Every year, the oceans, which contain about 50 times more carbon (39 000 PgC) than the atmosphere, take up about 25% of anthropogenically produced carbon (IPCC, 2007). Net CO2 transfer across the air–sea interface occurs whenever a partial pressure difference is established between atmosphere and ocean. The net flux (F) of CO2 across the air–sea interface is given by the product of the differences in partial CO2 pressure and a kinetic term (k), known as the transfer velocity (it has dimensions of length per time):where s is the CO2 solubility coefficient and pCO2w and pCO2a are the partial pressures of CO2 in water and air, respectively. Oceanic sinks and sources of CO2 are then highly variable and sensitive to climate and local environmental conditions. Besides temperature and biological processes, fluxes are driven by several different processes, such as near-surface turbulence at moderate wind speed (Zappa et al., 2007) and the formation of air bubbles (Woolf, 1997). However, wind is the primary forcing of the aqueous boundary layer which controls gas exchange. Quantitative relationships between gas exchange and wind speed have been suggested by several authors (Liss and Merlivat, 1986; Wanninkhof, 1992; Wanninkhof and McGillis, 1999). Air–sea gas exchange models and theory were exhaustively reviewed by Nightingale (2009), who discussed a wide range of processes from chemical biological enhancement, the surface renewal model, the surface penetration theory, the turbulence dissipation rate, the bubble-mediated gas transfer, the role of wave breaking, and so on.
During the last two decades, several authors have shown that the Southern Oceans and, in particular, the Circumpolar Antarctic and sub-Antarctic zones behave not only as sinks but also as sources of CO2, according to changing environmental conditions (Takahashi et al., 1993, 1997, 2002, 2009; Hoppema et al., 1995, 1999, 2000; Bakker et al., 1997; Metzl et al., 1999; De Baar and Boyd, 2000; Stoll et al., 2002). In these authors’ views, temperature is a very important factor affecting the air–sea exchange, although plankton blooms may sometimes be even more important. Roy et al. (2003) and Metzl et al. (2006) disputed the magnitude of the CO2 sink in that area and Le Quéré et al. (2007) estimated a weakening of the Southern Ocean sink between 1981 and 2004, attributing it to the observed increase in Southern Ocean winds in that period; Zickfeld et al. (2008) disagreed with these conclusions, and stated that, ‘Ekman-driven upwelling in the Antarctic Divergence exposes deep waters with elevated concentrations of dissolved inorganic carbon (DIC) to the surface, leading to outgassing of natural CO2 to the atmosphere’.
With the only exception of three carbon isotope measurements reported by Inoue and Sugimura (1986), direct measurements of atmospheric CO2 concentrations and its carbon isotope values in the area between New Zealand and Antarctica had never been reported until a few years ago, when Longinelli et al. (2007) reported the CO2 concentrations, measured in that area in the course of three different cruises, together with the carbon isotope composition of atmospheric CO2 from discrete air samples collected in 4L Pyrex flasks during four cruises. These results showed ‘several δ13C values of atmospheric CO2 considerably lighter than the background values measured over the oceans’. These quite negative values, ranging from about −8.5 to −10.3 ‰ (vs. Vienna Pee-Dee-Belemnite VPDB standard), were tentatively related to ‘partial contributions by different causes such as the southern negative gradient of δ13CDIC, the upwelling of deep waters and their northward transport, subsurface mixing processes and, finally, the possible contribution of biological heterotrophic activity and the production of isotopically light CO2 by remineralization processes of decomposing planktonic organic matter’ (Longinelli et al., 2007). Conclusions by several authors about the possibility of CO2 flowing from the ocean surface to the atmosphere in sections of the circumpolar Antarctic area agree with those by Longinelli et al. (2007) and with the reported δ13C values. It is quite difficult to state the precise relationships existing between the carbon isotope values of atmospheric CO2 and environmental conditions, which should be accurately monitored by research vessels. This was even more difficult in our case, since flask air samples represent instantaneous atmospheric conditions and were collected on board a cargo ship, where the lack of appropriate instrumentation prevented the collection of other data which would have been important in interpreting isotopic data. We report here atmospheric CO2 concentrations and δ13C values obtained from discrete air samples collected during four further cruises between New Zealand and Antarctica in 2004/2005, 2005/2006, 2009 and 2010, and compare them with previous data.
