ESTIMATES OF ANTHROPOGENIC CARBON UPTAKE FROM FOUR 3-D GLOBAL OCEAN MODELS

Reference
J. C. Orr, E. Maier-Reimer, U. Mikolajewicz, P. Monfray, J. L. Sarmiento, J. R. Toggweiler, N. K. Taylor, J. Palmer, N. Gruber, C. L. Sabine, C. LeQuere, R. M. Key, and J. Boutin, Estimates of anthropogenic carbon uptake from four 3-D global ocean models Global Biogeochemical Cycles, 15(1), 43-60, 2001.

Abstract
We have compared simulations of anthropogenic CO2 in four 3-D ocean carbon-cycle models as part of the Ocean Carbon-Cycle Model Intercomparison Project (OCMIP). Predicted global uptake agrees to within +/- 20%, falling within 1.85+/-0.35 Pg C/yr for the 1980-1989 average. The Southern Ocean dominates the present-day air-sea flux of anthropogenic CO2 in all models, with one third to one half of the global uptake occurring south of 30dg S. However, the highest estimates of total uptake in that region were 70% larger than the lowest. Future global uptake estimates diverged with time. The divergence is largest just after the peak in the rate of change in atmospheric pCO2 in the mid 1990's, when each model's atmospheric CO2 was constrained to follow scenario S450 from the Intergovernmental Panel on Climate Change (IPCC). One model predicted nearly the lowest uptake during the historical period but the largest uptake after 2250. To evaluate models, we compared simulated and data-based estimates for anthropogenic CO2 and bomb 14C. We also compared the relationship between these two tracers. All models showed that inventories of GEOSECS-era bomb 14C were poorly correlated with present-day inventories of anthropogenic CO2; conversely, the correlation improved for the region north of 30dg S when 14C was sampled during the WOCE era. Recent data-based estimates of anthropogenic CO2 suggest that most of the models substantially over-predict storage in the Southern Ocean; elsewhere they generally underpredict storage by less than 20%. Systematic errors associated with data-based estimates are potentially large in the Southern Ocean and should be assessed at the regional scale by applying data-based techniques to appropriate model output.

see related articles by Gruber et al. [1996], Gruber [1998], and Coatanoan et al. [2000]

download article as PDF file: