Review of "Ocean biogeochemistry in the coupled ocean-sea ice-biogeochemistry model FESOM2.1-REcoM3"

By Ö. Gürses, L. Oziel, O. Karakuş, D. Sidorenko, C. Völker, Y. Ye, M. Zeising and J. Hauck

Overall comments

The manuscript “Ocean biogeochemistry in the coupled ocean-sea ice-biogeochemistry model FESOM2.1-REcoM3” is a comprehensive analysis of a new iteration (version 2.1) of the FESOM global ocean model, coupled with the biogeochemical model REcoM3 (also an updated version). The authors set out to evaluate the model for a recent time period (1958-2021) to assess if the model can be used to evaluate climate change effects and CO₂ increases on a century timescale.

Overall, I think the authors have presented a clear and thorough manuscript that was generally well-written and easy to read. The authors presented model results that showed relatively good agreement with observations, and illustrated that the new iteration of the FESOM and REcoM models had some improvements for the model’s overall performance. I believe this manuscript will be a valuable contribution. I just have some relatively minor comments, which are detailed below.

Specific comments

1. I thought in general the authors had a really nice and detailed Methods section. For example, the description of the FESOM model grid was very clear and I appreciated the section explicitly listing changes to REcoM. However, I found there were a few details missing from the Methods. See specific lines below:

• Methods: Forcing (line 200ff): I’m curious about what river forcings you used, if any? Perhaps I missed it but it doesn’t seem to be detailed in your methods. I think this could be particularly important to note since you comment at line 564-565 that future work is looking at the role of rivers for carbon and nutrient transport.
• Line 208ff: I think a few more details should be specified about the initialization fields; in particular, I think it would be worthwhile to explicitly state the years that the various climatologies span. Additionally, you point to GLODAP for Alkalinity and DIC initialization but don’t describe what years from GLODAP were used?
• Line 246: I found it interesting that you only showed modelled mean fields for 2012-2021, given the long spin-up period as well as the authors stating in other locations (e.g. the abstract) that they are analyzing the period of 1958-2021. Why was this specific period chosen?

2. I found section 3.3.3 DIC Inventory Changes  a bit hard to follow (lines 522ff). I think part of it is just some awkward wording and potentially a need for it to be a bit more fleshed out (without adding too much more text!). Additionally, the authors state that “FESOM-2.1-REcoM3 is thus one of the few ocean biogeochemistry models that falls within the range of interior ocean anthropogenic carbon accumulation”, which is again emphasized in the Conclusions. I think this should be discussed a bit more here. For example, if the numbers are readily available, I wonder if adding in estimates of C inventory from other models into Table 4 will help illustrate this point and facilitate a bit more discussion.

In regards to Table 4, perhaps it’s personal preference, but I think the units should be in the column header as well as in the table caption? I also think it’s a bit confusing to report the total inventory in 1994 vs the increase in C inventory for the other periods under the same columns/without any delineation.

3. There are a few sentences that require more references/details about references. See specifics below:

• Line 270: What other works? Maybe also include specific numbers of some sort from other studies?
• Line 277: Please reference what “known other works” you are referring to
• Figure 19: You reference a “Globally integrated annual air-sea CO₂ flux from Global Ocean Biogeochemistry Models” but don’t state what these models are. My guess is that the GOBMs used are from the Global Carbon Budget 2022 (?) but as written it's not clear and I think you're missing the appropriate reference to 2022 Carbon Budget here. I also think that it could be useful to explicitly state in the figure caption the number of GOBMs used (10 I believe?) and where in the 2022 Carbon Budget the readers can find information about these models (i.e. reference what table they're listed in). Potentially more details about the pCO2-based data products could also be briefly given.
• Line 545: Is this a spatially averaged oxygen concentration or for a specific region? What were the values presented in Cocco et al. 2013 that you’re comparing to?

Other minor comments/suggestions

• Line 23: I'm a bit confused here... since a rate of 2.9 PgC per year was given but then the authors state it's in 2021. It's particularly confusing since the previous sentences refer to a 250 year trend in increasing atmospheric CO₂
  • Line 23-24 (about terrestrial/land fluxes) is also a bit confusing
• Line 55ff: I think I would suggest moving this paragraph up to right before introducing REcoM and move the introduction of REcoM down to where you introduce FESOM. That way, you've then fully described the problem you're trying to solve and followed by an explanation of how you're solving it.
• Line 260: I’m curious if you have any theories as to why FESOM is too saline at the surface compared to climatology that could be commented on in the text?
• Line 305: Any idea why the moderately high silicic acid values in the northern high latitudes is not reproduced well?
• Line 425: I’m not sure what is meant by “A more detailed description of zooplankton can increase NPP by 25%” or even how it fits in with the surrounding sentences
• Figure 13: I’m confused about what the difference between the orange dots and brown line are for MAREDAT? This should be more clearly defined in the figure caption.
• Figure 16: Panel A-B – could be nice to denote on the colourbar (or in the figure caption) what values (positive or negative) are sources/sinks of CO₂ for those unfamiliar
• Line 480: Perhaps outside the scope of this study but comparison to more regional studies could be interesting. For example, the North Atlantic shelves are shown to act as sinks of CO₂ (which is consistent with other global studies) but some regional studies have shown that the Scotian Shelf, for example, acts as a source of CO₂ (see Shadwick & Thomas 2014: https://doi.org/10.1016/j.marchem.2014.01.009; Rutherford et al. 2021: https://doi.org/10.5194/bg-18-6271-2021)
• Lines 502ff: I think this is a super interesting result! One would definitely assume that there is a similar bias in the control and historical simulations…
• Line 514: What years of the model simulation are these flux estimates calculated over?
• Line 516: is this an observationally based estimate?

Minor suggested wording changes

• Line 4: I think this should be slightly changed to: “Marine biogeochemical models are a useful tool but, as any model, are a simplification and need to be continually improved.”
• Line 13; “Dissolved oxygen is also added as a new tracer.”
• Line 14: remove comma after 1958-2021
• Line 21: sentence is a bit awkward
• Line 55: I think remove the colon and replace it with a period
• Line 60: Use either “such as” or “e.g.” – you don’t need both
• Line 75: add comma after “Here”
• Line 85: Add a comma after “FESOM1.4”
• Line 98: suggest adding a colon after “defined” so it reads “A pair of control volumes are defined: the vector control volumes are the prisms based on elements.”
• Line 141: maybe add some commas (and remove some of the “ands”) to make things clearer?
• Lines 324-325: both sentences are a bit awkward. Line 324, maybe say something like “The larger magnitude of dissolved iron in the model”? In line 325, maybe remove the word “probably”
• Line 326-327 is also a bit awkward
• Line 384: “The more severe than expected limitation in iron” is awkward
• Line 395: awkward sentence
• Line 422: other global modelling studies?
• Line 424: global ocean and the Southern Ocean

The study presents an updated ocean-sea ice BGC model with updates to the sea ice component as well as to both the biogeochemical and physical ocean model. The model runs include spinup and pre-industrial and historical-present atmospheric forcing, respectively. The model output is compared to observations and reanalysis products, first for ocean physics (AMOC, SST, SSS) and then for ocean biogeochemistry. The authors do a good job of introducing the model and describing updates. The comparisons to other data products covers a wide range, but I would like to see a bit more on the side of quantitative rather than qualitative comparisons. With that and the corrections or responses to the comments below, I would be happy to recommend for publication.

General Comments:

Throughout the paper, there are comparisons between two datasets without any real quantification. They can be found at lines 254, 289, 449, 453. One exception is line 435. Please quantify the other comparisons similarly or in an appropriate way.

There are also comparisons between model output for 2012-2021 with an observed/reanalysis product that doesn’t have a time period: PHC (line 254), WOA DIN and DSi (line 286), GLODAP (line 435), OC-CCI (338, actually noted in Fig caption, but should be included in text), 

Figure 3, A map of just SST and SSS (in addition to the difference plots) would be helpful, either for Clim or run A.

