A high-resolution, submesoscale-resolving ocean model of South East Queensland (ROMS v4.2)

Gwyther, David E.; Jeffers, Megan; Bostock, Helen

doi:10.5194/egusphere-2026-1852

Preprints

https://doi.org/10.5194/egusphere-2026-1852

Preprints

18 May 2026

| 18 May 2026

A high-resolution, submesoscale-resolving ocean model of South East Queensland (ROMS v4.2)

David E. Gwyther, Megan Jeffers, and Helen Bostock

Abstract. The ocean exhibits a continuum of motions spanning basin-scale circulation to micro-scale turbulence where between mesoscale eddies and small-scale turbulence lies the submesoscale (∼0.1–10 km, hours-days). The submesoscale is characterised by fronts, filaments and small coherent vortices that arise from instabilities of mesoscale currents, interactions with topography, and buoyancy forcing. These processes restratify the upper ocean while simultaneously enhancing vertical tracer exchange through intense vertical velocities concentrated at density gradients. In the East Australian Current, the western boundary current of the South Pacific Gyre, the jet accelerates and interacts with the continental shelf off South East Queensland, creating conditions favourable for submesoscale activity. Yet these processes remain unresolved and unexplored in existing regional simulations: mesoscale regional models do not resolve energetic frontal processes, high-resolution coastal models lack realistic offshore forcing, and observations alone are insufficient to constrain the three-dimensional subsurface circulation. Here, we present a submesoscale-permitting implementation of the Regional Ocean Modelling System (ROMS v4.2) for the EAC jet intensification region offshore of South East Queensland and northern New South Wales, Australia. We demonstrate that our model generates energetic fronts and filaments along the continental shelf edge, while maintaining surface and subsurface mesoscale hydrographic structure consistent with observations. Relative to mesoscale-resolving simulations, the configuration exhibits substantially increased hotspots of divergence, vorticity and vertical velocity variance, together with a flatter kinetic energy spectrum indicative of resolved submesoscale dynamics. A demonstration of the continental shelf break in the vicinity of K'gari shows the accelerating jet interacting with coastal promontories and the shelf break, producing localised and persistent submesoscale activity. This configuration enables investigation of submesoscale emergence and interaction with the mesoscale circulation, including jet-shelf modification, cross-shelf exchange and energy transfer across scales. This approach provides a pathway to quantify energy pathways and ocean dynamics, with applications extending beyond the East Australian Current to other western boundary current systems.

Received: 02 Apr 2026 – Discussion started: 18 May 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

David E. Gwyther, Megan Jeffers, and Helen Bostock

Status: final response (author comments only)

RC1:
'Comment on egusphere-2026-1852', Anonymous Referee #1, 14 Jul 2026
General comments
The manuscript, "A high-resolution, submesoscale-resolving ocean model of South East Queensland (ROMS v4.2)," presents a new high-resolution ocean model covering southeast Queensland and northern New South Wales, developed to realistically represent submesoscale processes in a region where the East Australian Current interacts with the continental shelf. The authors provide a comprehensive model evaluation using a wide range of both surface and subsurface observations. They also demonstrate that the high-resolution SEQld ROMS simulation captures substantially more kinetic energy within the mesoscale and submesoscale bands than coarser-resolution model and observation.
Although the evaluation identifies some model deficiencies, including a cold bias over the shelf and discrepancies in volume transport magnitude and phase relative to observations, these limitations are expected. As the model does not employ data assimilation, it is not intended to reproduce ocean features at their exact time and location. Rather, its purpose is to provide a dynamically consistent and realistic representation of the regional circulation and associated submesoscale processes.
Overall, this is a well-written and easy to follow manuscript describing a valuable modelling framework that will benefit future studies in the region. I recommend publication after the authors address the comments and suggestions provided below.

