• Antonov, J. I., and Coauthors, 2010: Salinity. Vol. 2, World Ocean Atlas 2009, NOAA Atlas NESDIS 69, 44 pp.

  • Arora, V. K., and Coauthors, 2011: Carbon emission limits required to satisfy future representative concentration pathways of greenhouse gases. Geophys. Res. Lett., 38, L05805, doi:10.1029/2010GL046270.

    • Search Google Scholar
    • Export Citation
  • Barbero, L., , J. Boutin, , L. Merlivat, , N. Martin, , T. Takahashi, , S. C. Sutherland, , and R. Wanninkhof, 2011: Importance of water mass formation regions for the air-sea CO2 flux estimate in the Southern Ocean. Global Biogeochem. Cycles, 25, GB1005, doi:10.1029/2010GB003818.

    • Search Google Scholar
    • Export Citation
  • Boé, J., , A. Hall, , and X. Qu, 2009: Deep ocean heat uptake as a major source of spread in transient climate change simulations. Geophys. Res. Lett., 36, L22701, doi:10.1029/2009GL040845.

    • Search Google Scholar
    • Export Citation
  • Böning, C. W., , A. Dispert, , M. Visbeck, , S. R. Rintoul, , and F. U. Schwarzkopf, 2008: The response of the Antarctic Circumpolar Current to recent climate change. Nat. Geosci., 1, 864869.

    • Search Google Scholar
    • Export Citation
  • Bonjean, F., , and G. Lagerloef, 2002: Diagnostic model and analysis of the surface currents in the tropical Pacific Ocean. J. Phys. Oceanogr., 32, 29382954.

    • Search Google Scholar
    • Export Citation
  • Cadule, P., , P. Friedlingstein, , L. Bopp, , S. Sitch, , C. D. Jones, , P. Ciais, , S. L. Piao, , and P. Peylin, 2010: Benchmarking coupled climate-carbon models against long-term atmospheric CO2 measurements. Global Biogeochem. Cycles, 24, GB2016, doi:10.1029/2009GB003556.

    • Search Google Scholar
    • Export Citation
  • Christian, J. R., and Coauthors, 2010: The global carbon cycle in the Canadian Earth System Model (CanESM1): Preindustrial control simulation. J. Geophys. Res., 115, G03014, doi:10.1029/2008JG000920.

    • Search Google Scholar
    • Export Citation
  • Collins, W. J., and Coauthors, 2011: Development and evaluation of an Earth-system model—HadGEM2. Geosci. Model Dev., 4, 10511077.

  • Conway, T. J., , P. P. Tans, , L. S. Waterman, , K. W. Thoning, , D. R. Kitzis, , K. A. Masarie, , and N. Zhang, 1994: Evidence of interannual variability of the carbon cycle from the NOAA/CMDL global air sampling network. J. Geophys. Res., 99, 22 83122 855.

    • Search Google Scholar
    • Export Citation
  • Cox, P., , P. Friedlingstein, , and P. Rayner, 2002: Modelling climate-carbon cycle feedbacks: A cross disciplinary collaboration priority. IGBP Global Change Newsletter, No. 49, IGBP Secretariat, Stockholm, Sweden, 1214.

    • Search Google Scholar
    • Export Citation
  • Dohan, K., , and N. Maximenko, 2010: Monitoring ocean currents with satellite sensors. Oceanography, 23, 94103.

  • Dohan, K., and Coauthors, 2010: Measuring the global ocean surface circulation with satellite and in situ observations. Proc. OceanObs’09: Sustained Ocean Observations and Information for Society, Venice, Italy, European Space Agency, WPP-306.

  • Downes, S. M., , A. Gnanadesikan, , S. M. Griffies, , and J. L. Sarmiento, 2011: Water mass exchange in the Southern Ocean in coupled climate models. J. Phys. Oceanogr., 41, 17561771.

    • Search Google Scholar
    • Export Citation
  • Dunne, J. P., and Coauthors, 2012: GFDLs ESM2 global coupled climate–carbon Earth system models. Part I: Physical formulation and baseline simulation characteristics. J. Climate, 25, 66466665.

    • Search Google Scholar
    • Export Citation
  • Dunne, J. P., and Coauthors, 2013: GFDL’s ESM2 global coupled climate–carbon Earth system models. Part II: Carbon system formulation and baseline simulation characteristics. J. Climate, 26, 22472267.

    • Search Google Scholar
    • Export Citation
  • Ferrari, R., , S. Griffies, , A. Nurser, , and G. Vallis, 2010: A boundary value problem for the parameterized mesoscale eddy transport. Ocean Modell., 32, 143156.

    • Search Google Scholar
    • Export Citation
  • Friedlingstein, P., and Coauthors, 2006: Climate–carbon cycle feedback analysis: Results from the C4MIP model intercomparison. J. Climate, 19, 33373353.

    • Search Google Scholar
    • Export Citation
  • Fung, I., , P. Rayner, , and P. Friedlingstein, 2000: Full-form earth system models: Coupled carbon-climate interaction experiment (the flying leap). IGBP Global Change Newsletter, No. 41, IGBP Secretariat, Stockholm, Sweden, 78.

    • Search Google Scholar
    • Export Citation
  • Geider, R. J., , H. L. Maclntyre, , and T. M. Kana, 1998: A dynamic regulatory model of phytoplanktonic acclimation to light, nutrients, and temperature. Limnol. Oceanogr., 43, 679694.

    • Search Google Scholar
    • Export Citation
  • Gent, P. R., , and J. C. McWilliams, 1990: Isopycnal mixing in ocean general circulation models. J. Phys. Oceanogr., 20, 150155.

  • Gille, S. T., 2008: Decadal-scale temperature trends in the Southern Hemisphere ocean. J. Climate, 21, 47494765.

  • Gnanadesikan, A., and Coauthors, 2006: GFDL’s CM2 global coupled climate models. Part II: The baseline ocean simulation. J. Climate, 19, 675697.

    • Search Google Scholar
    • Export Citation
  • Griffies, S. M., 1998: The Gent–McWilliams skew flux. J. Phys. Oceanogr., 28, 831841.

  • Holte, J., , and L. Talley, 2009: A new algorithm for finding mixed layer depths with applications to Argo data and Subantarctic Mode Water formation. J. Atmos. Oceanic Technol., 26, 19201939.

    • Search Google Scholar
    • Export Citation
  • Holte, J., , J. Gilson, , L. Talley, , and D. Roemmich, cited 2010: Argo mixed layers. Scripps Institution of Oceanography/UCSD. [Available online at http://mixedlayer.ucsd.edu/.]

  • Ilyina, T., , K. D. Six, , J. Segschneider, , E. Maier-Reimer, , H. Li, , and I. Nùñez-Riboni, 2013: Global ocean biogeochemistry model HAMOCC: Model architecture and performance as component of the MPI-Earth system model in different CMIP5 experimental realizations. J. Adv. Model. Earth Syst.,5,287315, doi:10.1029/2012MS000178.

  • Ito, T., , M. Woloszyn, , and M. Mazloff, 2009: Anthropogenic carbon dioxide transport in the Southern Ocean driven by Ekman flow. Nature, 463, 8083.

    • Search Google Scholar
    • Export Citation
  • Iudicone, D., , K. B. Rodgers, , I. Stendardo, , O. Aumont, , G. Madec, , L. Bopp, , O. Mangoni, , and M. Ribera d’Alcala, 2011: Water masses as a unifying framework for understanding the Southern Ocean carbon cycle. Biogeosciences, 8, 10311052.

    • Search Google Scholar
    • Export Citation
  • Jiang, C., , S. T. Gille, , J. Sprintall, , K. Yoshimura, , and M. Kanamitsu, 2012: Spatial variation in turbulent heat fluxes in Drake Passage. J. Climate, 25, 14701488.

    • Search Google Scholar
    • Export Citation
  • Johns, T. C., and Coauthors, 2006: The new Hadley Centre climate model (HadGEM1): Evaluation of coupled simulations. J. Climate, 19, 13271353.

    • Search Google Scholar
    • Export Citation
  • Jones, C. D., and Coauthors, 2011a: The HadGEM2-ES implementation of CMIP5 centennial simulations. Geosci. Model Dev., 4, 543570.

  • Jones, D. C., , T. Ito, , and N. S. Lovenduski, 2011b: The transient response of the Southern Ocean pycnocline to changing atmospheric winds. Geophys. Res. Lett.,38,L15604, doi:10.1029/2011GL048145.

  • Lagerloef, G. S. E., , G. T. Mitchum, , R. B. Lukas, , and P. P. Niiler, 1999: Tropical Pacific near-surface currents estimated from altimeter, wind, and drifter data. J. Geophys. Res., 104, 23 31323 326.

    • Search Google Scholar
    • Export Citation
  • Lenton, A., and Coauthors, 2012: The observed evolution of oceanic pCO2 and its drivers over the last two decades. Global Biogeochem. Cycles, 26, GB2021, doi:10.1029/2011GB004095.

    • Search Google Scholar
    • Export Citation
  • Le Quéré, C., and Coauthors, 2007: Saturation of the Southern Ocean CO2 sink due to recent climate change. Science, 316, 17351738.

  • Liss, P. S., , and L. Merlivat, 1986: Air-sea gas exchange rates: introduction and synthesis. The Role of Air-Sea Exchange in Geochemical Cycling, P. Buat-Menard, Ed., D. Reidel, 113–127.

  • Locarnini, R. A., , A. V. Mishonov, , J. I. Antonov, , T. P. Boyer, , H. E. Garcia, , O. K. Baranova, , M. M. Zweng, , and D. R. Johnson, 2010: Temperature. Vol. 1, World Ocean Atlas 2009, NOAA Atlas NESDIS 68, 52 pp.

  • Lovenduski, N. S., , N. Gruber, , S. C. Doney, , and I. D. Lima, 2007: Enhanced CO2 outgassing in the Southern Ocean from a positive phase of the southern annular mode. Global Biogeochem. Cycles, 21, GB2026, doi:10.1029/2006GB002900.

    • Search Google Scholar
    • Export Citation
  • Lovenduski, N. S., , N. Gruber, , and S. C. Doney, 2008: Toward a mechanistic understanding of the decadal trends in the Southern Ocean carbon sink. Global Biogeochem. Cycles, 22, GB3016, doi:10.1029/2007GB003139.

    • Search Google Scholar
    • Export Citation
  • Lumpkin, R., , and M. Pazos, 2007: Measuring surface currents with surface velocity program drifters: The instrument, its data, and some recent results. Lagrangian Analysis and Prediction of Coastal and Ocean Dynamics (LAPCOD), A. Griffa et al., Eds., Cambridge University Press, 39–67.

  • Marinov, I., , A. Gnanadesikan, , R. Toggweiler, , and J. L. Sarmiento, 2006: The Southern Ocean biogeochemical divide. Nature, 441, 964967.

    • Search Google Scholar
    • Export Citation
  • Marinov, I., , A. Gnanadesikan, , J. L. Sarmiento, , J. R. Toggweiler, , M. Follows, , and B. K. Mignone, 2008: Impact of oceanic circulation on the biological carbon storage in the ocean and atmospheric pCO2. Global Biogeochem. Cycles, 22, GB3007, doi:10.1029/2007GB002958.

    • Search Google Scholar
    • Export Citation
  • Marsland, S. J., , H. Haak, , J. H. Jungclaus, , M. Latif, , and F. Roske, 2003: The Max-Planck-Institute global ocean/sea ice model with orthogonal curvilinear coordinates. Ocean Modell., 5, 91127.

    • Search Google Scholar
    • Export Citation
  • McNeil, B. I., , N. Netzl, , R. M. Key, , R. J. Matear, , and A. Corbiere, 2007: An empirical estimate of the Southern Ocean air-sea CO2 flux. Global Biogeochem. Cycles, 21, GB3011, doi:10.1029/2007GB002991.

    • Search Google Scholar
    • Export Citation
  • Mikaloff Fletcher, S. E., and Coauthors, 2006: Inverse estimates of anthropogenic CO2 uptake, transport, and storage by the ocean. Global Biogeochem. Cycles,20, GB2002, doi:10.1029/2005GB002530.

  • Mikaloff Fletcher, S. E., and Coauthors, 2007: Inverse estimates of the oceanic sources and sinks of natural CO2 and the implied oceanic carbon transport. Global Biogeochem. Cycles, 21, GB1010, doi:10.1029/2006GB002751.

    • Search Google Scholar
    • Export Citation
  • Orr, J. C., and Coauthors, 2001: Estimates of anthropogenic carbon uptake from four three-dimensional global ocean models. Global Biogeochem. Cycles, 15, 4360.

    • Search Google Scholar
    • Export Citation
  • Pegion, P., , M. Bourassa, , D. M. Legler, , and J. J. O’Brien, 2000: Objectively derived daily “winds” from satellite scatterometer data. Mon. Wea. Rev., 128, 31503168.

    • Search Google Scholar
    • Export Citation
  • Raddatz, T. J., and Coauthors, 2007: Will the tropical land biosphere dominate the climate-carbon cycle feedback during the twenty-first century? Climate Dyn., 29, 565574.

    • Search Google Scholar
    • Export Citation
  • Reynolds, R. W., , and T. M. Smith, 1994: Improved global sea surface temperature analysis using optimum interpolation. J. Climate, 7, 929948.

    • Search Google Scholar
    • Export Citation
  • Russell, J. L., , K. W. Dixon, , A. Gnanadesikan, , R. J. Stouffer, , and J. R. Toggweiler, 2006: The Southern Hemisphere westerlies in a warming world: Propping open the door to the deep ocean. J. Climate, 19, 63826390.

    • Search Google Scholar
    • Export Citation
  • Sarmiento, J. L., , and N. Gruber, 2006: Ocean Biogeochemical Dynamics. Princeton University Press, 503 pp.

  • Sarmiento, J. L., and Coauthors, 2004: Response of ocean ecosystems to climate warming. Global Biogeochem. Cycles, 18, GB3003, doi:10.1029/2003GB002134.

