17 New Thoughts on the North Atlantic Carbon Budget (excerpted in part from Sarmiento et al., 1995)
J. L. Sarmiento
Princeton University
Princeton, NJ 08544-0710
A complete carbon budget for the Atlantic must consider input of carbon by rivers, burial in sediments, and gas exchange at the air-sea interface as well as transport of dissolved inorganic carbon (DIC) and dissolved organic carbon (DOC) across both the northern and southern boundaries of the basin. An assumption implicit in most studies is that the two major terms of the budget are DIC export to the south balanced by CO2 uptake from the atmosphere. We show here that a more likely scenario is that the southwards DIC export is balanced by DIC import from the Pacific via the Arctic with a possible contribution from rivers and DOC. We consider each term in turn beginning with the southwards DIC transport.
A number of studies have used hydrographic and total carbon measurements to estimate carbon transport across various sections in the Atlantic. Brewer et al., (1989) obtained an outflow of 0.26 Pg C/yr using observations at 25°N. Watson et al., (1995) note that Brewer subsequently proposed an anthropogenic carbon correction that gives a pre-industrial southwards transport of 0.41 Pg C/yr. Robbins (1994) has estimated a southwards transport of 0.3 Pg C/yr at 24°S in the Atlantic. His pre-industrial estimate is ~0.5 Pg C/yr. Schlitzer (1989b) gives the net southwards export from the Atlantic and Arctic as 0.1 to 1.1 Pg C/yr across 10°S. He reports this as an uptake from the atmosphere, but strictly speaking it is calculated as the divergence of the oceanic transport of dissolved inorganic carbon. If we leave out the highest transports of Schlitzer (1989), these estimates taken together give a southwards Atlantic Ocean DIC export of roughly 0.4 ± 0.3 Pg C/yr today, and 0.6 ± 0.3 Pg C/yr before the industrial revolution. This range includes the pre-industrial outflow of 0.6 Pg C/yr estimated by Broecker and Peng (1992) using an analysis of the properties of deep waters involved in the thermohaline overturning. Figure 17.1 shows all these Atlantic DIC transport estimates together with additional estimates by Martel and Wunsch (1993).
Could a large Atlantic DIC export, if it exists, be supplied by atmospheric CO2 uptake? Takahashi et al., (1995) use air-sea CO2 difference measurements and a simple model interpolation scheme to obtain an 18°S to 78°N atmospheric CO2 uptake of 0.25 to 0.53 Pg C/yr for 1990, with the range representing two estimates of the wind-speed dependent gas exchange coefficient (Table 17.1). However, a large portion of the 1990 uptake is due to the anthropogenic transient. We estimate from our models that the anthropogenic uptake between 18°S and 78°N is 0.38 Pg C/yr. The implication is that the pre-industrial air-sea uptake flux must have been extremely small, or that there may even have been an escape of CO2 to the atmosphere. This is most likely the case even if the thermal skin correction proposed by Robertson and Watson (1992) is included. Watson et al., (1995), who carefully analyze some of the potential problems with the Takahashi et al., (1995) technique, have used an alternative approach to estimate an uptake of 0.7 Pg C/yr for the region north of 15°N for the mid 1980's. Their uptake is at the upper limit of the uptake of 0.33 to 0.70 Pg C/yr proposed by Takahashi et al., (1995) for the region between 18°N and 78°N (cf. Table 17.1). We conclude that the air-sea flux for the Atlantic including the region
Figure 17.1. Northward transport of carbon in the Atlantic Ocean. The symbols are advective transport estimates obtained by analysis of zonal sections by Brewer et al., (1989), Schlitzer (1989a), Martel and Wunsch (1993), and Robbins (1994). Broecker and Peng (1992), and Watson et al., (1995), use a different approach for their calculations. All the transport estimates are for the present ocean except Broecker and Peng, which is for the pre-industrial ocean. The lines are model transports, including advection, diffusion and the virtual transport flux. Model diffusion represents processes on spatial scales smaller than 4°, most of which we presume would be captured by high resolution section measurements used for transport estimates. Model transport is shown for the pre-industrial steady state (labeled 1767), and for 1990. The difference between these two lines is the anthropogenic contribution.
between 18°S and 18°N is of order 0.4 ± = 0.2 Pg C/yr today, and that it was of order 0.0 ± 0.2 Pg C/yr before the industrial revolution. The pre-industrial air-sea flux is thus insufficient to supply the postulated large southwards transport of inorganic carbon of 0.6 ± 0.3 Pg C/yr.