Specific comments
Introduction
L66 and L73: Regarding "intensification region" and "intensification zone", I find this terminology confusing. First, because your model domain covers much more than just the so-called intensification zone. The intensification zone is a specific area in the northern part of your domain. Second, it has been shown that the EAC jet intensifies poleward from 27.5°S to 32°S, reaching its maximum at 32°S (Kerry, 2020). Since your simulation only covers the year 2012, you do not have sufficient seasonal, interannual, and decadal variability to conclude that this is consistently the region where the jet intensifies. I recommend specifying the latitude where the intensification occurs, e.g. "jet intensification at 25°S" or "northern jet intensification", and using this terminology consistently throughout the manuscript.
L73: "... numerical ocean model of the EAC jet intensification zone." Again, your model covers almost the entire EAC jet, not only the region where it intensifies around 25°S. Please revise this throughout the manuscript wherever it appears.

Model description
L88: For a more logical structure, begin with the model simulation period (move the first sentence of the paragraph starting at L124, leaving the "Model spin-up..." sentence to be included with the forcing description). Then describe the model domain (L95–97: "The model domain..."). Move the descriptive paragraph beginning at L105 (key model features) to before the description of the vertical stretching. Then present the horizontal resolution, vertical stretching, lateral and surface forcing, initial conditions, and finally the model spin-up.
You need a section describing the datasets used for model evaluation. Please include the spatial and temporal resolution, the number and spatial distribution of Argo floats, details of the mooring array, and whether the model output was extracted at the same latitude/location as the observations. This information should be clearly described in the manuscript.

Model evaluation and results
L147, L153 and L180: In all three cases, the manuscript overlooks the fact that the model exhibits considerably cooler temperatures on the shelf between 27°S and 31°S, while attributing significant errors mainly to the coarse-resolution products. Although you used RAMSSA SST, OSTIA would likely be a better observational product for this region. Since you state in L159 that the model "may in fact be closer to reality than implied by the calculated RMSE...", I suggest repeating the SST comparison using OSTIA to determine whether the observed shelf cooling is also present in that product.
I would like to see a map showing the mean circulation, SST and isobars with the four defined regions (shelf, slope, EAC, and offshore) overlaid. Unless there is a specific reason to use <100 m, I would also recommend adopting the more common definition of the shelf as ≤200 m.

For the TS diagram, you do not specify the number of Argo floats used, the time period (I assume they are from 2012), or whether they were colocated with the model output. It would also be useful to include a supplementary figure showing the spatial distribution of the Argo profiles used, so the reader can assess their coverage and potential sampling biases.

Submesoscale dynamics
To improve the logical flow, I suggest beginning this section with the SEQld ROMS vs BRAN comparison (Figures 6 and 7), followed by the whole-domain diagnostics (Figure 5), and finally the K'gari analysis (Figure 8). This sequence completes the model evaluation before moving into the diagnosis of submesoscale dynamics.

Technical comments
Abstract
L1–5: This section is rather long for an abstract. It could be shortened by providing a more concise explanation of what submesoscale processes are and why they are important. Some of this material could instead be moved to the Introduction.
L6: "... the jet accelerates and interacts with the ..." Please provide more context. Where does it accelerate? Does it only interact with the shelf when it accelerates?
L8: "...in existing regional simulations" — do you mean simulations for southeast Queensland specifically, or for the broader region?
L11: "...for the EAC jet intensification region" Again, the model is not limited to the jet intensification region; it covers the entire jet extension. Consider rephrasing to something like "for the EAC jet-dominated region (latitude range)."
L14: "... increased hotspots of divergence ..." Divergence of what? Please clarify.
L15: "... flatter kinetic energy spectrum" Flatter where? Please provide more context.
L16: "... vicinity of K'gari" Please include the latitude.
L17: "This configuration" Which configuration? Please be specific.
L18: "...jet-shelf modification..." This expression is unclear. Please clarify what is meant.
L19: "This approach" Which approach? Please state it explicitly.
L19: "pathway" is repeated twice in close succession.

