J. C. Orr(1), P. Monfray(2), J. L. Sarmiento(3), E. Maier-Reimer(4), J. R. Palmer(5), and R. G. Najjar(6)

(1) LMCE/CEA Saclay, Bât. 709 L'Orme, F-91191, Gif-sur-Yvette Cedex, France
(2) CFR, Labo. Mixte CNRS-CEA, L'Orme, Bat. 709/LMCE, CE Saclay, F-91198 Gif-sur-Yvette, France
(3) AOS Program, Princeton University, Sayre Hall, P.O.Box CN710, Princeton, NJ 08542-0710, USA
(4) Max Planck Institut für Meteorologie (MPIM), Bundestrasse 55, Hamburg, Germany
(5) Hadley Centre, UK Met. Office, London Rd., Bracknell RG12 2SZ, Great Britain
(6) Dept. of Meteorology, The Pennsylvania State University, University Park, PA 16802, USA

Identifying weaknesses of 3-D global ocean carbon-cycle models has been the focus of the IGBP/GAIM project OCMIP (Ocean Carbon-Cycle Model Intercomparison Project, Phase 1: 1995- 1997); improving these models is the long-term objective of OCMIP. Four modeling groups have participated (see Research GAIM, Newsletter 1). OCMIP has standardized related model simulations and analysis, as necessary for rigorous comparison, and has pinpointed certain critical regions where models disagree considerably.

Differences between the four OCMIP models are most prominent in the Southern Ocean for both natural and anthropogenic CO₂ . For example, all OCMIP models absorb between one third to one half of their global uptake of anthropogenic CO₂ south of 30ºS, but the total quantity absorbed there differs by nearly a factor of two (Figure 1); most of the differences in predicted global uptake (±20%) are attributable to model differences in the Southern Ocean. The majority of the remaining anthropogenic CO₂ flux is taken up in the tropics; the northern latitudes (>30ºN) absorb relatively little due to their small areal extent.

Model comparison is informative, but how do OCMIP models actually compare with the real ocean? In the Southern Ocean predictions from the four OCMIP models appear to roughly bracket real ocean distributions of anthropogenic CO₂, based on direct comparison of model results to new data-based estimates of anthropogenic CO₂ in the Southern Atlantic [Gruber et al., 1997] and Southern Indian Oceans [Sabine et al., 1997]. Uncertainties associated with such data-based estimates for oceanic anthropogenic CO₂ are not negligible and need closer attention [Wanninkhof et al., 1997]; however, differences between models are much larger, especially in the Southern Ocean.

The classical albeit indirect means of evaluating global ocean carbon-cycle model performance is through comparison of simulated vs. measured radiocarbon. OCMIP has studied both natural and bomb ¹⁴C (Orr, 1996). Uncertainties are associated with observational-based methods used to distinguish the bomb ¹⁴C component from the natural ¹⁴C background. These uncertainties are particularly large in the Southern Ocean. One simple way to reduce uncertainties involves comparing models to the measured WOCE (1990's) minus GEOSECS (1970's) ¹⁴C difference [Key, 1997]. Such observational ¹⁴C differences are beginning to be compared to simulated results from the OCMIP models.

But what has ¹⁴C to do with CO₂? The signals of transient ¹⁴C and anthropogenic CO₂ are nearly linearly related in some regions of the ocean as suggested by OCMIP model results. Because this relationship differs between models, the observed trend may provide a constraint with which to better evaluate ocean carbon-cycle models, if data-based uncertainties for both components can be reduced.

To study changes in future ocean uptake due to CO₂-induced shifts in ocean chemistry, all OCMIP models made simulations with IPCC scenario S450 (where atmospheric CO₂ is stabilized at 450 ppm) to year 2300. For this scenario, all models predict that the Southern Ocean air-sea flux grows in importance relative to other regions of the world ocean (Figure 2); however, model predictions diverge. Particularly, the model which absorbs the least anthropogenic CO₂ at present, absorbs the most from year 2200 onwards. Such changes must in part be due to differences in deep-ocean mixing between models, which become more apparent with time as anthropogenic CO₂ penetrates more deeply.

