D. E. Archer Department of Geophysical Sciences University of Chicago Chicago, Ill 60637 T. Takahashi, S. Sutherland, J. Goddard, D. Chipman, K. Rodgers, and H. Ogura Lamont-Doherty Earth Observatory of Columbia University Palisades, NY 10964
CO2), and alkalinity made during the JGOFS Survey I
(TT007, February-March 1992) and Survey II (TT011, August-September 1992)
expeditions. JGOFS data are compared with data from the Hawaii-Tahiti Shuttle
Experiment (HTSE, February, 1979 through June, 1980). The Survey I and II
expeditions took place during and after the el Ni
o
event of 1992, while HTSE occurred during mild el Ni
o
to near climatological conditions. The Survey I and II sea surface
temperatures are among the warmest and coldest, respectively, in the combined
JGOFS and HTSE data set, and sea surface concentrations of the biological
tracers NO3 and pCO2 from JGOFS bracketed the HTSE data
with lower concentrations during Survey I and higher values during Survey II.
However, the air-sea contrast in pCO2 was diminished in 1992 due to
rising atmospheric values. The variability of sea surface concentrations of biological tracers seems to be primarily controlled by the physical structure of the water column. In a comparison of HTSE and JGOFS data (decadal time scale) or Survey I and II data (seasonal / ENSO time scale), the concentrations of the tracers on constant-density (isopycnal) surfaces is nearly time invariant, so that the variation in sea surface concentrations is controlled by the outcropping of isopycnal surfaces. On the time scale of the station occupation (diurnal to a few days), variation in replicate measurements of pCO2 is correlated with variation in density, again indicating physical rather than biological control of pCO2 variability. These findings make an interesting contrast to JGOFS North Atlantic Bloom Experiment results [Chipman et al., 1993], where recent local biological forcing was found to dominate chemical variability. The implication of this finding is that a physical understanding of the equatorial Pacific circulation may be sufficient to make predictions of short-term variability in air-sea pCO2 fluxes in this region. Some minor exceptions to the rule of physical control of sea surface chemical properties include a freshwater cap just south of the equator which follows the sea surface, rather than any density surface, and Si, which appears to vary seasonally independently of the other biological tracers NO3, CO2, and O2.
Based on the relationship between the alkalinity and CO2 in
the upper 400 meters of the water column, about 13% of CO2
increase with depth is due to the dissolution of CaCO3, 65% to the
oxidation of biogenic debris, and 23% to increasing solubility of
CO2 in colder waters. The relationship between
CO2 and O2 in equatorial surface waters during
Survey II indicates a vigorous circulation with overturning timescale of only a
few days.
The JGOFS EqPac program was fortunate to find el Ni
o
conditions during Survey I and a return to more climatological conditions
during Survey II. We present an analysis of carbonate system measurements
including pCO2, CO2, and alkalinity from both
cruises. The data are compared with the observations made during the
Hawaii-Tahiti Shuttle Experiment (HTSE) program conducted from February, 1979
through June, 1980, a period of mild ENSO to near climatological conditions.
Ultimately, we will use JGOFS data in conjunction with numerical and
theoretical methods to attempt to answer the questions in the preceding
paragraph. Our immediate goal in this paper, however, is to present and
synthesize the data into a coherent descriptive framework.
CO2. In
addition, every second bottle was analyzed for pCO2 from virtually
every routine CTD, for a coverage of 20 m vertical resolution to a depth of 130
m, 4 times per day, averaging 6 casts per station. All of the data presented
here is available for public access on the JGOFS data archive system
(accessible at URL address http://www1.whoi.edu/jgofs.html). A detailed
discussion of the calibration and quality assurance of the JGOFS CO2
data is given by Takahashi et al. (in preparation).
Temperature, salinity, oxygen, nutrients, alkalinity and
CO2 were measured during the HTSE program by the staff of
the Physical and Chemical Oceanographic Data Facility of the Scripps
Institution of Oceanography [Williams, 1981a; 1981b; 1981c; 1981d]. The HTSE
program consisted of a total of 15 legs between Hawaii and Tahiti along
156oW, 153oW and 150oW. Nutrients and oxygen
were measured during all odd numbered legs plus Leg 14, and Alkalinity and
CO2 were measured during Leg 3 (April-May, 1979) and Leg 9
(November, 1979). Sea surface pCO2 was measured by the Lamont group
during five periods; Leg 3 (April-May, 1979), Leg 5 (June-July, 1979), Leg 9
(November, 1979), Leg 11 (January, 1980) and Leg 14 (April-May, 1980).
ln pCO2 / T = 4.23 % C-1
as determined by Takahashi et al. [1993]. Subsurface pCO2 values reported in this paper are corrected to potential temperature, and thus represent the pCO2 which the sample would have if it were adiabatically raised to the sea surface (potential pCO2).
