the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 14 Jul 2025
Sediment Biogeochemistry Model Intercomparison Project (SedBGC_MIP): motivation and guidance for its experimental design
Abstract. Benthic biogeochemical models are critical for understanding and predicting seafloor processes that regulate ocean chemistry, carbon sequestration, benthic habitat conditions, and climate feedbacks. However, current sediment models have limited predictive capabilities with widely variable complexity, structure, and underlying assumptions, highlighting a lack of consensus on essential process representations. To address this issue, this paper introduces the Sediment Biogeochemistry Model Intercomparison Project (SedBGC_MIP), a community-driven initiative aimed at systematically comparing existing benthic models against available observational constraints to refine key parameterizations and assess structural uncertainties. We review the state of sediment biogeochemical modeling, highlighting discrepancies in the representation of carbon cycling, burial, and redox remineralization processes across different model complexities. Through case studies, we demonstrate how varying model structures and ecosystem dynamics create uncertainty in global predicted biogeochemical feedbacks. We outline the objectives of SedBGC_MIP, including the need for standardized benchmarking, observational datasets, and cross-disciplinary collaboration to improve model skill and integration into Earth System Models. Ultimately, SedBGC_MIP aims to advance our ability to simulate benthic processes with greater accuracy, enhancing projections of ocean biogeochemistry under climate change scenarios with new capacity to address emerging living marine resource and geoengineering applications.
- Preprint
(3029 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
CEC1: 'Comment on egusphere-2025-1846 - No compliance with the policy of the journal', Juan Antonio Añel, 28 Jul 2025
-
AC1: 'Reply on CEC1', Samantha Siedlecki, 29 Jul 2025
I apologize for the confusion. We can update most of the repositories quickly (see below) but the ROMS model experiment described in Kearney et al (in review) requires a bit more time to accomplish as the co-author responsible is on vacation currently. She ensured us that the repository will be updated upon her return after August 9th. I hope this is alright. We will provide a completely updated and modified Code and Data Availability section at that time. Please instruct us as to how to submit a modified version of the manuscript with that section so it will be updated online. We appreciate the opportunity to correct this matter and appreciate your patience.ERSEM: Marine Systems Modelling group, P. M. L. (2022). ERSEM (22.11). Zenodo. https://doi.org/10.5281/zenodo.7300564FABM: Bruggeman, J., & Bolding, K. (2025). Framework for Aquatic Biogeochemical Models (v2.1.5). Zenodo.https://doi.org/10.5281/zenodo.14871484GOTM: Bolding, K., Bruggeman, J., Burchard, H., & Umlauf, L. (2021). General Ocean Turbulence Model - GOTM (v6.0.2). Zenodo. https://doi.org/10.5281/zenodo.4896611Global 1/2 degree MOM6+COBALT code from experiments alongside results: https://doi.org/10.5281/zenodo.15224381Citation: https://doi.org/
10.5194/egusphere-2025-1846-AC1 -
CEC2: 'Reply on AC1', Juan Antonio Añel, 29 Jul 2025
Dear authors,
Many thanks for addressing the outstanding issue so quickly. Please, let us know when you have a repository for the ROMS code.
Also, at this stage it is not necessary that you upload a new version of your manuscript. It is enough that you reply to this comment with the text for the new "Code and Data Availability" section, containing the requested information. Later, in potentially reviewed versions of your manuscript or before acceptation, you will have the opportunity to update the text in the manuscript.
Juan A. Añel
Geosci. Model Dev. Executive Editor
Citation: https://doi.org/10.5194/egusphere-2025-1846-CEC2 -
AC2: 'Reply on CEC2', Samantha Siedlecki, 19 Aug 2025
Please find the new code and data availability statements below. Let us know if you need anything further.