A Siemens Ultramat 5E analyser assembled for shipboard use (Ori et al., 1996; Lenaz et al., 2000; Longinelli et al., 2001, 2005, 2007) was used to record CO2 concentrations in the ocean atmosphere. The measurement system was based on the comparison between signals from two infrared absorbing cells, one filled with a flux of synthetic air with a constant CO2 concentration and the other filled with the air sample. The air was carefully dried by means of a cold trap (t<−50°C). The instrument was calibrated using the WMO-X85 scale by means of various working standards with concentrations of 345, 375, 378 and 435 ppmv carefully determined at the Monte Cimone Observatory (Sestola, Italy), a baseline site of the Italian Meteorological Service. Air samples were collected in 4L Pyrex flasks, evacuated to better than 5×10−3 mmHg. They were opened on the windward side of the ship at about 15 m above sea level, closed with greaseless high vacuum stopcocks and then stored in wooden boxes until the ship's return to Italy, 3–4 months later. Treatment of air samples for quantitative separation of CO2 was carried out in the laboratory. As reported in previous papers (Longinelli et al., 2007, 2010), the air samples from the flasks were pumped in a high vacuum line, at a rate of about 8–9 mL min−1, through a Swagelok microvalve and a liquid nitrogen–cooled spiral trap. When a final pressure lower than 5×10−3 mmHg was reached, the trap was heated at about −80°C by means of an ethyl alcohol/liquid nitrogen slush and the evolved CO2 was transferred in a sample tube and measured in a Finnigan Delta S mass spectrometer.
The isotopic values are reported in terms of delta units. Delta (δ) is defined by the relationship: δ=[Rsample−Rstand./Rstand.]×103, where R =13C/12C or 18O/16O
Measurements were carried out against a standard CO2 obtained from a very pure Carrara marble, the isotopic composition of which is −2.43 ‰ (δ18O vs.VPDB) and +2.45 ‰ (δ13C vs. VPDB), calibrated periodically versus NBS-19 and NBS-20. The values used for this calibration are: NBS-19, −2.20 (δ18O) and +1.95 (δ13C); NBS-20, −4.14 (δ18O) and −1.06 (δ13C).1
11NBS 20 standard material has not been available for a very long time. We had reasonable amounts of NBS 19 and NBS 20 inherited from various research centres where the senior author has worked in the past. All the NBS 20 has been consumed and we now face the serious problem of finding a reliable reference material to calibrate our laboratory standard. The existing LSVEC is isotopically too far from our Carrara laboratory standard, both in oxygen and carbon, and is consequently unreliable for calibration purposes.
All carbon isotope data are reported versus the VPDB standard. The accuracy of our procedure was tested, before running each set of flasks, on several flask air samples collected simultaneously outside the laboratory and treated with the same technique. The standard deviation of the carbon isotope values obtained from these samples in different years ranged from ±0.02 ‰ to ±0.04 ‰ (1σ). We believe that the reproducibility of the samples collected over the ocean is not far from these values. Although we are perfectly aware that this reproducibility is quite poor when compared, for instance, with that obtained by NOAA on land-based samples, lack of funding for this research prevented us from obtaining more accurate (and more expensive) extraction lines.
Carbon isotope results obtained from mass spectrometer analyses were corrected for the N2O effect according to the Mook and Van der Hoek (1983) equation:
where δ13C is the true isotopic value of the CO2 sample, δ13 is the measured isotopic value, calculated as if N2O were absent, ρ is the N2O/CO2 concentration ratio in the atmosphere and E is the ratio between the ionisation efficiencies for N2O and CO2 in the mass spectrometer. N2O concentrations along the ship's route could not be directly measured, as the equipment on board the M/V Italica, was totally inadequate. Consequently, to correct δ13C measured on CO2 from flask air samples, we calculated ρ using the mean value of N2O and CO2 concentrations reported by NOAA for the Cape Green (Tasmania) and South Pole stations during the same periods of our sample collection. This is obviously an oversimplification, but it was the only way of evaluating an average correction of the measured δ13C values. According to NOAA data, ρ values at the time of the 2004/2005, 2005/2006 and 2009 and 2010 cruises varied between 0.836×10−3 and 0.847×10−3. E was calculated as 0.6161 (as determined by Finnigan engineers on an instrument identical to our mass spectrometer). The value of the right member of eq. (1) may thus vary from 0.1735 to 0.1821, so that the possible overall error in the corrected δ13C value is, at most, in the hundredths of a unit. As the error of measurement is probably at least two times larger, to standardise the corrections, the mean value of 0.178 was used for all flask air samples (Tables 1–4). The correction value has not changed substantially during the last few years (Longinelli et al., 2005) because of the simultaneous increase of both N2O and CO2 concentrations and the use of the same mass spectrometer for isotopic measurements.