HTSE pCO2 was measured using a similar method. A 20-liter sea water sample was collected in a Pyrex jar, and about 2 liters of marine air was recirculated for about 15 minutes, until equilibration. A 200 ml gas sampling flask, equipped with a stopcock at each end, was inserted in the gas circulation circuit. At the end of equilibration, the gas sampling flask was isolated by closing the stopcocks and shipped to LDEO for CO2 measurements. Because of the large volume of water sample, its temperature was generally within 1oC of the in situ value and stayed constant during the equilibration process. During HTSE, only surface water pCO2 values were measured. Based on the analyses of replicate samples, the precision of the measurements was estimated to be +/-2.5 uatm. Since HTSE pCO2 data were calibrated on the same CO2 concentration scale as JGOFS, they are directly comparable.
CO2 (defined as the
sum of [CO2]aq, [HCO3-] and
[CO3=]). Water samples were poisoned with mercuric
chloride and sealed in 250-ml Pyrex bottles equipped with ground glass
stoppers. The results of a 200-day storage test showed that the CO2
concentration in bottled samples remained unchanged within our analytical
precision of +/-1.5 umol kg-1 [Takahashi et al., in preparation].
In addition, a reproducibility of +/-0.05% (or +/-0.94 umol kg-1)
was observed for 20 pairs of mixed layer samples. We have also determined a
number of the SIO Certified Reference Solutions provided by Andrew Dickson of
the Scripps Institution of Oceanography. The mean of our measurements for each
of two batches of the reference solutions (10 measurements for Batch #6 and 67
measurements for Batch #15) agrees with the respective manometric measurements
by C. D. Keeling within 0.6 umol kg-1.
During HTSE, the CO2 was measured by means of the
potentiometric acid titration method which was used for the GEOSECS program.
Although the precision of CO2 measurements was about +/-10
umol kg-1, they are subject to systematic errors of up to 26 umol
kg-1, attributable to instrumental blanks [Takahashi et al., 1981]
and imperfections of the aqueous solution model used for interpretation of
titration data [Bradshaw et al., 1981].
The HTSE alkalinity data were obtained by a similar method and calibrated against Na2CO3 solutions with an estimated precision of +/-4 ueq kg-1. However, due to minor differences in the data reduction procedures and treatment of calibration blanks, the HTSE data may be systematically lower than the JGOFS data by several ueq kg-1.
CO2,
alkalinity, and pH. In a separate manuscript, Takahashi et al. (in
preparation) compare the pCO2 data with computed values based on
measured alkalinity and CO2 for a total of 415 samples
collected down to 2000 meters depth. While there are subtle differences
between the various dissociation constants for carbonic and boric acids which
affect the comparison, Takahashi et al. find a broad consistency between the
pCO2, CO2, and alkalinity data, lending
confidence to the independent calibrations of the three analyses.
o
on the sea surface expression of transects of temperature, nitrate, and
pCO2 can be seen in Figure 1. SST during JGOFS brackets the HTSE
data, occupying the high range during ENSO condition of Survey I, and the low
range during Survey II. We see correspondingly depleted values of
NO3 and pCO2 characteristic of the warm oligotrophic
waters during Survey I and elevated "deep water" values during Survey II. It
is interesting to note that sea surface pCO2 data from JGOFS bracket
the HTSE data, consistent with temperature and NO3, in spite of the
nearly 30 uatm increase in atmospheric pCO2 between these two
expeditions. This similarity suggests that thermocline ventilation of
CO2 lags behind the atmospheric increase. JGOFS data are placed within the framework of the entire HTSE data set in Figure 2 and Table 1, where we compile mean surface water values, 10oN to 10oS and 140oW to 160oW, from HTSE and JGOFS. Open circles are from HTSE plotted on an axis of time at the bottom of the figure, and filled circles indicate JGOFS data with time along the top axis. Figure 2 shows that, with the exceptions of the nitrate concentration and AOU observed during August-September, 1992, JGOFS data are within the range of variation observed in February, 1979 through June, 1980. The exceptionally high nitrate and AOU values observed during JGOFS are attributable to the outcropping of especially dens subsurface waters and are discussed below. With an exception of AOU, the HTSE data show no clearly recognizable annual cycle.
The variation in the pCO2 value normalized to a constant temperature
of 20o C reflects mainly changes in the total CO2
concentration in seawater. The spring 1992 el Ni
o
value (262 uatm) is one of the lowest observed, reflecting an influx of low
CO2 western Pacific waters, and is similar to the November, 1979
value, which represents a weak el Ni
o
event. The value for fall, 1992 (297 uatm) is highest in the data set and is
attributed to the outcropping of dense waters rich in CO2 and
nutrients. In contrast, the mean pCO2 value of 373 uatm at in situ
temperature for the 1992 el Ni
o
is similar to that of 389 uatm observed in the fall, 1979. This is due
primarily to the effect of warm temperature counteracting the effect of low
CO2 waters of el Ni
o
conditions. Therefore, the effect of el Ni
o
on the equatorial source for atmospheric CO2 depends not only on the
CO2 concentration in the western equatorial Pacific water but also
on the temperature of water.
When the data are subjected to detailed examination, a picture emerges of primary control over sea surface biological tracer variability by the density structure of the water column and the outcropping of isopycnal surfaces. The time scale of outcropping control of chemistry ranges from the station occupation time (hours to several days) to seasonal (Survey I vs. Survey II) and decadal time scales (JGOFS vs. HTSE).
