.
Brezonik (1992), Increasing rates of atmospheric mercury deposition
in midcontinental North America, Science, 257(5071), 784–787,
doi:10.1126/science.257.5071.784.
GB4014
L. Jaeglé, Department of Atmospheric Sciences, University of Washington,
Box 351640, Seattle, WA 98195‐1640, USA. ([email protected].
edu)
S. Strode, SAIC, NASA Goddard Space Flight Center, Code 610.1,
Greenbelt, MD 20771, USA.
S. Emerson, School of Oceanography, University of Washington, Box
355351, Seattle, WA 98195‐5351, USA.
10 of 10