The carbon isotope composition of CO2 from flask air samples collected during the first four cruises from New Zealand to Antarctica (1998–1999, 2001–2002, 2002–2003 and 2003–2004) were reported graphically by Longinelli et al. (2007) (Fig. 1). Samples were always collected during the same period of the year, between late December and early January along similar routes (Fig. 2). The results show a rather large variability, an increasing frequency over time of isotopically light values and the lightest values concentrated in a latitudinal belt of about 15° across the Polar Front.
The 2004–2005 air samples yielded δ13C values which were totally different from those obtained from previous cruises. Twenty-four flask air samples were collected between 31 December 2004 and 6 January 2005, from about 44° S to about 71° S. The numerical results are listed in Table 1 and shown in Fig. 3a. Their saw-toothed distribution is clearly inexplicable and almost unbelievable. Systematic errors of measurement, although highly improbable, could hardly have produced the saw-toothed distribution of Fig. 3a for which we can find no acceptable explanation. During this cruise, no direct measurements of atmospheric CO2 concentrations could be carried out. To have at least a rough idea of the CO2 concentrations in each sample, we tried to calculate the amount of CO2 from the gas pressure of each sample in the mass spectrometer source, after calibration with known CO2 volumes. The method is obviously not quantitative, and the results obtained were quite surprising. Not only did the CO2 concentrations repeat the saw-toothed distribution of the δ13C values but also, strangely enough, higher CO2 concentrations corresponded to heavier δ13C values and vice-versa. The measured δ13C values and calculated CO2 concentrations are shown graphically in Fig. 3b. We believe that this self-contradictory distribution of values can hardly result from errors in measuring the two variables; however, we are unable to find a reliable interpretation for these puzzling data.
During the 2005–2006 cruise, 21 flask air samples were collected between 1 January 2006 and 6 January 2006: two of them south of New Zealand's South Island and 19 between 49.56.00 S and 68.03.11 S. The δ13C measurements of CO2 from the flask air samples are listed in Table 2 and shown in Fig. 4. The two results from the samples collected not far from the southern coast of New Zealand are close to −8.5 ‰ whereas almost all the others are more negative, four of them showing the lightest values ever measured between New Zealand and Antarctica. Sample 31, from 54.59.45 S, has a δ13C value of −11.04 ‰; sample 33, from 56.05.70 S, has a δ13C of −10.08 ‰; sample 37, from 60.01.00 S, shows −13.45 ‰ and sample 44, from 67.26.00 S, has −10.43 ‰. After the interval of the inexplicable results of the 2004–2005 cruise, these data seem to represent an evolution of the situation found during the previous 2002–2003 and 2003–2004 cruises, with more frequent and more negative isotopic events recorded. With the only exception of sample 29 (higher CO2 concentration and more negative δ13C), no major changes in CO2 concentrations were detected as quantitatively related to the measured changes in the δ13C values. This may be due to the instrumental method of measurement, since the Siemens Ultramat 5E analyser integrates the values measured every 2 seconds every 10 minutes, so that large transient changes may be practically obliterated by the integration process. The hypothesis of contamination from the ship's exhaust had already been discussed by Longinelli et al. (2007) and ruled out for several reasons. Consequently, these results may be essentially related to small-scale, positive air–sea fluxes with various causes, such as upwelling of deep waters with elevated concentrations of dissolved inorganic carbon, the possible contribution of biological heterotrophic activity, production of isotopically light CO2 by remineralization of decomposing Particulate Organic Matter (POM), and so on. As already reported in the introduction, many authors agree about the possibility of outgassing events of CO2 from sea-water to the atmosphere in the circumpolar areas; the statement by Zickfeld et al. (2008) on the possible contribution to the atmosphere of CO2 from upwelling deep waters being particularly interesting.
