Isotopic versus micrometeorologic ocean CO 2 fluxes: A serious conflict
Descrição do Produto
JOURNALOF GEOPHYSICALRESEARCH,VOL. 91, NO. C9, PAGES 10,517-10,527,
SEPTEMBER15, 1986
ISOTOPIC VERSUS MICROMETEOROLOGICOCEAN CO2 FLUXES: A
SERIOUS
CONFLICT
WallaceS. Broecker, 1 JamesR. Ledwell,1 Taro Takahashi, 1 RayWeiss,2 Liliane Merlivat,3 LaurentMemery, 3 Tsung-Hung Peng, • BerndJahne,5 andKarl Otto Munnich 5 Abstract.
Eddy correlation
measurements
portant than that driven by molecular through the diffusive boundary layer. not the first time the eddy correlation has yielded gas fluxes widely different
over
tbe ocean give CO2 fluxes an order of magnitude or more larger than expected from mass balance measurements using radiocarbon and radon 222. In particular, Smith and Jones (1985) reported large upward and downward fluxes in a surf zone at supersaturations of 15% and attributed them to the equilibration of bubbles at elevated pressures. They argue that even on the open ocean such bubble injection may create steady
state
CO2 supersaturations
transport This is method from
isotope studies. Jones and Smith [1977], Weseley et al. [1982], Bingham [1982], and Greenhut et al. [1983] have reported CO2 flux estimates far
greater
carbon.
than
Before
presenting
CO2 fluxes
and that inferences
those
based on radon
and radio-
ß
our case in
determined
the
that
not
wish
topic means: prebomb radiocarbon inventories; global surveys of mixed layer radon deficits; and oceanic uptake of bomb-produced radiocarbon. We argue that laboratory and lake data do not lead one to expect fluxes as large as reported from the eddy correlation technique; that the radon method of determining exchange velocities
daylight
showing that the energy loss estimates based on the net sensible heat flux (measured micrometeorologically) and net water flux (measured micrometeorologically) balance the net solar in-
is indeed useful
put (corrected
CO2 fluxes;
ing all tion
thesis
that
Smith
and
Jones
that
advance
that
neither
the
observed
the hypo-
concentrations
A strong
case has been made for
of CO2 fluxes
hours
challeng-
measured during
over vegetated
land
surfaces
for
soil
heat storage),
workers
carrying out these measurements clearly have established that the basic technique is valid for heat and water. By showing that the micro-
and
meteorologically
derived CO2 fluxes
agree with
independent agricultural estimates of net photosynthesis, these investigators make a convincing case for the validity of their CO2 fluxes. The
that the pCO2 values reported by Smith and Jones are likely to be systematically much too high. The CO2 fluxes for the ocean measured to date by the micrometeorological method can be reconciled with
technique.
we are
measured by the eddy correla-
[Alvo et al., 1984; Anderson et al., 1984; Ohtaki, 1984; Verma and Anderson, 1984]. By
cannot
account for the fluxes that they report;
CO2 fluxes
the reliability
supersaturations of CO2 due to bubble injection on the open ocean are negligible;
clear
of
methods we
of fluxes based on air-sea pCO2 differences and radon exchange velocities must be made with caution. We defend the global average CO2 exchange rate determined by three independent radioiso-
for estimating
to make it
defense
by isotopic
net CO2 fluxes
so measured range from 100 to
1000toolm-2 yr-1.
of
workers
radioisotopes of radon and carbon in the oceans nor the tracer experiments carried out in lakes and in wind/wave tunnels.
is
about
Theerror stated by these
3%, and the detection
limit
is
about10 toolm-2 yr-1. The situation
ent. Introduction
over
the
Based on isotopic
ocean
is
studies
quite
differ-
(see below),
the
average exchange rate of CO 2 int• yr andlOUt of the sea is estimated to be 20 tool m . The net --
In a recently
published
paper,
flux
Smith and
of CO2 for a specific
Jones [1985] cast doubt on the validity of air-
order
sea CO2 flux
the fractional
estimates
derived
from the radon
defined
method. Their conclusions are based on co2 flux measurementsmade by the eddy correlation technique. Their claim is that exchange driven by bubble
entrainment
is potentially'•far
of
the product
area must be of the
of this
exchange
supersaturation
rate
and
of dissolved
CO2,
as
(i)
S = ApCO2/pCO2(air)
more im-
where ApCO 2 is the excess of pCO2(sw), the CO2 pressure
1Lamont-Doherty GeologicalObservatoryof Columbia
University,
Palisades,
pCO2 (air),
California.
