4 Seasonal Variability of the Oceanic CO2 System and Air-Sea Flux of CO2 at

the BATS site in the Sargasso Sea

Nicholas R. Bates

Bermuda Biological Station for Research

Ferry Reach GE-01, Bermuda

There is a great interest in understanding the exchanges of carbon dioxide between ocean and atmosphere because of the predicted "enhanced greenhouse effect" on climate. At present, approximately 40% of the CO2 added to the atmosphere through fossil fuel combustion is thought to be absorbed by the oceans, although the magnitude of the global oceanic sink is difficult to quantify due to the relatively sparse data sets and poor seasonal coverage of large areas of the ocean (Tans et al., 1990; Siegenthaler and Sarmiento, 1993). Reducing the uncertainties in the estimates of the global and regional CO2 fluxes are thus important future goals of U.S. JGOFS.

The North Atlantic oceanic basin has a significant role in the global carbon cycle as a large sink for atmospheric CO2, from the high-latitude regions and sites of mode water formation to the subtropical oligotrophic ocean (Tans et al., 1990; Takahashi et al., 1993). In the gyre system between >15N and 50N, for example, the North Atlantic is thought to be a five fold larger sink for atmospheric CO2 (~0.30 GT C yr-1) compared to the North Atlantic basin, like global estimates of the oceanic sink, are poorly constrained due to the relative lack of seasonal surface pCO2 and pCO2 data (where pCO2 is the difference in partial pressures of CO2 of surface seawater and overlying air). The flux strength of CO2 in the mid-latitude North Atlantic gyre system is poorly known. Earlier estimates of CO2 fluxes in the Sargasso Sea around Bermuda from limited transect data suggested that the area was a source for CO2 (Tans et al., 1990; Takahashi, pers. comm., 1994). In contrast, recent observations at the U.S. JGOFS Bermuda Atlantic Time-series Study site (BATS) indicate that the Sargasso Sea is a significant sink for CO2 (greater than 1 moles CO2 m2 yr-1; Bates et al., 1993, 1994a, c) and that the gyre region may have a larger impact on the basin scale North Atlantic CO2 sink strength and global carbon flux estimates. The observations made at BATS offer significant improvements to our understanding of seasonal variability of the inorganic carbon cycle, the air-sea exchange of CO2 and the physical and biogeochemical processes operating in the North Atlantic oligotrophic ocean.

Monthly measurements at BATS indicate that there is an annual cycle in upper ocean hydrography and biological properties, comprised of deep convective mixing in winter and early spring, a pronounced spring bloom, followed by periods of strong thermal stratification in summer and fall (Michaels et al., 1994). Annual changes for the various components of the carbon dioxide system in the Sargasso Sea are significant and associated with distinct physical and biological processes (Bates et al., 1993, 1994a, b, c). The seasonal variability of total carbon dioxide (CT) is about 50 moles kg-1 and physical processes appear to modulate a significant proportion of this variability. Each winter, convective mixing entrains both nutrients and CO2-rich waters from the thermocline, with surface CT reaching a peak of approximately 2060-2065 moles kg-1 between March and April. Interannual variability of the spring CT maximum and spring bloom primary production appears to be related to the strength of winter convective mixing and the subsequent entrainment of nutrients and high concentration CO2 water into the mixed layer. Seasonal changes in alkalinity (TA) of 20 moles kg-1 are primarily associated with changes in salinity and those physical factors that influence salinity (i.e. water mass movements, mixing, evaporation and precipitation). Upper ocean pCO2 variability of approximately 90 to 100

Figure 4.1. Plot of surface ocean PCO2 at BATS against atmospheric pCO2 (low amplitude wave) measured on Bermuda (courtesy of T. Conway, NOAA). The small peak in PCO2 at the beginning of 1994 is a response to coccolithophore calcification.

atm (similar to the range observed at the NABE site) and changes in DpCO2 between -45 atm and +50 atm (Figure 4.1) appear to be controlled primarily by the annual cycle of temperature. During the fall pCO2 values decrease along a thermodynamic gradient as the mixed layer cools (~4.2% change in pCO2 per C temperature change; Figure 4.2). Interannual variability of the summer pCO2 maximum also appear to be ameliorated to a certain extent by the varying intensity and duration of rain inputs to the shallow, stratified mixed layer.