Introduction
L25: "... sustained observations of submesoscale processes remain challenging." Please briefly explain why they remain challenging.
L32: "... horizontal strain and extension" Please clarify what is meant by "extension".
L42–43: "... strongly influences ocean conditions ..." Please provide more context on how the EAC influences ocean conditions.
L44–50: Consider adding the circulation schematic to Figure 1 and referring to it here.
L54: The citation displayed as Sloyan et al. (2024) should be Chapman et al. (2024).
L65: "Yet" is repeated. Please reword.

Model description
L125: The hyperlink for Sloyan et al. (2024) points to Chapman et al. (2024). Please check this throughout the manuscript.
After introducing the model name SEQld ROMS, use this consistently throughout the manuscript instead of simply "ROMS" to avoid ambiguity.

Model evaluation and results
Figure 2
Label the columns as Satellite and SEQld ROMS, rather than simply Obs and Model.

Add panel titles indicating the variable (SSH, MKE, SST).

For SSH, include contour labels.

For SST, adjust the contour labels to improve visibility.

Add bathymetric contours to the SSH and SST panels.

L128: Since there is only one subsection, I recommend removing subsection 3.1.
L129: Remove "this" from "We evaluate the ability of this SEQld ROMS..."
L150: Include the region definitions in the text rather than only in the Figure 3 caption, and briefly justify the chosen boundaries.
L160: Add (Figure 1) at the end of the sentence.
Figure 3 caption
Remove the RMSE equation, as it already appears in the main text.

Adjust the RMSE annotation, as some text overlaps the plotted lines.

Figure 4
Add titles to panels (a) and (b): Argo and SEQld ROMS, respectively.

Salinity units are missing.

Add titles to panels (d) and (e): Mooring array and SEQld ROMS, respectively.

L179: Replace the parentheses around "The biases..." with commas, as this is relevant information and does not require parentheses.

Submesoscale dynamics
Figure 5
Add the year to the figure titles.

Increase the font size of the colour bar labels.

Add a top-row panel showing surface circulation and SST to help interpret the diagnostics.

Caption
Begin with "SEQld ROMS diagnostics..."

Specify horizontal flow convergence.

Include austral winter and austral summer, together with the corresponding dates (10 July 2012; 8 October 2012; 27 December 2012).

Add the year to all dates.

Paragraph beginning at L202
When describing Figure 5 (L203), because all features appear in all panels, I suggest removing the individual panel references and instead referring collectively at the end of the sentence, e.g. "...small submesoscale eddies (Figure 5a–c)."

L203: Delete "in" from "... (e.g. Figure 5b) and in small submesoscale eddies..."

L204: When stating that "...fields have patches of O(1) which shift through time", provide an example region (as you do later for K'gari) to help the reader understand where these shifting features occur.

L207: Add flow before divergence.

Figure 6
Add SEQld ROMS and BRAN labels to the right of the corresponding rows.

For the Ro panels (a) and (d), reducing the colour bar limits may improve the visibility of offshore features without compromising the shelf features.

L208: Replace our model with SEQld ROMS.
L211: Delete "the" from "to expect time slices from the SEQld ROMS".
L212: End the sentence after metrics. Begin the following sentence with "We compare annual-mean fields..."
Move the magnitude strain equation from L218 to its first mention in L213.
L213: Explain why the analysis is averaged over the upper 100 m.
L215: Add respectively in (Figure 6a,d, respectively).
Replace all occurrences of ROMS with SEQld ROMS.

L262: Add respectively to (Figure 8a,b, respectively).

Discussion and conclusions
Please refer to the earlier comment regarding the use of intensification zone and ensure consistent use of the model name SEQld ROMS throughout the manuscript.
Citation: https://doi.org/10.5194/egusphere-2026-1852-RC1
- AC1: 'Reply on RC1', David Gwyther, 01 Oct 2026
  
  We thank the reviewer for their comment.
  
  We have provided a response to all points in the attached PDF, and corrected the manuscript accordingly.
  
  Kind regards,
  
  David Gwyther (on behalf of the co-authors)
  
  Citation: https://doi.org/10.5194/egusphere-2026-1852-AC1
RC2:
'Comment on egusphere-2026-1852', Anonymous Referee #2, 22 Jul 2026

The authors have done a nice job of developing a high resolution simulation of the SEqld region, which indeed represents an important region and a gap between the well studied GBR region and the EAC eddy field off NSW. As such this model development/experiment description paper represents an important contribution to the field and I recommend publication after some major revisions.

