Climate Change 2001:
Working Group I: The Scientific Basis
Other reports in this collection

3.2.3 Ocean Carbon Processes

3.2.3.1 Background

The total amount of carbon in the ocean is about 50 times greater than the amount in the atmosphere, and is exchanged with the atmosphere on a time-scale of several hundred years. Dissolution in the oceans provides a large sink for anthropogenic CO2, due in part to its high solubility, but above all because of its dissociation into ions and interactions with sea water constituents (see Box 3.3).

Box 3.3: The varying CO2 uptake capacity of the ocean.

Because of its solubility and chemical reactivity, CO2 is taken up by the ocean much more effectively than other anthropogenic gases (e.g., chlorofluorocarbons (CFCs) and CH4). CO2 that dissolves in seawater is found in three main forms. The sum of these forms constitutes dissolved inorganic carbon (DIC). The three forms are: (1) dissolved CO2 (non-ionic, about 1% of the total) which can be exchanged with the atmosphere until the partial pressure in surface water and air are equal, (2) bicarbonate ion (HCO3-, about 91%); and (3) carbonate ion (CO32-, about 8%). As atmospheric CO2 increases, the dissolved CO2 content of surface seawater increases at a similar rate, but most of the added CO2 ends up as HCO3-. Meanwhile, the CO32- content decreases, since the net effect of adding CO2 is a reaction with CO32- to form HCO3- (Figure 3.1). There is therefore less available CO32- to react with further CO2 additions, causing an increasing proportion of the added CO2 to remain in its dissolved form. This restricts further uptake, so that the overall ability of surface sea water to take up CO2 decreases at higher atmospheric CO2 levels. The effect is large. For a 100 ppm increase in atmospheric CO2 above today's level (i.e., from 370 to 470 ppm) the DIC concentration increase of surface sea water is already about 40% smaller than would have been caused by a similar 100 ppm increase relative to pre-industrial levels (i.e., from 280 to 380 ppm). The contemporary DIC increase is about 60% greater than would result if atmospheric CO2 were to increase from 750 to 850 ppm.

The uptake capacity for CO2 also varies significantly due to additional factors, most importantly seawater temperature, salinity and alkalinity (the latter being a measurable quantity approximately equal to [HCO3-] + 2 x[CO32-]). Alkalinity is influenced primarily by the cycle of CaCO3 formation (in shells and corals) and dissolution (see Figure 3.1c).

The annual two-way gross exchange of CO2 between the atmosphere and surface ocean is about 90 PgC/yr, mediated by molecular diffusion across the air-sea interface. Net CO2 transfer can occur whenever there is a partial pressure difference of CO2 across this interface. The flux can be estimated as the product of a gas transfer coefficient, the solubility of CO2, and the partial pressure difference of CO2 between air and water. The gas transfer coefficient incorporates effects of many physical factors but is usually expressed as a non-linear function of wind speed alone. There is considerable uncertainty about this function (Liss and Merlivat, 1986; Wanninkhof, 1992; Watson et al., 1995). Improvements in the ability to measure CO2 transfer directly e.g., Wanninkhof and McGillis, 1999) may lead to a better knowledge of gas transfer coefficients.

Despite extensive global measurements conducted during the 1990s, measurements of surface water pCO2 remain sparse, and extensive spatial and temporal interpolation is required in order to produce global fields. Takahashi et al. (1999) interpolated data collected over three decades in order to derive monthly values of surface water pCO2 over the globe for a single "virtual" calendar year (1995). A wind speed dependent gas transfer coefficient was used to calculate monthly net CO2 fluxes. The resulting estimates, although subject to large uncertainty, revealed clear regional and seasonal patterns in net fluxes.

