Description: Global biogeochemical ocean models contain a variety of different biogeochemical components and often much simplified representations of complex dynamical interactions, which are described by many (≈10–≈100) parameters. The values of many of these parameters are empirically difficult to constrain, due to the fact that in the models they represent processes for a range of different groups of organisms at the same time, while even for single species parameter values are often difficult to determine in situ. Therefore, these models are subject to a high level of parametric uncertainty. This may be of consequence for their skill with respect to accurately describing the relevant features of the present ocean, as well as their sensitivity to possible environmental changes. We here present a framework for the calibration of global biogeochemical ocean models on short and long time scales. The framework combines an offline approach for transport of biogeochemical tracers with an Estimation of Distribution Algorithm (Covariance Matrix Adaption Evolution Strategy, CMAES). We explore the performance and capability of this framework by five different optimizations of six biogeochemical parameters of a global biogeochemical model. First, a twin experiment explores the feasibility of this approach. Four optimizations against a climatology of observations of annual mean dissolved nutrients and oxygen determine the extent, to which different setups of the optimization influence model's fit and parameter estimates. Because the misfit function applied focuses on the large-scale distribution of inorganic biogeochemical tracers, parameters that act on large spatial and temporal scales are determined earliest, and with the least spread. Parameters more closely tied to surface biology, which act on shorter time scales, are more difficult to determine. In particular the search for optimum zooplankton parameters can benefit from a sound knowledge of maximum and minimum parameter values, leading to a more efficient optimization. It is encouraging that, although the misfit function does not contain any direct information about biogeochemical turnover, the optimized models nevertheless provide a better fit to observed global biogeochemical fluxes.

Description: The Apparent Oxygen Utilisation (AOU) is a classical measure of the amount of oxygen respired in the ocean's interior. We show that AOU systematically overestimates True Oxygen Utilisation (TOU) in 6 coupled circulation-biogeochemical ocean models. This is due to atmosphere–ocean oxygen disequilibria in the subduction regions, consistent with previous work. We develop a simple, new, observationally-based approach which we call Evaluated Oxygen Utilisation (EOU). In this approach, we take into account the impact of the upper ocean oxygen disequilibria into the interior, considering that transport takes place predominantly along isopycnal surfaces. The EOU approximates the TOU with less than half of the bias of AOU in all 6 models despite large differences in the physical and biological components of the models. Applying the EOU approach to a global observational dataset yields an oxygen consumption rate 25% lower than that derived from AOU-based estimates, for a given ventilation rate.

Description: Although of substantial importance for marine tracer distributions and eventually global carbon, oxygen, and nitrogen fluxes, the interaction between sinking and remineralization of organic matter, benthic fluxes and burial is not always represented consistently in global biogeochemical models. We here aim to investigate the relationships between these processes with a suite of global biogeochemical models, each simulated over millennia, and compared against observed distributions of pelagic tracers and benthic and pelagic fluxes. We concentrate on the representation of sediment–water interactions in common numerical models, and investigate their potential impact on simulated global sediment–water fluxes and nutrient and oxygen distributions. We find that model configurations with benthic burial simulate global oxygen well over a wide range of possible sinking flux parameterizations, making the model more robust with regard to uncertainties about the remineralization length scale. On a global scale, burial mostly affects oxygen in the meso- to bathypelagic zone. While all model types show an almost identical fit to observed pelagic particle flux, and the same sensitivity to particle sinking speed, comparison to observational estimates of benthic fluxes reveals a more complex pattern, but definite interpretation is not straightforward because of heterogeneous data distribution and methodology. Still, evaluating model results against observed pelagic and benthic fluxes of organic matter can complement model assessments based on more traditional tracers such as nutrients or oxygen. Based on a combined metric of dissolved tracers and biogeochemical fluxes, we here identify two model descriptions of burial as suitable candidates for further experiments and eventual model refinements.

Description: This study presents results from 46 sensitivity experiments carried out with three structurally simple (2, 3, and 6 biogeochemical state variables, respectively) models of production, export and remineralization of organic phosphorus, coupled to a global ocean circulation model and integrated for 3000 years each. The models’ skill is assessed via different misfit functions with respect to the observed global distributions of phosphate and oxygen. Across the different models, the global root-mean square misfit with respect to observed phosphate and oxygen distributions is found to be particularly sensitive to changes in the remineralization length scale, and also to changes in simulated primary production. For this metric, changes in the production and decay of dissolved organic phosphorus as well as in zooplankton parameters are of lesser importance. For a misfit function accounting for the misfit of upper-ocean tracers, however, production parameters and organic phosphorus dynamics play a larger role. Regional misfit patterns are investigated as indicators of potential model deficiencies, such as missing iron limitation, or deficiencies in the sinking and remineralization length scales. In particular, the gradient between phosphate concentrations in the northern North Pacific and the northern North Atlantic is controlled predominantly by the biogeochemical model parameters related to particle flux. For the combined 46 sensitivity experiments performed here, the global misfit to observed oxygen and phosphate distributions shows no clear relation to either simulated global primary or export production for either misfit metric employed. However, a relatively tight relationship that is very similar for the different model of different structural complexity is found between the model-data misfit in oxygen and phosphate distributions to simulated meso- and bathypelagic particle flux. Best agreement with the observed tracer distributions is obtained for simulated particle fluxes that agree most closely with sediment trap data for a nominal depth of about 1000 m, or deeper.