Code Availability: The COBALTv2 model for the Global 1/2 degree MOM6+COBALT code from experiments alongside results are archived on Zenodo: https://doi.org/10.5281/zenodo.15224381. ERSEM model code is available at ERSEM: Marine Systems Modelling group, P. M. L. (2022). ERSEM (22.11). Zenodo. https://doi.org/10.5281/zenodo.7300564. FABM: Bruggeman, J., & Bolding, K. (2025). FABM framework available at Framework for Aquatic Biogeochemical Models (v2.1.5). Zenodo.https://doi.org/10.5281/zenodo.14871484. GOTM: Bolding, K., Bruggeman, J., Burchard, H., & Umlauf, L. (2021). General Ocean Turbulence Model - GOTM (v6.0.2). Zenodo.https://doi.org/10.5281/zenodo.4896611. Source code and supporting data for the Kearney et al. (in review) implementation of ROMS with 3 biogeochemical models: ROMS source code (with ocean, sea ice, and extended biological modules): beringnpz/roms: v2025.02 https://doi.org/10.5281/zenodo.16808934. ROMS Bering Sea application data: beringnpz/bering-Apps: v2023.10 https://doi.org/10.5281/zenodo.16808952. ROMS Communication Toolbox: beringnpz/romscom: v2023.04 https://doi.org/10.5281/zenodo.16808944
Data Availability: No new observations were generated as part of this work. The exact version of the model used in the case study surrounding the results described in Kearney et al. (in review), to produce the results used in this paper is archived on Zenodo under DOI (10.5281/zenodo.15015214, Kearney et al. 2025b), as are input data and scripts to run the model and produce the plots for all the simulations presented in this paper (Kearney et al. in review), compendium, and supporting data: kakearney/supplementary-data-bgcmip: v0.4 https://doi.org/10.5281/zenodo.15015215. The results used in this paper for the case study with ERSEM are archived on Zenodo under DOI link 10.5281/zenodo.15235658 (Lessin 2025). The results used in this paper for the case study with the global model experiment is archived on Zenodo under DOI link 10.5281/zenodo.15224380 (Rakshit & Luo, 2025).
Citation: https://doi.org/10.5194/egusphere-2025-1846-AC2
-
AC2: 'Reply on CEC2', Samantha Siedlecki, 19 Aug 2025
-
CEC2: 'Reply on AC1', Juan Antonio Añel, 29 Jul 2025
-
AC1: 'Reply on CEC1', Samantha Siedlecki, 29 Jul 2025
-
RC1: 'Comment on egusphere-2025-1846', Anonymous Referee #1, 17 Sep 2025
Summary:
Siedlecki, Nmor and colleagues introduce the Sediment Biogeochemistry Model Intercomparison Project and aim to provide guidance for its experimental design. The authors raise several valuable points about what such a MIP should deliver – for example, "a systematic characterization and comparison of current benthic models"; "identify and improve the key model parameterizations"; "identify observational data sets for evaluation" (lines 310-330) and create a protocol for intercomparison experiments (line 387). However, these recommendations remain vague and provide little concrete guidance on how to implement such a project. The three case studies presented do not help to clarify these issues, as it is unclear whether they are intended as blueprints for the MIP experiments, and they provide neither specific recommendations nor clear implications for the design of the MIP. Overall, the manuscript reads more like a general workshop summary, supplemented with a few simple model experiments, rather than a concrete guide to MIP design. As a result, the purpose of this publication, and the ways in which the information presented will benefit the community, remain unclear to me.At a minimum, I would a publication introducing a MIP to present a clear protocol for intercomparison experiments (as stated will be the next step, line 387 ff), to list specific standard tests for benchmarking participating models (mentioned in line 405), and to summariz specific observational data products that model results will be compared against. While these elements are mentioned in the manuscript, they are not developed in detail. In my view, it is self-evident that any MIP would need to include such methodology, and thus a dedicated publication to motivate this in a general way, as done here, seems unnecessary. As a result, the need for such a vague MIP proposal remains unclear to me.
General comments:
It remains unclear whether the authors envision intercomparison of benthic dynamics within coupled Earth system model (ESM) setups, or whether they propose stand-alone experiments with diagenetic models under (e.g., forced by the same boundary conditions), followed by recommendations for which representations should be included in ESMs. Perhaps a mixed approach is intended, but this is not stated explicitly. Table 1, which includes both ESM-embedded and stand-alone diagenetic models, does not help clarify this point.
Current Modeling approaches
Given that comparing diagenetic approaches is the main purpose of this publication, the summary of existing modeling approaches strikes me as far too brief. It lacks sufficient detail, omits discussion of important advantages and disadvantages, and overlooks several recent developments that are directly relevant. For example, models such as MEDUSA (Munhoven, 2021; Ye et al., 2021), OMEN-SED (Hülse et al., 2018; Pika et al., 2021), and BROM (Yakushev et al., 2017) have recently been implemented and, in some cases, coupled to ESMs. These studies not only exemplify current modeling efforts but also provide comprehensive overviews of global and regional approaches that could provide useful examples for the summary section of current modeling approaches.
Case studies
The three case studies do not help with providing clear implications or the authors do not draw specific suggestions from them.
3a: Experiment and Fig. 2 is exactly the same as in Kearney et al., in review,10.22541/au.174301705.52428579/v1). I am not sure if this is appropriate.3b: This is a useful experiment, comparing two diagenetic approaches of different complexities. However, how do we know that the more complex model provides more realistic results? A comparison with observations would be very useful here to show the benefit of the more complex model – if it does provide more realistic results.
3c: Unfortunately, these experiments do not compare different diagenetic models with each other. The Dunne et al. (2007) approach is a simple empirically-derived equation that scales OC burial to its sedimentation rate. The two end-member scenarios do not have a representation of sediment dynamics at all, but either keep all matter in the ocean (reflective boundary or allburial) or everything is lost at the bottom of the ocean (noburial). The described results are intuitive and not surprising and it is unclear what should be learned here.
