The riddle of eastern tropical Pacific Ocean oxygen levels:  the role of the supply by intermediate-depth waters

Duteil, Olaf; Frenger, Ivy; Getzlaff, Julia

doi:https://doi.org/10.5194/os-17-1489-2021

Articles | Volume 17, issue 5

https://doi.org/10.5194/os-17-1489-2021

© Author(s) 2021. This work is distributed under
the Creative Commons Attribution 4.0 License.

Special issue:

Ocean deoxygenation: drivers and consequences – past, present...

https://doi.org/10.5194/os-17-1489-2021

© Author(s) 2021. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 17, issue 5

Research article

|

27 Oct 2021

Research article |

| 27 Oct 2021

The riddle of eastern tropical Pacific Ocean oxygen levels: the role of the supply by intermediate-depth waters

Olaf Duteil, Ivy Frenger, and Julia Getzlaff

Download

Final revised paper (published on 27 Oct 2021)
Supplement to the final revised paper
Preprint (discussion started on 09 Mar 2020)

Interactive discussion

Status: closed

AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment

- Printer-friendly version

- Supplement

RC1: 'Review of "Intermediate water masses, a major supplier of oxygen for the eastern tropical Pacific ocean" by Duteil et al., 2020', Anonymous Referee #1, 30 Apr 2020
- AC1: 'Reply Referee 1', Olaf Duteil, 07 Sep 2020
RC2: 'Review for Duteil et al, “Intermediate water masses, a major supplier of oxygen for the eastern tropical Pacific ocean”', Anonymous Referee #2, 05 May 2020
- AC2: 'Reply Referee 2', Olaf Duteil, 07 Sep 2020

Peer-review completion

AR: Author's response | RR: Referee report | ED: Editor decision

AR by Olaf Duteil on behalf of the Authors (13 Oct 2020) Author's response Manuscript

ED: Referee Nomination & Report Request started (13 Oct 2020) by Minhan Dai

RR by Anonymous Referee #1 (08 Nov 2020)

Suggestions for revision or reasons for rejection

First of all, my apologies in the previous review comments that I made typos in acronyms,
"IWM" (Intermediate Water Mass) by IMW. Hope the comments was clear enough to
explain the points. I thank the authors for conducting a suite of additional simulations
addressing the impact of the atmospheric forcing, model integration time, and oxygen forcing simulations.

The authors have done substantial amount of additional simulations and revision and I think
the additional results (figures in response and supplementary materials) addressed the questions
from my comments. They showed that the impact of atmospheric forcing and spin-up state exists
but does not affect the question and conclusion at least in this study (i.e. strong coupling between subtropics and tropics).
The message in the summary session is organized and I think the results support the conclusions.

I still have few comments that I would like to ask the authors to address before publications.
The comments are mainly about the presentation and clarification of the paper and most of them are relatively minor comments.

1. I thank the authors for stating "conceptual reasoning" in the response. I understand why the authors choose to
utilize a suite of models and I think it is useful to combine a subset of the models and a single model sensitivity simulations
to dig into further mechanisms. The basic presentation style of this paper is to present analysis from a subset of models
(UVIC, NEMO2, and GFDL models) to explore the representations of IWM, EICS, and dissolved oxygen levels and
then show sensitivity simulations (NEMO2) to explore mechanisms, which I think is good.

The only part I have got stuck a bit is the section 4.3, which comes back to the result from the subset of models. The authors stated
in the comment "investigating mechanisms in a heterogeneous set of models by performing dedicated sensitivity simulations"
but the analysis and discussion I see is the analysis of MKE (Mean Kinetic Energy) and dissolved oxygen (i.e. more of the analysis rather than sensitivity simulations). The MKE discussion itself is good but I also had to look into the caption in Figure 9 carefully to see the results from UVIC GD13 (and this was not completely clear from just reading the section 4.3). I am not sure what the authors meant by "performing dedicated sensitivity simulations" from a heterogeneous set of models (because I only see sensitivity simulations from NEMO2) and I kind of see a slight jump from the previous sections.

As the authors stated in L413 in a short statement, a suite of tracer release and Lagrangian tracking simulations could help
interpreting the subset of the models so I suggest to include more statement on this part (perhaps including MKE and dissolved oxygen from NEMO2 sensitivity simulations help along with Figure 9?) Also, please make it more clear that you are showing the results from UVIC GD13 (perhaps pointing out the specific panel and mention blue contours, label all the panels alphabetically or just point out "UVIC" panel on the left column). I think the material presented is nice, it is more about the presentation and discussion to make a nice flow from sensitivity simulations back to a subset of models. I also think including results from UVIC GD13 is useful for discussion so it will be good to keep it in (it was just a bit unclear in the text and figure captions).

