A global model of the marine ecosystem for long-term simulations: sensitivity to ocean mixing, buoyancy forcing, particle sinking, and dissolved organic matter cycling

A new model of the marine ecosystem coupled into a global Earth System Climate Model suitable for long-term (multimillennial timescale) simulations is presented. The model is based on nitrate as the sole limiting nutrient. Prognostic equations for nutrients, phytoplankton, zooplankton, and detritus are solved online in the three-dimensional ocean circulation model component. Experiments with different parameterizations of vertical mixing, including a scheme of tidally driven mixing, changes in buoyancy forcing in the Southern Ocean, different particle sinking velocities, and the inclusion of dissolved organic matter are performed, and the results are compared with observations. The results reemphasize the roles of Southern Ocean freshwater forcing and diapycnal mixing in the low-latitude pycnocline in setting the global deep water circulation and properties. The influence of high mixing in the Southern Ocean as inferred from observations is much more limited. The deep water circulation also has a strong influence on the marine ecosystem and nutrient distributions. We demonstrate that larger values of vertical diffusion lead to a shallower nutricline due to increased upwelling. Export production and nutrient distributions respond sensitively to changes in mixing and to the ratio of particle sinking to remineralization in the upper ocean. The best fits to global measurements of temperature, salinity, deep ocean radiocarbon, mixed layer depth, nutrients, and chlorophyll are obtained for values of vertical mixing in the pycnocline of around 0.2–0.3 × 10?4 m2/s and for e-folding depth for particle remineralization of 100–200 m. A simple parameterization of dissolved organic matter dynamics increases primary production and nutrient concentrations in the upper ocean and improves chlorophyll distributions in the subtropical gyres but has no discernible influence on particulate export fluxes. Remaining model deficiencies are identified, and strategies for future model improvement are outlined