We developed a coupling scheme for the Community Earth System Model version 1.2 (CESM1.2) and the Model of Early Diagenesis in the Upper Sediment of Adjustable complexity (MEDUSA), and explored the effects of the coupling on solid components in the upper sediment and on bottom seawater chemistry by comparing the coupled model's behaviour with that of the uncoupled CESM having a simplified treatment of sediment processes.
CESM is a fully coupled atmosphere–ocean–sea-ice–land model and its ocean component (the Parallel Ocean Program version 2; POP2) includes a biogeochemical component (the Biogeochemical Elemental Cycling model; BEC).
MEDUSA was coupled to POP2 in an offline manner so that each of the models ran separately and sequentially with regular exchanges of necessary boundary condition fields.
This development was done with the ambitious aim of a future application for long-term (spanning a full glacial cycle; i.e.
For Earth system models, the simulation of biogeochemical cycles in the ocean is of fundamental importance.
Simulating biogeochemistry is important for the projection of unknown (e.g. future) climate states in a prognostic way, because the biogeochemical cycles play an active role in the climate system by changing greenhouse-gas concentrations in the atmosphere particularly through the carbon cycle.
Secondly, biogeochemical tracers are an important indicator of water masses, and thus a measure of the model quality in representing ocean structures when comparing model states with observations and reconstructions.
The distribution of biogeochemical matter in the ocean is determined by internal processes (e.g. physical volume transport, mixing of seawater, and the biological pump) and processes at the upper and lower boundaries.
The latter factors, boundary conditions in terms of numerical modelling, consist of two aspects
To simulate the sedimentary diagenesis, different modelling approaches with a variety of complexity have been used for paleoclimatological or global biogeochemical studies
For the climate part of the coupled model, we employed the Community Earth System Model (CESM;
The ocean model was also extended with the carbon-isotope component developed by
For the sediment model part, we adopted the Model of Early Diagenesis in the Upper Sediment of Adjustable complexity (MEDUSA;
In this study, MEDUSA was configured such that it treated explicitly eight solid components and nine solute components (Fig.
A schematic illustration of this study's coupling scheme. In the list of chemical species, “D” stands for “dissolved” and “I” for “inorganic”; for example, DO means dissolved oxygen and DIC dissolved inorganic carbon.
OM stands for organic matter.
Each of OM and calcite components had three categories (for
The communication between the two models was done in a so-called “offline” manner; that is to say, we kept the executables of both models separate and exchanged necessary information for their boundary conditions through file exchange.
We adopted the offline coupling considering the much longer characteristic timescale of the sediment model (e.g. a model time step in Modifying POP2 involves
introduction of new variables for the matter exchanged with MEDUSA as shown in Fig. adjusting writing/reading routines for boundary conditions describing the additional variables; modifying source/sink terms in the tracer prognostic equations for the bottom grid cells of the ocean model; and changing the formulation of the boundary conditions at the ocean floor for the particulate matter. Modifying MEDUSA involves
creating writing/reading routines for boundary conditions; and unit conversion for variables to be exchanged with POP2. Interfaces between the two models include
format adjustment for the input/output files to utilize the existing schemes of both models as much as possible; and automation of procedures: routines for each step of one-time coupling and a wrapper-level routine to repeat them.
For the coupling, CESM and MEDUSA were run sequentially as in the coupling between the atmosphere and ocean components of CESM
First, we spun up the two models.
CESM was initialized with the model state at the 507th year of the preindustrial run with the same resolution using the Community Climate System Model version 4 (CCSM4, the model preceding CESM1.2) by
Following the spin-up sequence, we made two experiments.
The first was a sequentially coupled CESM-MEDUSA run for another 100 surface years with the same acceleration method for the deep ocean as described above (EXCPL).
The second one was also run for 100 surface years but as a continuation of the “uncoupled” CESM spin-up run (EXORG).
The latter experiment was done to examine the effect of the coupling of the process-based sediment model at millennial timescales.
Again, it was not long enough for
In EXCPL, the two models communicated with each other 10 times during the 100 years; that is, the coupling interval was 10 surface years for CESM, which was equivalent to 200 years for the deepest ocean domain (i.e. deeper than 3500 m).
At the end of each 10-surface-year CESM simulation, the annual mean values of the necessary variables from the last surface year were passed to MEDUSA.
We ran MEDUSA for 200 years each time with a 10-year time step (see also the Supplement), and the model output at the last time step was used as input to CESM.
Giving priority to the deepest ocean domain that occupies as much as
Model performance was assessed by comparing the results to several observation-based datasets.
