the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 29 Jul 2026
A new ocean model configuration for management of central North Pacific Marine Resources (MOM6-COBALT-HIC6kv1.0)
Abstract. Despite often being referred to as an “ocean desert”, the North Pacific Subtropical Gyre (NPSG) supports economically valuable fisheries worth hundreds of millions of dollars annually, including Pacific bigeye tuna (Thunnus obesus) and other high-value pelagic species. The waters in and around the Hawaiian Islands support valuable and culturally significant nearshore fisheries and coral reef ecosystems and fuel ocean recreation and tourism valued in the billions. To be useful in understanding and managing these interacting resources, models must adequately represent ocean dynamics across multiple scales, from basin-wide circulation patterns to finer-scale island-ocean interactions. We evaluate the capacity of an ocean and biogeochemical model with 6 km horizontal grid spacing, a resolution considered “high” for climate-scale applications, across these scales. After a number of adjustments, including but not limited to calibrating light attenuation to subtropical observations and adjusting wind forcing to better reflect wind-current feedbacks, the model demonstrates strong skill in capturing basin-scale pelagic habitat features, including seasonal sea surface temperature anomalies (r > 0.95), accurate Transition Zone Chlorophyll Front migration and anomalies, and oxycline structure critical for bigeye tuna habitat compression. It also captures key features of the high seas ecosystems as represented by observations at station ALOHA, including exceptionally deep chlorophyll maxima (DCM) and seasonal variations and trends in acidification, though modest biases are evident. At the scale of island-ocean interactions, the model captures heightened eddy kinetic energy (EKE) downstream of the Hawaiian Islands and elevated productivity associated with the island mass effect. Finally, in nearshore regions, the model captures 88 % of the temperature variance observed at coral bleaching and habitat monitoring stations despite limited resolution. Overall, the configuration provides a robust foundation for a range of marine resource applications. Potential directions for future improvement include incorporating time-varying biogeochemical boundary conditions, refining salinity dynamics and carbonate chemistry representation, along with continued enhancement of nearshore circulation processes.
Competing interests: At least one of the (co-)authors serves as topic editor for the special issue to which this paper belongs.
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.- Preprint
(3430 KB) - Metadata XML
-
Supplement
(692 KB) - BibTeX
- EndNote
Status: open (until 22 Nov 2026)
- RC1: 'Comment on egusphere-2026-1877', Anonymous Referee #1, 18 Sep 2026 reply
-
RC2: 'Comment on egusphere-2026-1877', Anonymous Referee #2, 19 Oct 2026
reply
This manuscript presents HIC6k, a 6 km MOM6–COBALTv3 configuration for the Hawaiian Island Chain and the surrounding North Pacific Subtropical Gyre. The scope of the model and the multi-scale evaluation are useful. In particular, I appreciated the attempt to evaluate the model from basin-scale fields to island wakes and nearshore temperature records. The comparison with Station ALOHA data is also a valuable part of the paper.
The manuscript would benefit from a clearer description of the configuration development and a more careful distinction between results that are directly evaluated and intended future applications. Several inconsistencies in figures, tables, units, and line references also need to be resolved. These issues are important for a model-description paper because readers should be able to understand why the final configuration was selected and how its performance was assessed.
Major comments
1. Section 2.2 is one of the most useful parts of the paper, but the evidence for the individual adjustments is not presented consistently. For example, the current-feedback experiments are shown in the Supplement, the longwave adjustment includes a change in SST bias, and the phosphate adjustment is illustrated in Fig. S3. In contrast, the effects of the light-attenuation coefficient, background diffusivity, and gustiness are described mainly in qualitative terms.
Please add a concise table summarizing the adjustments. For each adjustment, it would be helpful to show the final value, the previous or reference value, the diagnostic used to select it, the simulation period or run length used for the comparison, and whether the changes were tested separately or added sequentially. If only limited development experiments are available, this can be stated clearly. Such a summary would make the configuration much easier to reuse or compare with other regional MOM6 systems.