•Centred'Etudes,Nucleairesde Saclay, Gif-sur-Yvette,
France.
by the seawater,
over
pressure of CO2 in the
overlying air. The absolute value of S is usually much less than one so that the net is much less than the exchange rate. For
New York.
2ScrippsInstitution of Oceanography, La Jolla,
exerted
the partial
example, values
•
Figure for
zonal
1 shows annual average bands
in
the
world
flux
ApCO2
oceans.
With values rangingbetween-38x10-6 and
•OakRidgeNational Laboratory,OakRidge,
+50x10-6 atto, and the current meanvalue for
Tenne s see.
pCO2(air)of about335x10 -6 atto,estimatesof
5Institute fur UmweltphySik der Universitat Heidelberg, Federal RepublicOf Germany.
zonal
Copyright 1986 by the AmericanGeophysicalUnion.
Local fluxes are not likely to be very much larger except in very special circumstances. CO2 supersaturations or undersaturations of more than 25% are seldom encountered in the open
Paper number 6C0236. 0148-0227 / 86/O06C-0236 $ 05. O0
average net flux
vary
from -3 (int_o the
sea) to +3 (out of the sea) toolm-2 yr-1.
10,517
10,518
Broeckeret al.:
Isotopic VersusMicrometeorologicOceanCO 2 Fluxes
80
80# 70
7
•0 •0 40
•50
:•0 I0
I0 0
0
I0
I0 :•0
gO
•,0
$0
40
40
5O
50 6O 70
80S
80S
Fig. 1.
Thedistribution of the pCO 2 difference(ApCO 2 in 10-B atm)
between surface water and air. The values represent mean value in each zone indicated by the heavy lines.
a zonal and seasonal Positive values
(hatched areas) indicate that the net CO2 flux is from sea to air, and negative values, air to sea. The global mean ApCO2 value is about -9 •atm, which is consistent with estimates of the invasion rate of fossil fuel CO2 into the ocean. The data were obtained during various expeditions including the GEOSECS,Transient Tracers in the Ocean (TTO), Weddell Polynya Experiment, Hawaii-Tahiti Shuttle, Atlantic Long-Lines (Ajax), NOAA Acid Rain, United State Coast Guard Polar Star, HMS Bransfield, ESSO Atlantic, Trans Freight Lines Jefferson, Japan Line Ltd. Japan Alliance and Icelandic Marine institute/Lamont-Doherty Geological Observatory joint seasonal
ocean, tions
tor
study
programs.
and radon measurements have
failed
of 3 greater
conditions studies
to
exchange
[Peng et al.,
have
not
in
show much
rates
detection
limits
of
average
The first
Laboratory
ultrasensitive
the
eddy
depen-
correlation
technique stated above (see Figure 2). Furthermore, corrections to the CO2 flux due to fluctuations
associated
with
heat
and
moisture
fluxes
are of this same order, again suggesting that accurate oceanic fluxes will be very difficult to extract from eddy correlation measurements. In
the
sections
which
follow
we first
summar-
ize the results of the CO2 flux measurements made by the eddy correlation technique. Then, we defend
data;
the exchange
rates
based on isotope
we show that evidence from wind/wave tun-
nels and lakes supports the isotope-based resuits rather than the eddy correlation results; we show
that
from bubble and we review
the
freedom
injection the
of
the
effects
evidence
for
radon
is
method
an advantage;
the
effect
of
bubbles on gas supersaturations in the open ocean. Finally, we discuss the bubble entrainment mechanism proposed by Smith and Jones to explain the high flux values obtained by their method, and we discuss a problem which may have affected the pCO2(sw) measurements of Smith and Jone s.
