Insights into Deep Mesoscale Eddy Variability in the Gulf of Mexico

Cooke, Justin P.; Donohue, Kathleen A.; Watts, D. Randolph

doi:10.5194/egusphere-2026-4505

Preprints

https://doi.org/10.5194/egusphere-2026-4505

Preprints

30 Jul 2026

| 30 Jul 2026

Insights into Deep Mesoscale Eddy Variability in the Gulf of Mexico

Justin P. Cooke, Kathleen A. Donohue, and D. Randolph Watts

Abstract. Deep (> 1000 m depth) barotropic mesoscale eddies contribute to Loop Current Eddy (LCE) separation events, but sparse deep observations in the Gulf of Mexico limit our dynamical understanding and forecast skill. Data from a regional HYbrid Coordinate Ocean Model (HYCOM) free-running simulation are used to analyze deep, mesoscale variability over an 18-year period in the Eastern Gulf. A spectral analysis reveals three notable frequency bands: 1/150–1/30 days^-1, 1/60–1/30 days^-1, and 1/30–1/10 days^-1. 60–150 day variance is concentrated in the central Eastern Gulf and the Deep Southeast Channel (DSC), whereas 30–60 day variance is strongest near the Mississippi Fan. Complex empirical orthogonal function analysis identifies three modes that contain 65 % of the total deep variance. The first mode is associated with 30–60 day eddies formed by baroclinic development near the Mississippi Fan and propagate into the Eastern Gulf basin, then westward; the second and third modes capture variance representative of eddies in the 60–150 day and 10–30 day bands. Periods of largest modal amplitudes coincide with times when the LC extends into the Gulf, indicating that the deep Eastern Gulf is energized by the LC via baroclinic development and topographic interactions. The presence of a strong deep cyclone in the DSC is associated with Modes 2 and 3; LCE separations coinciding with this feature occur south of 25° N. Insights from this work suggest greater emphasis in observing the DSC to improve the predictability of LCE separations, and our understanding of the physical mechanisms controlling these events.

Received: 26 Jul 2026 – Discussion started: 30 Jul 2026

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.

Download & links

Justin P. Cooke, Kathleen A. Donohue, and D. Randolph Watts

Status: final response (author comments only)

RC1:
'Comment on egusphere-2026-4505', Anonymous Referee #1, 17 Sep 2026
This manuscript investigates deep mesoscale eddy variability in the Gulf of Mexico using an 18‑year HYCOM simulation dataset. The authors apply spectral analysis and CEOF decomposition of eta_ref at 2000 m depth to identify three dominant frequency bands and three leading CEOF modes characterizing deep eddy variability, respectively. They further study the connections between these deep modes to the separation events of surface Loop Current Eddies (LCE). This study makes good use of multi-year high‑resolution model outputs to overcome the limitations of sparse in‑situ deep observations, and the findings provide new insights into the regional upper and deep circulation system. Nevertheless, several conceptual, methodological, and interpretive issues remain to be properly addressed and improved. Therefore, I recommend the submission for Major Revision.
Major Comments
The manuscript advances several interrelated claims that “deep barotropic mesoscale eddies contribute to LCE separation events”, that “deep eddy energy is LC‑driven”, and that “the DSC cyclone could provide a predictive measure for separation latitude”. Taken together, these statements create ambiguity regarding the causal direction between deep mesoscale eddies and LCE dynamics. The analysis, particularly results shown in Figure 5, largely demonstrates co‑occurrence rather than proven causation. The authors should therefore formulate objective, internally‑consistent descriptions of the interplay between deep eddy variability and LCE separation, clearly distinguishing correlational associations from causal effects throughout the text.

The authors compute frequency spectra from time‑series from six discrete locations and divide deep eddy variability into three timescale bands: 10-30 days, 30-60 days, and 60-150 days. Critical details regarding the spectral analysis workflow are absent in the text. It remains unclear whether the selection of these six sites influences the identification of the three frequency bands, nor is the physical or statistical basis for the band cut‑offs articulated. In the current presentation, these choices appear rather subjective and may exert a substantial impact on key results. The authors should perform corresponding sensitivity experiments and provide expanded methodological justification for site selection and frequency‑band boundaries.

The authors infer the relationship between eddy signals across three time windows and those associated with the three CEOF modes by evaluating spatial similarity between band-method and CEOF-derived fields, an approach that carries substantial subjectivity. Calculating frequency spectra of the CEOF temporal coefficients or performing CEOF decomposition on bandpass-filtered eddy signals, would establish more robust physical connections and enhance interpretability.