2. The spatially targeted 10% increase in downwelling longwave radiation is an important choice and needs a more transparent explanation. The manuscript attributes the cold bias to an inconsistency between ERA5 forcing and GLORYS12 boundary conditions. Please explain what diagnostic supports this interpretation, for example a surface heat-budget analysis, a comparison of relevant terms, or a short explanation of how other possible causes were assessed. The location and spatial extent of the correction should also be described more precisely.
Please also discuss the limitation of applying this adjustment outside the historical forcing setup. A correction tuned using ERA5 and GLORYS12 may not necessarily be appropriate for simulations using different boundary conditions or future forcing.
For the current-feedback scaling, please reconcile the references to Figs. S1 and S2 and state more clearly which EKE metric was used to select αcfb = 0.2. The wording is also inconsistent across the manuscript: Section 3.4 describes close agreement with the observational products, whereas the Conclusions still refer to EKE underestimation. These statements should be made consistent.
3. The manuscript states that the hindcast begins in 1993, but it is not clear whether an adjustment or spin-up period was used, or whether all statistics include the first simulation years. Please state this explicitly.
A brief diagnostic of temporal drift would strengthen the paper, especially because the biogeochemical boundary conditions are climatological. Time series or a short summary for selected temperature, salinity, oxygen, nitrate, and phosphate quantities would be sufficient. This would help readers judge whether the reported biases are stable over the integration period.
Please also clarify the treatment of biogeochemical tracers at the open boundaries and in the sponge region. Section 2.1.1 explicitly describes relaxation of momentum, temperature, and salinity, but the corresponding treatment of biogeochemical tracers is not clear.
4. The nearshore evaluation is potentially very useful, but the methods and interpretation need to be more explicit.
First, the manuscript should distinguish the correlation of absolute temperatures (mean r = 0.96) from the correlation of deseasonalized anomalies (mean r = 0.57). The statement in the Abstract that the model captures 88% of temperature variance is not sufficiently clear in relation to these reported statistics. Please use a consistent metric in the Abstract, Results, Discussion, and Conclusions, and report the anomaly skill alongside the raw temperature skill.
Second, please describe how model output was matched to each subsurface temperature recorder: the horizontal selection rule, the model depth used for 5, 15, and 25 m observations, and the temporal aggregation used in the statistics. This is particularly important given the 6 km grid spacing and the nearshore locations of the recorders.
Third, Fig. 16 shows monthly time series. The Discussion, however, refers to skill at 2–90 day timescales and uses this to support the discussion of degree-heating weeks (DHW). Please state the temporal resolution used in the quantitative comparison. If sub-monthly variability was not evaluated, this claim should be softened. A direct DHW comparison would be a useful extension, but the present conclusions should remain limited to the analysis that is shown.
Finally, the claimed advantage of the model's three-dimensional temperature field over satellite SST is reasonable in principle, but it is not demonstrated here. Either compare model temperatures at the recorder depths with the observations or state this more cautiously.
5. The Station ALOHA comparison is a strength of the paper, but some statements are more positive than the values in Table 1 suggest. For primary production, the modeled value is 298 mg C m⁻² d⁻¹, while Table 1 lists observational estimates of 180–220 and 492 mg C m⁻² d⁻¹. Please explain the differences in methodology or time period and avoid describing the agreement as simply “aligned well” without this context. The underestimation of particle export flux should be discussed together with this comparison.
Similarly, the modeled mesozooplankton carbon value (681 mg C m⁻²) appears substantially higher than the observational value listed in the table (236 ± 107 mg C m⁻²). Please discuss whether the model and observational quantities have the same depth integration and size-class definition before concluding that they are within observational ranges.
Please also check Table 1 carefully. The footnotes and superscripts are not fully consistent, the model names vary, and “Low Chlorophyll Biome” should be defined.
6. A careful revision is needed to correct several issues that affect reproducibility and interpretation. In particular, line 209 states that Fe initial and boundary conditions were taken from WOA18, but WOA does not provide an iron product. Please correct the source. The reference in Section 4.3 to a future “explicit representation of the iron cycle” should also be clarified, since the model description states that COBALTv3 includes iron cycling.