2.I would like to ask the authors to include a short statement (or suggestions) for analyzing CMIP class multi-models in summary section. I think the authors have done nice analysis on heterogeneous set of models (with the aid of sensitivity simulations exploring further mechanisms) and the results can possibly point out what we should analyze to explore the multi-model characteristics (or bias) understanding tropical OMZs. I understand the additional difficulties digging in multi-models since the models include biological effect (compensation authors discussed) but do you think analysis like in Figure 2d or 9 supports on understanding IWM and EICS impacts (or if authors have other suggestions or ideas from their simulations I would be interested to know).

3. I would like to ask for clarifications on "oxygen forcing" instead of "oxygen restoring". Actually the new term "forcing" raise additional question
(at least to me). Does this mean the authors experimental design is NOT "restoring" but just replacing the oxygen values, for example at 30N and 30S to observed values (and always fixed to observed values during the model integration)? What will be the difference? The forcing sounds to me like the atmospheric forcing (such as wind stress) changing with time but I assume that the dissolved oxygen forcing at the boundaries does not change with time (just fixed to the observed climatology) correct?

4. L98: Is the CORE-II climatological forcing "normal year" forcing or you constructed climatology based on CORE-II forcing from 1948-2007? Also, I think this does not impact the result but is there a reason why you used CORE-II forcing from 1948-2007 instead of 2009 (since CORE-II extend to 2009)?

5. What is the initial condition of sensitivity simulations (60 years simulations) by NEMO2? Do all these sensitivity simulations start from spun-up simulation from the mean state comparison (i.e. 1000 years integration from NEMO2)?

6. L39: Is it semicolon here? I would use period and separate the sentences but I will leave this to the author
(including checking with native speakers).

7. L126: "eNEMO" should be "NEMO"?

Hide

RR by Anonymous Referee #3 (06 Dec 2020)

Suggestions for revision or reasons for rejection

The authors present a variety of analyses of the controls on O2 at intermediate levels in the Equatorial Pacific. Despite some interesting results, the manuscript was remarkably scattered and unfocused, and thereby difficult to read. Part of the problem stems from an inadequate analysis of water masses, and part of the problem stems from a confusing set of experiments with a disparate set of models. The authors really need to state somewhere the abstract conclusions something like “biases in the modeled O2 concentrations in intermediate layers in the Equatorial Pacific (which we carefully define in density space) obviously reflect issues with the formation and fate of intermediate waters within the interior. This study uses A/B methods to disentangle the causes of unsatisfactory IWM O2 in the tropics, and arrives at the conclusion that the problem is X% formation, Y% fate/ventilation, and Z% bad biogeochemical modeling. The limitations of our analysis are C, and we recommend that D be used to look at this future work”. Otherwise this somewhat sprawling mess of a manuscript won’t be comprehensible and won’t contribute to broader community efforts.

The most important problem with the study is that it approached intermediate water masses (IWMs) with very little consideration of the associated water masses. Defining IWM as waters spanning 500m-1500m is scientifically flawed, and if the authors insist on having their analysis focused on that horizon then the words “intermediate water masses” should not appear in the title. There are a number of places where this problem arises. One confusing case was for the point made in Fig. 2b, namely the panel showing the observed and modeled vertical distribution of O2 at 30S in the Pacific. Fist and foremost, the authors really need to be clear about water masses. I believe that the O2 subsurface maximum is located near the boundary of SAMW and AAIW, rather than squarely in AAIW densities. This should be checked by averaging across the basin first in density at 30S. Second, with the text in lines 209-210, it wasn’t clear what the authors were saying about the “large role” of IWM. Are they referring to the “formation process” that sets the O2 content of IWM waters? Or something else?

There are a number of questions that also arise from the application of Lagrangian diagnostics. First, near line 170 in the text, it wasn’t clear whether the Lagrangian analysis included the bolus-velocities from the mesoscale parameterization? Or not? Second, it wasn’t clear from the text if the NEMO01 flow-fields were coarse-grained to the NEMO05 grid before running the trajectory analysis? But most importantly, over lines 395-401, it was puzzling that the authors invoke “qualitative mode” rather than “quantitative” mode in designing their Lagrangian experiments if their goal is to evaluate connectivity through source regions. The choice of 1000m as a release horizon isn’t justified. Why was this chosen? How does Fig. 7 help the reader to understand the critical scale and processes needed to represent IWMs in the Equatorial Pacific?