The most straightforward benchmark quantities in the context of model–data comparison relevant for this study are the weight fractions of the solid components in the upper sediment. Here, we focus on the surface sediment calcite, opal, and organic carbon (OC) for which
First, we evaluate the performance of the ocean component of the coupled model based on the average over the last CESM run (i.e. 10 surface years) in EXCPL.
The maximum transport of the Atlantic meridional overturning circulation (AMOC) is 16.6 Sv (1 Sv
Comparison of globally integrated biogeochemical quantities of this study with previous estimates available from observations. For EXCPL, fluxes to allow consistent comparison with the previous estimates were calculated from the time averages over the last CESM run (10 surface years).
To evaluate the model performance of the sediment part in the coupled model, we diagnostically obtained the weight fraction of solid components in the upper sediment from the outputs of MEDUSA.
The weight fraction for calcite (Fig.
The weight fraction of the
Saturation state for calcite of the bottom seawater (
We also analysed the model performance in seven geographical regions (Fig.
In the North Atlantic, the spatial distributions are not consistent, although the modelled region-mean weight fraction is comparable to that derived from the data.
For example, the calcite weight fraction is significantly higher in the western North Atlantic in the model results than in the observation-based data by
Noticeable discrepancies in the region-mean calcite weight fractions are found in the eastern South Pacific and in the Indian Ocean.
The Pacific anomaly, which shows a too-low modelled calcite weight fraction, is caused by too-corrosive bottom water.
The
Distribution of marine biogeochemical tracers:
The weight fraction of the organic-carbon component in the upper sediments:
The simulated weight-fraction fields for OC and opal show that they are minor components in general compared to calcite, and that is consistent with the observation-based data (Figs.
The weight fraction of the opal component in the upper sediments:
While the model performance with regard to the calcite weight fraction may be improved to some extent by changing the model parameters of MEDUSA that govern the calcite dissolution rate, we keep the default parameter values for EXCPL in this study, which helps to assess the model performance in a standard setting. We judge the general model performance including the reproduction of the approximate pattern of global solid weight-fraction fields to be adequate at this stage, at least for the following analyses and discussion that does not require an accurate reproduction of the observations.
Although the development of the coupled model in this study has been motivated by the aim of simulating the glacial–interglacial variations including the marine carbon cycle as an open system
The weight-fraction distribution for EXORG shows that the uncoupled model behaves differently.
The rough feature of the global distribution of the calcite weight fraction in EXORG is similar to that in EXCPL or the observation-based data because of the appropriate depth of the prescribed lysocline (Fig.
The noticeable differences in the weight fractions of OC and opal between EXCPL and EXORG are mostly caused by the different degrees of preservation of those two species in the upper sediment.
Burial ratios (the ratios of burial amount to the flux to the ocean bottom) of OM and opal calculated by MEDUSA in EXCPL are remarkably different from those given by the highly simplified parameterization in the original CESM (Fig.
Sediment burial ratios versus the particulate flux to the ocean floor for
As to the ocean state, EXORG has large-scale properties very similar to those for EXCPL; that is to say, in EXORG (EXCPL), the maximum strength of AMOC is 16.7 Sv (16.6 Sv), the global export production is 8.1 GtC yr
Globally integrated annual mean deposition flux of particulate matter to the sediment and their burial flux (in parentheses) at the end of EXCPL and EXORG.
Total inventories in the global ocean of DIC, ALK, and
However, the effect of the interactive coupling of MEDUSA on the local bottom-water chemistry is not negligible.
The difference in
The difference in the chemical composition of the deepest grid cells between EXCPL and EXORG that was obtained from the time averages over the last 10 surface years of the CESM run in each experiment.
A similar explanation, however, is not applicable to the eastern equatorial Pacific having positive
The flux of particulate organic carbon at the top of the sediment.
The time averages over the last CESM run (10 surface years) in EXCPL are shown.
The same scale as that in Fig. 5a of
Oxygen fluxes at the water–sediment boundary
On the other hand, the remarkable dipole structure of the
DIC fluxes at the water–sediment boundary at the last time step of the last MEDUSA run in EXCPL. The sign is downward positive; that is to say, positive values correspond to fluxes from ocean to sediments.
The most straightforward advantages of coupling CESM to MEDUSA are two-fold. First, the sediment model offers the explicit modelling of chemical and physical processes in the upper sediments, and second, modelled sediment stacks provided the climate model with sedimentary “archives”.