The nitrate discussion around lines 536–543 is difficult to follow because the seasons associated with the negative biases are not identified and the reported bias and RMSE units are inconsistent. Figure 15 and its description also need revision: the caption refers to the DCM, while the text refers to 100 m and surface chlorophyll. Please state exactly what is plotted and use consistent units.
Minor comments
- Use one acronym and version format consistently (HIC6k, HIC6km, and HIC 6km are all used). Define the version string in the model-description section if it is retained in the title.
- Please verify the gustiness value and its units at line 248, and explain how it differs from the reference configuration.
- The supplementary figure citation for the EKE sensitivity test is inconsistent: lines 272–273 cite Fig. S2, while later text refers to Fig. S1 for the same experiments.
- The OC-CCI period in lines 331–335 (1998–2019) does not appear to match the period stated for Fig. 8. Please reconcile the data periods.
- Please provide enough information to reproduce the Argo comparison in Fig. 6, including the data subset, period, number of profiles, and the definition of the oligotrophic/eutrophic classification.
- In Fig. 11, the panel references in the caption should be checked.
- Figure 16 requires careful revision. The STR record shown is only available for a few years around the 2015 event, not for the full three-decade period. The text and caption do not match the plotted colours in panel (c), and the site longitude differs between the text and figure caption.
- Table 2 lists HIC6k, GLORYS, and OISST columns, although the caption also refers to CRW. The duplicated phrase “Hawaiian Islands” in the caption should be removed.
- A few minor editorial points also need checking: “Phosphorous,” the notation in Section 2.2.4, “CDM” versus “DCM,” the use of “psu,” and the figure references in the Discussion. There are also a few typos around lines 900–902.
Citation: https://doi.org/10.5194/egusphere-2026-1877-RC2 -
RC3: 'Comment on egusphere-2026-1877', Anonymous Referee #3, 05 Nov 2026
reply
General Comments
This manuscript presents a comprehensive and well-organized description of the motivation, development, and evaluation of a coupled physical–biogeochemical regional ocean model for the central North Pacific. The authors clearly demonstrate the need for such a modeling framework and provide a thorough assessment of its performance across a wide range of physical and biogeochemical processes.
One of the notable strengths of this study is that it builds upon previous regional MOM6 configurations by extending model development and optimization to include not only physical ocean processes but also biogeochemical processes. I particularly appreciate the extensive use of diverse observational datasets to evaluate the model's ability to reproduce key physical and biogeochemical features. Furthermore, the authors provide a useful discussion of the remaining model limitations and potential directions for future improvements.
Overall, I believe that this study makes a valuable contribution to regional ocean modeling and will serve as a useful reference for the regional MOM6 modeling community, particularly for researchers interested in coupled physical–biogeochemical applications. Therefore, I recommend minor revisions before publication. However, I have several comments that should be addressed to further improve the clarity and scientific robustness of the manuscript, as outlined below.
Major Comments
1. Mixed-layer restratification parameterization
The physical model configuration appears to be largely based on Ross et al. (2023). However, I understand that a different mixed-layer restratification parameterization is employed in the present configuration. I suggest that the authors explicitly describe this difference and briefly discuss the rationale for adopting the current parameterization. This information would help readers better understand the physical model configuration and its differences from previous regional MOM6 configurations.
2. Spin-up of the physical and biogeochemical models
Could the authors clarify whether spin-up simulations were conducted for both the physical and biogeochemical components? Although GLORYS12v provides relatively high-resolution physical initial and boundary conditions, the model may still require an adjustment period to reach a dynamically consistent state. Similarly, the biogeochemical fields may require sufficient spin-up, particularly given the differences in spatial resolution and representation between the initialization datasets and the HIC6k configuration. If spin-up simulations were performed, I suggest providing information on their duration and the criteria used to determine whether the model had sufficiently adjusted. If no spin-up was performed, the authors should discuss the potential influence of initial adjustment on the model evaluation.
