A quasi-one-dimensional coupled climate-carbon cycle model 2. The carbon cycle component

A quasi-one-dimensional, coupled climate-carbon cycle model is presented which consists of two polar domains and one nonpolar domain. The model simulates the distribution of dissolved inorganic carbon (DIC), alkalinity, phosphate, dissolved oxygen, and temperature and contains a biological pump with production of organic tissue, calcite, and arogonite. Bottom water is conditioned in one polar domain through interaction with the atmosphere and convective mixing and is injected into the lower portion of the nonpolar domain. Intermediate water flows into the other polar domain and upwells. Successful simultaneous simulation of the observed distribution of all the tracers (including isotope ratios) requires (1) an upwelling velocity in the nonpolar domain that peaks around 2 m yr (-1) at a depth of 1 km, with a gradual decrease above and below this depth; (2) effective vertical diffusion coefficients for temperature and other tracers that are different in the upper 0.5 km; and (3) a carbonate carbon to organic carbon production ratio of only 0.09. These requirements are consistent with physical considerations and/or observational evidence. In particular, observational data combined with a consideration of mixing along isopycnal surfaces and model results both indicate that the effective vertical diffusion coefficient in the upper ocean should be smallest for temperature and largest for oxygen, with the values for alkalinity and phosphate modestly smaller than for DIC. The model parameters obtained by tuning the model to preindustrial tracer distributions also provide the best (and generally excellent) fit to observed transient isotope changes. Interactions between alkalinity and DIC modulate the effect on steady state atmospheric pCO(2) of changes in the model parameters. However, the model uptake of anthropogenic CO2 and the computed atmospheric CO2 variation to year 2200 are remarkably insensitive to the choice of model mixing parameters, given that these are assumed to be constant during a given simulation. Finally, the sensitivity of the model atmospheric pCO(2) to changes in the temperature of warm ocean surface matches that obtained by three-dimensional ocean carbon cycle models.