    • Search Google Scholar
    • Export Citation
  • Speer, K., , S. R. Rintoul, , and B. Sloyan, 2000: The diabatic Deacon cell. J. Phys. Oceanogr., 30, 32123222.

  • Sprintall, J., 2003: Seasonal to interannual upper-ocean variability in the Drake Passage. J. Mar. Res., 61, 2757.

  • Sprintall, J., , T. Chereskin, , and C. Sweeney, 2012: High-resolution underway upper ocean and surface atmospheric observations in Drake Passage: Synergistic measurements for climate science. Oceanography, 25, 7082.

    • Search Google Scholar
    • Export Citation
  • Stephenson, G. R., , S. T. Gille, , and J. Sprintall, 2012: Seasonal variability of upper ocean heat content in Drake Passage. J. Geophys. Res., 117, C04019, doi:10.1029/2011JC007772.

    • Search Google Scholar
    • Export Citation
  • Swart, N. C., , and J. C. Fyfe, 2011: Ocean carbon uptake and storage influenced by wind bias in global climate models. Nat. Climate Change, 2, 4752.

    • Search Google Scholar
    • Export Citation
  • Sweeney, C., , E. Gloor, , A. R. Jacobson, , R. M. Key, , G. McKinley, , J. L. Sarmiento, , and R. Wanninkhof, 2007: Constraining global air-sea gas exchange for CO2 with recent bomb 14C measurements. Global Biogeochem. Cycles, 21, GB2015, doi:10.1029/2006GB002784.

    • Search Google Scholar
    • Export Citation
  • Takahashi, T., , J. Olafsson, , J. G. Goddard, , D. W. Chipman, , and S. C. Sutherland, 1993: Seasonal variation of CO2 and nutrients in the high-latitude surface oceans: A comparative study. Global Biogeochem. Cycles, 7, 843878.

    • Search Google Scholar
    • Export Citation
  • Takahashi, T., and Coauthors, 2002: Global sea–air CO2 flux based on climatological surface ocean pCO2, and seasonal biological and temperature effects. Deep-Sea Res. II, 49, 16011622.

    • Search Google Scholar
    • Export Citation
  • Takahashi, T., and Coauthors, 2009: Climatological mean and decadal change in surface ocean pCO2, and net sea-air CO2 flux over the global oceans. Deep-Sea Res. II, 56, 554557, doi:10.1016/j.dsr2.2008.12.009; Corrigendum, Deep-Sea Res. I, 56, 2075–2076, doi:10.1016/j.dsr.2009.07.007.

    • Search Google Scholar
    • Export Citation
  • Taylor, K. E., , R. J. Stouffer, , and G. A. Meehl, 2012: An overview of CMIP5 and the experiment design. Bull. Amer. Meteor. Soc., 93, 485498.

    • Search Google Scholar
    • Export Citation
  • Toggweiler, J. R., , and B. Samuels, 1995: Effect of Drake Passage on the global thermohaline circulation. Deep-Sea Res., 42I, 477500.

  • Volodin, E. M., , N. A. Dianskii, , and A. V. Gusev, 2010: Simulating present-day climate with the INMCM4.0 coupled model of the atmospheric and oceanic general circulations. Izv. Atmos. Ocean. Phys., 46, 414431.

    • Search Google Scholar
    • Export Citation
  • Wanninkhof, R., , and W. M. McGillis, 1999: A cubic relationship between gas transfer and wind speed. Geophys. Res. Lett., 26, 18891892.

    • Search Google Scholar
    • Export Citation
  • Watanabe, S., and Coauthors, 2011: MIROC-ESM 2010: Model description and basic results of CMIP5-20c3m experiments. Geosci. Model Dev., 4, 845872.

    • Search Google Scholar
    • Export Citation
  • Woloszyn, M., , M. Mazloff, , and T. Ito, 2011: Testing an eddy-permitting model of the Southern Ocean carbon cycle against observations. Ocean Modell., 39, 170182.

    • Search Google Scholar
    • Export Citation
  • Wu, T., and Coauthors, 2013: Global carbon budgets simulated by the Beijing Climate Center Climate System Model for the last century. J. Geophys. Res., 118,43264347, doi:10.1002/jgrd.50320.

    • Search Google Scholar
    • Export Citation
  • Zahariev, K., , J. Christian, , and K. Denman, 2008: Preindustrial, historical, and fertilization simulations using a global ocean carbon model with new parameterizations of iron limitation, calcification, and N2 fixation. Prog. Oceanogr., 77, 5682.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Schematic of the overturning circulation in the Southern Ocean (Atlantic sector). At the PF, the upper layer is formed by the equatorward Ekman current and the poleward eddy-driven current (Speer et al. 2000; Iudicone et al. 2011). The deep water upwells DIC-rich water to the surface ocean south of the PF. North and south of the PF, the ocean takes up CO2 from the overlying atmosphere, as indicated from the shipboard measurements. Also shown are the Subantarctic Front (SAF) and Subtropical Front (STF).

  • View in gallery

    The cruise tracks of 111 shipboard transects (black lines) in the Drake Passage from 2002 to 2011. Note that the mean Polar Front is located around 58.5°S. The Drake Passage region in this study is defined as the triangle covered by these 111 transects, with vertices at 56°S, 65°W; 62°S, 65°W; and 62°S, 57°W.

  • View in gallery

    (a) Bias and standard error of the difference of transect-averaged seawater pCO2 from eight CMIP5 ESMs relative to shipboard measurements. Differences between the ESMs and shipboard measurements are shown in the austral winter (July–September; black bars), time mean (gray bars), and summer (January–March; white bars). Positive values in (a) indicate overestimation compared to the shipboard measurements. (b) The mean and standard error of transect-averaged surface upward sea-to-air CO2 flux from the eight CMIP5 ESMs. (c) As in (b), but for surface upward sea-to-air CO2 flux averaged in the entire circumpolar region (56°–62°S). No shipboard measurements are available for the surface upward CO2 flux. Positive values in (b),(c) indicate upward CO2 flux from the ocean to the overlying atmosphere. The standard error equals the standard deviation divided by square root of the number of the transects.

  • View in gallery

    Seasonal cycles of the 111-transect seawater pCO2 and its related variables from eight CMIP5 ESMs and shipboard measurements in Drake Passage. The variables include (a) seawater pCO2, (b) SST-forced seawater pCO2 computed from the temperature sensitivity of pCO2, and (c) DIC-forced seawater pCO2 computed by differencing (a) and (c). Gray shading denotes the austral summer period from January to March and winter period from July to September.

  • View in gallery

    As in Fig. 4, but for the seasonal cycles of the entire circumpolar (56°–62°S) seawater pCO2 and its related variables from eight CMIP5 ESMs.

  • View in gallery

    Spatial (lag) correlations of 10°-longitude-averaged seawater pCO2 anomalies along the circumpolar region (from 56° to 62°S) with the Drake Passage–averaged (57°–67°W) pCO2 anomalies for eight CMIP5 ESMs. The pCO2 anomalies are obtained by removing the time mean, the long-term trend, and the seasonal cycles. The dotted lines show the 95% significance level correlation values.

  • View in gallery

    July–September constructed seawater pCO2 against SST for (a)–(h) eight CMIP5 ESMs and (i) the shipboard observations in the Drake Passage region. Data points north of the Polar Front (56°–57.5°S) are shown in red, near the Polar Front (57.5°–59.5°S) are shown in black, and south of the Polar Front (59.5°–62°S) are shown in magenta. The thick black lines show the 0.3°C SST-binned averages of pCO2 as a function of SST, and the error bars denote the standard error of the mean. The multiple thick cyan lines in (a)–(h) show the constructed atmospheric pCO2 along the winter transects in eight ESMs and in (i) show the observed atmospheric pCO2 in Drake Passage by NOAA/ESRL for the years 2003–11.

  • View in gallery

    As in Fig. 7, but for July–September sea-to-air CO2 flux (mol m−2 yr−1) against SST. Positive values indicate sea-to-air upward CO2 flux.

  • View in gallery

    July–September constructed seawater pCO2 against SST for (a) CanESM2 and (c) INM-CM4.0 and sea-to-air CO2 flux against SST for (b) CanESM2 and (d) INM-CM4.0 in the control experiment simulated with preindustrial atmospheric pCO2 (278 μatm).

  • View in gallery

    (a)–(h) July–September-mean 10-m wind speeds (color; m s−1) and the upper 30-m ocean currents (vectors) in the Drake Passage region (50°–80°W, 52°–62°S) for eight CMIP5 ESMs. (i) The QuikSCAT winds (2002–09) and OSCAR ocean currents (2002–09).

  • View in gallery

    Latitude–depth profiles of potential temperature (°C) along 65°W for eight CMIP5 ESMs and WOA09 climatology.

  • View in gallery

    (a),(b) Amplitudes and (c),(d) phases of (top) the seasonal cycles of surface water pCO2 and (bottom) the SST-forced component of surface pCO2, for the eight CMIP5 ESMs (see legend) and the in situ measurements (black solid). Error bars denote the standard error of the mean. The phases indicate the peak months of the seasonal cycles of surface water pCO2.

  • View in gallery

    (a) Amplitudes and (c) phases of the seasonal cycles of surface water pCO2 associated with the DIC effect for the eight CMIP5 ESMs (see legend) and the in situ measurements (black solid). (b) Amplitudes and (d) phases of the seasonal cycles of MLD for four of the CMIP5 ESMs (see legend) and the shipboard measurements (black solid). The phases indicate the peak months of the seasonal cycles of surface water pCO2.

  • View in gallery

    The ratio of the SST-forced seasonal pCO2 variations to the DIC-forced seasonal pCO2 variations for the eight ESMs (see legend) and the shipboard measurements (black solid).

  • View in gallery

    July–September-mean MLDs (m), as determined from the model mixing schemes, for the Drake Passage region (50°–70°W, 52°–62°S) for (a) CanESM2, (b) MPI-ESM-LR, (c) GFDL-ESM2G, (d) GFDL-ESM2M, and (e) Argo climatology observations.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 70 71 10
PDF Downloads 44 44 2

Drake Passage Oceanic pCO2: Evaluating CMIP5 Coupled Carbon–Climate Models Using in situ Observations

View More View Less
  • 1 Earth & Space Research, Seattle, Washington
  • | 2 Scripps Institution of Oceanography, University of California, San Diego, La Jolla, California
  • | 3 CIRES, NOAA/ESRL/GMD, Boulder, Colorado
© Get Permissions
Full access

Abstract

Surface water partial pressure of CO2 (pCO2) variations in Drake Passage are examined using decade-long underway shipboard measurements. North of the Polar Front (PF), the observed pCO2 shows a seasonal cycle that peaks annually in August and dissolved inorganic carbon (DIC)–forced variations are significant. Just south of the PF, pCO2 shows a small seasonal cycle that peaks annually in February, reflecting the opposing effects of changes in SST and DIC in the surface waters. At the PF, the wintertime pCO2 is nearly in equilibrium with the atmosphere, leading to a small sea-to-air CO2 flux.

These observations are used to evaluate eight available Coupled Model Intercomparison Project, phase 5 (CMIP5), Earth system models (ESMs). Six ESMs reproduce the observed annual-mean pCO2 values averaged over the Drake Passage region. However, the model amplitude of the pCO2 seasonal cycle exceeds the observed amplitude of the pCO2 seasonal cycle because of the model biases in SST and surface DIC. North of the PF, deep winter mixed layers play a larger role in pCO2 variations in the models than they do in observations. Four ESMs show elevated wintertime pCO2 near the PF, causing a significant sea-to-air CO2 flux. Wintertime winds in these models are generally stronger than the satellite-derived winds. This not only magnifies the sea-to-air CO2 flux but also upwells DIC-rich water to the surface and drives strong equatorward Ekman currents. These strong model currents likely advect the upwelled DIC farther equatorward, as strong stratification in the models precludes subduction below the mixed layer.

Corresponding author address: ChuanLi Jiang, Earth & Space Research, 2101 4th Ave., Suite 1310, Seattle, WA 98121. E-mail: chjiang@esr.org

This article is included in the (C4MIP) Climate–Carbon Interactions in the CMIP5 Earth System Models special collection.

Abstract

Surface water partial pressure of CO2 (pCO2) variations in Drake Passage are examined using decade-long underway shipboard measurements. North of the Polar Front (PF), the observed pCO2 shows a seasonal cycle that peaks annually in August and dissolved inorganic carbon (DIC)–forced variations are significant. Just south of the PF, pCO2 shows a small seasonal cycle that peaks annually in February, reflecting the opposing effects of changes in SST and DIC in the surface waters. At the PF, the wintertime pCO2 is nearly in equilibrium with the atmosphere, leading to a small sea-to-air CO2 flux.

These observations are used to evaluate eight available Coupled Model Intercomparison Project, phase 5 (CMIP5), Earth system models (ESMs). Six ESMs reproduce the observed annual-mean pCO2 values averaged over the Drake Passage region. However, the model amplitude of the pCO2 seasonal cycle exceeds the observed amplitude of the pCO2 seasonal cycle because of the model biases in SST and surface DIC. North of the PF, deep winter mixed layers play a larger role in pCO2 variations in the models than they do in observations. Four ESMs show elevated wintertime pCO2 near the PF, causing a significant sea-to-air CO2 flux. Wintertime winds in these models are generally stronger than the satellite-derived winds. This not only magnifies the sea-to-air CO2 flux but also upwells DIC-rich water to the surface and drives strong equatorward Ekman currents. These strong model currents likely advect the upwelled DIC farther equatorward, as strong stratification in the models precludes subduction below the mixed layer.

Corresponding author address: ChuanLi Jiang, Earth & Space Research, 2101 4th Ave., Suite 1310, Seattle, WA 98121. E-mail: chjiang@esr.org

This article is included in the (C4MIP) Climate–Carbon Interactions in the CMIP5 Earth System Models special collection.