/z. The depth and stratification
intensity of the thermocline trace the shape of a large "W" in the water
column, with a shallower manifestation at the equator and in higher latitudes
(>12o). The sections reveal the influx of high salinity
Subtropical Underwater from the South, which has salinities as high as 36.5, in
a depth range of 50-300 m, and low salinity waters at the surface from the
North Pacific. The downward slope of isopycnal (constant sigma) surfaces away
from the equator in both hemispheres indicates the South Equatorial Current,
and the surfaceward return of the sigma surfaces north of 5o N is
the signature of the Counter Current.
Ocean water masses mix and flow most efficiently on surfaces of constant
density. In order to visualize the data within the context of density surfaces
we interpolated the data onto a vertical coordinate of sigma-theta
(
) in Figures 4 a-b. The top plot is contours of
pressure (depth), and below that are temperature and salinity. As with the
depth sections, we can see the mixing of salty water from the south at
of 23 - 25 in all transects (JGOFS and HTSE).
The boundary between the gray stippled areas at the tops of the plots and the
contoured areas below is the sea surface. In both JGOFS and HTSE, the data is
marked by a lower sea surface density in the north in boreal fall, caused
presumably by summer heating in the Northern hemisphere. However, this effect
is much stronger in the JGOFS data set. In the equatorial zone of
5oN - 5oS, the sea surface cuts most deeply into
space (the highest density water outcrops) during
JGOFS boreal fall. No corresponding effect is seen in the HTSE fall transect.
Thus the highest and lowest sea surface density values are found in the Survey
II transect, and the point of contact between these two regions is located at
2o N, the site of the "great convergent front" [Yoder et al., 1994],
(Archer et al., manuscript in preparation).
To first order, it appears as though the temperature structure of the water column as a function of density is unchanged from boreal spring to fall, and the changing expression of temperature at the sea surface is a result of outcropping higher sigma surfaces at the sea surface. This is not strictly correct; the salinity-temperature structure near the surface differs between boreal spring and fall. In Survey I (bottom left in Fig. 4-a), a 75-meter-thick low salinity water which has density less than 23.5 and salinity less than 34.25 caps the area north of about 8oN. During Survey II (bottom right in Fig. 4-a), waters in this area become much lighter (a density of about 21.5), and the 34.25 salinity contour follows the sea surface. The regional T/S plots for 4oN - 12oN in Fig. 5 show that the T/S relationship observed during Survey I is indistinguishable from that observed during Survey II for the densities greater than 22, but differs from the fall data only for waters with densities less than this value. The fall light water may be formed by warming and evaporation of the spring water, or may represent a warmer and more saline water transported laterally.
In the south of the equator, the T/S plots for 4oS - EQ in Fig. 5 show that the T/S relationships are unchanged from Survey I to II for waters more dense than 23.5, and that seasonal changes are confined to waters lighter than this density. Since the salinity for the light water remains also nearly constant from the spring to the fall, the fall water may be formed by cooling of the spring water.
The nutrient fields are shown interpolated into
space (see above) in Figure 6. The location of the NO3 contours in
space shows some variability from transect to
transect but little systematic pattern of
NO3/ variation, either between Survey I
and II or between JGOFS and HTSE. Thus the higher NO3 expressed at
the sea surface during Survey II (Figure 1) was a result of outcropping of
denser water during this time. Line plots of the interpolated JGOFS
NO3 values on density surfaces (Figure 7) are more revealing of the
differences between the sigma / NO3 relationship between Survey I
and II. While the values observed during the spring (el Ni
o
period) for north of about 5oN are similar to those obtained during
the fall after the el Ni
o,
those for the equatorial belt and southern latitudes are different from each
other. Within the equatorial belt, the spring values are greater than the fall
values for the densities less than 24.5, whereas in the southern latitudes, the
trend is not only reversed, but is also seen for all the density values shown
in Fig. 7. This indicates that while the effect of ENSO on the nitrate field
was confined to waters less dense than 24.5 in the equatorial belt
(5oN - 2oS), it reached down to the deep water regime
between 2oS and 12oS. The nitrate field north of
5oN was not affected. Since the nitrate concentration increases
rapidly with increasing density, the nitrate field is sensitive to changes in
vertical mixing. Therefore, the observed changes in the nitrate field further
suggest that the dynamic adjustment of the flow field that occured following el
Ni
o
involves not only zonal flows, but also vertical mixing in the area south of
about 5oN.
During HTSE, the sea surface density contrast between boreal spring and fall
was smaller, as was the contrast between sea surface NO3 values.
For SiO2, we do see something of a systematic difference between the
location of the contours in density space between boreal spring and fall. In
both JGOFS and HTSE, boreal fall SiO2 contours are found at higher
density than they are in the corresponding spring transects. We see no
systematic difference between the density of the constant SiO2
surfaces during the corresponding seasons from JGOFS and HTSE. During JGOFS,
two effects were conspiring to control the sea surface SiO2
concentration (Figure 1). During Survey I, in the south (the region of low sea
surface NO3 and high temperatures), the SiO2
concentration at a given density was relatively high, but the density of the
sea surface was low; these two effects canceled, and the result was that the
sea surface SiO2 did not undergo the same decrease as did the sea
surface NO3. In HTSE, there was no corresponding variation in sea
surface , and sea surface SiO2 values
were somewhat higher in boreal spring than fall. We are unable to determine,
based on the available data, the cause of the apparent seasonal cycle in
SiO2 relative to density.