3. Biogeochemical biases near the open boundaries (Figs. 9 and 10)
Figures 9 and 10 show relatively large biases in chlorophyll and nitrate concentrations near the open boundaries. The authors mention that sponge layers are applied to horizontal momentum (u, v) and physical tracers (temperature and salinity), but do not mention their application to biogeochemical tracers. Have the authors considered or tested the application of sponge layers to biogeochemical tracers? If not, applying sponge layers to these variables might be a potential approach to reducing the biases near the open boundaries and could be considered as a direction for future model improvement.
4. Possible causes of the SST cold bias (Section 2.2.3)
The authors attribute the SST cold bias primarily to an energy imbalance between the atmospheric forcing and ocean boundary conditions. However, I am not fully convinced that this imbalance alone explains the cold bias. Other regional MOM6 configurations have also employed GLORYS12v for open boundary conditions and ERA5 for atmospheric forcing, without necessarily exhibiting a predominantly negative SST bias. This suggests that other aspects of the model configuration may also contribute to the cold bias. For example, the choice of mixed-layer restratification parameterization and its associated coefficients could influence upper-ocean mixing and SST.
5. Selection of the current-feedback scaling parameter (Lines 285–291)
The authors selected a current-feedback scaling parameter of 0.2 based on the sensitivity experiments presented in Fig. S1. However, some of the statistical metrics appear to favor a different parameter value. Could the authors clarify why 0.2 was considered the optimal choice? A more detailed explanation of the selection criteria, including which aspects of model performance were prioritized, would help justify the chosen parameter value.
6. Calculation of temperature anomalies (Lines 690–699)
In Fig. 16d, the temperature anomalies for the different datasets are calculated relative to the CRW 5 km climatology. However, when evaluating the model's ability to reproduce interannual variability, anomalies are generally calculated by removing the climatology of each individual dataset over a common reference period. Could the authors explain why the CRW 5 km climatology was used as a common reference rather than calculating anomalies relative to each dataset's own climatology?
7. Representation of the 2015 warming event (Lines 697–699)
The authors state that HIC6k successfully captures the timing and magnitude of the 2015 warming event. However, Fig. 16 appears to show that HIC6k underestimates the amplitude of the 2015 warming event compared with the observational products. Furthermore, the simulated amplitude of the 2015 event appears to be smaller than that of the 2020 warming event. Given these differences, I am not fully convinced that the model accurately captures the magnitude of the 2015 event. I suggest that the authors reconsider this statement and provide a more balanced discussion of the model's ability to reproduce the timing and amplitude of extreme warming events.
Minor Comments
1. Line 273: "Fig. S2" should be corrected to "Fig. S1".
2. Figures 4, 5, and 6: Unlike other spatial figures, these figures do not include longitude and latitude labels. I suggest adding coordinate labels to maintain consistency throughout the manuscript.
3. Line 770: The reference to "Fig. 3a" appears to be incorrect, as Fig. 3a shows spatial anomalies of sea surface height (SSH). Should this instead refer to "Fig. 8a"?
4. Line 696: Please change "CRW5km" to "CRW 5km" for consistency with the terminology used elsewhere in the manuscript.
5. Throughout the manuscript: Both "HIC6k" and "HIC6km" are used to refer to the model configuration. Please use a consistent model name throughout the manuscript.
6. Figure 16d: The label "HIC 6km" should be revised to match the model name consistently used throughout the manuscript (e.g., "HIC6k").
Citation: https://doi.org/10.5194/egusphere-2026-1877-RC3
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 418 | 238 | 46 | 702 | 87 | 46 | 48 |
- HTML: 418
- PDF: 238
- XML: 46
- Total: 702
- Supplement: 87
- BibTeX: 46
- EndNote: 48
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This manuscript presents a new high-resolution coupled physical-biogeochemical model configuration (MOM6-COBALT-HIC6k) for the central North Pacific and Hawaiian Islands region. The stated objective is to provide an ocean forecasting and projection framework that can support living marine resource management, including applications to pelagic fisheries and ecosystem assessments.