To reiterate, what is missing here is a clear exposition of scientific objectives, or a clear motivation for the specific choice of models and dye/lagrangian tracer diagnostics. A fundamental limitation of the study is the lack of a water mass set of definitions for analysis with IWMs, without this the manuscript is in my opinion would clearly need to state “we ignore water masses except in a very hand-wavy qualitative sense, and focus instead on aggregated (AAIW, SAMW, etc.) properties over 500-1500m. “ In my opinion the manuscript would require major revisions to remedy these problems.

Hide

ED: Reconsider after major revisions (25 Jan 2021) by Minhan Dai

AR by Olaf Duteil on behalf of the Authors (23 Mar 2021) Author's response Author's tracked changes Manuscript

ED: Referee Nomination & Report Request started (04 Apr 2021) by Minhan Dai

RR by Anonymous Referee #1 (01 May 2021)

Suggestions for revision or reasons for rejection

I thank the authors for addressing the comments.

I have been in the "oxygen community" for a while so I received the core message and potential importance of this study but I also see the other reviewer's points that it could be confusing to understand the aim in the previous manuscript. I think revising the introduction and conclusion improved this part and addressed the points.

Along with the reply to other reviewers comments, the authors stated the aim and objective more clear and the diagnostics and metrics based on this study will be useful for further evaluating a suite of models (such as CMIP6, and statements included in the conclusion).

I have few minor comments on the revised manuscript.

Section 2.1: Sub-section title 2.1 Mean state : I think you should state a bit more in details such as "evaluations of simulated mean states".
Section 2.2: Sensitivity simulations sub-section 2.2.1 Forcing of oxygen to observed values in the subtropical regions
- NEMO2-30S30N: I think it is informative to also state that the oxygen boundaries are forced to observed concentrations at
the boundaries 30N and 30S for "the full depth (surface to the bottom)" (at least this is how I interpreted).
- NEMO2-30S30N1500m: Since you replied and discussed why you choose the depth interface of 1500m in the Reviewer #3's comment, I think you should include a short statement in this part justifying the choice of the depth interface.

L248: "restore oxygen": you should replace to "force oxygen" (as you explained in the response and to make the terminology consistent in the manuscript).

L350: a "primitive" EICS: could you clarify what you meant by "primitive" EICS?

Section 4.3: title "Model resolution and oxygen levels":
I thank for the authors for revising this section and replying why you choose not to include NEMO for the comparison here.
However, I still feel that there is a slight jump from the previous sensitivity sections to this section comparing a suite of models. Perhaps you should consider changing the sub-section title including words "discussion" or "implications" for example (something more informative to smoothly make transition from the previous sections). You might also consider including the NEMO MKE figures and short statements in the Annex A.

L472: take in account: this should be "take into account".

Hide

RR by Anonymous Referee #4 (07 May 2021)