262-263: I don’t think runs A minus B should be compared to A minus climatology. The former can be considered de-drifted output or anthropogenic signal, while the latter is how well your model matches to observations (or reanalysis product).

310-330: discussion of simulated iron. In this discussion, I don’t see any reference to the Aeolian iron deposition (other than intro at line 221-222) as an external (atmospheric) forcing. I see that the modelled value is about twice that of the Huang product. Could you please discuss the forcing here briefly?

Figure 10, and text 349-365. Fig 9 seems sufficient for your comparison. The high coastal values in OC-CCI of Fig 9, especiallly in the coastal argument, already supports the discussion of the issue with ocean color products and turbid but abiotic signal in coastal waters. I suggest removing Fig 10. I would still keep most of the turbid water discussion, but link it to Fig 9. I also suggest removing the lines 357 (from “Therefore”) to 362 up to “(Lee and Marr, 2022).”

Fig. 11, Sometimes the PP aren’t very limitted at all. Instead of just which is the most limitting factor, maybe use white for regions where all limitting factors are above 0.5 maybe? Also, should MLD be considered in concert with light limitation (i.e., a very deep MLD, despite high light at the surface, could limit PP because the plankton could sink deeper in the ML). 

Fig 14. Note that GLODAP is biased towards later in the sampling period when more measurements have been taken. For DIC in particular, that means GLODAP may have a higher value (DIC in surface ocean is increasing with time), but your model output is weighted equally over the time period. This would contribute to model-obs bias, but would not indicate a fault in the model. (This argument does not apply to Alk.)

454: “largely agree”: please quantify; “although the magnitude differs”: I’m not sure what this means, please clarify

499-505: note that the magnitude of drift is reduced, but also the sign. Both have drift that leads annual air-sea CO2 flux to increase in magnitude (FESOM1.4 is positive with a positive drift, and FESOm2.1 is neg with a neg drift), therefore, I don’t think it is self evident from that information that the flux will tend to zero with a longer spin-up.

Specific/technical comments

21: “preindustrial” should be “industrial”, or “end of preindustrial”

29: suggest changing “of the seasonal cycle” to “in accurately representing the seasonal cycle”

40: reference needed for “large interbasin gradient between the Pacific and Atlantic”

110: add “off” after parameterization

274: suggest changing “of both AMOC” to “, for both runs, of AMOC”

275: suggest adding to “a nearly constant” with “nearly constant, but with a small increase,” (I assume this is from continued model spinup drift?)

280: replace “pursues” with “is in the same area as” - This similarity is not necessarily surprising to me. In general, MLD differences are largest where MLD is largest and where horizontal gradients are largest (i.e. Labrador Sea and Southern Ocean). I assume the same for SST and SSS (I recommended to include those with Fig 3)

280: “Fig1” should be “Fig 3”

320 “quite a bit smaller” quantify please (e.g., “smaller by about a quarter magnitude”)

375 ”annal” should be “annual”

440: “(too low” please add “(which was too low in”

Fig 16, please add that positive indicates into the ocean in the caption

Fig. 17, use pCO2 instead of fCO2 to be consistent with rest of paper

488: suggest replacing “misfits” with “biases” (two times)

490 suggest changing “northern high latitudes” with “northern high latitudes negative bias”

491: suggest “Pacific” change to “Pacific positive bias”

Fig. 19, I can’t really tell the difference between the lines for pCO20products and for Models. Use a more distinct pair of line-colors please

508 (from 512-513), please move “with a constant atm...forcing (simulation B)” up to line 508, just after “control simulations”

524: change “is with 27.7 PgC” to “is 27.7 PgC, ”

525: I suggest removing “best”

533: remove one “also”

535: suggest change “too weak” to “too weak of uptake or too strong release”

545: Replace “Compared ... compared” with “Compared to a model intercomparison study of”, and also, what was the value of Cocco et al?

553: Change “sensitivities of” to “sensitivities to”