The model validation section is lacking description of the observations used (their resolution and limitations etc) and explanation of how the model is sampled for comparison with the observations. For example, the authors state they compare against satellite derived SSH and SST but have not described the products used.

Further, the authors state that the benefit of SEqld ROMS is its submesoscale resolving capability; however, no submesoscale resolving observations were drawn on the validate the SEqld model. The EAC mooring array and the SE Qld moorings both measure temperature and velocities at submesoscale-type time scales. SWOT observations now provide submesoscale resolving SSH fields.

Line comments:

Line 40: “In addition, submesoscale dynamics are further indicated by relative vorticity (spatial variation of velocity) that is comparable to the planetary vorticity (ζ/f ∼O(1));” This is the definition of the Rossby number – relative vorticity over planetary vorticity. Hence repetition from the sentence before.

Line 71: “ Moreover, it has been suggested that increasing model resolution to capture the submesoscale can reduce the ability of the data assimilation system to achieve skilful prediction of the mesoscale evolution (Sandery and Sakov, 2017). “ This is in contrast to findings by Kerry et al 2026 who showed that better representation of submesoscale energy resulted in improved mesoscale predictions, potentially through improved representation of the inverse energy cascade.

102: Explain rx0 or at least give a reference

108: What does ‘top out’ mean?

114: Not sure BRAN is dynamically – consistent (are you suing the EnKF version, or the EnOI version?). State version number.

Lines 129-145 and Figure 2 caption, please specify which satellite observations (or products) exactly you are referring to. If they are merged products please describe them and their resolution.

Figure 2 e,f: What is the SST product? It seems a coarse product has been used that does not capture the cool shelf waters. This is surprising as SST products are very high resolution and should capture this. Also there is a lot of structure in the SST modelled mean (wiggly contours). This seems odd to me. Is this an average over the full year simulation output as an average file from ROMS, or have the authors averaged discrete output times to give the not-so-smooth result I wonder?

148: Still unclear what the observations are, also is the model sampled at the observation times and locations, if so state this, if not describe how it is done.

156: Need details of the satellite product used and what its resolution is

159: You mean 3i? Satellite observations are typically unreliable for coastal regions (depths <100m). Again I don’t know what observations are being used. Alongtrack SSH can be trusted up to the coast but merged products not so much.

160: Not sure what this means.

163 remove ,

Figure 4b – Can the authors explain the tongue of cooler saltier water in the model? Was the model sampled at Argo float times and locations (I hope so)? Or is the tongue shelf waters not sampled by Argo? Can the authors explain the broader TS diagram of the model compared to the obs? Line 169 – suggests model is not sampled at Argo times and locations.

166-167: Are these water masses all evident in the TS plot? If so, describe, if not, suggest deleting.

Fig 4c: Surprising that the model is so off compared to the mooring array at 27S south where the EAC is most coherent. Is this an issue with getting the bifurcation of the westward currents correct? Have the authors looked at the spatial fields to understand the discrepancies. For example Jul-Sep the mooring array measures a strong southward current, but the model has weak southward and sometimes net northward flow.

Again were the modelled and observed sections the same? Was the method to compute transport and velocity the same from the observations and the model? Please add more detail around the methodology.

Figure 2d shows a clear weakening of the EAC MKE at 27S – which exists but is not as pronounced in the satellite derived MKE (2c)
This relates to 4d,e where the modelled EAC is too weak

Figure 4 caption: “In (b), only wintertime and summertime snapshots are combined to reduce data density.” Does this mean the model is not sampled at observation times and locations? Or does this mean something else? Please clarify.

Fig 5: “In (a), a length scale marker indicates the approximate boundary between mesoscale and submesoscale motions at this latitude.” This should vary with stratification and thus be different across summer versus winter

208 and Fig 6: Can you give the model a name (SEqld ROMS) rather than our model or this model?
It is customary to put the version rather than just BRAN i.e. BRAN2020?

