Optimization and sensitivity of a global biogeochemistry ocean model using combined in situ DIC, alkalinity, and phosphate data

Abstract

We present a systematic parameter optimization and sensitivity analysis of a three‐dimensional global ocean biogeochemistry model. We use the global data sets of dissolved inorganic carbon (DIC), alkalinity, and phosphate to constrain the parameters of a biogeochemistry model which include the stoichiometric ratios rC:P and rN:P, the fraction σ of organic material production allocated to dissolved organic matter (DOM), the lifetime 1/κ of DOM, the exponent α in the power law for the depth profile of the remineralization of particulate organic carbon (POC), the rain ratio R of CaCO3, and the e‐folding length scale d for the depth profile of CaCO3 dissolution. The data‐constrained parameter values are rC:P = 137 ± 11, σ = 0.74 ± 0.04, 1/κ = 1.7 ± 0.5 years, α = −0.97 ± 0.07, R = 0.081 ± 0.008, and d = 2100 ± 300 m. The postoptimization carbon export from POC is 15 ± 1 Gt/a and from CaCO3 is 1.2 ± 0.1 Gt/a of which 67 ± 4% dissolves above 2000 m. The ± ranges indicate an average 1% decrease in the fraction of the spatial variance in the observed tracer data captured by the model. The sensitivity of the model to its parameters is presented in terms of sensitivity patterns defined as the derivative of the model's equilibrium tracer distribution with respect to the parameters (S patterns). The soft‐tissue, carbonate, and gas exchange pump mechanisms responsible for the sensitivities are presented. The pump decomposition of the S patterns illustrates quantitatively how changes in organic and inorganic carbon fluxes are coupled with the large‐scale ocean circulation and how the gas exchange pump couples to the global ventilation patterns through changes in surface chemistry.