Multi-tracer constraints on ocean storage of anthropogenic CO2

During the second phase of the Ocean Carbon-Cycle Model Intercomparison, 13 models simulated oceanic uptake of anthropogenic CO2 for the period 1765 to 2000. For the 1980s, models agreed to within +/- 22% (1.99 +/- 0.43 Pg C yr-1, half the range over the mean). For the 1990s the OCMIP-2 models predict a 24% uptake increase (2.38 +/- 0.53 Pg C yr-1), in contrast with the 1980s-to-1990s decrease predicted by the most recent IPCC Third Assessment Report Chapter 3. However, the IPCC's estimates are based on atmospheric O2 measurements which are susceptible to error due to interannual variations in air-sea O2 fluxes. It appears likely that the OCMIP-2 range for the modern uptake of anthropogenic CO2 brackets real ocean uptake for four reasons: (1) the large model diversity; (2) the range of OCMIP-2 models bracket observed tracer constraints (CFC-11 along sections, global mean deep-ocean 14C); (3) the simulated global storage of anthropogenic CO2 correlates with the simulated global storage of CFC-11 and the simulated global-mean, deep-ocean natural C-14;(4) the simulated global inventories of anthropogenic CO2 bracket data-based estimates for that same tracer. In theory, the CFC-11 and C-14 data constraints should allow us to weight the models, and thus narrow uncertainties, based on how models perform in regards to matching ocean inventories of these independent tracers. Here we will discuss progress on this effort, in regards to global and regional inventories