What remains enigmatic is the role of biogeochemical processes in controlling the upper ocean carbon dioxide system and air-sea transfer of CO2. Two important processes are evident at the BATS site: (1) spring-summer depletion of CT due to non-Redfield biogeochemical processes and (2) coccolithophorid calcification.

(1) Spring-summer depletion of CT

At BATS, during the spring to summer period, mixed layer CT concentrations decrease by about 35 to 40 moles kg-1 (or ~25 moles of salinity normalized CT), coincident with production of O2 (Bates et al. 1994b, c). From April to December, there is a very consistent decline in the integrated stock (from 0 to 150m) of CT, an average drawdown of ~20 moles kg-1 or ~2.5 to 3.0 moles C m2yr-1 (Jenkins and Goldman, 1985; Michaels et al., 1994) while sediment trap fluxes average ~0.45 moles C m2yr-1. Thus the magnitude of the seasonal change in CT is in good agreement with previous estimates of new production at this site. Similar decreases in CT concentrations of between 40 and 70 moles kg-1 occur in the high-latitude North Atlantic at NABE and subpolar sites during the spring bloom (Takahashi et al., 1993). These CT high-latitude drawdowns are coincident with nutrient depletion and associated with photosynthetic fixation of CO2. However, at BATS, the seasonal CT drawdown occurs in the absence of measurable nitrate (less than 0.05 kg-1) and is an extreme example of the non-Redfield ratio CT drawdowns observed in the North Atlantic (see Sambrotto et al., 1993).

Figure 4.2. Plot of mixed layer PCO2 each cruise during 1994 shows seasonal cycle of PCO2 against temperature. Fall decrease follows thermodynamic gradient.

If the influence of temperature on PCO2 is removed by detrending the data (i.e. recalculating PCO2 at a constant temperature of 18°C while using in situ CT and TA values), the biological modulation of the annual cycle become evident (i.e. compare in situ pCO2, Figure 4.1; temperature detrended pCO2, Figure 4.3). A rapid decrease in PCO2 occurs between April and July, coincident with the decrease in mixed layer CT. Thus, the process controlling the depletion of CT has a direct impact on the air-sea flux of CO2. Reduced spring-summer utilization of CT leads to higher summer pCO2 and a lower annual net CO2 sink, whereas increased utilization of CT results in a larger CO2 sink.

(2) Coccolithophorid calcification

In February 1992, during the annual spring bloom, there was a significant non-conservative depletion of alkalinity (10 to 30 moles kg-1) and CT (5 to 12 moles kg-1) in the mixed layer (~180 m), which has been attributed to the effects of coccolithophore calcification (Bates et al., 1994a). Mixed layer PCO2 values were significantly higher by 20-30 atm compared to normal winter conditions, reaching a value of 340 to 355 atm in the surface 20 m and at the chlorophyll maximum. As a result the pCO2 gradient was reduced by about 25 to 30 atm compared to the preceeding and following months when mixed layer PCO2 values ranged from 320 to 325 atm. Although the perturbation was short-lived, the biogeochemical signature of coccolithophorid calcification significantly changed the annual air-sea flux of CO2, reducing the oceanic sink of CO2 by approximately 25% (Bates et al., 1992, 1994a).

Figure 4.3. Plot of corrected surface PCO2 at BATS beginning on 1st January 1991 (units are in atm.). PCO2 values are calculated at constant 19°C. alternates between winter values of ~300 atm. and summer values of ~270 atm. Arrows denote biological influences; spring-summer depletion and coccolithophore calcification.

In summary, the annual cycle of the inorganic carbon system and air-sea flux of CO2 observed at BATS appears to be primary controlled by physical processes such as temperature and convective mixing. However, coccolithophorid calcification and the spring-summer non-Redfield depletion of CT are clearly important processes that significantly influence the exchange of CO2 between atmosphere and ocean in oligotrophic regions. These complex biogeochemical mechanisms may provide a balance between whether the Sargasso Sea is a source or a sink for atmospheric CO2.