However, some of the 2005–2006 results are puzzling, such as the rather negative values obtained while the ship was sailing near the edge of the Campbell Plateau, at latitudes between about 50° and 52° S (samples 27, 28 and 29: Table 2 and Fig. 4), where the ocean depth is between 500 and 1000 m.
The δ13C results obtained from 1998 to 2006, shown in Fig. 5, seem to reinforce the hypotheses already suggested by Longinelli et al. (2007), with the only exception of the anomalous 2004–2005 data. After 2006, we were unable to obtain berths on the M/V Italica for 3 yr, and collection of discrete air samples and measurement of CO2 concentrations between New Zealand and Antarctica were compulsorily discontinued. We lack information which might have shown some connection between the results of 2006 and those of 2009, when we were again able to develop our work on board the Italica and repeat air sample collections twice (20–24 December 2009 and 25–29 January 2010), taking advantage of a double crossing from New Zealand to Antarctica. During the former voyage, 10 flask air samples were collected from 50.18.82 S to 70.08.05 S, and 20 samples during the latter one, from 45.03.30 S to 69.28.50 S. CO2 concentrations were recorded during both cruises, so that we can now compare the two variables along the ship's course. These flask air samples were treated according to the usual procedure, and the results are shown in Tables 3 and 4, respectively. To our surprise, the air samples collected during these two cruises showed extremely homogeneous carbon isotopic compositions: −8.69 ± 0.07 ‰ for the 10 samples in 2009 and −8.44 ± 0.12 ‰ for the 20 samples in 2010. The mean value of the 2010 samples was close to the background values of recent years which ranged from about −8.0 to about −8.3 in the decade 2002–2012 (values measured at the NOAA stations of Mahè Island, Indian Ocean, Cape Grim, Tasmania, Crozet Island, Southern Indian Ocean and the Japanese Syowa station, the Antarctica coast, from the beginning to the end of that decade).
The isotopic results are shown in Figs. 6 and 7, and in this case the homogeneity of the results is even more striking, in sharp contrast with both the variability of isotopic results found during previous cruises and the expected tendency to a progressive increase in negative events and an evolution of isotopic values towards 13C-depleted values. In both cases, the CO2 concentrations show very homogeneous and very low values. When we calculate their mean value measured in 2006 (377.5 ppmv) and that for the 2010 cruise (383.2 ppmv), the calculated growth rate is 1.4 ppmv y−1. This is lower than the values calculated at the two closest land-based stations: the NOAA Cape Grim station (Tasmania) close to 1.7 ppmv y−1 and the Japanese Syowa station (North Antarctic coast), also close to 1.7 ppmv y−1.
The δ13C from the air samples collected over five expeditions (Fig. 5) reasonably match the conclusions of Takahashi et al. (2009), who located a large ‘positive’ area between New Zealand and Antarctica (between about 55° and 70° S) which is to be considered a probable source of CO2. As already suggested by Longinelli et al. (2007), the contribution to the atmosphere of isotopically light CO2 may have several causes. Upwelling of deep waters particularly rich in isotopically light-dissolved CO2 normally occurs at relatively high latitudes, but the upwelled water may then be transported north by Ekman transport and, after mixing with Antarctic Surface Water, partially liberate its CO2 to the atmosphere, thus modifying the background δ13C, particularly in the lower layers of the atmosphere where our flask samples were collected. The carbon isotope composition of the CO2 dissolved in deep waters is largely derived by remineralization of POM and may be extremely negative, the δ13C of marine POM reaching values as low as −16 to −35 ‰ (VPDB) (Fontugne and Duplessy, 1981; Fontugne et al., 1991; François et al., 1993; Goericke and Fry, 1994; Kennedy and Robertson, 1995; Bentaleb et al., 1998; Lourey et al., 2004).
Near-surface turbulence and wave breaking may enhance outgassing. The possibility of a CO2 increase in the surface layer due to the influx of CO2 related to eddy-diffusive influx from underlying water layers was already discussed by Stoll et al. (2002). At high latitudes, the recorded number of negative isotopic events tends to decrease and CO2 isotopic values tend to increase, while surface waters at high latitudes are generally characterised by high concentrations of CO2 with low δ13C values because of the contribution of upwelling water masses. The decrease in isotopically light events may be due to surface water temperatures which, south of the Polar Front, are generally close to or even below zero (Craig et al., 1981; Rintoul et al., 1997; Russo et al., 1999). These temperature conditions may prevent, or at least drastically reduce, positive air–sea fluxes because of the increasing solubility of CO2 at low temperatures.
