the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 21 May 2025
Hidden vortices: Near-equatorial low-oxygen extremes driven by high-baroclinic-mode vortices
Abstract. Long-term time series of dissolved oxygen (DO) measurements from the upper 500 m depth of the eastern tropical North Atlantic (ETNA), collected over a period of up to 15 years at three different mooring sites, reveal recurring extreme low-oxygen events lasting for several weeks. Similarly, observations from 15 individual meridional ship sections between 6° N and 12° N along 23° W show DO concentrations far below 60 µmol kg⁻¹ in the upper 200 m – significantly lower than the climatological values at this depth (>80 µmol kg⁻¹). Two-third of these low-oxygen events could be related with high-baroclinic-mode vorticies (HBVs) with their cores located well below the mixed layer. Despite the energetic equatorial circulation and the expected dominance of wave-like structures in the near-equatorial region, these HBVs persist as relatively long-lived and coherent features. Based on moored and shipboard observations from the ETNA, and supported by an eddy-resolving ocean-biogeochemistry model, we characterize their dynamics and DO distribution. Observed water mass properties and model analyses suggest that most HBVs originate from the eastern boundary and can persist for more than six months. As they propagate westward into regions of higher potential vorticity (PV), anticyclonic HBVs with low-PV cores remain more effectively isolated and have longer lifespans compared to cyclonic HBVs with high-PV core. The vertical structure of the dominant anticyclonic HBVs corresponds to baroclinic modes 4–10, with associated Rossby radii ranging from 34 km to 13 km, respectively. This is consistent with observed eddy sizes and is well below the corresponding 1st baroclinic Rossby radius of deformation (> 100 km). Since none of the observed HBVs exhibit a surface signature, a substantial portion of the near-equatorial eddy field may remain undetected by satellites, yet still exert significant influence on ocean ecosystems and biogeochemical cycles.
- Preprint
(4678 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 16 Jul 2025)
-
RC1: 'Comment on egusphere-2025-2175', Anonymous Referee #1, 11 Jun 2025
reply
The authors describe a vertically and horizontally spatially constrained low dissolved oxygen (DO) event observed in the near-equatorial North Atlantic using moored and repeated ship-based observations, and investigate its origin using output from a high-resolution coupled climate model with a simplified Biogeochemical component. The manuscript provides a detailed description of the observational data and analysis methods employed. I find the observational documentation of the phenomenon the authors refer to as HBV to be of sufficient scientific value to merit publication.
One major concern is that the authors argue that the long lifespan is one of the key characteristics of the observed HBVs, yet this conclusion appears to rely solely on the model output. As evidenced in the comparison between the model and observations presented in the manuscript, there may be non-negligible biases in the spatiotemporal variability of DO anomalies (see also my later comments). It should also be noted that general ocean circulation models tend to exhibit reduced dissipation of mesoscale/submesoscale structures, potentially leading to artificially prolonged features. Furthermore, the MiniBLING model employed here does not account for the diverse and complex remineralization processes that drive oxygen consumption in the mesopelagic zone.
To strengthen the argument for the longevity of HBVs based on observations, it would be beneficial to incorporate additional evidence, such as analyses using Apparent Oxygen Utilization (AOU), which carries information related to water mass age. If available, supplementary water mass diagnostics using other tracers (e.g., nitrate, phosphate) would also be valuable in corroborating the persistence of low-DO waters associated with HBVs, at least as circumstantial evidence.
Comments:
Does the MIMOC dataset include dissolved oxygen? I could not find oxygen data in the source referenced by the authors (https://www.pmel.noaa.gov/mimoc/). If my understanding is correct, what dataset was used to generate Fig. 2a?
Regarding Fig. 2a and 2b, it would further strengthen the manuscript if the authors could include a comparison map of the depth at which the oxygen minimum occurs.
Line 335: Is it valid to assume that the barotropic component is zero? In fact, the flow estimated from SLA (e.g., the gray arrows in Fig. 6a) may be significantly influenced by the barotropic component. If the barotropic flow cannot be assumed to be zero, wouldn't it be appropriate to subtract the barotropic velocity (approximated by vertically averaged velocity) from the observed velocity profiles as part of the preprocessing?