Results
for CO 2
a fac-
dencies of gas exchange on dynamical conditions for CO2 or other gases, nor, as we argue later, are such dependencies expected from studies of bubble phenomenology. In summary, the range of net fluxes expected on the open ocean lies almost entirely within the
Summary of Ocean Eddy Correlation
condi-
than
than in
1979].
indicated
stormy
more
attempt
to measure the CO2 flux
tween the ocean and atmosphere relation
method
was
made
on
the
be-
by the eddy corbeach
at
Sable
Island, Nova Scotia, by Jones and Smith [1977]. The weather at the time of measurement (July 1976) was characterized by high humidity, fog, and relatively gentle winds. Three measurements were made over a 2-day period. The result
yielded netyrupward CO 2 fluxesrangingfrom16 to -2 25 mol m ß As the ApCO2 between air and sea was not
measured,
there
is
no way to convert
this result to an exchange rate. If for example the CO2 partial pressure in the water were 20% higher than that for the air, then their measurement would correspond to a CO2 exchange rate
in the rangeof 80-125molm-2 yr-1 (i.e.,
4-6
times the radioisotope-based mean global CO2 exchange rate). The second published CO2 net flux by the eddy correlation
method
coastal waters Miami, Florida,
was
measured
1980 [Weseley et al. ' 1982].
flux
over
shallow
above a submerged reef near during a 2-week period in March
was reported
to be 215-250
The net CO 2 2 urnward • mol m-
yr-
at
a windspeedof 10 ms-1. Again,no pCO 2 data are available. If the surface ocean pCO2 were 20% above the atmospheric value, these results
would yield CO•yr exchange ranging to 1250 mol m-1 (i.e.,rates = 50 times from the 1075 radioisotope-based mean). Bingham [1982] reported the first aircraftbased CO2 net flux measurement by the eddy cor-
Broeckeret al.: 2O
I
.028
i
Summary of Radioisotope-Based
'E 021
od
.007 LIMIT
-5
X
u
been made from radioisotope
•
'•
o
-.007
- I
n-
-
:D -I0 --
I-z
_15
m,m I-' z
m,mZ
n,'o
•w
•
o•o z _20 ø
esti-
CO2 exchange rate have inventories.
The
reader is referred to Broecker for a summary of these studies of the basic papers containinM
and Peng [1982] and for a listing the measurements.
The
the
first
estimate
of natural
LL X
,014 .J
-
near
is
based
radiocarbon.
on
distribution
The assumption
is made
steady
-,021 U
i.uz
o•-•o•
w -.028
200 400 pSEA -6• co2(io tm)
z
600
state.
In particular,
the assumption
is madethat the decayof 14Cwithin the sea was balancedby the entry of I•C atomsthroughthe air-sea boundary. Through a knowledge of the volume of the ocean, the average ZCO2 content
of
seawaterandthe averageI•C/C ratio for the CO 2
speg.ies
of •C
0
radioisotope-based
mates of the mean global
0
.ooo
M ICRO-
rr
independent
Gas Exchange Rates
that prior to 1850 the distribution of 14Cwas
METEOROLOGY
o
Three
o
DETECTION
o
10,519
Isotopic VersusMicrometeorologic OceanCO 2 Fluxes
dissolved
in
ocean water,
the
in the sea can be determined.
decay
rate
The amount
of i•C enteringthe seadepends on the CO 2 exchange rate andthe differencebetween the I•C/C ratio in atmospheric and surface ocean carbon (corrected for the equ.ilibrium isotope fraction-
ation).