1. Introduction

The global ocean takes up more than a quarter of the total anthropogenic carbon dioxide (CO2) that is released into the atmosphere, and the Southern Ocean is thought to be responsible for more than 40% of the global ocean’s uptake of anthropogenic CO2 (e.g., Toggweiler and Samuels 1995; Orr et al. 2001; Sarmiento et al. 2004; Russell et al. 2006; Marinov et al. 2006; Mikaloff Fletcher et al. 2006). Southern Ocean net uptake of anthropogenic and natural CO2 is thus an important factor controlling future CO2 levels in the atmosphere. Fluxes of CO2 through the air–sea interface are controlled by winds and by differences in the partial pressure of CO2 (pCO2) in the surface ocean compared with the overlying atmosphere (e.g., Takahashi et al. 2002). Since geographical variations of atmospheric pCO2 are relatively small (Conway et al. 1994; Takahashi et al. 2002), temporal and spatial variations of Southern Ocean pCO2 are key to assessing projections for future atmospheric CO2 concentrations (e.g., Marinov et al. 2008; Cadule et al. 2010).

Variations of surface water pCO2 are governed by surface ocean temperature, salinity, dissolved inorganic carbon (DIC), and alkalinity (Alk). Surface ocean temperature and salinity are controlled by the coupled ocean–atmosphere physical processes. DIC and alkalinity are controlled by the air–sea gas exchange, horizontal and vertical transport, and mixing, as well as biological processes (e.g., Sarmiento and Gruber 2006). Surface water pCO2 is therefore determined by a complex interplay of biological, chemical, and physical processes.

Recent studies have shown that, since 1990, pCO2 in the surface waters of the Southern Ocean has increased at a rate that is similar to or slightly faster than the mean atmospheric rate of increase (e.g., Le Quéré et al. 2007; Lenton et al. 2012). Natural CO2 is outgassed to the atmosphere in the westerly driven upwelling region of the Southern Ocean (e.g., Mikaloff Fletcher et al. 2007; Lovenduski et al. 2007). Takahashi et al. (2009) and Lovenduski et al. (2007, 2008), among others, have speculated that the strong poleward shift of the westerly winds since the 1960s, which is associated with the positive index of the southern annular mode, has produced stronger upwelling at and south of the fronts in the Antarctic Circumpolar Current (ACC) system. This strong poleward shift of the winds has also brought DIC-rich water from the deep ocean to the surface (see the schematic in Fig. 1). This hypothesis suggests that the total amount of DIC upwelled to the ocean surface south of the fronts and its location after being upwelled play a key role in predicting the CO2 uptake rate by the Southern Ocean. A poleward shift of the fronts would reduce the area over which upwelling occurs and, together with the increased upwelling velocity, would likely change the total DIC in the Southern Ocean (Russell et al. 2006). The upper 1000 m of the ACC have warmed during the last 50–60 yr (e.g., Gille 2008), which would be expected to have increased seawater pCO2 and decreased oceanic CO2 uptake. Furthermore, net cross-frontal transport, which is largely wind-driven equatorward Ekman transport but is partly compensated by the eddy-driven poleward transport, is expected to export upwelled DIC equatorward (Ito et al. 2009). These processes in the Southern Ocean are intimately tied to the global ocean through the meridional overturning circulation (Speer et al. 2000; Iudicone et al. 2011; Downes et al. 2011; D. C. Jones et al. 2011). Thus, examining the north–south variations of the Southern Ocean surface pCO2 is a key step toward understanding pCO2 variations associated with long-term climate change.

Fig. 1.
Fig. 1.

Schematic of the overturning circulation in the Southern Ocean (Atlantic sector). At the PF, the upper layer is formed by the equatorward Ekman current and the poleward eddy-driven current (Speer et al. 2000; Iudicone et al. 2011). The deep water upwells DIC-rich water to the surface ocean south of the PF. North and south of the PF, the ocean takes up CO2 from the overlying atmosphere, as indicated from the shipboard measurements. Also shown are the Subantarctic Front (SAF) and Subtropical Front (STF).

Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-12-00571.1

As a choke point of the ACC system, the Drake Passage has become a key location for investigating the Southern Ocean and its role in global climate. The decade-long (2002–11) underway surface water pCO2 measurements in Drake Passage (Takahashi et al. 2009; Sprintall et al. 2012) provide valuable in situ observations for examining pCO2 variations in this otherwise undersampled region. Given the importance of the Southern Ocean surface water pCO2 variations to the global carbon cycle, using the available in situ measurements to evaluate the Drake Passage pCO2 concentrations in the Coupled Model Intercomparison Project, phase 5 (CMIP5) Earth system models (ESMs) will help to assess projections of future atmospheric concentrations. In earlier coupled model intercomparisons, the majority of ESMs prescribed CO2 concentration scenarios, derived from relatively simple offline carbon cycle models (Friedlingstein et al. 2006). In CMIP5, the ESMs couple the ocean carbon cycle to the climate models (Fung et al. 2000; Cox et al. 2002; Friedlingstein et al. 2006).

In this study, we first examine the spatial mean, meridional, and seasonal variations of the underway decade-long (2002–11) shipboard pCO2 measurements across Drake Passage. We then use these in situ measurements to evaluate pCO2 variations in eight CMIP5 ESMs that were available for this study (Taylor et al. 2012). To further investigate discrepancies between the observations and the ESMs, we also compare the westerly winds, ocean currents, and stratification between the models and the satellite and in situ measurements. We found that six of the eight CMIP5 ESMs show spatial-mean surface water pCO2 concentrations that are comparable to shipboard measurements in the Drake Passage. The seasonal cycles of all ESMs except one, however, have larger seasonal cycles than the observations. Moreover, in the Polar Front region, four of the eight ESMs simulate seawater partial pressures that are elevated by 15–60 μatm relative to atmospheric partial pressures, resulting in a significant sea-to-air CO2 flux to the overlying atmosphere.

2. Observations and CMIP5 ESMs

The ice-strengthened Antarctic support ship R/V Laurence M. Gould traverses Drake Passage approximately 20 times per year through all seasons, taking about 2 days for each crossing. Since March 2002, the ship has measured surface water pCO2 with an underway sampling frequency of 2.5 min. As of March 2011, 178 transects had been completed. Of these, we selected 111 transects that have relatively straight trajectories for direct comparison with the gridded ESM results (Fig. 2). This quality control approach for selecting transects is similar to that used by Jiang et al. (2012) in a study of turbulent heat fluxes. In this study, we refer to the Drake Passage region as the triangle covered by these 111 transects, with vertices at 56°S, 65°W; 62°S, 65°W; and 62°S, 57°W (Fig. 2).

Fig. 2.
Fig. 2.

The cruise tracks of 111 shipboard transects (black lines) in the Drake Passage from 2002 to 2011. Note that the mean Polar Front is located around 58.5°S. The Drake Passage region in this study is defined as the triangle covered by these 111 transects, with vertices at 56°S, 65°W; 62°S, 65°W; and 62°S, 57°W.

Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-12-00571.1

Seawater pCO2 was measured using a showerhead equilibration method (Takahashi et al. 2002). The seawater was equilibrated continuously with a carrier gas in this water–air equilibrator. The equilibrated air samples were then dried, and the seawater CO2 concentration was determined every 2.5 min. The reproducibility of measurements of the dry mole fraction [moles of CO2 (moles)−1 of dry air] in the atmospheric samples taken every 2 h have been evaluated by direct comparison with discrete flask measurements taken as the ship crosses the Polar Front on each cruise (roughly every 15 days) in the Drake Passage. Both flask and underway measurements were taken from the same inlet line mounted on the bow mast of the Gould. The flasks were then measured by National Oceanic and Atmospheric Administration (NOAA)/Earth System Research Laboratory (ESRL) using standards whose mole fractions are directly linked to the World Meteorological Organization. Since the strong westerly winds result in the atmospheric pCO2 being well mixed across the Drake Passage, in the following analysis we will use the discrete flask measurement as representative of atmospheric pCO2 across the whole passage. Comparisons between the discrete and underway measurements yielded differences of less than 0.15 μatm. Given temperature corrections and other artifacts associated with converting the dry mole fraction of CO2 in air to pCO2 in surface seawater, we estimate the accuracy of the pCO2measurements to be ±1 μtam. Temperatures were measured at 4 m from a thermosalinograph on the Gould to help assess the temporal and spatial variations in oceanic pCO2. In this study, these 4-m shipboard ocean temperatures will be referred to as sea surface temperatures (SSTs). The SST precision is estimated to be about 0.005°C.

As noted above, CMIP5 Earth system models are directly coupled to ocean carbon models (Taylor et al. 2012). The model design of the coupled carbon–climate models in CMIP5 is based on CMIP3 (Fung et al. 2000; Cox et al. 2002; Friedlingstein et al. 2006). For this analysis we focus on the time period from 2002 to 2011. The first part of this period, from 2002 to 2005, is considered historical in the CMIP5 simulations. For the period from 2006 to 2011, we use the representative concentration pathway (RCP) 8.5 scenario, which is defined such that the radiative forcing of all direct and indirect agents reaches 8.5 W m−2 near 2100 (http://www.pik-potsdam.de/~mmalte/rcps/). We use results from eight climate modeling groups selected because, as of May 2012, they offered the only ESMs that provided surface water pCO2 output for historical and RCP 8.5 experiments. The eight climate modeling groups are 1) the Beijing Climate Center Climate System Model, version 1.1 (BCC_CSM1.1; Wu et al. 2013); 2) the second generation Canadian Earth System Model (CanESM2; Arora et al. 2011; Zahariev et al. 2008); 3) the Hadley Centre Global Environment Model, version 2–Earth System (HadGEM2-ES; C. D. Jones et al. 2011; Collins et al. 2011), 4) the Institute of Numerical Mathematics Climate Model, version 4.0 (INM-CM4.0; Volodin et al. 2010); 5) the Model for Interdisciplinary Research on Climate, Earth System Model (MIROC-ESM; Watanabe et al. 2011), 6) the Max Planck Institute Earth System Model, low resolution (MPI-ESM-LR; Raddatz et al. 2007; Marsland et al. 2003; Ilyina et al. 2013); 7) the Geophysical Fluid Dynamics Laboratory Earth System Model with Generalized Ocean Layer Dynamics (GOLD) component (GFDL-ESM2G; Dunne et al. 2012); and 8) the Geophysical Fluid Dynamics Laboratory Earth System Model with Modular Ocean Model 4 (MOM4) component (GFDL-ESM2M; Dunne et al. 2013). The Program for Climate Model Diagnosis and Intercomparison (PCMDI) at Lawrence Livermore National Laboratory (http://www-pcmdi.llnl.gov/) provides some documentation for these ESMs.

In general, with typical resolution of 2° × 2°, most of the CMIP5 models represent the effects of eddies using eddy parameterizations (e.g., Gent and McWilliams 1990). However, as summarized by Gnanadesikan et al. (2006), GFDL-ESM2M uses the skew flux approach of Griffies (1998), in which the quasi-Stokes streamfunction is computed (Ferrari et al. 2010). These models also differ in the bathymetry dataset they used. For instance, BCC_CSM1.1 used the 5-min digital bathymetric database and MIROC-ESM used a spatially smoothed ETOPO5 topography. In contrast to the smoothed ETOPO5 topography used in GFDL-ESM2M, GFDL-ESM2G uses a depth average of the high-resolution ETOPO5 bathymetry (Dunne et al. 2012). The bathymetry used in HadGEM2-ES is derived from the ETOPO5 1/12° at high latitudes (Johns et al. 2006).

Although the majority of the ESMs do not restore the ocean surface tracers (temperature, salinity, and alkalinity) to the climatologies, these models differ in the specification of their ocean biogeochemistry. For instance, GFDL-ESM2G and GFDL-ESM2M include all elemental cycles, including limiting nutrients (nitrogen, phosphorus, iron, and silicate) and other substances cycled by the biosphere (carbon, oxygen, lithic material, calcium carbonate, opal, and sulfur). These two models also include three explicit phytoplankton classes—small, large, and diazotrophs—comprising four phytoplankton functional types: small, nitrogen fixers, large nondiatom, and large diatom. HadGEM2-ES includes carbon, oxygen, iron, and sulfur cycles and two phytoplankton classes (diatoms and nondiatoms). In CanESM2 and MIROC-ESM, only carbon and nitrogen cycles and one phytoplankton class (single species) are included. Carbon, oxygen, and phosphorus cycles and no phytoplankton class are included in BCC_CSM1.1. MPI-ESM-LR includes carbon, nitrogen, oxygen, and phosphorus cycles and one phytoplankton class with a diagnostic discrimination between diatoms and coccolithophores. INM-CM4.0 has only the carbon cycle and no phytoplankton class. These model differences in the quantification of the model ocean biogeochemistry are expected to result in differences in model performance in representing the ocean biogeochemical cycles.

We use a sampling strategy similar to that of Jiang et al. (2012), who compared shipboard transects of turbulent flux-related variables to gridded flux products. Gridded CMIP5 ESMs provide synoptic Eulerian maps, while shipboard measurements are not strictly synoptic. Thus, we linearly interpolate the gridded CMIP5 ESM output to the longitude, latitude, and time of the ship observations along the 111 transects used in this study. To evaluate the mechanisms governing model–data pCO2 differences, in addition to pCO2 and air-to-sea CO2 flux, we examine three physical variables in both the CMIP5 ESMs and in the observational fields: surface currents, westerly winds, and stratification.

