Alkalinity and Sensitivity to Alkalinity Enhancement in CMIP6 Earth System Models

We evaluated the large-scale alkalinity distribution in 14 CMIP6 models against the observational data set GLODAPv2 and found that most models as well as the multi-model-mean underestimate alkalinity at the surface and in the upper ocean, while overestimating alkalinity in the deeper ocean. The dissection of the global mean alkalinity biases into their contribution from physical processes (preformed alkalinity), remineralization, and carbonate formation and dissolution showed that the bias stemming from the physical redistribution of alkalinity is dominant. However, below the upper few hundred meters the bias from carbonate dissolution can become similarly important as physical biases, while the contribution from remineralization processes is negligible. In light of ongoing and planned projects involving ocean alkalinity enhancements experiments using ESMs, a back-of-the-envelope calculation was conducted with each model’s global mean surface alkalinity and dissolved inorganic carbon (DIC) as input parameters. It was shown that the degree of compensation of DIC and alkalinity biases at the surface matters more than the alkalinity biases themselves for additional CO₂ uptake capacity. The global mean surface alkalinity bias relative to GLODAP ranges from -3.6% to +2.1% with a mean of -1.1%, while for DIC the relative bias ranges from -2.6% to +2.5% with a mean value of -0.6%. Because of this partial compensation, all but two of the here evaluated CMIP6 models overestimate the Revelle factor at the surface and thus overestimate the CO₂-draw-down after an alkalinity addition of 100 µmol kg^-1  by up to 13% and the pH increase by up to 7.2%. This probable overestimate has to be taken into account when reporting on efficiencies of ocean alkalinity enhancement experiments using CMIP6 ESMs.