The air-sea I•C/C difference has been
estimated
from
existing
measurements
to be
46+10ø/oo (seeFigure3). TheCO 2 exchan_•e -1
Fig. 2. Relationship between net CO2 flux across the air-sea interface and the CO2 partial pressure for surface water based on the radio-
rate estimated in this way is 17+4 mol m yr (the reader is reminded of the difference between exchange rate and net flux introduced earlier). This rate applies to the preanthropo-
yr
ß Also shown is the relationship
genicCO 2 pressure in the atmosphere (=280x10 -6
yr
) and the units used by Smith and Jones
isotope-based CO 2 exchange rate of 20 molm-2 -1
between
the units usedby geochemists (i.e., molm-2 -1
[1985](mgCO 2 m-2 s-l).
Although surfacewater
atm). Correcting for the anthropogenic increase in-2 atmos.pheric CO2 content this becomes 20+5 mol
m
yr -l for the period 1970-1980 whenthe atmo-
pCO 2 values as high as 550 x 10-ø atm (in upwel-
sphericpCO 2 averaged about330x10 -6 atm.
the ice edges)
of the 222Rn to 226Ra ratio in the surface ocean
ling areas)andas lowas 150x 10-6 atto(near have been observed
locally,
valuesrangeMostcommonly between 200x 10-6 and 400 x 10-6 atm.
The important point is that
the range of expected net CO2 flux corresponding to the commonlyobserved pCO2 values in global surface
limits
ocean
water
lies
within 2
of +0.01 mg CO2 m-
and Jones
s-
the
1
detection
stated by Smith
[1985].
The
July
method.
1981 off
His
measurements
San Diego,
were
California
(at
122øW). The net CO2 upward flux, for
humidity-induced
density
made
in
30øN,
uncorrected
fluctuations,
was
about100molm-2 yr-1. Again,no pCO 2 results are available. If the ocean had a CO2 pressure 20% higher than the air, this value would cur2 -1 respond to an exchange rate of 500 mol m- yr . In their most recent paper, Smith and Jones [1985] report eddy correlation measurements of CO2 made at the Sable Island beach site in late autumn. Results of their measurements range from
a maximum
net
air-to-sea
flux
of
about
32
toolm-2 yr-1 to a maximum net sea-to-air flux of about 17 mol m-2 yr-1. They reported that during the entire observation period the partial pressure of CO2 in the seawater remained above that in the atmosphere by 10-15%. Because of the rapid changes in the sign of the flux these authors do not attempt to calculate a mean CO2 flux. Hence to date, all meteorologic measurements
CO2 exchange rates estimated
from
the reported microover the ocean give
which are far
radioisotope
higher
studies.
than
estimate
is
based
on
mixed layer. For the open ocean, present in this layer is generated
measurements
the radon by the in
situ decayof 226Radissolvedin seawater. The radon content in the air over the open ocean is too small to contribute significantly to the surface water burden. By measuring the ratio of 222Rnactivity to 226Ra activity in samples from the mixed layer, the fraction of radon which escapes to the atmosphere and, in turn, the evasion
relation
second
rate
station where made that the
of
radon
can
be
assessed.
At
each
this is done the assumption is system is at steady state on a
time scale of several days (the half-life of Rn is 3.82 days). However, as both the mixed layer thickness and radon evasion rate vary with meteorological conditions, this assumption is rarely fulfilled. The argument is that if the results from enough stations (or times) are averaged, the biases resulting from departures from steady state will cancel. 222
Penget al. [1979]use the 222Rnand226Ra results from stations world ocean to obtain
located throughout the an average radon evasion
rate
If
(see Figure
4).