For this study, near-surface (30 m) currents are from the 5-day, ° × ° Ocean Surface Current Analyses–Real Time (OSCAR) from 2002 to 2009, which uses scatterometer-derived vector winds to compute the wind-driven velocity from an Ekman/Stommel formulation (Lagerloef et al. 1999; Bonjean and Lagerloef 2002; Dohan et al. 2010; Dohan and Maximenko 2010). The end date is determined by the fact that Quick Scatterometer (QuikSCAT) winds are not available after November 2009. OSCAR uses the gradient of ocean surface topography fields from an altimetric gridded Archiving, Validation, and Interpretation of Satellite Oceanographic (AVISO) data product (http://www.aviso.oceanobs.com). OSCAR also includes a contribution from horizontal temperature gradients using the Reynolds objectively interpolated SST product (OI.v2; Reynolds and Smith 1994), which influences velocity via the thermal wind relation. The drifter currents from the Global Drifter Program (Lumpkin and Pazos 2007) and the time–space interpolated OSCAR currents indicate correlation coefficients greater than 0.8 (K. Dohan 2012, personal communication).

We compare CMIP5 ESM winds with wind fields from the four times daily 1° × 1° Center for Ocean-Atmospheric Prediction Studies (COAPS) QuikSCAT wind speed from 2002 to 2009 (Q-COAPS; Pegion et al. 2000). Wind speed at 10 m for Q-COAPS uses a direct minimization approach with tuning parameters determined from generalized cross-validation and QuikSCAT satellite observations filtered by the normalized objective function rain flag. We selected Q-COAPS winds rather than ship winds to evaluate the winds in CMIP5 ESMs for three primary reasons. One reason stems from the high regression slope (0.71 ± 0.17) and correlation coefficient (0.75 ± 0.16) between the QuikSCAT level 2B swath winds at 25-km resolution (which Q-COAPS uses to produce the gridded fields) and the 1-min-interval shipboard wind measurements in the Drake Passage (Jiang et al. 2012). The second reason is that the gridded winds are less noisy and can provide large spatial coverage compared to the ship measurements. Finally, OSCAR relies on the Q-COAPS winds and so, for consistency, we use the same.

Finally to evaluate the vertical stratification of CMIP5 ESMs and its possible links with the surface water pCO2, we use monthly World Ocean Atlas 2009 (WOA09) potential temperature calculated from temperature and salinity (Locarnini et al. 2010; Antonov et al. 2010) along 65°W. To examine the meridional variation of the seasonal cycle of mixed layer depths (MLDs), we also use the surface ocean MLDs derived from the expendable conductivity–temperature–depth (XCTD) sampling in Drake Passage (Sprintall 2003; Stephenson et al. 2012). This MLD is defined as the depth at which potential density differs from the potential density at 11-m depth by 0.03 kg m−3 (e.g., Stephenson et al. 2012). To examine the MLD variation over a broader zonal extent within the Drake Passage region, we use climatological MLDs derived from Argo float measurements (Holte et al. 2010; Holte and Talley 2009).

3. Drake Passage variations

In this section, we investigate the variations of seawater pCO2 and related variables in the Drake Passage using the decade-long shipboard measurements. We then use these measurements to evaluate the overall performance of the eight CMIP5 ESMs.

a. Time mean

Bias and standard error of the 111-transect-averaged seawater pCO2 from eight CMIP5 ESMs relative to the shipboard measurements are shown in Fig. 3a (gray bars). Positive values indicate that in the spatial average, model estimates exceed observed values. In the Drake Passage, six out of the eight CMIP5 ESMs show a spatially averaged time-mean bias smaller than 6 μatm. Among these six ESMs, pCO2 values in BCC_CSM1.1, GFDL-ESM2G, and GFDL-ESM2M underestimate the Drake Passage’s pCO2 by up to 6 μatm, while the pCO2 values in CanESM2, INM-CM4.0, and MIROC-ESM overestimate by up to 4 μatm. The other two ESMs, HadGEM2-ES and MPI-ESM-LR, underestimate the surface ocean pCO2 by 25 and 19 μatm, respectively.

Fig. 3.
Fig. 3.

(a) Bias and standard error of the difference of transect-averaged seawater pCO2 from eight CMIP5 ESMs relative to shipboard measurements. Differences between the ESMs and shipboard measurements are shown in the austral winter (July–September; black bars), time mean (gray bars), and summer (January–March; white bars). Positive values in (a) indicate overestimation compared to the shipboard measurements. (b) The mean and standard error of transect-averaged surface upward sea-to-air CO2 flux from the eight CMIP5 ESMs. (c) As in (b), but for surface upward sea-to-air CO2 flux averaged in the entire circumpolar region (56°–62°S). No shipboard measurements are available for the surface upward CO2 flux. Positive values in (b),(c) indicate upward CO2 flux from the ocean to the overlying atmosphere. The standard error equals the standard deviation divided by square root of the number of the transects.

Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-12-00571.1

In addition to the annual means (gray bars), Fig. 3a also shows seasonal-mean biases for austral winter (July–September; black bars) and austral summer (January–March; white bars). These seasons are chosen because they represent extremes in surface pCO2 (see section 3b). For the eight CMIP5 ESMs, the winter pCO2 biases are less than 14 μatm, with GFDL-ESM2M, CanESM2, and INM-CM4.0 underestimating the winter pCO2 by up to 5 μatm. While HadGEM2-ES and MPI-ESM-LR have the largest time-mean biases (19–25 μatm) in Drake Passage, their winter biases are less than 10 μatm. Their summer biases, however, are larger than 40 μatm. The summer biases for the other six ESMs are less than 17 μatm.

The net CO2 flux out of the ocean F is a function of the sea-to-air pCO2 difference δpCO2, the CO2 gas transfer velocity k, and the solubility of CO2 in seawater α, F = kαδpCO2 (Sweeney et al. 2007; Takahashi et al. 2009). At fixed temperature and salinity, the gas transfer velocity k is thought to be roughly a power function of wind speed: that is, kUx, where U is wind speed and x has been proposed to have a value between 1 (Liss and Merlivat 1986) and 3 (Wanninkhof and McGillis 1999). The Gould does not directly measure the air–sea CO2 flux, and its estimation depends on interpolating not only the pCO2 (which changes slowly) but also synoptic winds (which change rapidly); therefore, comparing monthly air–sea fluxes with in situ values is not straightforward. In addition, its estimation through the bulk algorithm described above (Sweeney et al. 2007; Takahashi et al. 2009), using the measured pCO2 and the other available data products, results in large variations and uncertainties. This further complicates the evaluation of the CMIP5 model performance. Hence, in the following we present only an intermodel comparison of the sea-to-air CO2 flux from the eight CMIP5 ESMS (Fig. 3b) with no comparison to the CO2 flux estimated using the bulk algorithm. Positive values represent the upward CO2 flux into the atmosphere. In the time mean (gray bars), GFDL-ESM2G has the smallest air–sea CO2 flux in the Drake Passage. CanESM2 fluxes CO2 into the overlying atmosphere, while in the other seven CMIP5 ESMs the ocean takes up CO2 from the overlying atmosphere. During the austral winter period, however, two of these seven ESMs (MIROC-ESM and MPI-ESM-LR) release CO2 into the overlying atmosphere.

To examine the extent to which the results found in the Drake Passage (Fig. 3b) represent the entire circumpolar region, we show in Fig. 3c the sea-to-air CO2 flux averaged in the entire circumpolar region (56°–62°S) from eight CMIP5 ESMs. In both the annual and seasonal means, the sea-to-air CO2 flux in the entire circumpolar region compares well with the CO2 flux in the Drake Passage, implying that the Drake Passage’s pCO2 variations are representative of the entire circumpolar region.

b. Seasonal cycle

From the previous subsection, we know that six of the eight CMIP5 ESMs show spatially averaged surface water pCO2 values that are comparable to the shipboard measurements in Drake Passage (with time-mean biases smaller than 6 μatm). We now examine the seasonal cycles by fitting the observations and constructed ESM transects to a sinusoidal seasonal cycle using a least squares approach.

Examining the seasonal variations of seawater pCO2 requires examining its dependent variables. As discussed earlier, surface water pCO2 depends on surface temperature, salinity, DIC, and alkalinity. That is, pCO2K2/(K0K1)(2DIC − Alk)2/(Alk − DIC), where K2/(K0K1) is the equilibrium constant, which is a function of temperature and salinity. To investigate SST-forced surface ocean pCO2 changes, we compute the effect of SST on pCO2 by perturbing the mean pCO2 with the difference between the observed and mean temperature (Takahashi et al. 2002). That is, , where and are the time-mean pCO2 and temperature. Since salinity-forced seasonal pCO2 variations are negligible (Sarmiento and Gruber 2006), we interpret the residual component after subtracting the SST-forced pCO2 changes as pCO2 changes forced by DIC and alkalinity, pCO2DIC,Alk = pCO2pCO2SST. Alkalinity is more homogeneous than DIC in Drake Passage, with a typical fixed value of 2268 μmol kg−1, as empirically estimated by McNeil et al. (2007). As a result, pCO2DIC,Alk is mainly controlled by DIC variations, and henceforth we refer to it as pCO2DIC.

Figure 4a shows the seasonal cycles of the seawater pCO2 from the shipboard measurements and eight CMIP5 ESMs. The observed Drake Passage–averaged seawater pCO2 (thick black line) is highest in austral winter (August) and lowest in austral summer (February). Note that the time-mean pCO2 value (368 μatm) is included in Fig. 4a, and the summer-to-winter change clearly shows the seasonal variations of pCO2 (Table 1). Positive signs in Table 1 indicate that the peak month occurs in austral winter, and negative signs indicate the peak month is in austral summer. Observed pCO2 increases 9 μatm from a minimum in austral summer to a maximum in winter (Fig. 4a), as a result of the compensation between the SST-forced pCO2 drawdown of 51.0 μatm (Fig. 4b) and the DIC-forced pCO2 increase of 60.3 μatm (Fig. 4c). SST-forced seasonal pCO2 variations are approximately 6 months out of phase with DIC-forced pCO2 changes both for the observations and for the eight CMIP5 ESMs: SST-forced pCO2 changes reach their highest values in austral summer, and DIC-forced pCO2 are highest in austral winter when wintertime mixing with DIC-rich deep waters and respiration of organic matter have peaked.

Fig. 4.
Fig. 4.

Seasonal cycles of the 111-transect seawater pCO2 and its related variables from eight CMIP5 ESMs and shipboard measurements in Drake Passage. The variables include (a) seawater pCO2, (b) SST-forced seawater pCO2 computed from the temperature sensitivity of pCO2, and (c) DIC-forced seawater pCO2 computed by differencing (a) and (c). Gray shading denotes the austral summer period from January to March and winter period from July to September.

Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-12-00571.1

Table 1.

The summer-to-winter magnitudes of the seawater pCO2, the SST-forced pCO2 change (pCO2SST), and the DIC-forced changes (pCO2DIC) for the shipboard measurements and eight CMIP5 ESMs. Positive values indicate pCO2 increase from the austral winter (July–September) to the austral summer (January–March).

Table 1.

Four of the eight CMIP5 ESMs (HadGEM2-ES, INM-CM4.0, MIROC-ESM, and MPI-ESM-LR) are consistent with shipboard measurements in showing high pCO2 values in austral winter (positive signs). However, three of these four ESMs (HadGEM2-ES, MIROC-ESM, and MPI-ESM-LR) have much larger summer-to-winter pCO2 differences than the observations. These large summer-to-winter differences stem mainly from the fact that DIC-forced seasonal pCO2 changes dominate SST-forced seasonal pCO2 changes (Table 1). Consistent with this finding, Cadule et al. (2010) used an atmospheric transport model in conjunction with output from three ESMs (including a previous version of HadGEM2-ES) and found larger seasonal air–sea flux variations in the Southern Ocean than the station observations. The other four CMIP5 ESMs (CanESM2, BCC_CSM1.1, GFDL-ESM2G, and GFDL-ESM2M) are 6 months out of phase with the shipboard measurements, largely because SST-forced pCO2 changes in these models dominate DIC-forced pCO2 changes (Table 1).

Figure 5 shows the seasonal cycles of the entire circumpolar (56°–62°S) seawater pCO2 and its related variables (pCO2|SST and pCO2|DIC) from the eight CMIP5 ESMs. The seasonal cycles of the SST- and DIC-forced pCO2 changes in the entire circumpolar region (Figs. 5b,c) in all ESMs significantly correlate with the pCO2 changes in the Drake Passage (Figs. 4b,c), with correlation coefficients above 0.85. The entire circumpolar pCO2 variations (Fig. 5a) in six of the eight ESMs significantly correlate with the variations in the Drake Passage (Fig. 4a), with correlation coefficients above 0.83. The exceptions are INM-CM4.0 (R = 0.66, R95%CI = 0.72) and GFDL-ESM2M (R = 0.12, R95%CI = 0.75), mainly because their peak months in the Drake Passage pCO2 variations greatly differ from their peak months in the circumpolar region.

Fig. 5.
Fig. 5.

As in Fig. 4, but for the seasonal cycles of the entire circumpolar (56°–62°S) seawater pCO2 and its related variables from eight CMIP5 ESMs.

Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-12-00571.1

c. Anomalies

The previous two subsections show that the time mean (Fig. 3) and seasonal cycles (Fig. 5) of seawater pCO2 in Drake Passage largely represent the entire circumpolar region in most of the eight ESMs. To further examine the extent to which the pCO2 variations in Drake Passage resemble the other areas of the circumpolar region, we show in Fig. 6 the spatial (lag) correlations of the 10°-longitude-averaged seawater pCO2 anomalies for the circumpolar region (56°–62°S) with the Drake Passage–averaged (57°–67°W) pCO2 anomalies for eight CMIP5 ESMs. The pCO2 anomalies here are obtained by removing the time mean, the long-term trend, and the seasonal cycles from the original time series. In calculating the correlation coefficients in Fig. 6, we take the time series of the pCO2 averaged in Drake Passage and correlate it at zero time lag with the pCO2 time series at other 10°-longitude bands around the Southern Ocean.

Fig. 6.
Fig. 6.

