Using Preformed and Remineralized Nutrients to Map Spatial Variation in Remineralization

Production of organic matter by marine phytoplankton plays a key role in the regulation of the Earth's climate. Model studies have demonstrated that atmospheric CO₂ concentrations can be very sensitive to small changes in the Remineralization Lengthscale (RL), the depth at which organic material is remineralized into CO₂ and nutrients. Therefore, observed nutrient fields offer a significant test for global biogeochemical models. However, nutrient vertical distribution is the result of local one-dimensional biogeochemistry and processes occurring remotely in areas of water mass formation. We use an approach based on preformed (N_pre) and remineralized (N_rem) nutrients to disentangle these influences and to map the horizontal distribution of RL. Using a global coupled physical-biogeochemical model we performed a set of simulations where nutrient distributions were alternately restored to observations in different regions of water mass formation in order to constrain the global distribution of N_pre. By varying RL over a wide range of values we determined the optimal RL to reproduce observations of N_rem in individual biomes. This provided an emergent global map of RL. The influence of biases due to the representation of water masses was assessed using two different circulations and the emergent map of RL was tested on a third independent representation of the ocean circulation.