Developing models that accurately predict carbon sources and sinks requires accurate representation of ocean circulation. To this end, researchers in the Bender and Sarmiento teams have been using spatial gradients in oxygen, argon, radiocarbon and CFC’s to help decipher ocean circulation patterns and evaluate the skill of ocean general circulation models.

One study has validated existing models by showing that an air-sea exchange feature predicted for years in modeling studies can be detected in observations. Computer simulations by Sarmiento’s group and others had predicted that concentrations of an air-sea gas exchange indicator called atmospheric potential oxygen, or APO (approximately O2+1.1CO2), would exhibit a positive bulge across the equator. Until recently, O2/N2 latitudinal cross-sections across the tropics did not exist and it was not clear if the prediction of a large APO signal merely reflected deficiencies in ocean models, particularly unrealistic upwelling rates in the tropics.

Manuel Gloor and colleagues in Jorge Sarmiento’s group have been comparing new O2/N2 data from Y. Tohjima (NIES, Japan) with predictions of atmospheric APO based on ocean interior data. The work has revealed excellent agreement with the earlier predictions of both the seasonal cycle of the signal as well as of the annual mean latitudinal distribution (Figure 11). The results also support the realism of ocean transport simulations as well as, to some extent, model representation of ocean biology.

Fig. 11. Latitudinal distribution of annual mean (a) CO2, (b) O2/N2, and (c) APO. Circles represent shipboard data, and squares and triangles represent the data at Ochi-ishi and Hateruma, respectively. Open and solid symbols indicate averages for 2002 and 2003, respectively. (d) Comparison of observed annual mean APO with the model simulation results of Gruber et al. (2001). Individual profiles of the observed APO are shifted to visually fit the model-simulated APO profile.

The presence of an oxygen bulge is also supported by new data from Michael Bender and colleagues. One paper, first-authored by Mark Battle (a former Princeton postdoc now at Bowdoin College), analyzed the team’s results together with those of Ralph Keeling (Scripps Institute of Oceanography) to determine the meridional gradient of the atmospheric O2/N2 ratio. Battle showed that, as predicted by Sarmiento’s group, there is a maximum at the equator that is due to high rates of photosynthesis supported by the supply of nutrients to tropical waters by upwelling.

In a related project, Michael Bender’s team has worked on updating the global database of argon/nitrogen ratios. A previous modeling study had indicated a significant discrepancy in timing of observed and model gas fluxes between the atmosphere and ocean, indicating a problem with mixed layer physics in the GFDL ocean model. The model was improved in response to this finding, and the team hopes that incorporating new data to almost double the number of measurements will provide a more accurate diagnostic to compare with model simulations

Finally, an important analysis by Sarmiento’s group showed that estimates of contemporary oceanic uptake of anthropogenic carbon cluster quite tightly if models that do not fit radiocarbon and CFC observations are first eliminated. Taken together with other recent research such as the air-sea fluxes obtained by ocean inversions carried out by Sarmiento’s group, there is now a convergence of oceanic uptake estimates of anthropogenic carbon around a value of 2.2 ± 0.3 Pg C yr-1 for the early 1990s, a number consistent with observation-based estimates made by Michael Bender’s group (Table 2). The study highlights the need for standard data-based metrics for testing ocean carbon cycle models, and for sensitivity testing to determine the reason for the discrepancies among simulations.

Table 2. Converging estimates of yearly uptake of anthropogenic carbon by the ocean sink.