Description: In the aftermath of an earthquake and tsunami on 11 March 2011 radioactive 137Cs was discharged from a damaged nuclear power plant to the sea off Fukushima Dai-ichi, Japan. Here we explore its dilution and fate with a state-of-the-art global ocean general circulation model, which is eddy-resolving in the region of interest. We find apparent consistency between our simulated circulation, estimates of 137Cs discharged ranging from 0.94 p Bq (Japanese Government, 2011) to 3.5 ± 0.7 p Bq (Tsumune et al., 2012), and measurements by Japanese authorities and the power plant operator. In contrast, our simulations are apparently inconsistent with the high 27 ± 15 p Bq discharge estimate of Bailly du Bois et al. (2012). Expressed in terms of a diffusivity we diagnose, from our simulations, an initial dilution on the shelf of 60 to 100 m2 s−1. The cross-shelf diffusivity is at 500 ± 300 m2 s−1 significantly higher and variable in time as indicated by its uncertainty. Expressed as an effective residence time of surface water on the shelf, the latter estimate transfers to 43 ± 16 days. As regards the fate of 137Cs, our simulations suggest that activities up to 4 mBq l−1 prevail in the Kuroshio-Oyashio Interfrontal Zone one year after the accident. This allows for low but detectable 0.1 to 0.3 m Bq l−1 entering the North Pacific Intermediate Water before the 137Cs signal is flushed away. The latter estimates concern the direct release to the sea only.

Description: Global models of the oceanic nitrogen cycle are subject to many uncertainties, among them type and form of biogeochemical processes involved in the fixed nitrogen cycle, and the spatial and temporal scales, on which the global nitrogen budget is regulated. We investigate these aspects using a global model of ocean biogeochemistry, that explicitly considers phosphorus and nitrogen, including pelagic denitrification and nitrogen fixation as sink and source terms of fixed nitrogen, respectively. The model explores different parameterizations of organic matter sinking speed, oxidant affinity of oxic and suboxic remineralization, and regulation of nitrogen fixation by temperature and different stoichiometric ratios. Examination of the initial transient behaviour of different model setups initialized from observed tracer distributions reveal changes in simulated nitrogen inventories and fluxes particularly during the first centuries. Millennial timescales have to be resolved in order to bring all biogeochemical and physical processes into a dynamically consistent steady state, for which global patterns of biogeochemical tracers and fluxes are reproduced quite well. Analysis of global properties suggests that particularly particle sinking speed, but also the parameterization of denitrification determines the extent of oxygen minimum zones, global nitrogen fluxes, and hence the oceanic nitrogen inventory. However, the ways and directions, in which different parameterizations of particle sinking, nitrogen fixation and denitrification affect the global diagnostics, are different, suggesting that these may, in principle, be constrained independently from each other. Analysis of the model misfit suggests a particle flux profile close to the one suggested by Martin et al. (1987). Simulated pelagic denitrification best agrees with the lower values between 59 and 84 Tg N yr−1 recently estimated by other authors.

Description: To describe the underlying processes involved in oceanic plankton dynamics is crucial for the determination of energy and mass flux through an ecosystem and for the estimation of biogeochemical element cycling. Many planktonic ecosystem models were developed to resolve major processes so that flux estimates can be derived from numerical simulations. These results depend on the type and number of parameterizations incorporated as model equations. Furthermore, the values assigned to respective parameters specify a model's solution. Representative model results are those that can explain data; therefore, data assimilation methods are utilized to yield optimal estimates of parameter values while fitting model results to match data. Central difficulties are (1) planktonic ecosystem models are imperfect and (2) data are often too sparse to constrain all model parameters. In this review we explore how problems in parameter identification are approached in marine planktonic ecosystem modelling. We provide background information about model uncertainties and estimation methods, and how these are considered for assessing misfits between observations and model results. We explain differences in evaluating uncertainties in parameter estimation, thereby also discussing issues of parameter identifiability. Aspects of model complexity are addressed and we describe how results from cross-validation studies provide much insight in this respect. Moreover, approaches are discussed that consider time- and space-dependent parameter values. We further discuss the use of dynamical/statistical emulator approaches, and we elucidate issues of parameter identification in global biogeochemical models. Our review discloses many facets of parameter identification, as we found many commonalities between the objectives of different approaches, but scientific insight differed between studies. To learn more from results of planktonic ecosystem models we recommend finding a good balance in the level of sophistication between mechanistic modelling and statistical data assimilation treatment for parameter estimation