Figure 2: (a) History of the flux of anthropogenic CO₂ from atmosphere to ocean as predicted by the four OCMIP models, all forced by observed atmospheric CO₂ from 1765 to 1995, followed by IPCC future scenario S450. Before 1990, models fall into 2 categories. It is interesting that within each group, models with explicit mixed layer dynamics (Hadley and IPSL models) respond more rapidly to the sharp maximum in the S450 scenario's rate of change of atmospheric CO2 with time (not shown), which occurs in the 1990's; air-sea fluxes lag this peak due to the slow response time of the ocean.. A larger effect is the later divergence between MPI and IPSL models. (b) Although there are many differences between models, sensitivity tests in the IPSL model show that large differences in carbon uptake are caused simply by changing the advection scheme. The advection scheme used in each of the OCMIP models is indicated by consistency in line signatures between subplots; Hadley also uses CTCS.
 

While results from the original four OCMIP models have been undergoing analysis and the first attempts have been made to determine causes of model-data and model-model discrepancies, numerous other groups have been busy developing their own global ocean carbon-cycle models. These models will be welcomed in the second phase of OCMIP: 1998 -2000. OCMIP-2 is a joint effort led by IGBP/GAIM and IGBP/JGOFS and will take advantage of insights gained from OCMIP-1. Furthermore, OCMIP-2 will benefit from greater model diversity, which offers better insurance that a range of model predictions brackets real ocean behavior.

OCMIP-2 will include 13 models: 4 from the U.S., 7 from Europe, 1 from Australia, and 1 from Japan. The U.S. effort, funded by NASA via the JGOFS Synthesis and Modeling Project, will add number and strengthen OCMIP's link with JGOFS and WOCE data synthesis communities. The European group of models is funded through the EC Environment and Climate Programme via the project GOSAC (Global Ocean Storage of Anthropogenic Carbon). GOSAC will also lead a related model comparison effort, jointly funded by the IEA Greenhouse Gas R&D Programme, to assess global aspects of the proposal which offers to artificially accelerate ocean storage of CO₂ by diverting CO₂ emissions from fossil-fuel fired power plants directly into the abyss, thereby short-circuiting the slow process of air-sea CO₂ exchange.

OCMIP-2 will look in more detail at both natural and anthropogenic components of CO₂ and ¹⁴C. New model-data comparison will help to further diagnose problems and improve participating ocean carbon-cycle models. Additionally, circulation fields will be diagnosed through new simulations for CFC's and ³He. Simulations for O₂ will help pinpoint deficiencies in different formulations for ocean biogeochemistry. Models will be run with identical ocean biogeochemistry in order that differences due to ocean circulation fields can be isolated. At first, a simple diagnostic phosphate restoring model will be used by all OCMIP models. Subsequently, OCMIP groups will employ a more sophisticated prognostic model developed by JGOFS 1-D modelers.

A third phase of OCMIP will begin in 2001. Currently on the OCMIP-3 agenda are model comparisons focused on (1) interannual variability of the air-sea CO₂ flux and (2) changing climate in the 21st century and beyond, through coupled atmosphere-ocean simulations which attempt to model how air-sea CO₂ fluxes will be affected by changing ocean circulation and circulation-induced changes in ocean biogeochemistry.

References:

Gruber, N., J. L. Sarmiento, and T. F. Stocker, 1997. Anthropogenic CO₂ in the Atlantic Ocean, 5^th International CO2 Conference, Cairns, Australia, Sept 8-12, 1997.

Key, R. M., 1997. Changes in Pacific Ocean Distribution of Radiocarbon, U.S. WOCE Report, No. 9, 5- 8,U.S. WOCE Office, Texas A&M University, College Station, TX, 77843

Orr, J.C., 1996. The Ocean Carbon-Cycle Model Intercomparison Project of IGBP/GAIM, in Ocean Storage of Carbon Dioxide, Workshop 3: International Links and Concerns, 33-52, ISBN 1 898373 04 3, CRE Group Ltd., Cheltenham, GL52 4RZ, UK.

Sabine, C. L., A.G. Dickson, C. Goyet, P. Guenther, K. M. Johnson, R. M. Key, F. J. Millero, J.L. Sarmiento, D.R.W. Wallace, and C. D. Winn, 1997. An evaluation of the anthropogenic CO₂ inventory of the Indian Ocean, The Oceanography Society Meeting, Seattle, WA, USA, April, 1997.

Wanninkhof, R., S. Doney, T.-H. Peng, R. Feely, and J. Bullister, 1997. Penetration of Anthropogenic CO₂ into the Atlantic Ocean, 5^th International CO₂ Conference, Cairns, Australia, Sept 8-12, 1997.