CO2, temperature, and salinity through carbonate
equilibrium chemistry. Since the equilibrium relations are nonlinear,
pCO2 is not conservative to mixing (for example, combining a parcel
of 200 uatm pCO2 water with equal parts of a parcel of 400 uatm
pCO2 water will not necessarily yield a parcel of 300 uatm
pCO2 water). Thus alkalinity and CO2 data are
necessary to understand the factors that control pCO2.
The sea surface distributions of salinity-normalized alkalinity
(Alk35) and CO2 (CO2 35) are
shown in Fig. 8. Both the JGOFS alkalinity and CO2 values
are within the respective ranges observed during the HTSE program, with an
exception of Survey II data located between the equator and 3oN.
CO2 35 shows an equatorial peak similar in shape
to NO3, indicating the importance of biological soft-tissue control
of CO2 values. Survey II values between the equator and
about 3oN are higher than Survey I and HTSE values by about 50 umol
kg-1, consistent with sea surface NO3 values (Figure 1).
Sea surface Alk35 is nearly constant across the study area. The
"potential alkalinity" (sum of total alkalinity and nitrate concentration
[Brewer et al., 1976]) eliminates the effect of photosynthesis on alkalinity;
sea surface values are plotted against salinity in Figure 9. Survey I and II
trends may be approximated by the following linear regression lines;
Survey I: TALK (ueq kg-1) = 70.63 . (Sal) - 154.1
with a RMSD of 3.3 ueq kg-1; and
Survey II: TALK (ueq kg-1) = 73.96 . (Sal) - 270.3
with a RMSD of 4.1 ueq kg-1. Within the observed salinity range
from 33.8 to 36.0, these two trends are indistinguishable. Also, there is no
deviation of this relation from waters near the equator (at salinity values
near 35), in spite of the outcropping of high alkalinity deep waters and the
potential for high rates of production of calcareous organisms in equatorial
waters. Apparently the signatures of circulation and biology are less obvious
in alkalinity than in CO2.
Depth sections of alkalinity and CO2 are presented in
Figure 10 (a). Storage or analysis problems ruined some of the
CO2 data; missing values were filled with calculated
estimates based on alkalinity and pCO2 for 125 samples from stations
2, 5, 12, and 14 (TT007) and 14 and 15 (TT011). The contours of these
carbonate tracers closely followed the depth of the thermocline, indicated as
before by shading scaled to /z.
Anomalously low CO2 values occurred at the equator at 150 -
250 m depth (just below the thermocline), corresponding to the equatorial
undercurrent. We see below that this chemical anomaly appears to reflect
differences in atmospheric "ventilation" (gas exchange with the atmosphere) and
to some extent biological production. Most of the observed variation in
alkalinity is generated by freshwater dilution and evaporation, and can be
removed by normalizing alkalinity to a constant salinity value of 35.0. Depth
sections of normalized alkalinity and CO2 are presented in
Figure 10 (b). The nearly 100 ueq variation in Alk from Figure 10 (a)
collapses to a variation of only 20 ueq variation in Alk35 in Figure
10 (b), a value which approaches the analytical uncertainty in the
determination of alkalinity (4 ueq kg-1). In order to highlight the
real pattern of variation above the analytical uncertainty in Alk35,
a single pass of an averaging smoother was performed prior to contouring. The
normalized Alk35 and CO2 35 are still seen to be
closely tied to the depth of stratification
/z.
The factors that control the variability in alkalinity and
CO2 can best be visualized in
space (Figure 11 a). Alk35 was higher in the intermediate waters
from the north (very old North Pacific Intermediate Water which also contains
high concentrations of dissolved SiO2). As was observed for
NO3, contours of CO2 35 in
space are similar between Survey I and II, with
the changing sea surface expression controlled by changing outcropping
structure of the isopycnal surfaces. The equatorial undercurrent is found at a
density of 26 - 26.5 at this longitude between 2oN and
2oS; the chemical anomaly of the undercurrent, i.e. about 50 umol
kg-1 lower CO2 than its surroundings, can be
seen in the 2150 umol kg-1 contour in Figs. 10-a and -b. The lower
CO2 values reflect its source waters in the western
equatorial Pacific, which are depleted in nutrients and low in
CO2 (Takahashi et al., 1990).