Line 376: Castelao and Johns (2011) and Castelao et al. (2013) are not included in the reference list. Please double-check that all cited works are properly listed in the bibliography.
Line 385 (“The optimal eddy center allows…”): The meaning of this sentence, particularly the latter part, is unclear. Please revise for clarity.
Line 407: The method for estimating the propagation speed is not clearly described. Please revise this section to clarify how the speed was calculated.
Figure 3c,d: What does the x-axis represent? Please add labels or clarify in the caption.
Line 585: In the discussion of discrepancies between ship-based and satellite-derived observations, spatial resolution is indeed important, but temporal resolution is also critical. Note that the raw satellite data used for gridding does not have daily temporal resolution. In addition, what is the reason for omitting near-surface velocities in Fig. 6d and 6e? If such data are available, do satellite-derived velocities correspond better to near-surface velocities from ship-based observations, or to vertically averaged velocities?
Line 626: Does this refer to Equation 9? Please clarify.
Line 626 (“The solid black line represents…”): In Fig. 6f, I can only identify the line representing HBV speed. Could you clarify what this sentence is referring to?
Line 603: Since Fig. 6 only displays modes 6 and 10, it is not possible to assess the behavior across modes 4 to 10. Please revise this sentence or clarify with additional figures if necessary.
Line 692 (regarding Fig. 4a and 4b): There appears to be a stark difference in the zonal distribution of the strength of low-DO extremes (i.e., the lower end of the dots) between the observations and the model. The observations show the most pronounced low DO events offshore (around 24–21°W), whereas the model indicates such events occur closer to the coast. This inconsistency could point to a potentially significant model bias—possibly arising from the model misrepresentation of HBV origins, trajectories, or associated biogeochemical processes along the path toward 23W. Even though the mean fields (e.g., Fig. 2 and the medians in Fig. 4c) appear consistent, this does not guarantee that the model correctly reproduces the variability targeted in this study (e.g., the dot distribution in Fig. 4c, d). While observational data may be limited and subject to sampling bias—potentially explaining some of the discrepancies—this possibility should be explicitly considered. I encourage the authors to justify the model’s suitability for this analysis, for example, by performing a model–observation comparison using pseudo-observations from the model. Alternatively, as noted earlier, the conclusions should not rely solely on the model results and should incorporate more observation-based evidence.
Lines 606, 776, 947: Please define “SCV” upon first usage.
Line 825 (“Rossby numbers were below 1”): Is there a corresponding figure showing the Rossby number distribution? If so, please reference it.
Line 859: Please define “SACW” and “NACW” when first mentioned.
Citation: https://doi.org/10.5194/egusphere-2025-2175-RC1 -
RC2: 'Comment on egusphere-2025-2175', Anonymous Referee #2, 12 Jun 2025
reply
This manuscript presents a compelling study of high-baroclinic-mode vortices (HBVs) in the eastern tropical North Atlantic, combining shipboard and moored data with eddy-resolving model output. The authors describe how HBVs (whose subsurface cores are isolated from surface turbulent processes) transport low-oxygen water masses offshore from the eastern boundary. The study both provides evidence for the physical advection of low-O₂ water but also considers ongoing oxygen consumption via remineralization along the vortices’ trajectory. These dynamics are discussed in the context of their potential implications for biogeography and biogeochemical cycling in the region.
The observational challenge of capturing HBVs, given their relatively small spatial scale and intermittent frequency of generation, is well acknowledged. In that light, the dataset compiled and analyzed here is impressive and already provides a valuable contribution to the literature. The use of numerical modeling to complement the observations is also appreciated, though I raise a few questions below regarding the model’s ability to resolve these features. Overall, I find the manuscript suitable for publication, pending some minor considerations centered around the following suggestions:
- The authors might consider supplementing their HBV identification and tracking with additional water mass tracers such as spiciness and apparent oxygen utilization (AOU). These metrics, particularly spiciness (which is conserved along isopycnals), can provide clearer insight into the origins and evolution of the anomalies. For example, a panel showing spiciness in Figure 8 could strengthen the interpretation.