Spatial (lag) correlations of 10°-longitude-averaged seawater pCO2 anomalies along the circumpolar region (from 56° to 62°S) with the Drake Passage–averaged (57°–67°W) pCO2 anomalies for eight CMIP5 ESMs. The pCO2 anomalies are obtained by removing the time mean, the long-term trend, and the seasonal cycles. The dotted lines show the 95% significance level correlation values.

Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-12-00571.1

In general, the Drake Passage pCO2 anomalies significantly correlate with the other areas of the Southern Ocean for all ESMs. Surface water pCO2 anomalies from 56° to 62°S in INM-CM4.0 shows little longitudinal variations, including a fixed dominant linear trend and several fixed low frequencies, and the Drake Passage pCO2 anomalies therefore highly correlate with the other circumpolar areas (Fig. 6d). The little zonal variations in INM-CM4.0 are likely due to the fact that there is no biology in its ocean biogeochemistry. These findings, in conjunction with the results in the time mean and seasonal cycles as detailed in the previous two subsections, confirm the representativeness of the Drake Passage in the surface water pCO2 variations in the entire circumpolar region from 56° to 62°S for most of the eight ESMs.

4. Meridional variations

In this section we examine the meridional variations of the surface water pCO2, sea-to-air CO2 flux, and the pCO2 seasonal cycles in the Drake Passage. We then use these observations to evaluate meridional variations of pCO2 in the eight CMIP5 ESMs.

The Polar Front, one of the major frontal features that defines the ACC, has a mean latitude around 58.5°S in Drake Passage (Sprintall 2003; Jiang et al. 2012). Differences in SST-forced and DIC-forced processes are observed north and south of the Polar Front (Takahashi et al. 2002; Downes et al. 2011; Iudicone et al. 2011). Given the importance of SST and DIC in the oceanic pCO2 variations (section 3b), seawater pCO2 is also expected to differ on either side of the Polar Front. As discussed in the introduction, examining the meridional variations of the Southern Ocean pCO2 is a key step toward understanding pCO2 variations associated with long-term climate change. Therefore, the relative roles of DIC- and SST-forced pCO2 variations north and south of the Polar Front are crucial to the projections of the future atmospheric CO2 levels in the CMIP5 ESMs. With a typical resolution of 2° × 2°, however, these CMIP5 ESMs generally only have a few grid points in the Drake Passage region and thus do not have sufficient resolution to allow a full evaluation of their meridional gradients. The evaluations in this section help define requirements for future improvements of these CMIP5 ESMs.

Here we refer to the region 57.5°–59.5°S as the Polar Front, the region 56°–57.5°S as north of the Polar Front, and the region 59.5°–62°S as south of the Polar Front. These regional definitions are for descriptive purposes only, as the Polar Fronts in the CMIP5 ESMs are not consistently located around 58.5°S because of model biases that may stem in part from coarse model resolution. Because of this, we show in Figs. 7 and 8 the meridional variation of surface pCO2 and sea-to-air CO2 flux with respect to SST instead of latitude. This removes biases in the Polar Front position but shows biases in temperature and meridional gradients, reflecting a range of processes that contribute to pCO2 and sea-to-air CO2 flux variations. We also examine westerly winds, surface ocean currents, and potential temperature profiles in order to probe the data–model discrepancies.

Fig. 7.
Fig. 7.

July–September constructed seawater pCO2 against SST for (a)–(h) eight CMIP5 ESMs and (i) the shipboard observations in the Drake Passage region. Data points north of the Polar Front (56°–57.5°S) are shown in red, near the Polar Front (57.5°–59.5°S) are shown in black, and south of the Polar Front (59.5°–62°S) are shown in magenta. The thick black lines show the 0.3°C SST-binned averages of pCO2 as a function of SST, and the error bars denote the standard error of the mean. The multiple thick cyan lines in (a)–(h) show the constructed atmospheric pCO2 along the winter transects in eight ESMs and in (i) show the observed atmospheric pCO2 in Drake Passage by NOAA/ESRL for the years 2003–11.

Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-12-00571.1

Fig. 8.
Fig. 8.

As in Fig. 7, but for July–September sea-to-air CO2 flux (mol m−2 yr−1) against SST. Positive values indicate sea-to-air upward CO2 flux.

Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-12-00571.1

a. Seawater pCO2

During austral winter, biological consumption is low and physical processes dominate pCO2 variability. We thus show only data from the austral winter period between July and September in sections 4a4c. The impact of biological consumption on seasonal pCO2 variations will be addressed in section 4d. In Fig. 7, we plot constructed wintertime oceanic pCO2 against SST for the eight ESMs and the shipboard observations: pCO2 values north of the Polar Front are shown as red, near the Polar Front are shown as black, and south of the Polar Front are shown as magenta. The 0.3°C SST-binned averages of pCO2 as a function of SST are shown as the thick black lines with the error bars denoting the standard error of the mean. In computing the degrees of freedom, each transect is treated as an independent realization, because the transects cover all seasons of the year with consecutive transects typically separated in time by 2–6 weeks and each transect takes about 2 days to complete.

As noted above, the strong westerly winds in the ACC system tend to homogenize the atmospheric pCO2 in the planetary boundary layer, and so the discrete in situ atmospheric pCO2 (Fig. 7i) is considered representative of atmospheric pCO2 variations across the entire Drake Passage. Similarly, the atmospheric pCO2 concentrations in the eight ESMs also show little meridional variation (cyan lines in Figs. 7a–h). Note that the discontinuous cyan lines in GFDL-ESM2G and GFDL-ESM2M (Figs. 7g,h) are due to the missing data near the ice edge in the Southern Ocean.

The shipboard observations (Fig. 7i) show that the winter-averaged surface water pCO2 near the Polar Front (2°–4°C; 385 μatm; thick black line) is nearly in equilibrium with the time-mean atmospheric pCO2 (380 μatm). The observed atmospheric pCO2 values increase with time from 2002 to 2011 (cyan lines in Fig. 7i). The atmospheric pCO2 growth rate is estimated to be 1.95 μatm yr−1, close to the growth rate of 1.9 μatm yr−1 in the Southern Ocean reported by Lenton et al. (2012). The growth rate of the spatially averaged observed oceanic pCO2 is estimated to be 2.4 μatm yr−1, close to previous reported values of 2.5 ± 0.7 μatm yr−1 by Lenton et al. (2012) and 2.2 ± 0.5 μatm yr−1 by Takahashi et al. (2009). Although the decadal-mean surface water pCO2 is nearly in equilibrium with the decadal-mean atmospheric pCO2, over the 10-yr time series the growth rate of the ocean pCO2 increases faster compared to that of the atmospheric pCO2. This faster oceanic growth rate means that sea-to-air fluxes of CO2 are higher by the end of the time series in 2011.

The region just south of the Polar Front (SST < −1°C) has lower seawater pCO2 concentrations (360 μatm) than the overlying atmosphere (Fig. 7i), indicating that the ocean acts as a sink for atmospheric CO2 within the Southern Ocean seasonal ice zone (Fig. 1). Sprintall et al. (2012) showed a strong correlation of δpCO2 and upper ocean temperature just south of the Polar Front (PF) in winter in the Drake Passage, suggesting that SST plays a role in seawater pCO2 variations there. From the equilibrium with the atmosphere at 2°C to the under saturation at −1°C, seawater pCO2 (thick black lines in Fig. 7i) decreases from 385 to 360 μatm. The SST dependence of seawater pCO2 (8 μatm °C−1) south of the Polar Front is half the SST dependence (16 μatm °C−1) that would be estimated based on the relation developed by Takahashi et al. (1993), who found δpCO2/δSST ≈ 0.0423 × pCO2, assuming SST to be the sole factor controlling variations in pCO2. The fact that the observed seasonal cycle in pCO2 is less dependent on SST than predicted by its change in solubility implies that, south of the Polar Front, DIC-forced seawater pCO2 changes may also play an important role. The DIC-forced processes might include the air-to-sea CO2 flux and the upwelling of high DIC water from the deep ocean, as indicated in Fig. 1.

North of the Polar Front, for SSTs around 6°C, observed seawater pCO2 is about 370 μatm, which is lower than in the Polar Front (385 μatm at 2°–4°C) and is a departure from equilibrium with the atmosphere. On the basis of temperature dependence alone, pCO2 would be expected to increase with warmer temperatures, so the observed departure from equilibrium has the wrong sign to be temperature controlled. This departure implies that the DIC sink terms, such as subduction into the deep ocean and uptake by phytoplankton, may play a significant role north of the Polar Front. Alternatively, source waters from the subtropics could contribute to lower DIC water, which, when cooled to 6°C, results in a significantly lower pCO2 than found at the Polar Front. Iudicone et al. (2011) argued that water mass transformations can have an important impact in setting the surface carbon properties north of the Polar Front. We speculate that the effective subduction into the deep ocean in the convergence zone, the wind-driven equatorward meridional transport, and the DIC uptake by photosynthesis likely play a role in removing carbon from surface water. This is due to the fact that other processes, such as air–sea gas flux and possible cross-frontal equatorward transport from the south, all should increase surface water pCO2 (Fig. 1).

Six of the eight ESMs are consistent with observations that show a maximum of seawater pCO2 near the Polar Front. The exceptions are INM-CM4.0 (Fig. 7d) and to a lesser extent HadGEM2-ES (Fig. 7c). All ESMs except INM-CM4.0 show a pCO2 reduction as temperatures decrease south of the Polar Front. However, in other respects, the eight available ESMs show significantly different meridional variations than the observations (Figs. 7a–h). While all ESMs except INM-CM4.0 and GFDL-ESM2M show larger values of oceanic pCO2 than in the overlying atmosphere near the Polar Front, the seawater pCO2 in four ESMs (CanESM2, MIROC-ESM, MPI-ESM-LR, and to some extent GFDL-ESM2G) appears to exceed atmospheric pCO2 values by 15–60 μatm, suggesting that the sea-to-air pCO2 gradients are larger in these four models than in the observations (Figs. 7b,e–g). If we assume the winds are the same in all ESMs and in the shipboard measurements (we will show in section 4c that the model winds are generally too strong), then the Polar Fronts in these four models lose much more CO2 to the overlying atmosphere than suggested by the shipboard observations or the other ESMs because of the excessive sea-to-air gradient. The minimum values of pCO2 south of the Polar Front in these four models, however, are consistent with the shipboard measurements. This leads to a much larger south-to-north increase in pCO2 between the −1°C water south of the PF and the 4°C water near the Polar Front, implying excessive DIC sources by the oceanic physical processes near the Polar Front in these four models. These processes could include an overestimate in the upwelling of DIC-rich water and excessive biological drawdown of DIC as the waters move north of the PF or rapid mixing of subtropical waters before waters are subducted into the deep ocean, as indicated in Fig. 1. In addition, CanESM2 appears to have a narrower SST range (4°–5°C) over which the oceanic pCO2 is larger than the overlying atmosphere (Fig. 7b). This narrower SST range implies less total CO2 uptake from the overlying atmosphere.

b. Sea-to-air CO2 flux

Figure 8 shows austral winter transects of the CO2 flux from the ocean to the atmosphere as a function of SST, where positive values indicate a sea-to-air CO2 flux out of the ocean. The sea-to-air CO2 flux in INM-CM4.0 (Fig. 8d) is about an order of magnitude larger than in the other ESMs. The meridional variations of sea-to-air CO2 flux (Fig. 8) are generally proportional to the sea-to-air pCO2 differences (Fig. 7) for all ESMs, implying that in these models the air–sea pCO2 differences play an important role in the air–sea CO2 flux.

Near the Polar Front, the oceans of four models (BCC_CSM1.1, HadGEM2-ES, INM-CM4.0, and GRDL-ESM2M; Figs. 8a,c,d,h) are nearly in equilibrium with the overlying atmosphere. In these models, the ocean acts as a sink for atmospheric CO2 south of the Polar Front for water temperatures less than 1°C. In contrast, in CanESM2 (Fig. 8b), MIROC-ESM (Fig. 8e), MPI-ESM-LR (Fig. 8f), and to some extent GFDL-ESM2G (Fig. 8g), the oceans release CO2 to the overlying atmosphere, because seawater pCO2 exceeds atmospheric pCO2. In these same models south of the Polar Front, oceanic pCO2 is nearly in equilibrium with the atmosphere, and as a result the oceans exchange little CO2 with the overlying atmosphere. The near equilibrium state illustrated by the shipboard pCO2 observations (Fig. 7i) at the Polar Front stands in contrast to these four ESMs that show a CO2 source to the overlying atmosphere.

To further investigate the excessive sea-to-air CO2 flux near the Polar Front, we show in Fig. 9 the oceanic pCO2 and sea-to-air CO2 flux for CanESM2 and INM-CM4.0 in the preindustrial control experiments. These two models are the only models that provide the oceanic pCO2 output for the ESM control experiments. In these control experiments, the oceanic pCO2 variations and air–sea CO2 flux are merely driven by the natural carbon variation and flux, as constant preindustrial atmospheric pCO2 values (278 μatm) are used. As a result, the differences of pCO2 variations (Figs. 7, 9) and sea-to-air CO2 flux (Figs. 8, 9) indicate the effect of anthropogenic carbon variations that are input into the ocean.

Fig. 9.
Fig. 9.

July–September constructed seawater pCO2 against SST for (a) CanESM2 and (c) INM-CM4.0 and sea-to-air CO2 flux against SST for (b) CanESM2 and (d) INM-CM4.0 in the control experiment simulated with preindustrial atmospheric pCO2 (278 μatm).

Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-12-00571.1

For both CanESM2 and INM-CM4.0, the natural pCO2 variations and air–sea natural carbon flux show a meridional variation similar to that seen under contemporary conditions. In particular, under preindustrial conditions, seawater pCO2 values in CanESM2 exceed atmospheric pCO2 values (278 μatm) by about 50 μatm (Fig. 9a), and the Polar Front CO2 outgassing to the atmosphere is about 2 mol m−2 yr−1 (Fig. 9b). These variations are larger than values under contemporary conditions, with oceanic pCO2 exceeding atmospheric pCO2 by about 15 μatm (Fig. 7b) and the sea-to-air CO2 flux estimated to be 1.5 mol m−2 yr−1 (Fig. 8b). This implies that in CanESM2, natural carbon outgassing plays a larger role (1.4–4 times) than anthropogenic carbon uptake in determining the excessive pCO2 bias.

