LAUSR

Iron is represented in biogeochemical oceanmodels by a variety of structurally different approaches employing generally poorly constrained empirical parameterizations. Increasing the structural complexity of ironmodules also increases computational costs and introduces additional uncertainties, with as yet unclear benefits. In order to demonstrate the benefits of explicitly representing iron, we calibrate a hierarchy of ironmodules and evaluate the remainingmodel-data misfit. Thefirstmodule includes a complex iron cycle withmajor processes resolved explicitly, the secondmodule applies iron limitation in primary production using prescribedmonthly iron concentration fields, and the thirdmodule does not explicitly include iron effects at all. All three modules are embedded into the same circulationmodel.Models are calibrated against global data sets ofNO3, PO4 andO2 applying a state-of-the-artmulti-variable constraint parameter optimization. The model with fully resolved iron cycle ismarginally (up to 4.8%) better at representing global distributions ofNO3, PO4 andO2 compared tomodels with implicit or absent parameterizations of iron.We also found a slow downof global surface nutrient cycling by about 30%and a shift of productivity from the tropics to temperate regions for the explicit ironmodule. The explicit iron model also reduces the otherwise overestimated volume of suboxic waters, yielding results closer to observations.