Using frequency‑band analysis and CEOF decomposition, the study examines deep eta_ref variability and compares it with surface LCE separation to explore their relationship. At present, the comparison is restricted largely to deep eddy signals at 2000 m versus surface LC activities. Given that the HYCOM simulation provides full four‑dimensional outputs, why do the authors not investigate eddy signatures across the entire water column? A full‑water‑column analysis would yield a more complete physical picture and advance our understanding of upper‑deep layer coupling.

The manuscript simply compares the time series of the three deep-ocean CEOF modes with surface LCE separation events and finds that they occur roughly in phase. The paper only states that topographic interactions with the Mississippi Fan play a key role in the generation of deep mesoscale eddies and contribute to nearly all LCE separations. How exactly does this process operate? What dynamical evidence supports this claim? As such, the manuscript lacks in-depth mechanistic investigation and discussion.

The abstract and conclusions claim that a strong deep cyclone in the DSC is linked to Modes 2 and 3 and can modulate LCE separation. Nevertheless, only Figure 2a, showing mean eta (representing mean flow), presents this cyclonic feature. Neither var(eta) (representing eddy variability) nor the modal decomposition results confirm the existence of this cyclone. From my experience with EOF analyses, the mean field is routinely subtracted before decomposition, producing modes with eddy structures of positive and negative anomalies. I don't know whether CEOF decomposition produces similar signed anomaly structures.

Minor Comments
Lines 46-47: This sentence only describes the surface LC separation and is irrelevant to the deep eddies focus of this paragraph.

Line 53: Change “play” to “plays”.

Line 77: This study doesn’t reveal how the LC energizes the deep eddy field.

Line 86: Remove “of” in the phrase “south off of”.

Line 90: Since this study focuses on deep-ocean eddy variability, the vertical resolution of the model should be explicitly introduced to support the reliability of deep-layer analyses.

Line 102: eta_ref is not pressure.

Lines 132-134: This deep cyclone featured in this section should be introduced in the Introduction to establish prior background and improve logical coherence.

3 caption: Change “Bathymetry are” to “Bathymetry is”.

Line 187: Remove the duplicated phrase “of the”.

4 caption: How is the normalized amplitude computed?

4 caption: How do features propagate? From pink to purple to blue to green? Several statements in the main text conflict with this; please clarify.

Line 213: Remove the unnecessary comma before “appears.”

Line 240: Remove the unnecessary comma in “deep, mesoscale eddies”.

