WHOI PO

Estimating ocean carbon uptake: Global integrals to mCDR additionality

GalenMcKinley

ColumbiaUniversity and Lamont-Doherty Earth Observatory

Since the preindustrial era, the ocean has removed about40% of fossil CO₂ from the atmosphere, and it will eventually absorbat least 80% of human CO₂ emissions. At the global scale, threeindependent approaches - numerical models, machine learning based dataproducts, and interior observations - allow for quantification of the oceancarbon sink of the last few decades to within 25% uncertainty. This impressiveagreement is first-order attributable to the strong forcing from rapidly risingatmospheric pCO₂. This level of agreementacross methods would likely be acceptable if scientific curiosity were the solemotivator. However, since managing the global climate means managing the globalcarbon cycle, it is critical to further drive down uncertainties. I demonstratehow machine learning can be applied to combine hindcast models with surfaceocean pCO₂ data, providing improved quantification. I also show howCMIP models can serve as a testbed to identify areas where additional ocean pCO₂observations are most needed. Given the outsized role of the ocean in mitigatinganthropogenic CO₂ emissions, the concept of engineering marineCarbon Dioxide Removal (mCDR) to enhance the ocean sink is now of greatinterest. mCDR projects will need to demonstrate that any carbon absorbed is“additional” to what would have occurred without intervention. By comparison toindependent observations at 1x1 degree monthly resolution, I illustrate thattoday’s models and products will need substantial refinement before expectedsignals from mCDR could be distinguished from the uncertainty of theunperturbed state.