this
average
is used to
obtaina meanglobalCO•exchange rate, the re-
suit is about 16molm6 yr-1 fortheatmospheric pCO2 of 330x10- atm. The reader unfamiliar
with
the
models
used
to
convert
from
exchange rate for one gas to that of another is referred to Broecker and Peng [1982] and Jahne [1985] for discussions of this procedure. While it
is
difficult
to
estimate
the
error
with this result, it is likely about Because the higher-latitude portions
associated
+25%. of GEOSECS
10,520
Broecker et al.:
0
l.p
I
IsotopicVersusMicrometeorologic Ocean CO 2 Fluxes
I
: *
I
I
-
tweenthe bombi•C to C ratio for atmospheric CO2 and surface water ZC02, prior to this As shown by Broecker et al. [1980; 1985],
time. the
given
the
•" '•%", ONSET - water columninventoryof bombi•C atomsat any 14CFROM '
-or
-..:-%... ,/_
ocean
station
can be estimated
from
profiles of i•C/C, ZCO 2 and 3Hmeasured during the GEOSECS programand the i•C/C for thesestations
estimated
from measurements on prebomb
samples. Thehistory of the 14C/Cfor atmo-
•--•-40
x x
-=____._ß: __ .. .-... -60
ß
spheric
CO2 is well
documented from direct
meas-
urements. A complete history for the I•C/C for CO2 dissolved in the surface ocean is available from measurements on corals in a few places. However, as shown by Broecker et al. [1985], adequate estimates of this history can be made at
all
other
places
in
the
ocean based
on the
prebomb i•C/C value and the i•C/C value measured 1850
I
I
I
I
1875
1900
1925
1950
x
YEAR
Fig. 3. Summary of radiocarbon an estimate of the preindustrial
data yielding (i.e., presum-
2 duringthe GEOSECS survey. In this wa{, mean
CO2 exchange rates
of 19.2 mol m-
yr
for
the
Indian Ocean 19.4 mo12m-2 Pacific Ocean, and 2•.3 mol m- yr-{r-1 forfor thethe Atlantic Ocean
are
estimated.
From
these
an
ocean
aver-
ably steady state) difference betweenthe 14C/C
age close to 20 molm-2 yr-1 is obtainedin
ratio
natural i•C discussedabove. The error in this average is probably no more than 3 mol m-2 yr-1.
in atmospheric CO2 and surface ocean ZCO2
(corrected
for
the equilibrium
ation). The small atmosphere obtained
isotope
fraction-
dots show the trend for the from measurements on tree
close
agreement
with
the
estimate
based on
rings [Stuiver and Quay, 1981]. As the 14C/C ratio in the atmosphere is geographically uniform, this record is typical of the entire atmosphere. The large dots show the trend for surface
water
from
the
Florida
Straits
from measurements on growth r•ng
as
I
obtained
dated corals
[Druffel and Linick, 1978]. As the 14C/Cratio in surface location,
ocean carbon varies this
record
cannot
with
geographic
be taken
as the
ocean average. Only a limited number of prenuclear water samples were collected and analyzed for radiocarbon. Most of these were in
the Atlantic
[Broecker
et al.,
1960].
the
data
for
more sparse, significantly
Straits
site
becomes 43O/oo
between 40øN and 40øS. the
Pacific
there
and
different
from
those
I I I
III
THICKNESS
• 5 SPEED o.o
for
oceans
that for
ocean was 46O/oo (i.e.,
are
the At-
Two pieces
of
information
the
were wind likely
distribu-
are
needed:
the inventory of bombI•C atomsin the oceanfor one time
(in
this
case
survey);
and the history
the
time
of
¬
-
0.4
0.6
0.8
t.o
I
I
I
I
40øS) shows lower values in all three oceans. We estimate that the global average pre-
and surface
-
Although
Indian
is no evidence
I
•
½r' ='l.7cm/hr rn _ MEAN WIND
These
of 46O/oo obtained for the 1850-1900 period at Florida
I
:• =0.77 I-- 15 - MEANMIXEDLAYER 0 I0-VELOCITY =Z.er./•oy
value for 41 such samples collected between 40øN and 40øS in the Atlantic Ocean is -53O/oo as compared to-56O/oo for four Florida Straits coral samples from the same time interval (i.e., 1955-1957). Thus the preindustrial difference the
I
O3 MEAN PISTON
results are shown by crosses. TheaverageA14C
the Atlantic
I
(/320 - MEAN 222Rn-Z26Ra ACTIVITITY RATIO •1
the
GEOSECS
of the difference
be-
4.