Line 259: Change “using 18-years from” to “using 18-year data from”.
Citation: https://doi.org/10.5194/egusphere-2026-4505-RC1
RC2:
'Comment on egusphere-2026-4505', Anonymous Referee #2, 02 Oct 2026
This is a review of "Insights into Deep Mesoscale Eddy Variability in the Gulf of Mexico" by Cooke et al. This manuscript characterizes deep mesoscale eddy activity in the Gulf of Mexico by decomposing the mesoscale variability into frequency bands and spatial modes, and relating it to the evolution of the Loop Current (LC). The study's relevance lies in the direct link between the intensification of these mesoscale eddies and the timing of Loop Current Eddy (LCE) separation. The manuscript is very well written and structured, the figures are of high quality, and the methods and results are clearly presented. However, some points require attention — in particular, the authors claim that the deep eddies identified are barotropic, and use a measure of the depth-integrated pressure as an indication of barotropicity, which is not sufficient to establish this. Additionally, it is not clear how these (barotropic) eddies relate to Loop Current Frontal Eddies and the deep eddies with a signature at 2000 m. I therefore recommend major revisions before publication.
Major Comments
1) "Deep" eddies vs. barotropic eddies
The authors claim that the "deep eddies" in this study are barotropic. There are three issues with this claim.
First, the claim of barotropicity is only based on a quantity associated with the depth-integrated pressure, which is not a direct measure of barotropicity — a strong baroclinic eddy would also produce a strong signal in η_ref. I recommend that the authors either expand the analysis by performing a vertical modal decomposition to isolate the barotropic signal prior to applying the spectral analysis, or refrain from claiming that these features are barotropic.
Second, the strong signal in η_ref could originate from surface or upper-ocean eddies, which are expected to carry a strong pressure signature (available potential energy is generally larger in the upper ocean than in the deep ocean, due to geostrophic adjustment of isopycnals being proportional to eddy strength). In other words, it is not clear that the dominant contribution to η_ref is not coming from surface- or upper-ocean-intensified eddies, rather than from barotropic features or from deep eddies intensified near 2000 m. A longitude–depth plot of the depth-cumulative pressure in a vertical cross-section across these eddies would help clarify the origin of the η_ref signal (barotropic, surface-intensified, or depth-intensified).
Third, LCFEs are not mentioned in the manuscript, yet they are a well-documented feature of the Loop Current system, and it is unclear how they relate to the modes identified here. The literature has largely treated LCFEs in the upper ocean (0–1000 m) and deep eddies (≥2000 m) as distinct features, and it is not straightforward that they are vertically coherent (i.e., barotropic); vertical cross-sections from both models and observations suggest they are not barotropic, but rather upper-ocean- and deep-ocean-intensified. Are these features vertically aligned? Are they dynamically linked? The analyses of Donohue et al. (2016) and Morey et al. (2020) show depth-intensified eddies corresponding to higher baroclinic modes rather than a barotropic mode. Donohue et al. (2016) did look at the link between upper-ocean meanders and deep eddies, but not the relationship between LCFE and deep eddies. Additionally, iIn the literature review, I suggest the authors clearly distinguish between studies that have examined motions intensified at depth or in deeper layers, and those that have specifically identified cyclonic motions around the LC as barotropic — either through vertical modal decomposition or by demonstrating unambiguous barotropic characteristics in cross-sections. Since the authors use 'deep' as a synonym for 'barotropic,' some sentences in the introduction becomes confusing, and sentences referring to the (actual) deep eddies may mislead the reader into thinking they are about barotropic cyclones.
2) Implications and broader significance: What are the applications and implications of these results beyond the broad goal of better understanding deep mesoscale eddies and their impact on the timing of the LC shedding? More clearly articulating this in the introduction (as the research gap), as well as in the conclusions would increase the scientific relevance of this manuscript.
Minor Comments
33–34: "two vertical modes constitute >90% of the variance" — variance of? Of the three-dimensional flow (ocean velocity) in the eastern Gulf of Mexico?

The 20 vertical layers, even if hybrid, is low and is a limitation for the correct resolution of deep (intensified at depth) eddies, as those require the resolution of higher vertical modes.
Citation: https://doi.org/10.5194/egusphere-2026-4505-RC2

Justin P. Cooke, Kathleen A. Donohue, and D. Randolph Watts

Data sets

Bottom pressure and temperature collected in the Gulf of Mexico, 2019-06-14 - 2021-05-19 K. Donohue et al. https://doi.org/10.7266/BZ9B3C54

HYCOM + CFSR Gulf of Mexico 1/25° 54-year Experiment (GOMl0.04/expt_02.2) D. S. Dukhovskoy et al. https://www.hycom.org/data/goml0pt04/expt-02pt2

Interactive computing environment

JCooke188/ceofs: Deep_Eddy_Modal_Analysis_v1.0 J. P. Cooke https://doi.org/10.5281/zenodo.21608468

Justin P. Cooke, Kathleen A. Donohue, and D. Randolph Watts

Viewed

Total article views: 312 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
205	77	30	312	26	27

HTML: 205
PDF: 77
XML: 30
Total: 312
BibTeX: 26
EndNote: 27

Views and downloads (calculated since 30 Jul 2026)

Month	HTML	PDF	XML	Total
Jul 2026	49	9	4	62
Aug 2026	68	18	9	95
Sep 2026	58	33	11	102
Oct 2026	30	17	6	53

Cumulative views and downloads (calculated since 30 Jul 2026)

Month	HTML	PDF	XML	Total
Jul 2026	49	9	4	62
Aug 2026	68	18	9	95
Sep 2026	58	33	11	102
Oct 2026	30	17	6	53

Viewed (geographical distribution)

Total article views: 275 (including HTML, PDF, and XML) Thereof 275 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 09 Oct 2026

Short summary

Abyssal ocean storms play a key role in the separation of Loop Current (LC) eddies. Sparse deep observations limit our understanding; this work uses a simulation to study deep circulation. Deep energy coincides with LC extension, indicating the eastern Gulf is energized by the LC via topographic interactions. Three patterns appear. Notably, when a deep cyclone is present in the southern Gulf, LC separation occurs south of 25N. Strategic deep measurements could improve prediction of separations.


Total:	0
HTML:	0
PDF:	0
XML:	0