CO2 concentration with depth is maintained
against mixing and circulation by export of biologically produced organic
matter and calcium carbonate, and by gas exchange (colder subsurface waters
outcrop and exchange with the atmosphere at higher latitudes). These effects
have been called the "soft tissue, hard tissue, and solubility pumps". We
estimate the relative contributions of these mechanisms as follows. The
surface - 400 m deep increase in CO2 35 is roughly 300 umol
kg-1 (Figure 12). If we assume that for an abiological ocean, 400 m
subsurface water would be at 12o C, have 2350 ueq kg-1
alkalinity, and be in equilibrium with atmospheric pCO2, then the
resulting CO2 would be 2120 ueq kg-1. At the
surface, pCO2 of 420 uatm at T=25oC and the same
alkalinity gives CO2 of 2050 umol kg-1, with a
resulting "solubility pump" CO2 contrast of something like
70 umol kg-1. The increase in potential Alk35 of 80-90
ueq kg-1 corresponds to a CO2 gradient of 40 umol
kg-1 generated by the production and dissolution of
CaCO3. Finally, assuming the classical Redfield ratio of
NO3:C of 16:106, the 30 umol kg-1 contrast in
NO3 implies a "soft tissue" biological pump contribution of 200 umol
kg-1 CO2. The combined CO2
gradient predicted by this analysis (70 + 40 + 200 = 310) compares well with
the observed total of ~300 umol kg-1, and the respective
contributions of the soft- and hard-tissue and solubility pumps work out to be
65%, 13%, and 23% respectively. The ratio of organic carbon to calcium
carbonate production is therefore about 5:1. This ratio, or perhaps our
temporal resolution of the ratio, appears to be insensitive to El Ni
o
conditions.
In the upper 75 meters, CO2 increases by about 100 umol
kg-1, whereas potential alkalinity values (2317 +/- 4 ueq
kg-1) stay virtually unchanged. This indicates that net production
or dissolution of CaCO3 in the upper layer is undetectably small,
and hence the increase in CO2 is primarily due to oxidation
of biogenic debris. In deeper waters down to 400 meters, alkalinity increases
linearly with increasing CO2 concentration. The saturation
depth for aragonite is located at about 200 meters depth, while calcite
saturation is only reached at about 1000 meters [Millero, 1982]. This suggests
that the alkalinity increase in the upper 400 meters can more easily be
attributed to the dissolution of aragonite.
CO2 distributions, both the pCO2 and AOU in the
subsurface waters appear to be controlled by the position of the thermocline.
There is a clear signature of the equatorial undercurrent in both fields
between about 150 and 250 meters deep with lower AOU and pCO2
values. The data are gridded in space in Figure
14. The contours of pCO2 in density space appear to be independent
of proximity to the sea surface, as was observed for nutrients, alkalinity, and
CO2. However, the distribution of AOU provides an
interesting contrast to these previous results. The zero contour for AOU (i.e.
saturation with atmospheric oxygen) follows the density of the sea surface
(Fig. 14). This difference must be due to the fast gas exchange equilibration
time for oxygen relative to nutrient uptake rates by organisms or the gas
exchange equilibration time of the buffered gas CO2.
CO2 data
with the AOU values from the same samples, from bottles < 30 m depth from
JGOFS Survey I and II. In some of the Survey II samples, the oxygen
concentration is seen to depart from atmospheric saturation (the AOU becomes
positive). The slope of the observed AOU / CO2 relation is
close to the classical Redfield respiration ratio -O2 : C value of
-1.3 : 1. If we take the conclusion from the previous section that only 65% of
the CO2 change in the upper 400 meters is accounted for by
the soft tissue biological cycle, then the expected slope for a closed
(subsurface) system would be -0.85 : 1. Both gases, oxygen and CO2, are seen to depart from atmospheric saturation equilibrium in these sea surface samples. This disequilibrium allows us to take advantage of a difference in exchange times for the two gases to estimate the exposure timescale of the waters to the atmosphere. Upon exposure to the atmosphere, both oxygen and CO2 will begin to equilibrate toward atmospheric saturation values. A canonical gas exchange piston velocity is about 3 meters day-1. Given a piston velocity, the gas exchange rate of oxygen can be calculated as
Flux O2 [mol m-2 d-1] = k [m d-1]
. 
O2 [mol m-3]
where k is an exchange coefficient, and O2 is the deviation
from oxygen saturation. For total CO2, the exchanging species is
CO2(aq), which constitutes about 0.5% of the total dissolved
CO2 concentration. Under conditions of constant temperature and
salinity, the exchange flux of CO2 can be calculated from the total
CO2 concentration as
Net Flux CO2 [mol m-2 d-1] = k [m
d-1] . 
.
CO2 / CO2
ref . [CO2](aq)ref
where is the Revelle buffer factor, calculated to be 12 for the
exposed = 23.6 surface water,
CO2 is the deviation of total CO2 of
the saturation value, CO2ref (taken here to 1990 uM
kg-1), and [CO2](aq) ref is the
dissolved CO2 gas concentration at atmospheric saturation, taken
here to be 10.2 uM kg-1. The diagonal lines in the shaded region of
Fig. 15 indicate the time evolution of the total CO2 / oxygen
signature of a suite of recently exposed surface waters. The initial ratio of
total CO2 / oxygen variability is labeled "Initial (Subsurface)".
In the course of 5-20 days, the faster equilibration of oxygen drives the
covariation toward a flat ratio, as observed in the rest of the Survey I and II
data sets. The exposure time could be a factor of two longer if we assume a
mean mixed layer depth of 60 m rather than 30 meters.