As previously mentioned, the results from the 2004 to 2005 cruise (Table 1, Fig. 3) do not allow any reasonable interpretation, whereas the 2009 and 2010 results indicate environmental conditions which exclude any positive contribution of CO2 to the atmosphere from the sea surface. During these two last cruises, technical problems prevented the recording of wind speed and direction. However, in the unanimous opinion of the ship's crew, these cruises, particularly that of 2010, were particularly quiet, with almost calm seas and low wind, environmental conditions which may have contributed to the results obtained.
The lack of sampling from 2006 to 2009 does not help us to understand whether the results obtained during 2009 and 2010 cruises are due to substantial annulment of the processes which caused the negative episodes recorded in previous years or due to fortuitous conditions of the ocean and atmospheric dynamics. As already noted, the impossibility of using an oceanographic ship for these cruises prevented us from collecting data on many other variables, such as vertical profiles of seawater temperature, seawater–atmosphere temperature gradients, concentration of dissolved CO2 in surface water, biological autotrophic versus heterotrophic activity in surface waters, and so on, which may have been of importance to relate the results to definite environmental conditions.
Discrete air samples collected during eight cruises between New Zealand and Antarctica during austral summers between 1998 and 2010 revealed a large variety of δ13C values of atmospheric CO2. Samples from five cruises (1998–1999, 2001–2002, 2002–2003, 2003–2004 and 2005–2006) showed increasing episodes of isotopically light CO2 in the lower atmosphere, δ13C values ranging from slightly lighter than background values (−8.0 to about −8.4 ‰) down to −13.45 ‰. The lack of direct measurements of several variables of environmental conditions prevented us from suggesting precise causes for these episodes. Isotopic results from the 2004 to 2005 cruise are a sort of improbable puzzle; they could hardly have been caused by systematic errors of measurement but are definitely beyond any reasonable interpretation. The data from the last two sets of samples (December 2009 and January 2010) greatly differ from the results from previous cruises, showing absolutely flat δ13C values, paralleled by flat CO2 concentrations. We hope that further air sampling in the same area (samples have been collected in January 2012) will show whether they should be viewed as constituting only a fortuitous episode or as the beginning of a substantially different trend possibly caused by changes in the oceanic circulation related to changes in the climatic conditions.
Thanks are due to the crew of the M/V Italica for their continuous help and collaboration and to two anonymous reviewers for their comments and suggestions. This study is contribution No. 1773 of the ISMAR-CNR, Bologna branch-marine geology.
Bakker D. C. E , de Baar H. J , Bathman U. V . Changes of carbon dioxide in surface waters during spring in southern ocean . Deep-Sea Res. II . 1997 ; 44 : 91 – 127 . https://doi.org/10.3402/tellusb.v64i0.17472 .
Bentaleb I , Fontugne M , Descolas-Gros C , Girardin C , Mariotti A , co-authors . Carbon isotopic fractionation by phytoplancton in the Southern Indian Ocean: relationship between δ13C of particulate organic carbon and dissolved carbon dioxide . J. Mar. Syst . 1998 ; 17 : 39 – 58 . https://doi.org/10.3402/tellusb.v64i0.17472 .
De Baar , H. J. W , Boyd , P. W . 2000 . The role of iron in plankton ecology and carbon dioxide transfer of the global oceans . In: The Dynamic Ocean Carbon Cycle: a Midterm Synthesis of the Joint Global Ocean Flux Study . R. B. Hanson , H. W. Ducklow and J. G. Field 5 : IGBP Book Series Cambridge University Press : Cambridge , pp. 61 – 140 .
Fontugne M. R , Descolas-Gros C , de Billy G . The dynamics of CO2 fixation in the Southern Ocean as indicated by carboxylase activities and organic carbon isotopic ratios . Mar. Chem . 1991 ; 35 : 371 – 380 . https://doi.org/10.3402/tellusb.v64i0.17472 .