Regional net CO2 transfers estimated from contemporary surface water pCO2 data should not be confused with the uptake of anthropogenic CO2. The uptake of anthropogenic CO2 is the increase in net transfer over the pre-industrial net transfer, and is therefore superimposed on a globally varying pattern of relatively large natural transfers. The natural transfers result from heating and cooling, and biological production and respiration. Carbon is transferred within the ocean from natural sink regions to natural source regions via ocean circulation and the sinking of carbon rich particles. This spatial separation of natural sources and sinks dominates the regional distribution of net annual air-sea fluxes.

CO2 solubility is temperature dependent, hence air-sea heat transfer contributes to seasonal and regional patterns of air-sea CO2 transfer (Watson et al., 1995). Net cooling of surface waters tends to drive CO2 uptake; net warming drives outgassing. Regions of cooling and heating are linked via circulation, producing vertical gradients and north-south transports of carbon within the ocean (e.g., of the order 0.5 to 1 PgC/yr southward transport in the Atlantic Basin; Broecker and Peng, 1992; Keeling and Peng, 1995; Watson et al., 1995; Holfort et al., 1998).

Biological processes also drive seasonal and regional distributions of CO2 fluxes (Figure 1c). The gross primary production by ocean phytoplankton has been estimated by Bender et al. (1994) to be 103 PgC/yr. Part of this is returned to DIC through autotrophic respiration, with the remainder being net primary production, estimated on the basis of global remote sensing data to be about 45 PgC/yr (Longhurst et al., 1995; Antoine et al., 1996; Falkowski et al., 1998; Field et al., 1998; Balkanski et al., 1999). About 14 to 30% of the total NPP occurs in coastal areas (Gattuso et al., 1998). The resulting organic carbon is consumed by zooplankton (a quantitatively more important process than herbivory on land) or becomes detritus. Some organic carbon is released in dissolved form (DOC) and oxidised by bacteria (Ducklow, 1999) with a fraction entering the ocean reservoir as net DOC production (Hansell and Carlson, 1998). Sinking of particulate organic carbon (POC) composed of dead organisms and detritus together with vertical transfer of DOC create a downward flux of organic carbon from the upper ocean known as "export production". Recent estimates for global export production range from roughly 10 to 20 PgC/yr (Falkowski et al., 1998; Laws et al., 2000). An alternative estimate for global export production of 11 PgC/yr has been derived using an inverse model of physical and chemical data from the world's oceans (Schlitzer, 2000). Only a small fraction (about 0.1 PgC) of the export production sinks in sediments, mostly in the coastal ocean (Gattuso et al., 1998). Heterotrophic respiration at depth converts the remaining organic carbon back to DIC. Eventually, and usually at another location, this DIC is upwelled into the ocean's surface layer again and may re-equilibrate with the atmospheric CO2. These mechanisms, often referred to as the biological pump, maintain higher DIC concentrations at depth and cause atmospheric CO2 concentrations to be about 200 ppm lower than would be the case in the absence of such mechanisms (Sarmiento and Toggweiler, 1984; Maier-Reimer et al., 1996).

Marine organisms also form shells of solid calcium carbonate (CaCO3) that sink vertically or accumulate in sediments, coral reefs and sands. This process depletes surface CO32-, reduces alkalinity, and tends to increase pCO2 and drive more outgassing of CO2 (see Box 3.3 and Figure 3.1). The effect of CaCO3 formation on surface water pCO2 and air-sea fluxes is therefore counter to the effect of organic carbon production. For the surface ocean globally, the ratio between the export of organic carbon and the export of calcium carbonate (the "rain ratio") is a critical factor controlling the overall effect of biological activity on surface ocean pCO2 (Figure 3.1; Archer and Maier-Reimer, 1994). Milliman (1993) estimated a global production of CaCO3 of 0.7 PgC/yr, with roughly equivalent amounts produced in shallow water and surface waters of the deep ocean. Of this total, approximately 60% accumulates in sediments. The rest re-dissolves either in the water column or within the sediment. An estimate of CaCO3 flux analogous to the export production of organic carbon, however, should include sinking out of the upper layers of the open ocean, net accumulation in shallow sediments and reefs, and export of material from shallow systems into deep sea environments. Based on Milliman's (1993) budget, this quantity is about 0.6 PgC/yr (± 25 to 50 % at least). The global average rain ratio has been variously estimated from models of varying complexity to be 4 (Broecker and Peng, 1982), 3.5 to 7.5 (Shaffer, 1993), and 11 (Yamanaka and Tajika, 1996). (It should be noted that rain ratios are highly depth dependent due to rapid oxidation of organic carbon at shallow depth compared to the depths at which sinking CaCO3 starts to dissolve.) If one accepts an organic carbon export production value of 11 PgC/yr (Schlitzer, 2000), then only Yamanaka and Tajika's (1996) value for the rain ratio approaches consistency with the observation-based estimates of the export of CaCO3 and organic carbon from the ocean surface layer.