We conclude that the surface exposure time of the high nutrient, high pCO2 equatorial surface waters during JGOFS Survey II was only on the order of 5-20 days at the time of their observation. Since the return of the equatorial cold tongue was observed remotely several months before the JGOFS Survey II cruise (based on TOGA-TAO sea surface temperature data), the undersaturation was driven by a short residence time of water at the sea surface, rather than recent exposure of waters of this density. This conclusion is significant for two reasons. Firstly, the rest of the JGOFS Survey II data set, including the dissolved thorium and the biological and ecological data, were also taken during these conditions, and knowledge of this special circulation may affect their interpretation. Secondly, the subduction of nutrient-rich waters is a process which has special significance to understanding the dynamics of the carbon cycle in the thermocline. Consider a parcel of water at the sea surface, with zero nutrients and in atmospheric equilibrium in CO2. The parcel is subducted and gains nutrients and CO2 by respiration of sinking biological particles. If this parcel is brought to the surface again and held there until biological activity has depleted the nutrient stock completely, the excess CO2 typical of thermocline water is incorporated into biogenic particles and pumped to depth, resulting in zero net flux to the atmosphere (neglecting any temperature change and its effect on CO2 solubility). If on the other hand the parcel is exposed to the atmosphere but subducted before its nutrients are completely utilized, a net flux of CO2 from the thermocline to the atmosphere is allowed to occur. In other words, ventilation and subduction of high-nutrient waters may constitute a "leak" in the biological pump. Unfortunately, without knowledge of the spatial or temporal extent of this "leak", it is difficult to assess its impact on the steady state chemistry of the thermocline.
units, rather than by depth. The intervals in
NO3 or were chosen to be appropriate
for comparison with the depth-binned results. When the increment value is too
small (the binning is too fine), many of the data points fall into bins with
n=1, leading to their elimination from the calculation. When the interval
value is increased, the total number of populated bins decreases, biasing the
calculation to higher variance (with fewer populated bins, a wider range of
pCO2 values will be found in a given bin).
The depth variance from both data sets ranged from 6-7% of the mean, while the
nitrate and density binned variance was 2-3% of the mean (which is still larger
than the single cast mixed layer variance of ~0.7%, taken to be a measure of
analytical uncertainty). The NO3 binning of the replicate
pCO2 data resulted in a halving of the data set variance. This is
not surprising, since variations in NO3 are thought to reflect the
effect of biological export on pCO2. The interesting observation is
that the variance is nearly identical between the NO3 and the
binning exercises, which indicates that
is as good a predictor of pCO2 as is
NO3. In other words, since density might control NO3 but
no mechanism exists for NO3 control of density, the primary
controller of local variability is not local biological processes, but rather
the local sea surface expression of the regional density structure of the water
column. This can be contrasted with observations during the North Atlantic
Spring Bloom experiment [Chipman et al., 1993], during which local, recent
biological production played a significant role in the evolution of sea surface
pCO2 values.
o
on the Revelle Buffer Factor
CO2 concentration and
pCO2 in sea water may be expressed in terms of the Revelle factor
():
= ( ln pCO2 / ln
CO2)
at constant temperature, salinity and other chemical properties. The Revelle
factor is inversely related to the capacity of sea water to take up
CO2 for a given increase in pCO2. In global surface
ocean waters, is about 10 on the average ranging from 8 in low
CO2 tropical surface waters to 12 in high CO2 polar
waters. In deep water containing high CO2 concentrations, it
reaches as high as 17 [Takahashi et al., 1980; Takahashi et al., 1993]. The
value of can be calculated from a plot of the natural logarithm of
pCO2 measured at 20oC and 35.00 salinity against
ln(CO2 35) (Figure 16). The slope of the plots represents
the Revelle factor value evaluated at a constant temperature and salinity.
Since the pCO2 values were measured at a constant temperature, no temperature correction is needed. On the other hand, the measured pCO2 values must be normalized to a constant salinity of 35 by considering the effect of salinity on the solubility of CO2 in sea water, the dissociation constants of carbonic and boric acids, and the concentration of boric acid. The correction is generally not large: a 1% increase in salinity would cause about 1% increase in pCO2 at constant temperature. We have derived the following relationship:
pCO2 (S=35) = pCO2 (S) . (35.00 /
S) 
where pCO2 (S=35) is the value at 35.00 salinity, pCO2
(S) is the value at salinity S, and where is defined
(analogously to ) as
= ( ln pCO2 / ln Sal) = 1.15 -
6.45 x 10-4 . pCO2 (S).
A linear regression of the data yields a Revelle factor for the tropical
surface water between 12oN and 12oS of 7.9+/-0.5 for
Survey I and 9.3+/-0.1 for Survey II. Survey I data yield a greater
uncertainty because of their smaller range in the data. Nevertheless, the
difference between Survey I and II values is greater than two sigma and hence
is statistically significant. The lower Revelle factor for the Survey I data
can be accounted for by the influx of warm waters from the western equatorial
Pacific during the 1991-92 el Ni
o
period. These western Pacific waters contain lower CO2
concentrations (1880 to 1900 umol kg-1) and have a Revelle factor of
about 8 [Takahashi et al., 1988]. During Survey II a subsurface signature is
observed in the sea surface value of .