Lines 209-212 As BRAN …. To ….metrics, Thgis whole statement seems out of place and uncessary. Delete and start with: We compare annual means…

216 better resolved

228, 246 and throughout: Call it SEqld ROMS

246 across scales

247 Please explain the resolution of the satellite derived geostrophic velocities to explain why they are less than BRAN. I would expect BRAN KE to match the satellite derived KE (see previous BRAN comparison papers)

250 again what product, what smoothing? KE spectra is best computed alongtrack to preserve satellite resolution

251: This sentence is unclear

252: ‘roll off’ seems colloquial - normally the spectra would be cropped for wavenumber smaller than what is effectively resolved

252-254: I don’t feel the authors have justified this conclusion in the text and the figure does not tell me this. Please revise the conclusions of this analysis.

264 “dying out” colloquial

280-285: Speculation? Or include justification in more detail. Seems SEqld gets too much separation north of 27S – hence the current too weak compared to the EAC array observations (Figs 2 and 3). Is this related to this statement?

291 - be more general, all global circulation models, not just BRAN. Reanalyses will be time-accuarate. “Additionally, larger-scale models which represent this study area, such as BRAN, provide time-accurate representations of the mesoscale circulation,” could be replaced with “Additionally, near-global reanalyses, such as BRAN, provide time-accurate representations of the mesoscale circulation, ….”

297: I feel this model validation is really lacking the submesoscale – what about using the high frequency moorings to show frequency spectra and how the fine scale, high frequency motions are represented in SEqld ROMS compared to coarser products or models. Or better still SWOT KE spectra. Can you add a SWOT line to a and b (obviously not for 2012 but for the available period)?

304-305: Yes – if you don’t validate against submesoscale resolving obs directly, perhaps reference some of the values observed further south and show your model looks more like these… i.e. what Ro are observed by the Coffs radar – do they look similar to your model values…?

314 comparable slope? 315-316, good

329 BARRA is hourly, correct?

Citation: https://doi.org/10.5194/egusphere-2026-1852-RC2
- AC2: 'Reply on RC2', David Gwyther, 01 Oct 2026
  
  We thank the reviewer for their comment.
  Like with RC1, the comments and feedback from RC2 have substantially improved this manuscript.
  We have provided a response to all points in the attached PDF, and corrected the manuscript accordingly.
  Kind regards,
  David Gwyther (on behalf of the co-authors)
  
  Citation: https://doi.org/10.5194/egusphere-2026-1852-AC2

David E. Gwyther, Megan Jeffers, and Helen Bostock

Viewed

Total article views: 510 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
297	172	41	510	29	29

HTML: 297
PDF: 172
XML: 41
Total: 510
BibTeX: 29
EndNote: 29

Views and downloads (calculated since 18 May 2026)

Month	HTML	PDF	XML	Total
May 2026	133	41	12	186
Jun 2026	25	12	5	42
Jul 2026	39	26	6	71
Aug 2026	24	14	3	41
Sep 2026	43	50	6	99
Oct 2026	33	29	9	71

Cumulative views and downloads (calculated since 18 May 2026)

Month	HTML	PDF	XML	Total
May 2026	133	41	12	186
Jun 2026	25	12	5	42
Jul 2026	39	26	6	71
Aug 2026	24	14	3	41
Sep 2026	43	50	6	99
Oct 2026	33	29	9	71

Viewed (geographical distribution)

Total article views: 479 (including HTML, PDF, and XML) Thereof 479 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 12 Oct 2026

Short summary

The Australian eastern seaboard is dominated by the East Australian Current, a fast flowing jet that shapes coastal conditions, marine ecosystems and heat distribution. Within this system, small short-lived swirling features – too tiny for most computer models to capture – play an important role in moving nutrients and heat. We present a high-resolution ocean model that simulates these fine-scale processes, to be used for investigating cross shelf exchange, nutrient transport and ocean dynamics.


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