The ocean plays an important role in the climate system by regulating the amount of carbon dioxide in the atmosphere. Carbon dioxide entering the deep waters of the ocean is removed from interaction with the atmosphere for periods of hundreds of years. A significant fraction of the excess carbon dioxide released by man's activities enters the deep waters of the ocean and plays no further part in global warming over century timescales.

The role of the Ocean Carbon Cycle Modelling Group is to represent the fate of carbon dioxide (more generally carbon) in the ocean in computer models of climate. These models are used to better understand the role of the ocean in the global carbon cycle and are being developed to be included in the next generation of climate prediction models.

This work is carried out in collaboration with scientists in the George Deacon Division at National Oceanography Centre.

Carbon dioxide from the atmosphere dissolves in the surface waters. On entering the ocean, carbon dioxide undergoes rapid chemical reactions with the water and only a small fraction remains as carbon dioxide. The carbon dioxide and the associated chemical forms are collectively known as dissolved inorganic carbon or DIC. This chemical partitioning of DIC ('buffering') affects the air–sea transfer of carbon dioxide, as only the unreacted carbon dioxide fraction in the sea water takes part in ocean–atmosphere interaction.

The dissolved inorganic carbon (DIC) is transported by ocean currents. Near the poles, cold dense waters sink towards the bottom of the ocean and subsequently spread through the ocean basins. These waters return to the surface hundreds of years later. As more carbon dioxide can dissolve in cold water than in warm, these cold dense waters sinking at high latitudes are rich in carbon and act to move large quantities of carbon from the surface to deep waters. This mechanism is known as the 'solubility pump'.

As well as being transported around the ocean, dissolved inorganic carbon is also used by ocean biology. In the surface waters, drifting microscopic oceanic plants known as phytoplankton grow. As with land based plants, phytoplankton take in carbon dioxide during growth and convert it to complex organic forms. The phytoplankton are eaten by drifting oceanic animals known as zooplankton, which themselves are preyed upon by other zooplankton, fish or even whales. During these biological processes, some of the carbon taken in during growth of the phytoplankton is broken down from the organic forms of the biology back to inorganic forms (DIC). If between the carbon uptake by phytoplankton and the subsequent return of the carbon to DIC, the biological material has been transported to depth, for example by the sinking of large biologically formed particles, there is a net transfer of carbon from the surface to depth. This process is termed the 'biological pump'. The carbon can also sink as skeletal structures of the biology which is known as the 'carbonate pump'.

It is clear from the above description of the carbon cycle in the ocean that representing the system in a computer model is a complex task. The chemical reactions of carbon dioxide in the surface waters need to be represented to provide a good estimate of the air–sea flux of carbon dioxide. There are also several mechanisms for the carbon dioxide entering at the surface to reach deep ocean waters. The physics of the ocean needs to be well represented in the model to capture the solubility pump. Ocean biology needs to be included in the model to capture the biological pump, including the carbonate pump. To represent the physics of the ocean, a version of the Hadley Centre climate model HadCM3 is used. The ocean component is presently run at a horizontal resolution of 3.75° longitude by 2.5° latitude, compared to the version used in climate prediction which has a 1.25° resolution. The reduced resolution means the model represents the ocean physics less well, but it does require less computing resources and hence greater testing of the carbon cycle in the model can be performed.

Although the model makes many simplifications of what actually occurs in the ocean, the results are encouraging. Below, the zonally integrated air–sea flux of carbon dioxide from the model is compared to the zonal integral from a global ocean climatology compiled from available observations (Takahashi et al. 1997, Proc. Natl. Acad. Sci. USA 94,8292-8299). In both the Atlantic and the Pacific the model generally compares well with the observations. The major difference between the model and observations occurs around the equator in the Pacific. The physics of the model brings too much deep carbon-rich water to the surface in the equatorial Pacific, producing a large flux of carbon dioxide to the atmosphere.

Ongoing research topics in ocean carbon cycle modelling include;

• analysing and improving the ocean component of global carbon cycle simulations which include the ocean, atmosphere and terrestrial carbon cycles;
• implementing the ocean carbon cycle model at the increased horizontal model resolution of 1.25° as used in the climate prediction models;
• coupling the production of DMS into the ocean ecosystem model (DMS is a gas produced by phytoplankton which, when it enters the atmosphere, has the potential to alter properties of clouds);
• include multiple types of phytoplankton in the ecosystem model;
• intercomparison of our ocean carbon cycle model with models from other groups around the world through the global OCMIP-2 project and the EC-funded European component, GOSAC.