(as opposed to depth) coordinates show little
systematic variability between Survey I and II or between JGOFS and HTSE. The
variability in sea surface expression in those properties is therefore
primarily caused by fluctuations in the outcropping structure of the density
surfaces. This observation can be contrasted with the distribution of oxygen
(AOU), which appears to be strongly influenced by proximity to the sea surface,
with the exception of the Survey II equatorial surface waters, which appear to
have been replenished by subsurface waters on times scales of only 5-20 days.
We conclude that while the biological pump is responsible for the surface
depletion in nutrients and CO2, mixing and circulation of water are
fast enough in this region that local biological forcing is not detectable in
local chemical signatures. If the biological pump were faster relative to
circulation, the nutrient and carbon sections would resemble the oxygen
distribution (and reflect proximity to the sea surface in addition to density
control). Local chemistry is determined by local physics and regional (and
time averaged) biology. There are several exceptions to this generalization. First, in the equatorial region, especially just south of the equator, there is a freshening of the surface water that appears to follow the sea surface rather than traveling with a density surface. Second, there appears to be an annual cycle, observed in both JGOFS and HTSE transects, in the concentration of dissolved SiO2 as a function of density, with higher values in boreal spring than in fall. Given a constant outcropping structure, we would expect higher sea surface SiO2 concentrations in boreal spring, and in fact observe this in HTSE data. During JGOFS Survey I, lighter density surfaces outcropped, counteracting the tendency for higher boreal spring SiO2 concentrations, resulting in a nearly constant sea surface SiO2 between Survey I and II (in contrast to NO3).
We estimate based on the CO2, alkalinity, and temperature
structure of the water column that the time averaged molar ratio of organic
carbon to calcium carbonate productivity is about 5:1.
Arrhenius G. (1988) Rate of production, dissolution, and accumulation of biogenic solids in the ocean. Palaeo., Palaeo., Palaelo., 67, 119-146.
Bradshaw A. L., P. G. Brewer and D. K. Shafer (1981) Measurements of total carbon dioxide and alkalinity by potentiometric titration in the GEOSECS program. Earth Planet. Sci. Letters, 55, 99-115.
Brewer P. G. and J. C. Goldman (1976) Alkalinity changes generated by phytoplankton growth. Limn. Oceanorg., 21, 108-117.
Chavez F. P. and R. T. Barber (1987) An estimate of new production in the equatorial Pacific. Deep Sea Res., 34, 1229-1243.
Chipman D. W., J. Marra and T. Takahashi (1993) Primary production at 47oN and 20oW in the North Atlantic Ocean: a comparison between the 14C incubation method and mixed layer carbon budget observations. Deep Sea Res., 40, 151-169.
Feely R. A., R. Wanninkhof, C. E. Cosca, P. P. Murphy, M. F. Lamb and M. E. Steckely (1995) CO2 distribution in the Equatorial Pacific during the 1991-91 ENSE event. Deep-Sea Res. 42., 365-386.
Feely R.A., R.H. Gammon, B.A. Taft, P.A. Pullen, L.S. Waterman, T.J. Conway, J.F. Gendron, and D.P Wisegarver (1987) Distribution of chemical tracers in the eastern equatorial Pacific during and after the 1992-83 El Nino/Southern Oscillation Event. J. Geophys. Res. 90, 6545-6558.
Garside C. and J. C. Garside (1995) Euphotic zone nutrient algorithms for the NABE and Eqpac study sites. Deep-Sea Res. 42, 335-348.
Millero F. J. (1982) The effect of pressure on the sulubility of minerals in water and seawater. Geochim. Cosmochim. Acta, 46, 11-22.
Murray J. W., E. Johnson and C. Garside (1995) A U.S. JGOFS process study in the equatorial Pacific (EQPAC): Introduction. Deep-sea Res. 42, 275-294.
Murray J. W., M. W. Leinen, R. A. Feely, J. R. Toggweiler and R. Wanninkhof (1992) EqPac: A process study in the Central Equatorial Pacific. Oceanography, 5, 134-142.
Murray R. W., M. Leinen and A. R. Isern (1993) Biogenic flux of Al to sediment in the central equatorial Pacific Ocean: Evidence for increased productivity during glacial periods. Paleocean., 6, 651-670.
Takahashi T., D. Archer and D. Chipman (in preparation) Observations of CO2 Chemistry during the JGOFS Equatorial Pacific Expeditions, 1992: Evaluation of the dissociation constants of carbonic acid.
Takahashi, T., D.W. Chipman, J. Goddard, G. Mathieu, and L.-M. Ma (1990) Carbon dioxide in the surface water of the western equatorial Pacific Ocean. In Sea-Air Interaction in Tropical Western Pacific, J.-P. Chao and J.A. Young, ed., China Ocean Press, PRC, 511-539.
Takahashi T. and A. E. Bainbridge (1981) GEOSECS Carbonate data. In GEOSECS Atlantic Expedition, Vol. 1, Hydrographic Data, A. E. Bainbridge, U.S. Government Printing Office, Washington, D.C., pp. 7-10.