Suggestions for revision or reasons for rejection

Duteil et al. examined the role of intermediate depth water (IDW, 500-1500m) in affecting oxygen level in the eastern Pacific Ocean using a set of ocean models. They found the correct oxygen level at extra tropical boundaries (30oS and 30oN) and correct EICS simulation are important factors in simulating oxygen in the eastern Pacific Ocean. These findings could help to explain model bias in the climate models. The work is interesting. However, the numerical experiments are not well designed to convincingly obtain the conclusion and the oxygen analysis based on a certain depth without considering specific water mass are confusing. I suggest a major revision before considering the publication of this manuscript.
General comments
1. IDW is not a specific water mass, but the text makes readers feel like this is a water mass. Lines 83-84 reads “They (models) generally display too shallow and thin IDW”. If the IDW refers to a water in a certain depth, it could not be shallow and thin. This would make the readers confused. There are some similar descriptions in the text and these confused descriptions should be corrected.
2. The numerical experiment NEMO2-30S30N and NEMO2-30S30N1500M are not well designed that use a correct oxygen level at 30N and 30S without considering density level. If the water mass is not at the correct depth in the model, the interpolation of oxygen at 30N and 30S may bring bias to the experiment. It would be more convincing if the authors at least could discuss this in the manuscript, such as a simple analysis of oxygen in the water mass to prove the bias is small. A further suggestion that the authors may use a regional model to do such an experiment instead of a global ocean model. For example, one is forced by global ocean model results (salt, temp, o2), the other is forced by WOA results (salt temp o2).
3. The conservative tracer release experiment (S2.2.2) cannot convincingly help readers to reach the conclusion that EICS is important in transporting oxygen from western Pacific to eastern Pacific. It would help to reduce confusion if the authors could give more explanation from result to conclusion. Except for tracer release, I would also suggest analyzing the volume transport or other convincing variables. If possible, remove the part that talking about surface circulation which makes the manuscript unfocused.
4. It would be clearer if the author could give more introduction to NEMO in the Section 2.1. The NEMO is explicitly used in this work and it need a detailed introduction. Also the two-way nest should be introduced here. More configurations should be written, such as mixing scheme, vertical coordinate, initial conditions.
5. Section 4.3 is an extended section of 4.2 with oxygen instead of tracer release. However, this section 4.3 is confused to me. I would suggest rewriting with a focus that directly connect with section 4.2.
Specific comments
1. Line 9, “intermediate depth waters (IDW)”, give a specific depth (500-1500m).
2. Line 10, “test” --> analyze?
3. Line 29, either it should say the euphotic zone or biological processes.
4. Lines 40-41, what is upper oxygen level?
5. Line 49, which water mass is formed in North Pacific at depth 500-1500m? If there is, please describe it in the introduction part. If not, please delete the “North Pacific”.
6. Line 106, title in section 2.1 should be model description, model introduction, or similar things.
7. Lines 123-133, I suggest using the official name instead the name given by the authors. It is up to the authors if they prefer a given name in the text. But at least they should tell readers the official names of GFDL1, GFDL025, GFDL01.
8. Line 145, the authors should discuss or show if the experiment reach an equilibrium status or not.
9. Line 160, “same as NEMOO2-30S30N” should be “same as NEMO2-30S30N”
10. Line 160. The experiment description of NEMO2-30S30N1500 is not clear. Is it forced from 0-1500m, from bottom to 1500m, or only at 1500 m?
11. Line 173, “In a second set” -- > “in the second set”?
12. Lines 176-178. What is the 0.1o two-ways nest? The authors should introduce them in the model description part. In addition, it should be “two-way.”
13. Lines 189-190. “They (Lagrangian) are not affected by subgrid scale diffusive and advective processes.” Why the Lagrangian is not affected by these two processes? This is confused. May be you want to say these processes are not considered in this work?
14. Lines 208-210. It there any reason that make authors conclude or propose that the high oxygen is due to diffusion from mixing layer?
15. Lines 217-219. Is there any paper describing this circulation? Please cite one or two papers here.
16. Line 238. I would suggest moving this sentence “A correlation …” before “reasons for this weak”. This would make the readers easy to understand.
17. Line 246. Why do you choose 1997-2007 for NEMO2-30S30N? In line 374, you choose 2002-2007 for tracer release experiments. If possible, a specific reason like equilibrium status should be discussed or given in the supporting information.
18. Line 286, I would not consider the circulation below subtropical gyre as a branch. You could say below the subtropical gyre at depth in 500-1500m. In addition, it should be gyre instead of gyres if you are only talking about south Pacific Ocean gyre.
19. Line 296, right bracket is missed. In addition, I cannot see the changes in advective terms (I think you are talking about zonal advection here) are found along the equator. Is it along the 30S?
20. Line 294-295. Why is the imbalance apparent in the tropics and why is related to isopycnal diffusion? The authors need to give more explanation here. In addition, what I see about the change of isopycnal diffusion is largest in the sub-tropics instead of tropics.
21. Line 315, Why do you say the equatorward propagation is due to small scale isopycnal processes? I did not see it from the figures. Any citations?
22. Line 375, where is the release location? I tried to find it in the introduction part but cannot find it.
23. Line 432. Is there any way that could quantify the contribution of ventilation due to EICS? Currently, the result and conclusion are very descriptive. I suggest digging more on it.
24. Line 496. Not consistent. Line 496 says equatorward transport is due to isopycnal subgrid scale mixing processes. However, line 315 says the equatorward transport is due to western boundary current and isopycnal process.
25. Figure 4 and other figures, a name should be given at each panel. Otherwise, it is really hard to know which panel you are talking about in the manuscript. Figure resolution is not high. The authors should provide a high-resolution version.
26. Fig. 1a, this panel shows the circulation at the 500-1500m. Could the author confirm the SEC arrow here? In the surface, the SEC has a different path. If this is true in the 500-1500m, please cite a paper.