The overall productivity of the ocean is determined largely by nutrient supply from deep water. There are multiple potentially limiting nutrients: in practice nitrate and/or phosphate are commonly limiting (Falkowski et al., 1998; Tyrell, 1999). Silicate plays a role in limiting specific types of phytoplankton and hence in determining the qualitative nature of primary production, and potentially the depth to which organic carbon sinks. A role for iron in limiting primary productivity in regions with detectable phosphate and nitrate but low productivity (HNLC or "high nutrient, low chlorophyll regions") has been experimentally demonstrated in the equatorial Pacific (Coale et al., 1996) and the Southern Ocean (Boyd et al., 2000). In both regions artificial addition of iron stimulated phytoplankton growth, resulting in decreased surface-water pCO2. In HLNC regions, the supply of iron from deep water, while an important source, is generally insufficient to meet the requirements of phytoplankton. An important additional supply of iron to surface waters far removed from sediment and riverine sources is aeolian transport and deposition (Duce and Tindale, 1991; Fung et al., 2000; Martin, 1990). This aeolian supply of iron may limit primary production in HNLC regions, although the effect is ultimately constrained by the availability of nitrate and phosphate. Iron has been hypothesised to play an indirect role over longer time-scales (e.g., glacial-interglacial) through limitation of oceanic nitrogen fixation and, consequently, the oceanic content of nitrate (Falkowski et al., 1998; Broecker and Henderson, 1998; Box 3.4). The regional variability of oceanic nitrogen fixation (Gruber and Sarmiento, 1997) and its temporal variability and potential climate-sensitivity have recently become apparent based on results from long time-series and global surveys (Karl et al., 1997; Hansell and Feely, 2000).

Carbon (organic and inorganic) derived from land also enters the ocean via rivers as well as to some extent via groundwater. This transport comprises a natural carbon transport together with a significant anthropogenic perturbation. The global natural transport from rivers to the ocean is about 0.8 PgC/yr, half of which is organic and half inorganic (Meybeck 1982, 1993; Sarmiento and Sundquist 1992; Figure 3.1). Additional fluxes due to human activity have been estimated (Meybeck, 1993) to be about 0.1 PgC/yr (mainly organic carbon). Much of the organic carbon is deposited and/or respired and outgassed close to land, mostly within estuaries (Smith and Hollibaugh, 1993).The outgassing of anthropogenic carbon from estuaries can be a significant term in comparison with regional CO2 emissions estimates (e.g., 5 to 10% for Western Europe; Frankignoulle et al., 1998). The natural DIC transport via rivers, however, is part of a large-scale cycling of carbon between the open ocean and land associated with dissolution and precipitation of carbonate minerals. This natural cycle drives net outgassing from the ocean of the order 0.6 PgC/yr globally, which should be included in any assessment of net air-sea and atmosphere-terrestrial biosphere transfers (Sarmiento and Sundquist, 1992) and ocean transports (e.g., Holfort et al., 1998).



Other reports in this collection