Impact of Ocean Physical Conditions on Ocean Carbon Pumps and Atmospheric CO2

During glacial periods, atmospheric CO₂ concentrations are known to have been about 90 ppmv lower than during interglacial. However, climate models have not been able to fully reproduce this decrease, partly due to large uncertainties in changes in ocean physical fields. In this study, we evaluate the impact of uncertainties in ocean physical fields on atmospheric pCO₂ during the Last Glacial Maximum (LGM) using a single offline ocean biogeochemical model forced by 12 ocean physical states derived from PMIP.

The simulated glacial atmospheric pCO₂ reduction is 40.3 ± 7.8 ppmv on average, with a large inter-model spread. This reduction mainly comes from the SST-dependent solubility effect (−30.1 ± 5.6 ppmv) and the enhanced efficiency of the organic matter pump (−21.6 ± 6.6 ppmv), cancelled somewhat by the response of the gas-exchange pump (+6.2 ± 9.4 ppmv). Our analysis suggests that the enhanced efficiency of the organic matter pump is associated with the older deep-water age in the glacial ocean and the response of the gas-exchange pump appears controlled by the SST contrast between the North Atlantic and the Southern Ocean.

We find that models with older radiocarbon deep-water ages exhibit more efficient sequestration of carbon transported by the organic matter pump into the deep ocean, leading to a larger glacial reduction in atmospheric pCO_2. However, all models used in this study underestimate the deep-water radiocarbon ages suggested by Δ¹⁴C paleoclimate records. In addition, both the global mean SST and the global mean ocean temperature are tend to be underestimated in the model compared to paleoclimate proxy reconstructions, leading to the smaller contribution of the SST-dependent solubility effect to the pCO₂ reduction. If such model biases (i.e. underestimation of deep-water ages and the SST cooling) are corrected, we estimate that the corrected model estimate of the glacial pCO₂ reduction becomes up to ~65ppmv which is still not enough for 90 ppmv reduction obtained from ice core record. Our results imply that the improvement in the reproducibility of the glacial ocean physical field alone are insufficient to fully account for the glacial atmospheric CO₂ reduction and further improvements in the representation of ocean biogeochemical processes are also required under constraints including carbon isotope records.