Hide

ED: Reconsider after major revisions (12 May 2021) by Minhan Dai

AR by Olaf Duteil on behalf of the Authors (09 Jul 2021) Author's response Author's tracked changes Manuscript

ED: Referee Nomination & Report Request started (20 Jul 2021) by Minhan Dai

RR by Anonymous Referee #4 (05 Aug 2021)

Suggestions for revision or reasons for rejection

I thank the authors for responding to my previous remarks in the revised manuscript. Largely, the changes that have been made are satisfactory. The revised version is clear and well structured. However, there are some typos in the manuscript, and I would like to suggest the authors really go through the manuscript several times including the text and figure. I still have some minor comment that might help to improve the manuscript. I suggest a minor revision before considering publication of this manuscript.
1. Where do the authors download the GFDL model results? Please add it in the data and code availability.
2. Line 251, the models presenting the poorest oxygenated water at 30S are GFDL025 and GFDL1. If I understand correctly, these models should be GFDL025 and UVIC model according to Fig. 2b?
3. Line 285, NEMO2-30S30N1500M – NEMO2-30S30N should be NEMO2-30S30N1500M – NEMO2-ref. Please double check other typos.
4. Y tick label of figs. 3d and 3f is covered and missed. Please double check other figures.
5. Is the oxygen restored at boundary of 30S and 30N or in the region beyond 30S-30N? Because I saw there are large changes between 30S-40S and 30N-40N in Figs. 3c-3f. If the oxygen is only restored at the section of 30S and 30N, it is worth to discuss the poleward influence (at least a simple discussion). This should be clarified in the model experiment introduction part.
6. The simulation is from 1948-2007 in NEMO2-ref, NEMO2-30N30S, NEMO2-30S30N1500M. Which year results are you using for reanalysis in Fig. 3 and Fig. 4? Is it the whole simulation period? Please add this information to the introduction part or where you start to analyze the model results.
7. Line 300. Since the advection is separated into x and y directions, I think the equation should be in the cartesian coordinate. If I understand correct, Diff_iso should also be separated into x and y directions. Please double check this. I know the separation of diff_iso would not influence the figure 3g and Figure 4 and conclusion (you combine diff_iso_x and diff_iso_y together in the analysis). However, it should be clarified in this equation.
8. Line 339, I cannot see deep oxygen anomaly is upwelled in the eastern equatorial part of basin from Fig 3g. I would suggest rephrasing the word here. The text should rely on the figure. In addition, a vertical advection like Fig. 4g should be provided for experiment of NEMO2-30S30N1500M (you could plot it in the supporting information). I am curious about the vertical advection difference between nemo2_30S30N_1500m and nemo2_ref (similar to fig 4g).
9. Lines 349-350. I would suggest the authors rephrase the conclusion here. In my understanding, the NEMO2-ref shows the equatorward transport of oxygen should be dominated by zonal and meridional advection (Fig. 4a-d). In the experiment of NEMO2-30N30S, the contribution of small-scale isopycnal processes increase very significantly in the region of 30s-5s. This indicates the role of small-scale isopycnal processes might be larger than what we expect if we have a correct oxygen level at 30S. If I understand correctly, please reconsider how to conclude a more scientific results here. Similar re-consideration is suggested in the lines 481-483.
10. Line 387: 10 cm-1 should be 10 cm/s. Also in Line 386, 5 cm.s-1, should the dot be in the middle (i.e., ·) instead of period? Please follow the requirement of journal.
11. Fig. 8, What’s the meaning of negative and positive values? The word in line 457-458 of final version and line 533-535 of track version are different. Please double check.
12. Line 410. I am still confused about the release location even you answer it previous. This experiment should set the tracer in the initial status, then run the simulation. Is there any release location that continue releasing tracer during the simulation?
13. Is there any specific reason that the authors plot the current at 1000 m in Fig. 5? Most of analysis in figs. 3 and 4 is based on the mean between 500-1500m.
14. Lines 361-363. A reference or figure should be provided to show the model well reproduce the upper current structure.
15. Line 424. I cannot understand the shadow zone here.
16. Line 247, pacific should be Pacific. Double check other similar typos.
17. In Fig. 4a and 4b, a vector length-scale bar for current vector should be provided? Otherwise, it is hard to know the current size.
18. Title in section 5 is suggested to be “discussion and conclusions” or “Summary and implications”. It is okay if the author insists on the current one.
19. Line 475: at 1500 m depth --> below 1500 m depth
20. Line 238 in track version. The region change from 5S-5N to 10S-10N. Is the simulation re-run or a typo in the previous version?

Hide

ED: Publish subject to minor revisions (review by editor) (06 Aug 2021) by Minhan Dai

AR by Olaf Duteil on behalf of the Authors (24 Aug 2021) Author's response

ED: Publish as is (01 Sep 2021) by Minhan Dai

Short summary

The large oxygen minimum zones in the tropical Pacific Ocean are still not well represented by typical climate models. We analyze a set of ocean models and highlight the fact that an oxygen concentration that is too low at intermediate depth in the subtropical regions associated with a sluggish representation of the intermediate equatorial current system may be responsible for the overly large extension of the modeled oxygen minimum zones, potentially hampering future projections.