We also examined longitudinal (east–west) variations of sea-to-air CO2 flux for the entire Drake Passage region (50°–80°W, 52°–62°S) in the eight ESMs (not shown). Consistent with Fig. 8, the oceans of CanESM2, MIROC-ESM, MPI-ESM-LR, and GFDL-ESM2G release CO2 to the atmosphere at a rate of 0.5–1.5 mol m−2 yr−1 at all longitudes near the Polar Front. In contrast, the oceans of BCC_CSM1.1 and HadGEM2-ES take up CO2 at a rate of 0.5 mol m−2 yr−1 at all longitudes. In addition, south of the Polar Front the oceans in BCC_CSM1.1 and HadGEM2-ES have a higher CO2 uptake rate than they do near the Polar Front.

c. Winds and ocean currents

Figures 7 and 8 indicate that the sea-to-air pCO2 differences δpCO2 are large enough to play an important role in the sea-to-air CO2 flux in all ESMs. In particular, for all winter transects from 2002 to 2011, the pCO2 values in four of the eight ESMs (CanESM2, MIROC-ESM, MPI-ESM-LR, and to some extent GFDL-ESM2G) near the Polar Front coincide with large CO2 releases to the atmosphere. To illustrate the relative roles of ocean currents and winds in the sea-to-air CO2 flux, in Figs. 10a–h we show the austral winter-mean wind speed and the near-surface (30 m) ocean currents in all eight ESMs. Winds are dynamically determined features in all the eight CMIP5 ESMs. In this study we time average the winter wind speeds and currents to be consistent with Figs. 7 and 8. For comparison, we show in Fig. 10i the austral winter-mean QuikSCAT wind speed (2002–09) and the austral winter-mean OSCAR ocean currents (2002–09) derived from satellite measurements.

Fig. 10.
Fig. 10.

(a)–(h) July–September-mean 10-m wind speeds (color; m s−1) and the upper 30-m ocean currents (vectors) in the Drake Passage region (50°–80°W, 52°–62°S) for eight CMIP5 ESMs. (i) The QuikSCAT winds (2002–09) and OSCAR ocean currents (2002–09).

Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-12-00571.1

West of 65°W, the wind speeds in CanESM2, INM-CM4.0, MIROC-ESM, and MPI-ESM-LR are on average 2–3 m s−1 stronger than QuikSCAT winds; however, east of 65°W, winds in these models are 3–4 m s−1 stronger than QuikSCAT winds. This stronger wind bias in coupled general circulation models has also been reported by Swart and Fyfe (2011) and by Russell et al. (2006). As suggested by the schematic in Fig. 1, the strong eastward winds in these models in the entire Drake Passage region may 1) enlarge the magnitude of the CO2 release from the ocean to the atmosphere; 2) enhance upwelling of higher DIC from the deep ocean to the surface at and south of the Polar Front; and 3) drive the Ekman transport in the surface layer of the Southern Ocean strongly northward (to the left of the eastward winds), which also moves DIC-rich water northward. Observations and models suggest that, near the Polar Front, eddy-driven poleward transport compensates to some extent the equatorward Ekman transport (Böning et al. 2008; Ito et al. 2009). In coarse-resolution models, overly strong winds can disrupt this balance and drive a northward net meridional transport. North of the Polar Front where intermediate water forms through convective processes (Speer et al. 2000), meridional transport is expected to be small in the near-surface layer of the ocean. However, in the four models with stronger winds east of 65°W, the austral winter-mean Eulerian ocean surface currents are consistently equatorward and stronger throughout all of Drake Passage compared to the observations. This implies that the upwelled DIC-rich water from south of the Polar Front is likely transported farther equatorward in the surface layer instead of being subducted into the deep ocean, leading to a longer exchange time with the atmosphere. It also seems likely that because of the low resolution the poleward eddy-driven meridional currents in these models are too weak to compensate the strong equatorward Ekman currents.

Vertical stratification can serve as a measure of how easily water can be subducted or upwelled. For this study we use the vertical gradient of potential temperature as an approximate measure of stratification since more detailed stratification information, such as potential density, is not readily available from the ESMs. The impact of the salinity on the stratification needs further study, as salinity potentially plays an important role in the mixed layer calculation in the Southern Ocean (e.g., Stephenson et al. 2012). Compared to the WOA09 climatology (Fig. 11i), vertical profiles of potential temperature along 65°W in CanESM2, HadGEM2-ES, INM-CM4.0, and MPI-ESM-LR (Figs. 11b–d,f) imply much stronger stratification at and just north of the Polar Front, making the upwelled DIC harder to subduct to the deep ocean in these regions. As shown in Fig. 1, four physical processes likely act together to reduce the model inventory of DIC in the Drake Passage. These processes include the strong westerly winds and the consequent strong equatorward Ekman transports, strong net meridional currents (implying weak poleward eddy transport compensation), strong stratification preventing subduction of CO2 to the deep ocean, and excessive CO2 flux to the atmosphere. A full analysis of the carbon budget in the Drake Passage would be needed to examine the effect of these physical processes on the DIC inventory in these models and is beyond the scope of this analysis.

Fig. 11.
Fig. 11.

Latitude–depth profiles of potential temperature (°C) along 65°W for eight CMIP5 ESMs and WOA09 climatology.

Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-12-00571.1

d. Seasonal cycle

We now explore the relative roles of SST forcing and DIC forcing in governing the seasonal cycle of pCO2 as a function of meridional position within Drake Passage. We fitted the 1°-latitude-binned observations and the constructed ESM transects to a sinusoidal seasonal cycle (Fig. 12) using a least squares approach, A cos(2πt + φ), where A and φ represent amplitude and phase. Here the amplitudes have positive sign and differ from the summer-to-winter magnitudes in Table 1, in which the magnitudes carry some of the phase information. Consistent with the Drake Passage–averaged results (Fig. 4 and Table 1), meridionally varying seawater pCO2 concentrations in the shipboard measurements and in all eight ESMs except INM-CM4.0 show a significant seasonal cycle. The amplitude of the seasonal cycle of observed pCO2 in the Drake Passage can reach up to 7.5 μatm, which is much smaller [(10−1)] than the seasonal pCO2 in other parts of the global ocean (Takahashi et al. 2009; Christian et al. 2010). In contrast, all of the ESMs, with the exception of INM-CM4.0, have much larger seasonal amplitudes (Fig. 12a). GFDL-ESM2G, GFDL-ESM2M, and BCC_CSM1.1 have seasonal cycle amplitudes for pCO2 that are most consistent with the shipboard measurements. The amplitude of the seasonal cycle of pCO2 in MPI-ESM-LR reaches up to 75 μatm at 56.5°S, and in HadGEM2-ES it reaches up to 47 μatm at 59.5°S.

Fig. 12.
Fig. 12.

(a),(b) Amplitudes and (c),(d) phases of (top) the seasonal cycles of surface water pCO2 and (bottom) the SST-forced component of surface pCO2, for the eight CMIP5 ESMs (see legend) and the in situ measurements (black solid). Error bars denote the standard error of the mean. The phases indicate the peak months of the seasonal cycles of surface water pCO2.

Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-12-00571.1

The amplitudes of the seasonal cycles in surface water pCO2 show significant meridional variations (Fig. 12a, left). Just north of the Polar Front, the amplitude of the seasonal cycle of observed pCO2 is triple the amplitude south of the front (about 7.5 μatm compared to 2.5 μatm). Jiang et al. (2012) found the amplitude of the seasonal cycle of SST north of the Polar Front to be half the amplitude south of the front (about 1°C compared to 2°C). This suggests that the meridional gradients of the seasonal cycle amplitudes have opposite signs for pCO2 and SST. This confirms the results from the spatially averaged analysis that surface temperature is not the only factor controlling the seasonal cycle of the surface pCO2; DIC-related processes also contribute to seasonal variations in surface pCO2.

Like the amplitudes, the months when seawater pCO2 peaks vary with latitude (Fig. 12c). The observed seawater pCO2 consistently peaks in austral winter (August) north of the Polar Front and peaks in the austral summer (February) just south of the Polar Front. In the eight ESMs, pCO2 peaks between February and August; however, they differ greatly with latitude. In HadGEM2-ES, MIROC-ESM, and MPI-ESM-LR the phases of seawater pCO2 show little meridional variations and are mainly locked to austral winter. This agrees with the shipboard measurements north of the Polar Front but not south of the Polar Front. In the models BCC_CSM1.1, CanESM2, and GFDL-ESM2G, the phases indicate a maximum between November and January north of the Polar Front, which is several months too late compared to the observations. Consistent with Woloszyn et al. (2011), the phases south of the Polar Front in the ESMs differ substantially from the observations, especially near the seasonally ice-covered regions (near 62°S). Overall, the meridional variation of the phase of pCO2 in GFDL-ESM2M best matches the observations but shows a gentler transition from south to north across the Polar Front, which is probably a result of the low horizontal resolution of the model.

We know from the spatially averaged seasonal cycle analysis in Fig. 4 that the compensation between the SST-forced and DIC-forced seasonal pCO2 changes determines the seasonal pCO2 variations. Further, Figs. 12b,d show the amplitudes and phases of the seasonal cycles of the pCO2 as predicted, based on the SST effect pCO2|SST. North of the Polar Front the amplitude of the seasonal cycles of pCO2 variations associated with SST is half the amplitude that it is south of the front (about 16 μatm to the north compared to 31 μatm to the south; Fig. 12b). This north–south gradient is by definition consistent with the seasonal cycles of SST (Jiang et al. 2012), but it is opposite to the seasonal cycles of oceanic pCO2 (Fig. 12a). While the amplitude of the seasonal cycle is larger in the eight ESMs compared to the observations, the meridional gradients of the model and observed amplitudes are similar. In shipboard observations, the SST effect on pCO2 peaks from February to March, while in the eight ESMs it peaks later, from March to May.

In the top panels of Fig. 13, we show the seasonal cycles of the pCO2 due to the DIC effect pCO2|DIC obtained by removing the SST effect from the pCO2 variations. In shipboard observations, the DIC-induced changes in pCO2 show a smaller north–south gradient, with about 23 μatm to the north compared to 29 μatm to the south. The observed DIC effect on pCO2 peaks during austral winter from July to August, about 6 months after the observed austral summer peak, due to changes in the SST effect on pCO2. This out-of-phase relation between the SST- and DIC-forced effects explains the small amplitudes of the seasonal cycle of pCO2 variations (2.5 μatm to the south and 7.5 μatm to the north; Fig. 12a), since the effects compensate for each other. The amplitudes of the eight ESMs show meridional gradients that differ from the observations, with larger magnitude seasonal cycles. The majority of the eight ESMs show peaks from July to November, about 1–2 months later than the observations.

Fig. 13.
Fig. 13.

(a) Amplitudes and (c) phases of the seasonal cycles of surface water pCO2 associated with the DIC effect for the eight CMIP5 ESMs (see legend) and the in situ measurements (black solid). (b) Amplitudes and (d) phases of the seasonal cycles of MLD for four of the CMIP5 ESMs (see legend) and the shipboard measurements (black solid). The phases indicate the peak months of the seasonal cycles of surface water pCO2.

Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-12-00571.1

For both the observations and the eight ESMs, the SST-forced seasonal pCO2 variations are approximately 6 months out of phase with DIC-forced pCO2 changes (Figs. 12b, 13a). As a result, the ratio of amplitudes of the seasonal pCO2 predicted by the SST effect to those predicted by the DIC effect indicates their relative importance in the seasonal pCO2 variations (Takahashi et al. 2002). Because the seasonal amplitudes have positive values in Figs. 12 and 13, their ratios also have positive signs (Fig. 14). Ratios much larger than 1 indicate a leading role for surface warming in February, and ratios much smaller than 1 indicate that DIC effects play a leading role in August. Figure 14 shows that the ratio from shipboard measurements is less than 1 (about 0.7) north of the Polar Front, suggesting that the DIC effect plays a slightly more important role than the SST effect. Because of this, the observed seawater pCO2 variations peak the same months as the DIC-driven pCO2 in August. South of the Polar Front, the SST and the DIC effects seem to play equal roles. As a result, seawater pCO2 peaks between October and February, in between the peak months for SST (February–March) and DIC (August–September).

Fig. 14.
Fig. 14.

The ratio of the SST-forced seasonal pCO2 variations to the DIC-forced seasonal pCO2 variations for the eight ESMs (see legend) and the shipboard measurements (black solid).

Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-12-00571.1

Of the eight ESMs, GFDL-ESM2G and GFDL-ESM2M generally simulate realistic ratios of the seasonal SST-forced to DIC-forced pCO2 amplitude variations at all latitudes in the Drake Passage. The role of the DIC effect is slightly underestimated north of the Polar Front in these two models. North of the Polar Front, the ratios in three ESMs (BCC_CSM1.1, CanESM2, and HadGEM2-ES) compare well with the observations, while the DIC effect in MIROC-ESM and MPI-ESM-LR seems to be overestimated in this region. South of the Polar Front, BCC_CSM1.1 and CanESM2 overestimate the role of SST while HadGEM2-ES, MIROC-ESM, and MPI-ESM-LR appear to underestimate the SST effect.