2
4 K(m6/d(3y) 8 I0 12
In the top panel
is shown a histogram
of the activity ratios of 222Rnto 226Rafound
in the mixed layer for 90 stations occupied during the GEOSECSexpeditions [Peng et al., 1979]. The average ratio is 0.77 with no values below 0.50. In the lower panel, a histogram for the iston velocities (k) calculated from the
22Rn/226Ra activity ratios andmixedlayer thicknesses
(h) are shown.
velocity is 2.8 m d-1 (i.e.
values higherthan7.0md-•
The average piston
11 7 cmh-l).
'
were
observed.
No
Broeckeret al.:
What Would the Radioisotope Distribution Like if the Gas Exchange Rate Were 10 Times Larger? Starting with the radon method, let sider how the radioisotope distributions
3.5
Look
N•3.0
us conwould
,
ß
ß ,
•t
10 times the best estimatesß This
rate corresponds to net CO 2 fluxesof order30
• 2.0
toolm-2 yr- at 15%supersaturation,still at the low end of fluxes reported from the eddy correlation techniqueß For the average mixed
layer
O'
•_ 2.5
differ if the CO 2 exchange rate were200toolm-2 yr-1 i e
10,521
IsotopicVersusMicrometeorologic OceanCO 2 Fluxes
thickness
(ißeß,
, •
•
= 54 m) and the average
0)
•
radioisotope exchange (i.e. a piston velocity of 2.b•sed md-l), therate 222Rn to 2•BRa activity
ratio
is 0.77.
Were the piston
veloc-
ity instead10 timesas great (ißeß, 28 md-l), then the expected ratio would drop to 0.25. Ninety radon stations were taken as part of the GEOSECSAtlantic and Pacific programsß No ratio
less that 0ß50 was found (see Figure 41• •/C The mean preanthropogenic air-sea
15
Fig. 6.
curves) and of the mean surface water bomb 1•C
In addi-
result obtainedusing the meanbomb1•C excess
summary), more recent high-quality measurements on ring-dated corals exclude the possibility that this difference was so small (see Figure
in surface
Finally, if the CO•exchange rate were 200
rather than 20 toolm- yr-1, the increase in
from the distribution
the atmospheric
14C. At the time of the GEOSECS survey the atmospherici•C/C ratio was400O/oohigher than its
prebomb value,
while
the surface
than its
(160ø/oo,
see Figure
5) and
obtained by Broecker et al. [1985] for the time of the GEOSECS survey. •e mean oceanic vertical mixing rate obtained by Li et al. [1984] is given for comparison. This diagram demonstrates the sensitivity of the CO2 exchange rate derived
surface ocean 14C due to the bomb tests would
was only 160ø/oo higher
water
the total bomb1•C inventoryof 2.9x1028 atoms
3).
behind
curves) at the time of
the GEOSECS surveys for various combinations of the mean ocean CO2 exchange rate and the mean ocean vertical •xing rate. •e results were calculated for a one-dimensional ocean [see Broecker et al., 1980]. The black dot shows the
tion to several tens of measurements on prenuclear surface water (see Broecker [1963] for
so far
30
Plot showingconto•s of the ocean
excess (in O/oo, light
exchange ra•e200toolm -2 yr-1 rather than20 would be
not have lagged
25
inventoryof bomb 1•C (in 10'• atoms,heavy
ratio
difference (corrected for equilibrium isotope fractionation) is 46+10ø/oo. Were the true
tool m- yr- , the expected difference one-tenth this value (ißeß, 5ø/oo)ß
20
CO• EXCHA•E RATE(toolm'•yr-I)
ties
in
of bomb1•C to uncertain-
the mean surface
water
excess
(i.e.,
•
+15ø/oo•sand in theThe oceanic inventory (i.e., is •0.2x10 atoms). range of uncertainty
ocean ratio
prebomb value
sho•
by the dashed ellipse.
As can be seen,
the average CO2 exchange rate is tightly
con-
strained. 3o
I
o_ 20
Lihat lebih banyak...
Comentários