François R , Altabet M. A , Goericke R , McCorckle D. C , Brunet C , Poisson A . Changes in the δ13C of surface water particulate organic matter across the subtropical convergence in the SW Indian Ocean . Global Biogeochem. Cy . 1993 ; 7 ( 3 ): 627 – 644 . https://doi.org/10.3402/tellusb.v64i0.17472 .
Goericke R , Fry B . Variations of marine plankton δ13C with latitude, temperature, and dissolved CO2 in the world ocean . Global Biogeochem. Cy . 1994 ; 8 ( 1 ): 85 – 90 . https://doi.org/10.3402/tellusb.v64i0.17472 .
Hoppema M , Fahrbach E , Schroeder M , Wisotzki A , de Baar H. J. W . Winter-summer differences of carbon dioxide and oxygen in the Weddell Sea surface layers . Mar. Chem . 1995 ; 51 : 177 – 192 . https://doi.org/10.3402/tellusb.v64i0.17472 .
Hoppema M , Fahrbach E , Stoll M. H. C , de Baar H. J. W . Annual uptake of atmospheric CO2 by the Weddell Sea derived from a surface balance, including estimations of entrainment and new production . J. Mar. Syst . 1999 ; 19 : 219 – 233 . https://doi.org/10.3402/tellusb.v64i0.17472 .
Hoppema M , Stoll M. H. C , de Baar H. J. W . CO2 in the Weddell Gyre and Antarctic circumpolar current: austral autumn and early winter . Mar. Chem . 2000 ; 72 : 203 – 220 . https://doi.org/10.3402/tellusb.v64i0.17472 .
IPCC ., 2007 . Climate change 2007: the physical science basis . In: Contribution of Working Group I to the Fourth Assessment Report of Intergovernmental Panel on Climate Change . S. Solomon , D. Qin , M. Manning , Z. Chen , M. Marquis , et al. . Cambridge University Press : Cambridge , 996 . pp.
Kennedy H , Robertson J . Variations in the isotopic composition of particulate organic carbon in surface waters along an 88°W transect from 67°S to 54°S . Deep-Sea Res. II . 1995 ; 42 : 1109 – 1122 . https://doi.org/10.3402/tellusb.v64i0.17472 .
Lenaz R , Giovanelli G , Ori C , Colombo T , Masarie K . co-authors Shipboard report on atmospheric CO2 concentrations recorded on continuous from Mediterranean Sea to Antarctica . II. Nuovo Cimento . 2000 ; 23C : 507 – 514 .
Le Quéré C , Roedenbeck C , Buitenhuis E. T , Conway T. J , Langenfelds R . co-authors . Saturation of the Southern Ocean CO2 sink due to recent climate change . Science . 2007 ; 316 : 1735 – 1738 . https://doi.org/10.3402/tellusb.v64i0.17472 .
Liss , P. S , Merlivat , L . 1986 . Air-sea gas exchange rates: introduction and synthesis . In: The Role of Air-Sea Gas Exchange in Geochemical Cycling . P. Buat-Menard , Springer : Dordrecht , pp. 113 – 127 .
Longinelli A , Colombo T , Giovanelli G , Lenaz R , Ori C . co-authors . Atmospheric CO2 concentrations and δ13C measurements along a hemispheric course (1998/99), Italy to Antarctica) . Earth Planet. Sci. Lett . 2001 ; 191 : 167 – 172 . https://doi.org/10.3402/tellusb.v64i0.17472 .
Longinelli A , Giglio F , Langone L , Lenaz R , Ori C . co-authors . Atmospheric CO2 concentrations and δ13C values across the Antarctic circumpolar current between New Zealand and Antarctica . Tellus . 2007 ; 59B : 130 – 137 .
Longinelli A , Lenaz R , Ori C , Langone L , Selmo E . co-authors . Decadal changes in atmospheric CO2 concentration and δ13C over two seas and two oceans: Italy to New Zealand . Atmos. Envir . 2010 ; 44 : 4303 – 4311 . https://doi.org/10.3402/tellusb.v64i0.17472 .
Lourey M. J , Trull T. W , Tilbrook B . Sensitivity of δ13C of Southern Ocean suspended and sinking organic matter to temperature, nutrient utilization, and atmospheric CO2 . Deep-Sea Res. I . 2004 ; 51 : 281 – 305 . https://doi.org/10.3402/tellusb.v64i0.17472 .