Barbero et al. (2011) found high DIC values associated with deep MLDs in the Pacific sector of the Southern Ocean. Given the important role of the seasonal DIC variations throughout the Drake Passage and to further isolate the impact of MLD on seasonal pCO2 variations through ocean stratification, we show the amplitudes (Fig. 13b) and phases (Fig. 13d) of the seasonal cycles of the MLD, in the shipboard measurements and four of the eight ESMs: CanESM2, MPI-ESM-LR, GFDL-ESM2G, and GFDL-ESM2M. These models are chosen because, as of July 2012, they were the only ESMs that had released MLD estimates based on the model mixing schemes. Consistent with the observed MLDs derived from the shipboard XCTD observations, the MLDs in the four ESMs peak in austral winter (July–September). Three of the ESMs (CanESM2, GFDL-ESM2G, and GFDL-ESM2M) have seasonal magnitudes that are comparable to the observations. North of the Polar Front, the amplitudes and phases of the MLD in these four ESMs seem to coincide with the DIC-induced pCO2 variations (Fig. 13a), suggesting the important role of the MLD variations in the seasonal pCO2 changes in these models. Deep MLDs of ESMs in high latitudes have also been found in the CMIP3 ESMs (Boé et al. 2009). The finding in this study also supports Boé et al.’s (2009) speculation that intermodel differences in oceanic mixing (i.e., MLDs) can be associated with a spread in the modeled oceanic uptake of CO2. Besides MLDs, other factors, such as biological activity, also play a role in the seawater pCO2 seasonal variations. Cadule et al. (2010) pointed out that the phytoplankton growth rate defined by Geider et al. (1998), which is widely used in the CMIP5 ESMs, will affect the seasonality of the oceanic pCO2.

North of the Polar Front, MPI-ESM-LR shows a much larger seasonal cycle in MLD than the observations and the other three ESMs (CanESM2, GFDL-ESM2G, and GFDL-ESM2M), with amplitudes up to 180 m (Fig. 13b). To illustrate the spatial variations of the MLD, we show in Figs. 15a–d the July–September-mean MLD in the Drake Passage for the same ESMs as in Fig. 13b. We also show the climatological MLD derived from Argo float measurements in Fig. 15e (Holte et al. 2010). Consistent with the Argo MLD, all four ESMs show deeper MLDs north of the PF than south of the PF. While the four ESMs show different spatial features from each other, MLDs in GFDL-ESM2G and GFDL-ESM2M show the best agreement with climatology. The MPI-ESM-LR MLD in the winter can reach up to 1000 m north of the Polar Front and these deep MLDs probably play a key role in the overly large seasonal pCO2 variations in this model (Fig. 12a), although the causal relationship requires further study. Dunne et al. (2012) pointed out that an atmospheric air–sea surface shortwave bias may cause both seasonal MLD and SST biases in the Southern Ocean. Besides the variability in the biological productivity, we speculate that another possibility is that the large seasonal cycle of the MLD in MPI-ESM-LR brings CO2 from the deep ocean to the surface layer north of the Polar Front. Because the DIC effect dominates the seasonal pCO2 variations north of the PF (Fig. 14), these deep MLDs seems to explain the overly large amplitudes of the seasonal cycles of the pCO2 variations in MPI-ESM-LR north of the Polar Front (Fig. 12a).

Fig. 15.
Fig. 15.

July–September-mean MLDs (m), as determined from the model mixing schemes, for the Drake Passage region (50°–70°W, 52°–62°S) for (a) CanESM2, (b) MPI-ESM-LR, (c) GFDL-ESM2G, (d) GFDL-ESM2M, and (e) Argo climatology observations.

Citation: Journal of Climate 27, 1; 10.1175/JCLI-D-12-00571.1

5. Conclusions

In this study, we have utilized the decade-long (2002–11) underway shipboard measurements from Drake Passage to examine spatial-mean, meridional (north–south), and seasonal variations in seawater pCO2. These year-round underway repeat measurements provide a means to assess the performance of the available CMIP5 ESMs, which for the first time directly couple the ocean carbon cycle to climate models. Our analysis has focused on assessing seawater pCO2 and the air-to-sea CO2 flux in the context of winds, ocean currents, and vertical stratification in the Drake Passage. Of the eight available ESMs, six ESMs reproduce the observed annual-mean pCO2 values averaged over the Drake Passage region, and all ESMs but one show the significant meridional variations that are consistent with shipboard observations.

The amplitudes of seasonal cycles of the observed pCO2 show significant meridional variations, with the amplitude north of the Polar Front (7.5 μatm) triple the amplitude south of the Polar Front (2.5 μatm). Our analysis shows that the SST-forced and DIC-forced seawater pCO2 variations play an equal role south of the Polar Front, while the DIC-forced pCO2 variations play a slightly more important role north of the Polar Front. In addition, these two effects are out of phase at all latitudes, with the SST-forced pCO2 variations peaking from February to March and the DIC-forced pCO2 peaking from July to August. The compensation of these two effects explains the observed meridional variations of the oceanic pCO2. In contrast with the observations, all ESMs but INM-CM4.0 show larger seasonal cycle magnitudes at all latitudes, due to biases in the relative roles of SST-forced and DIC-forced pCO2 variations. North of the Polar Front, the deepening of the mixed layer in austral winter in four of the eight ESMs, possibly together with the biological consumption in the austral spring/summer (Sprintall et al. 2012), seems to play a significant role in the seasonal surface pCO2 variations. For ESMs that realistically simulate the seasonal deep mixed layer north of the Polar Front, the seasonal surface pCO2 variations compare well with observations. In contrast, ESMs with mixed layers that are too deep or too shallow north of the Polar Front directly influence surface pCO2 seasonal variations, because of the contribution by the vertical mixing of DIC-rich subsurface water to the surface.

Contrary to observations, which show atmospheric and surface water pCO2 concentrations to be near equilibrium near the Polar Front (58.5°S; 2°–4°C), in four of the eight ESMs (CanESM2, MIROC-ESM, MPI-ESM-LR, and to some extent in GFDL-ESM2G) surface water pCO2 values exceed atmospheric pCO2 values by 15–60 μatm. This causes large sea-to-air CO2 fluxes in these four models, leading to an increase of atmospheric pCO2. In other words, in these ESMs the Polar Front acts as a CO2 source to the overlying atmosphere. Because surface water pCO2 values far south of the Polar Front in these models are consistent with shipboard measurements, this leads to much larger gradients in seawater pCO2 between the −1°C water in the upwelling region and the 2°C water near the Polar Front. This implies that physical oceanic processes supply excessive pCO2 to the Polar Front region in these four models. This could occur as a result of strong net meridional CO2 transport and/or overly weak subduction to the deep ocean.

Overall, compared to observations, most CMIP5 ESMs reproduce the annual-mean pCO2 values averaged in the Drake Passage region. However, there remain several distinct discrepancies. The ESMs typically show much larger seasonal cycles than the shipboard measurements. Other issues include excessive surface water pCO2 concentrations near the Polar Front, strong equatorward Ekman currents driven by the overly intensified westerlies, strong net equatorward meridional currents owing to weak poleward eddy transport that is insufficient to compensate the equatorward Ekman currents, excessively strong vertical stratification, and winter mixed layers that are either too deep or too shallow north of the Polar Front. We hypothesize that within the Drake Passage region these models release too much CO2 to the overlying atmosphere, which reduces the inventory of DIC in the model Drake Passage. Further studies will be needed to examine whether the processes in Drake Passage are representative of the entire Southern Ocean and further whether these processes significantly impact the atmospheric CO2 concentrations in the ESM projections.

The underway shipboard pCO2 measurements used in this study provide a valuable benchmark for assessing the performance of the CMIP5 ESMs in the Drake Passage. However, the sparse in situ measurements prohibit us from examining zonal (east–west) variations of the surface ocean pCO2 throughout the whole Southern Ocean. Lenton et al. (2012) analyzed the growth rate of the seawater pCO2 across the ACC system and found that different mechanisms influence the surface ocean pCO2 in the Atlantic compared to the Pacific–Indian sectors. In this study, we found the spatially averaged sea-to-air CO2 flux and seasonal pCO2 variations in the entire circumpolar region (56°–62°S) agree well with their variations in the Drake Passage for most of the eight ESMs. In addition, the ESM pCO2 anomalies (removing the annual mean, long-term trend, and seasonal cycles) in Drake Passage significantly correlate with the other areas of the Southern Ocean. Thus, despite the limited zonal spatial coverage in the Drake Passage measurements, they appear to provide a useful means to assess the representation of CO2 in long-term climate projections.

Acknowledgments

We gratefully acknowledge support from the National Science Foundation (OCE Award 0850350 and Office of Polar Programs Awards 0943818, 1129005, and 0944761) and NASA Award NNX08AR63G. Jiang was supported as a part of a diagnostic analysis effort proposed by U.S. CLIVAR (OPP Award 1129005) and by the Scripps Postdoctoral fellowship and NASA Award NNX08AR63G. The authors would like to acknowledge Teresa K. Chereskin and Sharon Escher for providing the LMG datasets. We also thank all the technical and scientific support from the staff of eight ESM groups and the Program for Climate Model Diagnostic and Intercomparison (PCMDI) for making the ESM output available to the public. Jiang would like to acknowledge Dr. François W. Primeau for his constructive and valuable comments during her stay at University of California, Irvine. She would also like to thank Camisa Carlson at Earth & Space Research for her thorough English language editing. The authors thank three anonymous reviewers and the editor for their insightful and constructive comments and suggestions. Q-COAPS was obtained online (from http://coaps.fsu.edu/scatterometry/gridded/). The OSCAR ocean current data from the Physical Oceanography Distributed Active Archive Center (PO.DAAC) at the NASA Jet Propulsion Laboratory, Pasadena, California, was obtained online (from ftp://podaac.jpl.nasa.gov/).

REFERENCES

  • Antonov, J. I., and Coauthors, 2010: Salinity. Vol. 2, World Ocean Atlas 2009, NOAA Atlas NESDIS 69, 44 pp.

  • Arora, V. K., and Coauthors, 2011: Carbon emission limits required to satisfy future representative concentration pathways of greenhouse gases. Geophys. Res. Lett., 38, L05805, doi:10.1029/2010GL046270.

    • Search Google Scholar
    • Export Citation
  • Barbero, L., , J. Boutin, , L. Merlivat, , N. Martin, , T. Takahashi, , S. C. Sutherland, , and R. Wanninkhof, 2011: Importance of water mass formation regions for the air-sea CO2 flux estimate in the Southern Ocean. Global Biogeochem. Cycles, 25, GB1005, doi:10.1029/2010GB003818.

    • Search Google Scholar
    • Export Citation
  • Boé, J., , A. Hall, , and X. Qu, 2009: Deep ocean heat uptake as a major source of spread in transient climate change simulations. Geophys. Res. Lett., 36, L22701, doi:10.1029/2009GL040845.

    • Search Google Scholar
    • Export Citation
  • Böning, C. W., , A. Dispert, , M. Visbeck, , S. R. Rintoul, , and F. U. Schwarzkopf, 2008: The response of the Antarctic Circumpolar Current to recent climate change. Nat. Geosci., 1, 864869.

    • Search Google Scholar
    • Export Citation
  • Bonjean, F., , and G. Lagerloef, 2002: Diagnostic model and analysis of the surface currents in the tropical Pacific Ocean. J. Phys. Oceanogr., 32, 29382954.

    • Search Google Scholar
    • Export Citation
  • Cadule, P., , P. Friedlingstein, , L. Bopp, , S. Sitch, , C. D. Jones, , P. Ciais, , S. L. Piao, , and P. Peylin, 2010: Benchmarking coupled climate-carbon models against long-term atmospheric CO2 measurements. Global Biogeochem. Cycles, 24, GB2016, doi:10.1029/2009GB003556.

    • Search Google Scholar
    • Export Citation
  • Christian, J. R., and Coauthors, 2010: The global carbon cycle in the Canadian Earth System Model (CanESM1): Preindustrial control simulation. J. Geophys. Res., 115, G03014, doi:10.1029/2008JG000920.

    • Search Google Scholar
    • Export Citation
  • Collins, W. J., and Coauthors, 2011: Development and evaluation of an Earth-system model—HadGEM2. Geosci. Model Dev., 4, 10511077.

  • Conway, T. J., , P. P. Tans, , L. S. Waterman, , K. W. Thoning, , D. R. Kitzis, , K. A. Masarie, , and N. Zhang, 1994: Evidence of interannual variability of the carbon cycle from the NOAA/CMDL global air sampling network. J. Geophys. Res., 99, 22 83122 855.

    • Search Google Scholar
    • Export Citation
  • Cox, P., , P. Friedlingstein, , and P. Rayner, 2002: Modelling climate-carbon cycle feedbacks: A cross disciplinary collaboration priority. IGBP Global Change Newsletter, No. 49, IGBP Secretariat, Stockholm, Sweden, 1214.

    • Search Google Scholar
    • Export Citation
  • Dohan, K., , and N. Maximenko, 2010: Monitoring ocean currents with satellite sensors. Oceanography, 23, 94103.

  • Dohan, K., and Coauthors, 2010: Measuring the global ocean surface circulation with satellite and in situ observations. Proc. OceanObs’09: Sustained Ocean Observations and Information for Society, Venice, Italy, European Space Agency, WPP-306.

  • Downes, S. M., , A. Gnanadesikan, , S. M. Griffies, , and J. L. Sarmiento, 2011: Water mass exchange in the Southern Ocean in coupled climate models. J. Phys. Oceanogr., 41, 17561771.

    • Search Google Scholar
    • Export Citation
  • Dunne, J. P., and Coauthors, 2012: GFDLs ESM2 global coupled climate–carbon Earth system models. Part I: Physical formulation and baseline simulation characteristics. J. Climate, 25, 66466665.

    • Search Google Scholar
    • Export Citation
  • Dunne, J. P., and Coauthors, 2013: GFDL’s ESM2 global coupled climate–carbon Earth system models. Part II: Carbon system formulation and baseline simulation characteristics. J. Climate, 26, 22472267.

    • Search Google Scholar
    • Export Citation
  • Ferrari, R., , S. Griffies, , A. Nurser, , and G. Vallis, 2010: A boundary value problem for the parameterized mesoscale eddy transport. Ocean Modell., 32, 143156.