The methodology is also unclear and potentially flawed. First, 60 years are not enough to spinup the seafloor-sediment interface – ocean overturning alone would require thousands of years, and the biogeochemistry would take even longer. Second, how do the authors account for the change in the lower boundary condition? Is the ocean consistently loosing (allburial) or gaining (noburial) carbon and nutrients (and gaining and loosing O2) or is this somehow accounted for in how dissolved species are restored via weathering? Hence, I don’t believe this is a good/realistic example to show the effect of using different diagenetic model representations.
Model intercomparison proposal
I find this section very vague, and I miss clear recommendations or guidelines for the MIP.
For instance (lines 309 – 312) “A first step to facilitate this development is a systematic characterization and comparison of current benthic models ...”
Lines 322 – 324: “The goal […] is to identify and improve the key model parameterizations and formulations to upscale and extrapolate site-specific models to the resolution of current ESMs.”
Lines 325 – 326 “provides a path for achieving consensus and reducing model structural uncertainty”
line 330: “utilize and identify observational data sets for evaluation”
I wonder why (at least part of) this is not done as part of the presented publication? Such an analysis would be the minimum I would expect from a paper introducing a MIP.Moreover, the manuscript raises several important questions (lines 332–338) but provides no methodology for how these could actually be addressed.
Importance of observations
I am surprised that >50% of this section focuses on CMIP-type model output, i.e., ESM results rather than observations. Would it not be more useful to compile real world observational datasets that can be used to force the stand-alone diagenetic models or to compare the model output against? Moreover, CMIP-type models generally do not resolve sediment dynamics in detail (as the authors themselves note), since their high computational cost prevents sufficiently long integrations to spin up the deep ocean and sediments. The strong focus on CMIP output therefore seems suprising.The second half of the section mentions a few long-term, single-site observations. However, I wonder how useful such limited data are for evaluating diagenetic models that are supposed to be applied in an ESM context? But I can see it as one part of the MIP puzzle although it is not clear if/how it will be integrated into the broader methodology.
It would likely be more valuable to compile large-scale datasets (ideally global) or to explore ways of extrapolating limited observations to the global scale, in order to drive or benchmark the different diagenetic models.
While Table 2 lists the types of observations needed, this remains very abstract; I would have expected the authors to collate concrete datasets for these variables.
VISION FOR CONSISTENT BENCHMARKS AND PROTOCOLS
line 387 “The next step for a SedBGC_MIP would be to create a protocol for intercomparison experiments.”
Again, such a protocol, in my view, represents the minimum I would expect from a publication introducing a MIP.The first and last paragraphs of this section describe some technical challenges, which are typical for MIPs and may not need detailed discussion here.
Unfortunately, this section still does not clarify whether the authors intend to perform standardized tests with stand-alone diagenetic models, with models coupled to global or regional pelagic models, or some combination of both.
Additionally, not all references cited in the text appear in the reference list. For example, checking lines 27–29, Burdige and Gieskes (1983), Munhoven (2007), Couture et al. (2010), Yakushev (2017), and Hülse et al. (2018) are missing.
Citation: https://doi.org/10.5194/egusphere-2025-1846-RC1 -
RC2: 'Comment on egusphere-2025-1846', Sebastiaan van de Velde, 29 Sep 2025
This manuscript by Siedlecki and Nmor proposes a model intercomparison project for existing benthic/sediment models. While the idea in itself is interesting and I would be supportive of this effort, I am a bit confused by the presentation of the project. I am not sure that the manuscript in its current form presents an addition to the literature, nor whether it presents an actual SMIP, or merely the intention of perhaps starting to organize a SMIP. This level of vagueness should be addressed, and I would also ask consider my other comments listed below, which hopefully will help to improve this manuscript.
Kind regards
Sebastiaan van de Velde
General comments:Every model has a purpose and based on that purpose you should establish what level of complexity you need (and more is not always better in this case). This is not really addressed in this paper, but it is quite essential – what is the point of comparing two models that have different purposes? Yet the case studies presented do exactly that, so it is not really clear what the point of these are. It would be helpful to start by defining what the purpose of the SMIP would be – it seems to center on carbon cycling, but would this be to couple to GCMs, or to simulating impacts on the foodweb, or …?
The paper also fails to truly consider what other considerations one need to make beyond what processes to include, which is the computational efficiency, it is briefly mentioned in the intro, but not afterwards. As far as I am aware, coupling complex 1D models to GCMs is not feasible, which is why most GCMs use 0D meta-models (and since most of the discussions seems to center on large scale ocean cycling, I am assuming that this is maybe the main aim of a SMIP?). Through most of the discussion and model experiments, I get the feeling that the pervading sentiment is that more model complexity is better (what else was the purpose of case study 3b?). Yet, what is the point of building an extremely complex model with extensive parameterisation of all these processes, if you then have to bring it back to a 0D integrated model?