Description: Idealised and hindcast simulations performed with the stand-alone ocean carbon-cycle configuration of the Norwegian Earth System Model (NorESM-OC) are described and evaluated. We present simulation results of two different model versions at different grid resolutions and using two different atmospheric forcing data sets. Model version NorESM-OC1 corresponds to the version that is included in the fully coupled model NorESM-ME1, which participated in CMIP5. The main update between NorESM-OC1 and NorESM-OC1.2 is the addition of two new options for the treatment of sinking particles. We find that using a constant sinking speed, which has been the standard in NorESM's ocean carbon cycle module HAMOCC (HAMburg Ocean Carbon Cycle model) does not transport enough particulate organic carbon (POC) into the deep ocean below approximately 2000 m depth. The two newly implemented parameterisations, a particle aggregation scheme with prognostic sinking speed, and a simpler scheme prescribing a linear increase of sinking speed with depth, provide better agreement with observed POC fluxes. Additionally, reduced deep ocean biases of oxygen and remineralised phosphate indicate a better performance of the new parameterisations. For model version 1.2, a re-tuning of the ecosystem parameterisation has been performed, which (i) reduces previously too high primary production in high latitudes, (ii) consequently improves model results for surface nutrients, and (iii) reduces alkalinity and dissolved inorganic carbon biases at low latitudes. We use hindcast simulations with prescribed observed and constant (pre-industrial) atmospheric CO2 concentrations to derive the past and contemporary ocean carbon sink. For the period 1990–1999 we find an average ocean carbon uptake ranging from 2.01 to 2.58 Pg C yr-1 depending on model version, grid resolution and atmospheric forcing data set.

Description: The assessment of the ocean biota's role in climate climate change is often carried out with global biogeochemical ocean models that contain many components, and involve a high level of parametric uncertainty. Examination the models' fit to climatologies of inorganic tracers, after the models have been spun up to steady state, is a common, but computationally expensive procedure to assess model performance and reliability. Using new tools that have become available for global model assessment and calibration in steady state, this paper examines two different model types – a complex seven-component model (MOPS), and a very simple two-component model (RetroMOPS) – for their fit to dissolved quantities. Before comparing the models, a subset of their biogeochemical parameters has been optimised against annual mean nutrients and oxygen. Both model types fit the observations almost equally well. The simple model, which contains only nutrients and dissolved organic phosphorus (DOP), is sensitive to the parameterisation of DOP production and decay. The spatio-temporal decoupling of nitrogen and oxygen, and processes involved in their uptake and release, renders oxygen and nitrate valuable tracers for model calibration. In addition, the non-conservative nature of these tracers (with respect to their upper boundary condition) introduces the global bias as a useful additional constraint on model parameters. Dissolved organic phosphorous at the surface behaves antagonistically to phosphate, and suggests that observations of this tracer – although difficult to measure – may be an important asset for model calibration

Description: Various functions have been suggested and applied to represent the sedimentation and remineralisation of particulate organic matter (POM) in numerical ocean models. Here we investigate some representations commonly used in large-scale biogeochemical models: a constant sinking speed, a sinking speed increasing with depth, a spectrum of particles with different size and different size-dependent sinking velocities, and a model that assumes a power law particle size distribution everywhere in the water column. The analysis is carried out for an idealised one-dimensional water column, under stationary boundary conditions for surface POM. It focuses on the intrinsic assumptions of the respective sedimentation function and their effect on POM mass, mass flux, and remineralisation profiles. A constant and uniform sinking speed does not appear appropriate for simulations exceeding a few decades, as the sedimentation profile is not consistent with observed profiles. A spectrum of size classes, together with size-dependent sinking and constant remineralisation, causes the sinking speed of total POM to increase with depth. This increase is not strictly linear with depth. Its particular form will further depend on the size distribution of the POM ensemble at the surface. Assuming a power law particle size spectrum at the surface, this model results in unimodal size distributions in the ocean interior. For the size-dependent sinking model, we present an analytic integral over depth and size that can explain regional variations of remineralisation length scales in response to regional patterns in trophodynamic state.