In future applications, those two advantages will facilitate a direct comparison between the climate model and (paleoceanographic) data taken from sediments, which will provide a valuable constraint on the model from a paleoclimatological/paleoceanographic viewpoint.
Otherwise, one would need to translate records obtained from sediments in an empirical way to corresponding variables of the ocean model, which would introduce an additional source of uncertainty to the model–data comparison.
Those advantages are clearly demonstrated in the comparison of the solid weight-fraction distribution among EXCPL, EXORG, and the observation-based data.
Additionally, the state of the upper sediments at a certain time has a vertical structure reflecting the “memory” of past states because the vertical mixing of the solid phase occurs by means of bioturbation.
MEDUSA has an adequate model structure including interphase biodiffusion
In addition to the direct advantages, the sediment model will influence the simulated ocean biogeochemistry by providing more realistic boundary conditions.
The early diagenetic processes in the upper sediments produce the chemical fluxes to the ocean and hence directly affect the chemical composition of the bottom water.
The results of this study suggest that the feedback from the upper sediments would have substantial impacts on the bottom-water chemistry even at millennial timescales.
Consequently, it would be worth considering carefully how to model the sediment feedback in an ocean or climate model, and a prognostic sediment model that simulates the early diagenetic processes explicitly will have an advantage, especially if it is used for a climate simulation covering different climate states and the transitions between them.
In this study, the MEDUSA coupling produces
As a future application of the coupled model, we aim at investigating the role of sedimentary diagenesis in the climate changes at glacial–interglacial timescales.
In this context, one of the future tasks will be a simulation of the evolution of the atmospheric carbon dioxide concentration (
We consider that using comprehensive models, given their higher computational cost, for that purpose has at least three advantages over using EMICs.
First, EMICs typically use more empirical parameterizations than process-based representations of physical (and other) phenomena in their model components to realize a more efficient computation.
For many EMICs, this applies in particular to the atmospheric component.
Such model representations cannot properly capture the feedback from variations in model input if it is beyond the range of the underlying empirical relationship.
From this viewpoint, comprehensive models would be more advantageous to simulate the response of the atmosphere or the ocean to the variation in the sediment component in a long-term transient “paleo” simulation that explores climate states very different from the present day.
Second, the ocean component of some EMICs is of lower dimension
We coupled a dynamical model of early diagenesis in ocean sediments (MEDUSA) to the ocean component including a biogeochemical module of an advanced comprehensive climate model (CESM1.2). A simulation for the modern climate state demonstrated that the coupled CESM-MEDUSA model is able to approximately simulate the observed global patterns of solid composition in the upper sediments.
The comparison between the coupled and uncoupled models shows that the coupling of MEDUSA only has minor effects on the bulk properties of the global ocean in millennial-timescale climate simulations, as expected from the characteristic timescale of sedimentary processes. This study, however, reveals that the sediment–model coupling is significant in two aspects even at such a timescale. First, the simulated sediments provide an additional measure of model performance, and the observation-based global distributions of sediment properties are much better reproduced by CESM coupled to MEDUSA than by the uncoupled CESM. Secondly, some immediate effects of the sediment–model coupling are found in the chemical composition of the bottom water. The difference in the chemical composition of the bottom water between the MEDUSA-coupled model and the uncoupled model is large in the regions of high POC flux to the sediment, which suggests that it would be important to simulate the remineralization of POC in the upper sediments appropriately depending on the bottom-water chemical composition (e.g. oxygen availability). Additionally, the different treatments of the sediment processes can result in some visible displacement of the water masses in the deep ocean, which causes the different distributions of chemical tracers.
The MEDUSA coupling will yield another remarkable advantage over the original model with regard to the
The newly developed model source codes to tailor CESM1.2 and MEDUSA (version 359 or newer) for the coupling and the routines to make input files for either model from output files of the other are available in
The supplement related to this article is available online at:
TK developed the model code for the coupling with input from AP and GM. TK and AP designed the experiments, and TK carried them out. TK interpreted and discussed the results with contributions from all co-authors. AP, UM, and MS conceptualized the overarching research goal and acquired the financial support leading to this publication. TK prepared the manuscript with contributions from all co-authors.
The authors declare that they have no conflict of interest.
The authors would like to thank the reviewers for their insightful comments and suggestions.
This research was funded by the PalMod project (
This research has been supported by the German Federal Ministry for Education and Research (BMBF) (grant no. FKZ: 01LP1505D). The article processing charges for this open-access publication were covered by the University of Bremen.
This paper was edited by Paul Halloran and reviewed by two anonymous referees.