I am left a bit puzzled as to what the proposition for a SMIP model experiment is. There are a few outputs mentioned, but what would be your model benchmark? Or at least the initial trial? Sure the data might be lacking, but as you mention there are a few stations with multiyear data, so you could be concrete for your first step. To be truly a model experiment description paper, you should actually propose a SMIP model experiment…
Specific comments:
L22: would do good to have a citation for this statement
L31: ‘sequestration in the benthos’ sounds a bit awkward – do you mean in the sediment or in the benthic fauna?
L37: Mention the need to account for all different processes (e.g. iron reduction), but then none of your models do account for it, and you never show how including these processes affect simulating benthic carbon cycling
L40: ‘Furthermore, … global carbon budget’ – this sentence is meaningless. Heterogeneity in what? I see later that you used ChatGPT for an initial draft, might pay to double check for other meaningless sentences that sound good …
L53: Models will always be a simplification, and how simple your model needs to be depends on the question and purpose. So stating that ‘the processes currently oversimplified in the models’ is meaningless without saying what the actual purpose of the model is in the first place.
L62: most of the 1D sediment models (e.g. OMEXDIA from (Soetaert et al. 1996)) are able to simulate transient conditions, and have already been used to do so (see for example (van de Velde et al. 2018)). The real question to address is whether this transience is important for the context. It is not because the porewater distribution changes over time that this significantly affects the dynamics of what you are interested in. See for example the review of (Soetaert et al. 2000): you are perfectly able to use a 0D integration assuming steady-state to simulate seasonal cycles within the sediment.
L94: should the funder be mentioned here?
L97: what do you mean with benthos?
Case 3a: Was any of the three approaches better at reproducing the data? Also, has any effort been made to adapt the parameters to the local system? If you apply a parametrized setup with global parameters without adapting the parameters, it is not surprising that you end up with poor performance.
Case 3b: the extra information you get will also be highly dependent on the parametrisation of the model setup, so I would be careful with selling this as ‘complex models are better’. Has there been any comparison to data, or is this just a model exercise? In the latter case, what was the point of it? You knew in advance that the model without the biological pools would be unable to give information about them.
Case 3c: It is not very clear to me what the point of that exercise is? None of these results are unexpected, but what are we supposed to learn from it?
Refs:
Soetaert, K., P. M. J. Herman, and J. J. Middelburg. 1996. A model of early diagenetic processes from the shelf To abyssal depths. Geochimica et Cosmochimica Acta 60: 1019–1040.
Soetaert, K., J. J. Middelburg, P. M. J. Herman, and K. Buis. 2000. On the coupling of benthic and pelagic biogeochemical models. Earth Science Reviews 51: 173–201. doi:10.1016/S0012-8252(00)00004-0
van de Velde, S. J., V. Van Lancker, S. Hidalgo-Martinez, W. M. Berelson, and F. J. R. Meysman. 2018. Anthropogenic disturbance keeps the coastal seafloor biogeochemistry in a transient state. Scientific Reports 8: 1–10. doi:10.1038/s41598-018-23925-y
Citation: https://doi.org/10.5194/egusphere-2025-1846-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 1,936 | 52 | 24 | 2,012 | 24 | 25 |
- HTML: 1,936
- PDF: 52
- XML: 24
- Total: 2,012
- BibTeX: 24
- EndNote: 25
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Dear authors,
Unfortunately, after checking your manuscript, it has come to our attention that it does not comply with our "Code and Data Policy".
https://www.geoscientific-model-development.net/policies/code_and_data_policy.html
You have archived the code of the models used in your manuscript and part of the data on GitHub. However, GitHub is not a suitable repository for scientific publication. GitHub itself instructs authors to use other long-term archival and publishing alternatives, such as Zenodo. Therefore, the current situation with your manuscript is irregular. Please, publish your code in one of the appropriate repositories and reply to this comment with the relevant information (link and a permanent identifier for it (e.g. DOI)) as soon as possible, as we can not accept manuscripts in Discussions that do not comply with our policy. Also, please include the relevant primary input/output data.
Also, you must include modified Code and Data Availability sections in a potentially reviewed manuscript, containing the information of the new repositories.
Additionally, please, remove all the references to GitHub sites from the Code and Data Availability sections, as they do not serve the purpose of ensuring the replicability of your work, and only create confussion.
I must note that if you do not fix this problem, we cannot continue with the peer-review process or accept your manuscript for publication in our journal.
Juan A. Añel
Geosci. Model Dev. Executive Editor