Metzl N , Brunet C , Jabaud-Jan A , Poisson A , Schauer B . Summer and winter air-sea CO2 fluxes in the Southern Ocean . Deep-Sea Res. I . 2006 ; 53 : 1548 – 1563 . https://doi.org/10.3402/tellusb.v64i0.17472 .
Ori , C , Lenaz , R , Colombo , T , Giovanelli , G . 1996 . Atmospheric CO2 concentration measured continuously from the Mediterranean to the Bellingshausen Sea: technology and methodology . In: Conference Proceedings of the 6th Workshop Ital. Res. Antarc. Atmos., Firenze 1995 . SIF : Bologna , pp. 361 – 367 .
Rintoul S. R , Donguy J. R , Roemmich D. H . Seasonal evolution of upper ocean thermal structure between Tasmania and Antarctica . Deep-Sea Res . 1997 ; 44 ( 7 ): 1185 – 1202 . https://doi.org/10.3402/tellusb.v64i0.17472 .
Russo , A , Artegiani , A , Budillon , G , Paschini , E , Spezie , G . 1999 . Upper occean thermal structure and fronts between New Zealand and the Ross Sea- Austral Summer 1994–1995 and 1995–1996 . In: Oceanography of the Ross Sea Antarctica . G. Spezie and G. M. R. Manzella . Springer-Verlag : ItaliaMilano , pp. 67 – 75 .
Stoll M. H. C , Thomas H , de Baar H. J. W , Zondervan I , De Jong E , co-authors . Biological versus physical processes as drivers of large oscillations of the air-sea flux in the Antarctic marginal ice zone during summer . Deep-Sea Res . 2002 ; 49 : 1651 – 1667 . https://doi.org/10.3402/tellusb.v64i0.17472 .
Takahashi T , Feely R. A , Weiss R. F , Wanninkhof R , Chipman D. W , co-authors A . Global air-sea flux of CO2. An estimate based on measurements of sea-air pCO2 difference . Proceed. Nat. Acad. Sci . 1997 ; 94 : 8292 – 8299 . https://doi.org/10.3402/tellusb.v64i0.17472 .
Takahashi T , Olafsson J , Goddard J. C , Chipman D. W , Sutherland S. C . Seasonal variation of CO2and nutrients in the high-latitude surface oceans: a comparative study . Global Biogeochem. Cy . 1993 ; 7 : 1431 – 1438 . https://doi.org/10.3402/tellusb.v64i0.17472 .
Takahashi T , Sutherland S. C , Sweeney C , Poisson A , Metzl N . co-authors . Global air-sea CO2 flux based on climatological surface ocean pCO2, and seasonal biological and temperature effects . Deep-Sea Res. II . 2002 ; 49 : 1601 – 1622 . https://doi.org/10.3402/tellusb.v64i0.17472 .
Takahashi T , Sutherland S. C , Wanninkhof R , Sweeney C , Feely R. A . co-authors . Climatological mean and decadal change in surface ocean pCO2, and net sea-air CO2 flux over the global oceans . Deep-Sea Res II . 2009 ; 56 : 554 – 577 . https://doi.org/10.3402/tellusb.v64i0.17472 .
Wanninkhof R . Relationship between wind speed and gas exchange over the ocean . J. Geophys. Res . 1992 ; 97 : 7373 – 7382 . https://doi.org/10.3402/tellusb.v64i0.17472 .
Wanninkhof R , McGillis W. R . A cubic relationship between air-sea CO2 exchange and wind speed . Geophys. Res. Lett . 1999 ; 26 : 1889 – 1892 . https://doi.org/10.3402/tellusb.v64i0.17472 .
Zappa , C. J , McGillis , W. R , Raymond , P. A , Edson , J. B , Hintsa , E. J . and co-authors . 2007 . Environmental turbulent mixing controls on air-water gas exchange in marine and aquatic systems . Geophys. Res. Lett . 34 : 6 . DOI: https://doi.org/10.3402/tellusb.v64i0.17472 .
Zickfeld K , Fyfe J. C , Eby M , Weaver A. J . Comment on “Saturation of the Southern Ocean CO2 sink due to recent climate change” . Science . 2008 ; 319 : 570 . https://doi.org/10.3402/tellusb.v64i0.17472 .