Takahashi T., W. S. Broecker, A. E. Bainbridge and R. F. Weiss (1980) Carbonate chemistry of the Atlantic, Pacific, and Indian Oceans: The results of the geosecs expeditions, 1972-1978. National Science Foundation, Washington D.C., pages.
Takahashi T., J. Goddard, S. Sutherland, D. Chipman and C. Breeze (1986) Seasonal and geographic variability of carbon dioxide sink/source in the oceanic areas: observations in the north and equatorial Pacific Ocean, 1984-1986 and global summary. U.S. Dep. of Energy, Washington, D.C., 66 pages.
Takahashi T., J. Goddard, S. C. Sutherland, G. Mathieu and D. W. Chipman (1988) Assessment of carbon dioxide sink/source in the north Pacific Ocean: seasonal and geographic variability, 1986-1987. DOE, CO2 Research Division, Washington DC, pages.
Takahashi T., J. Olafsson, J. G. Goddard, D. W. Chipan and S. C. Sutherland (1993) Seasonal variation of CO2 and Nutrients in the high-latitude surface oceans: a comparative study. Global Biogeochemical Cycles, 7, 843-878.
Tans P. P., I. Y. Fung and T. Takahashi (1990) Observational constraints on the global atmospheric carbon dioxide budget. Science, 247, 1431-1438.
Williams R. T. (1981a) Hawaii-Tahiti Shuttle Experiment, Hydrographic Report, Leg 1, 4-25 February, 1979, and Leg 3, 3 April - 2 May, 1979, Vol. I, Scripps Reference No. 81-5. Scripps Institution of Oceanography, La Jolla, CA, pages.
Williams R. T. (1981b) Hawaii-Tahiti Shuttle Experiment, Hydrographic Report, Leg 13, 18 March - 10 April 1980, and Leg 15, 17 May - 14 June, 1980, Vol IV, Scripps Reference No. 81-8. Scripps Institution of Oceanography, La Jolla, CA, pages.
Williams R. T. (1981c) Hawaii-Tahiti Shuttle Experiment, Hydrographic Report, Leg 5, 17 June - 15 July, 1979, and Leg 7, 19 August - 11 September, 1979, Vol. II, Scripps Reference No. 81-6. Scripps Institution of Oceanography, La Jolla, CA, pages.
Williams R. T. (1981d) Hawaii-Tahiti Shuttle Experiment, Hydrographic Report, Leg 9, 31 October - 24 November, 1979, and Leg 11, 8 January - 3 February, 1980, Vol. III, Scripps Reference No. 81-7. Scripps Institution of Oceanography, La Jolla, CA, pages.
Yoder J. A., S. G. Ackleson, R. T. Barber and P. Flament (1994) A line in the sea. Nature, 371, 689-693.
Legs Year/ Julia Temp Sal. PO4 NO3 SiO2 AOU TCO2 TALK pCO2 2
n
Month Date1 (oC) ---------- umol kg-1 ---------- (ueq (uatm)
kg-1)
1979-90 Hawaii Tahiti Shuttle Experiment
1 2/79 47 26.51 34.83 0.44 2.72 2.64 -3.2
3 4/79 107 27.86 35.08 0.42 2.46 2.49 -10.9 19903 2321 401/290
5 7/79 182 27.99 35.11 0.43 2.38 1.94 -3.4 393/280
7 9/79 242 27.90 35.10 0.41 2.14 1.56 -6.1
9 11/79 316 28.07 35.06 0.39 2.03 2.01 -3.9 19613 2306 364/260
11 1/80 21 27.79 35.12 0.37 2.12 1.66 -1.9 397/288
13 3/80 89 27.99 35.07 0.40 2.12 2.06 -4.4
14 4/80 123 28.46 389/272
15 6/80 152 28.57 35.12 0.38 1.75 2.12 -8.1
1992 JGOFS Equatorial Pacific Experiment
7 2/92 52 28.29 34.96 0.40 1.63 2.48 -2.9 1960 2313 373/262
11 9/92 241 26.55 34.89 0.37 3.06 1.73 +4.1 1980 2305 389/297
1Middle
date for each expedition, which is generally 30 days long.2(pCO2 in sea water at in situ temperature) / (pCO2 in sea water normalized to 20.0oC).
3These CO2 values were obtained using the
potentiometric acid titration method, and are not on the same calibration scale
used for the JGOFS program.
Table 2. A statistical analysis of the variation in surface ocean pCO2: the variance of the replicate pCO2 data set when binned by station and by depth or by density. The number of samples differs between the depth and the sigma binning because samples in bins where n=1 were excluded. The number of bins is smaller for the density binning, which by itself should tend to make the variance larger, since the data are compared with fewer common mean values. The variance is calculated as the root mean square difference between each data point and its bin mean. Relative total variance is the total variance divided by the total data set mean value.
TT007 TT011
depth NO3 density depth NO3 density
(sigma (sigma
units) units)
Bin Interval 20 m 1.0 umol 0.25 20 m 1.0 umol 0.25
kg-1 kg-1
N Samples 212 151 183 462 324 424
N Bins 79 28 40 79 43 64
Variance, 27.0 8.05 8.6 30.1 9.8 13.2
(uatm)
variance, 6.25 2.08 2.03 7.09 2.4 3.16
(%)