Table 17.1. Integrated air-sea fluxes (Pg C/yr) from Takahashi compared to model.
The column labeled C-14 uses a gas exchange coefficient based on calibration with oceanic observations of bomb radiocarbon (Tans et al., (1990)), and that labeled L & M uses Liss and Merlivat (1986) . Our study uses the bomb radiocarbon calibrated gas exchange coefficient of Wanninkhof (1992) . A positive sign signifies loss from the ocean to the atmosphere.
Latitude Takahashi et al., (1995) This studyC-14 L & M
42°N to 78°N -0.48 -0.23 -0.34
18°N to 42°N -0.22 -0.10 -0.14
18°S to 18°N +0.17 +0.08 +0.14
Total -0.53 -0.25 -0.34
We conclude that if there is a large export of dissolved inorganic carbon from the Atlantic it must be supplied by another mechanism such as an inflow of dissolved organic carbon or riverine carbon, or an input from the Pacific through the Bering Straits. Rintoul and Wunsch (1991) estimate a dissolved organic nitrogen to inorganic nitrogen conversion of 119±35 kmol s-1 in the region between 24°N and 36°N. If we use a Redfield ratio of 117 organic carbon atoms to 16 nitrogen atoms, this would imply a net DOC import of 0.33 Pg C/yr. These results suggest that it is critical to include DOC measurements in the sections used to estimate carbon transports.
What about river input? Sarmiento and Sundquist (1992) summarized studies estimating that the total riverine carbon input to the world oceans is 0.4 to 0.7 Pg C/yr. A substantial fraction of this must occur in the Atlantic, where most of the rivers of the world discharge. We are not aware of a breakdown of the riverine carbon budget that would allow us to estimate the Atlantic input directly. Watson et al. (1995) suggest that 10% of the riverine flux plays a role in the Atlantic DIC budget, but there is nothing to substantiate this number. The total riverine carbon input is certainly large enough to supply the southwards flux suggested by the ocean transport calculations.
Finally we have the possibility of transport of DIC through the Arctic from the Bering Straits. Lundberg and Haugan (1995) have estimated a carbon through flow of 0.63 Pg C/yr from the product of the net water flow into the Arctic of 0.77 Sverdrups (Aagaard and Carmack, 1989) times the DIC concentration of 2020 to 2040 mmol kg-1 (Anderson et al., 1990) and the density of 1.026 kg l-1 (Stigebrandt, 1984). This is a very large number indeed, and would have quite a substantial impact on the Atlantic carbon budget. Our model has essentially no flow through the Bering Strait. Additional inputs of freshwater and carbon to the Arctic, estimated to be of order 0.18 Sv and 0.2 Pg C/yr (Anderson et al., 1990) respectively, would also have to be considered. Of 0.95 Sv net water flow coming across the Norwegian sill only 0.13 Sv leave the South Atlantic (Wijffels et al., 1992). This gives a carbon input of about 0.8 Pg C/yr and a loss of about 0.11 Pg C/yr associated with the net water transport. The difference between the input and output, 0.72 Pg C/yr, must be gotten rid of another way. The most likely candidate, given the low air-sea flux, is geostrophic transport.
We conclude that there may exist a large southwards transport of carbon out of the Atlantic Ocean. However, this transport does not appear to be supplied by atmospheric CO2 uptake, but rather by some unknown combination of northwards DOC transport, riverine input, and almost certainly a substantial flux from the Arctic due to input from the Pacific via the Bering Strait. Any future studies of the North Atlantic carbon budget should attempt to constrain all of these terms simultaneously.