    • Search Google Scholar
    • Export Citation
  • Friedlingstein, P., and Coauthors, 2006: Climate–carbon cycle feedback analysis: Results from the C4MIP model intercomparison. J. Climate, 19, 33373353.

    • Search Google Scholar
    • Export Citation
  • Fung, I., , P. Rayner, , and P. Friedlingstein, 2000: Full-form earth system models: Coupled carbon-climate interaction experiment (the flying leap). IGBP Global Change Newsletter, No. 41, IGBP Secretariat, Stockholm, Sweden, 78.

    • Search Google Scholar
    • Export Citation
  • Geider, R. J., , H. L. Maclntyre, , and T. M. Kana, 1998: A dynamic regulatory model of phytoplanktonic acclimation to light, nutrients, and temperature. Limnol. Oceanogr., 43, 679694.

    • Search Google Scholar
    • Export Citation
  • Gent, P. R., , and J. C. McWilliams, 1990: Isopycnal mixing in ocean general circulation models. J. Phys. Oceanogr., 20, 150155.

  • Gille, S. T., 2008: Decadal-scale temperature trends in the Southern Hemisphere ocean. J. Climate, 21, 47494765.

  • Gnanadesikan, A., and Coauthors, 2006: GFDL’s CM2 global coupled climate models. Part II: The baseline ocean simulation. J. Climate, 19, 675697.

    • Search Google Scholar
    • Export Citation
  • Griffies, S. M., 1998: The Gent–McWilliams skew flux. J. Phys. Oceanogr., 28, 831841.

  • Holte, J., , and L. Talley, 2009: A new algorithm for finding mixed layer depths with applications to Argo data and Subantarctic Mode Water formation. J. Atmos. Oceanic Technol., 26, 19201939.

    • Search Google Scholar
    • Export Citation
  • Holte, J., , J. Gilson, , L. Talley, , and D. Roemmich, cited 2010: Argo mixed layers. Scripps Institution of Oceanography/UCSD. [Available online at http://mixedlayer.ucsd.edu/.]

  • Ilyina, T., , K. D. Six, , J. Segschneider, , E. Maier-Reimer, , H. Li, , and I. Nùñez-Riboni, 2013: Global ocean biogeochemistry model HAMOCC: Model architecture and performance as component of the MPI-Earth system model in different CMIP5 experimental realizations. J. Adv. Model. Earth Syst.,5,287315, doi:10.1029/2012MS000178.

  • Ito, T., , M. Woloszyn, , and M. Mazloff, 2009: Anthropogenic carbon dioxide transport in the Southern Ocean driven by Ekman flow. Nature, 463, 8083.

    • Search Google Scholar
    • Export Citation
  • Iudicone, D., , K. B. Rodgers, , I. Stendardo, , O. Aumont, , G. Madec, , L. Bopp, , O. Mangoni, , and M. Ribera d’Alcala, 2011: Water masses as a unifying framework for understanding the Southern Ocean carbon cycle. Biogeosciences, 8, 10311052.

    • Search Google Scholar
    • Export Citation
  • Jiang, C., , S. T. Gille, , J. Sprintall, , K. Yoshimura, , and M. Kanamitsu, 2012: Spatial variation in turbulent heat fluxes in Drake Passage. J. Climate, 25, 14701488.

    • Search Google Scholar
    • Export Citation
  • Johns, T. C., and Coauthors, 2006: The new Hadley Centre climate model (HadGEM1): Evaluation of coupled simulations. J. Climate, 19, 13271353.

    • Search Google Scholar
    • Export Citation
  • Jones, C. D., and Coauthors, 2011a: The HadGEM2-ES implementation of CMIP5 centennial simulations. Geosci. Model Dev., 4, 543570.

  • Jones, D. C., , T. Ito, , and N. S. Lovenduski, 2011b: The transient response of the Southern Ocean pycnocline to changing atmospheric winds. Geophys. Res. Lett.,38,L15604, doi:10.1029/2011GL048145.

  • Lagerloef, G. S. E., , G. T. Mitchum, , R. B. Lukas, , and P. P. Niiler, 1999: Tropical Pacific near-surface currents estimated from altimeter, wind, and drifter data. J. Geophys. Res., 104, 23 31323 326.

    • Search Google Scholar
    • Export Citation
  • Lenton, A., and Coauthors, 2012: The observed evolution of oceanic pCO2 and its drivers over the last two decades. Global Biogeochem. Cycles, 26, GB2021, doi:10.1029/2011GB004095.

    • Search Google Scholar
    • Export Citation
  • Le Quéré, C., and Coauthors, 2007: Saturation of the Southern Ocean CO2 sink due to recent climate change. Science, 316, 17351738.

  • Liss, P. S., , and L. Merlivat, 1986: Air-sea gas exchange rates: introduction and synthesis. The Role of Air-Sea Exchange in Geochemical Cycling, P. Buat-Menard, Ed., D. Reidel, 113–127.

  • Locarnini, R. A., , A. V. Mishonov, , J. I. Antonov, , T. P. Boyer, , H. E. Garcia, , O. K. Baranova, , M. M. Zweng, , and D. R. Johnson, 2010: Temperature. Vol. 1, World Ocean Atlas 2009, NOAA Atlas NESDIS 68, 52 pp.

  • Lovenduski, N. S., , N. Gruber, , S. C. Doney, , and I. D. Lima, 2007: Enhanced CO2 outgassing in the Southern Ocean from a positive phase of the southern annular mode. Global Biogeochem. Cycles, 21, GB2026, doi:10.1029/2006GB002900.

    • Search Google Scholar
    • Export Citation
  • Lovenduski, N. S., , N. Gruber, , and S. C. Doney, 2008: Toward a mechanistic understanding of the decadal trends in the Southern Ocean carbon sink. Global Biogeochem. Cycles, 22, GB3016, doi:10.1029/2007GB003139.

    • Search Google Scholar
    • Export Citation
  • Lumpkin, R., , and M. Pazos, 2007: Measuring surface currents with surface velocity program drifters: The instrument, its data, and some recent results. Lagrangian Analysis and Prediction of Coastal and Ocean Dynamics (LAPCOD), A. Griffa et al., Eds., Cambridge University Press, 39–67.

  • Marinov, I., , A. Gnanadesikan, , R. Toggweiler, , and J. L. Sarmiento, 2006: The Southern Ocean biogeochemical divide. Nature, 441, 964967.

    • Search Google Scholar
    • Export Citation
  • Marinov, I., , A. Gnanadesikan, , J. L. Sarmiento, , J. R. Toggweiler, , M. Follows, , and B. K. Mignone, 2008: Impact of oceanic circulation on the biological carbon storage in the ocean and atmospheric pCO2. Global Biogeochem. Cycles, 22, GB3007, doi:10.1029/2007GB002958.

    • Search Google Scholar
    • Export Citation
  • Marsland, S. J., , H. Haak, , J. H. Jungclaus, , M. Latif, , and F. Roske, 2003: The Max-Planck-Institute global ocean/sea ice model with orthogonal curvilinear coordinates. Ocean Modell., 5, 91127.

    • Search Google Scholar
    • Export Citation
  • McNeil, B. I., , N. Netzl, , R. M. Key, , R. J. Matear, , and A. Corbiere, 2007: An empirical estimate of the Southern Ocean air-sea CO2 flux. Global Biogeochem. Cycles, 21, GB3011, doi:10.1029/2007GB002991.

    • Search Google Scholar
    • Export Citation
  • Mikaloff Fletcher, S. E., and Coauthors, 2006: Inverse estimates of anthropogenic CO2 uptake, transport, and storage by the ocean. Global Biogeochem. Cycles,20, GB2002, doi:10.1029/2005GB002530.

  • Mikaloff Fletcher, S. E., and Coauthors, 2007: Inverse estimates of the oceanic sources and sinks of natural CO2 and the implied oceanic carbon transport. Global Biogeochem. Cycles, 21, GB1010, doi:10.1029/2006GB002751.

    • Search Google Scholar
    • Export Citation
  • Orr, J. C., and Coauthors, 2001: Estimates of anthropogenic carbon uptake from four three-dimensional global ocean models. Global Biogeochem. Cycles, 15, 4360.

    • Search Google Scholar
    • Export Citation
  • Pegion, P., , M. Bourassa, , D. M. Legler, , and J. J. O’Brien, 2000: Objectively derived daily “winds” from satellite scatterometer data. Mon. Wea. Rev., 128, 31503168.

    • Search Google Scholar
    • Export Citation
  • Raddatz, T. J., and Coauthors, 2007: Will the tropical land biosphere dominate the climate-carbon cycle feedback during the twenty-first century? Climate Dyn., 29, 565574.

    • Search Google Scholar
    • Export Citation
  • Reynolds, R. W., , and T. M. Smith, 1994: Improved global sea surface temperature analysis using optimum interpolation. J. Climate, 7, 929948.

    • Search Google Scholar
    • Export Citation
  • Russell, J. L., , K. W. Dixon, , A. Gnanadesikan, , R. J. Stouffer, , and J. R. Toggweiler, 2006: The Southern Hemisphere westerlies in a warming world: Propping open the door to the deep ocean. J. Climate, 19, 63826390.

    • Search Google Scholar
    • Export Citation
  • Sarmiento, J. L., , and N. Gruber, 2006: Ocean Biogeochemical Dynamics. Princeton University Press, 503 pp.

  • Sarmiento, J. L., and Coauthors, 2004: Response of ocean ecosystems to climate warming. Global Biogeochem. Cycles, 18, GB3003, doi:10.1029/2003GB002134.

    • Search Google Scholar
    • Export Citation
  • Speer, K., , S. R. Rintoul, , and B. Sloyan, 2000: The diabatic Deacon cell. J. Phys. Oceanogr., 30, 32123222.

  • Sprintall, J., 2003: Seasonal to interannual upper-ocean variability in the Drake Passage. J. Mar. Res., 61, 2757.

  • Sprintall, J., , T. Chereskin, , and C. Sweeney, 2012: High-resolution underway upper ocean and surface atmospheric observations in Drake Passage: Synergistic measurements for climate science. Oceanography, 25, 7082.

    • Search Google Scholar
    • Export Citation
  • Stephenson, G. R., , S. T. Gille, , and J. Sprintall, 2012: Seasonal variability of upper ocean heat content in Drake Passage. J. Geophys. Res., 117, C04019, doi:10.1029/2011JC007772.

    • Search Google Scholar
    • Export Citation
  • Swart, N. C., , and J. C. Fyfe, 2011: Ocean carbon uptake and storage influenced by wind bias in global climate models. Nat. Climate Change, 2, 4752.

    • Search Google Scholar
    • Export Citation
  • Sweeney, C., , E. Gloor, , A. R. Jacobson, , R. M. Key, , G. McKinley, , J. L. Sarmiento, , and R. Wanninkhof, 2007: Constraining global air-sea gas exchange for CO2 with recent bomb 14C measurements. Global Biogeochem. Cycles, 21, GB2015, doi:10.1029/2006GB002784.

    • Search Google Scholar
    • Export Citation
  • Takahashi, T., , J. Olafsson, , J. G. Goddard, , D. W. Chipman, , and S. C. Sutherland, 1993: Seasonal variation of CO2 and nutrients in the high-latitude surface oceans: A comparative study. Global Biogeochem. Cycles, 7, 843878.

    • Search Google Scholar
    • Export Citation
  • Takahashi, T., and Coauthors, 2002: Global sea–air CO2 flux based on climatological surface ocean pCO2, and seasonal biological and temperature effects. Deep-Sea Res. II, 49, 16011622.

    • Search Google Scholar
    • Export Citation
  • Takahashi, T., and Coauthors, 2009: Climatological mean and decadal change in surface ocean pCO2, and net sea-air CO2 flux over the global oceans. Deep-Sea Res. II, 56, 554557, doi:10.1016/j.dsr2.2008.12.009; Corrigendum, Deep-Sea Res. I, 56, 2075–2076, doi:10.1016/j.dsr.2009.07.007.

    • Search Google Scholar
    • Export Citation
  • Taylor, K. E., , R. J. Stouffer, , and G. A. Meehl, 2012: An overview of CMIP5 and the experiment design. Bull. Amer. Meteor. Soc., 93, 485498.

    • Search Google Scholar
    • Export Citation
  • Toggweiler, J. R., , and B. Samuels, 1995: Effect of Drake Passage on the global thermohaline circulation. Deep-Sea Res., 42I, 477500.

  • Volodin, E. M., , N. A. Dianskii, , and A. V. Gusev, 2010: Simulating present-day climate with the INMCM4.0 coupled model of the atmospheric and oceanic general circulations. Izv. Atmos. Ocean. Phys., 46, 414431.

    • Search Google Scholar
    • Export Citation
  • Wanninkhof, R., , and W. M. McGillis, 1999: A cubic relationship between gas transfer and wind speed. Geophys. Res. Lett., 26, 18891892.

    • Search Google Scholar
    • Export Citation
  • Watanabe, S., and Coauthors, 2011: MIROC-ESM 2010: Model description and basic results of CMIP5-20c3m experiments. Geosci. Model Dev., 4, 845872.

    • Search Google Scholar
    • Export Citation
  • Woloszyn, M., , M. Mazloff, , and T. Ito, 2011: Testing an eddy-permitting model of the Southern Ocean carbon cycle against observations. Ocean Modell., 39, 170182.

    • Search Google Scholar
    • Export Citation
  • Wu, T., and Coauthors, 2013: Global carbon budgets simulated by the Beijing Climate Center Climate System Model for the last century. J. Geophys. Res., 118,43264347, doi:10.1002/jgrd.50320.

    • Search Google Scholar
    • Export Citation
  • Zahariev, K., , J. Christian, , and K. Denman, 2008: Preindustrial, historical, and fertilization simulations using a global ocean carbon model with new parameterizations of iron limitation, calcification, and N2 fixation. Prog. Oceanogr., 77, 5682.

    • Search Google Scholar
    • Export Citation
Save