- I have questions around the decision to define HBV events using an arbitrary 10th percentile threshold (e.g., in Figures 3 and 4). This approach may flag low-oxygen "anomalies" even in relatively quiescent regions with little true HBV activity. In Figure 3a, for instance, I find only the event between 2023–2024 particularly convincing. It may be worth considering alternative detection criteria, such as thresholds based on standard deviations or interquartile ranges, which could provide a more statistically grounded definition of outliers.
- While the focus on anticyclonic eddies (ACEs) is understandable given their higher detection frequency in observations, the manuscript would benefit from a (slightly) more symmetric treatment of cyclonic eddies (CEs). Figures 9 and 10 do a good job of characterizing ACE dynamics and evolution via model output; a similar analysis of a representative CE from the model could be similarly instructive. For instance, this could be used to tie in the discussion around CE instability and decay mechanisms (e.g., interaction with high-PV water, lines 884–893). Including this could both reinforce the contrast between eddy types while also showing their similarities, at least from model output.
I also have a additional minor comments, labeled with specific line numbers:
[L. 45-46]: The authors could cite the recent study from Deutsch et al. (2020) (https://doi.org/10.1038/s41586-020-2721-y). That study convincingly shows that temperature and O2 shape the biogeography of marine organisms.
[L. 117]: The authors could introduce the acronym ‘SCV’ here after the first mention of submesoscale coherent vortices.
[L. 124]: Did the authors mean to write “mesoscale-permitting”?
[L. 212]: This makes it seem like authors are only showing WOA oxygen during the model validation, but the authors frequently cite MIMOC data in the text. If MIMOC includes oxygen, please mention that here.
[L. 219]: CM2.6 only has a resolution of 0.1 degree, is that high enough to resolve HBVs? It may be helpful to briefly discuss the model resolution (both horizontal and vertical) in the context of HBV scales, especially if the model is close to the margin of resolving such structures.
[L. 235]: What do the authors mean by “five daily model outputs”? I’m assuming they mean to say the output resolution is every 5-days. Please clarify.
[L. 260]: Could the upper OMZ in observations be caused by HBV advection? If so, doesn’t that say that the model is not accurately capturing their influence?
[L. 261]: Just a suggestion, but since the authors mention depths deeper than 500m, then panels in Figure 2-f could be extended to at least 700m (the deepest depth mention during the validation).
[Section 3.1.1 - 3.1.1 & 3.2]: While very useful to include, these sections could be moved to a Supplementary material. The methods section is quite long as currently presented, and these sections broadly introduce standard oceanographic methodologies (e.g., methods introduced in physical oceanography textbooks). However, if manuscript length is not a concern, feel free to keep them in since they are very useful to frequently reference during discussion of results (Section 4).
[L. 392]: “The horizontal eddy center at each model time step”. The authors don’t mean the computation time-step here, but the output frequency of the model (5 days?). It could be helpful to clarify this.
[L. 421]: You could present this additional time-series in the supplementary material.
[L: 483]: Can the authors speculate on what is driving the events not linked to subsurface eddies (#5, #6, #8-10)?
[L. 523]: Just a suggestion, but the authors could use sea-surface height anomaly products (e.g. Satellite or re-analysis), mapped to the location of the mooring, to determine if there was a pronounced surface signature of these events. That could help determine if the feature is driven by surface-intensified ACEs (if strongly positive) or subsurface-intensified ACEs (if no or weak signature). The authors choose to do this for CTD profiles around L. 585, so why not extend this here? If their arguments from L. 585-588 hold, then mention this for the mooring data as well.
[L. 696]: In this section, the references for specific Figure 6 panels are incorrect. Please update them.
[L. 748]: This should be referencing Figure 9, not Figure 8.
[L. 862]: Why do the authors report the model having Ro of roughly 0.4 when the time-series in Figure 10 clearly shows lower values near 0.1?
Typos (note there were several more, but I forgot to write their locations, so a second read-through is warranted):
[L: 170]: “…additionally a DO sensor…” (an —> a)
[L: 424]: There a few typos in this sentence.
[L. 914]: Africa.
Citation: https://doi.org/10.5194/egusphere-2025-2175-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 126 | 23 | 6 | 155 | 3 | 4 |
- HTML: 126
- PDF: 23
- XML: 6
- Total: 155
- BibTeX: 3
- EndNote: 4
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1