The manuscript's principal strengths are the large model domain for the particular study area, relatively high horizontal resolution, comprehensive physical and biogeochemical validation effort, and the multi-scale evaluation strategy spanning basin-scale, mesoscale, and nearshore environments. The model reproduces many ecosystem-relevant physical and biogeochemical variables, including temperature, salinity, oxygen, chlorophyll, nutrients, mixed-layer properties, and sea-surface height. Particularly notable strengths include the realistic representation of the oxycline and oxygen minimum zone structure, the reproduction of mesoscale variability patterns, and the assessment of model performance across seasonal to decadal timescales. The authors also demonstrate generally realistic vertical structure and temporal variability across a wide range of observational datasets, providing confidence that the model captures key physical and biogeochemical processes relevant to regional ecosystem dynamics.
My main concern is the suitability of the evaluation for the intended fisheries management applications. In particular, the manuscript devotes extensive attention to chlorophyll and nitrate-phosphate state variables while providing little information on zooplankton biomass, despite motivating the model in the context of pelagic fisheries and tuna habitat applications. Certain aspects of the biogeochemical model configuration could also be described in greater detail. In addition, further discussion of nutrient limitation, particularly the roles of iron and silicate, and information on computational performance would strengthen the manuscript.
Overall, I find the model configuration valuable and relevant for the GMD readership. The model development is solid, the evaluation is extensive, and the manuscript is generally well written.
General Comments
1. Model suitability for fisheries applications
The manuscript is motivated partly by applications to bigeye tuna and other pelagic fishes. However, fish production and fisheries yields are not solely determined by primary productivity. Several studies have shown that mesozooplankton productivity, trophic transfer efficiency, and secondary production are stronger predictors of fisheries production than phytoplankton biomass indicators such as chlorophyll concentration alone (e.g., Friedland et al., 2012; Stock et al., 2017). In addition, fisheries and ecosystem models such as those described by Kearney et al. (2012) and Dueri et al. (2014) rely on the representation of zooplankton and mesozooplankton prey fields to connect lower-trophic-level ecosystem dynamics with fish production and habitat suitability. I appreciate that Table 1 includes standing-stock estimates for microzooplankton and mesozooplankton, which provide useful information on the mean state of the food web. However, standing stocks alone provide only a partial assessment of model skill. Given the fisheries-oriented motivation of the study, I encourage the authors to provide, where observational datasets are available, a more comprehensive evaluation of zooplankton biomass and secondary production, including information on their spatial distributions and temporal variability. Such diagnostics would help demonstrate whether the model reproduces prey fields that are directly relevant to higher trophic levels and fisheries applications.
2. Biogeochemical model description
The physical model configuration is generally well described. However, the description of the biogeochemical model would benefit from additional information on the zooplankton component. How many zooplankton groups are represented and how is zooplankton mortality parameterized? Given the relevance of zooplankton to both grazing control of phytoplankton and prey availability for higher trophic levels, a brief discussion of whether mortality includes a density-dependent "higher-predation closure" term representing unresolved predators would be particularly useful for the fisheries applications outlook of the model.
3. Nutrient limitation
The manuscript evaluates nitrate and phosphate but provides little discussion of iron and silicate. Iron modulates diazotroph activity (as mentioned in the discussion) and silicate controls diatom growth. Even if iron and silicate are not dominant limiting nutrients within the focal region, the authors should justify why they were excluded from the evaluation.
Specific comments
Introduction
Line 102: “The computational expense of these models must also be manageable enough to enable sufficient simulations”
For a model development and evaluation paper in GMD, I would appreciate information.
Such information is useful for readers considering adoption of the configuration and reproducibility.
Methods
Line 214: The statement regarding lower phosphate concentrations provides an opportunity to mention diazotroph dynamics and their role in regional nutrient cycling earlier in the article.
Line 215: Please clarify what is meant by "biological tracers which equilibrate rapidly." How quickly do these tracers converge?
Lines 343, 382: Please justify the omission of iron and silicate from the nutrient evaluation.
Results
Figure 4i: Could the authors comment on the relatively large temperature differences near 200 m depth within the subtropical interior?
Lines 507-508: The model appears to maintain excessive chlorophyll concentrations at depth relative to observations. Could this issue be revisited in the discussion?
Figure 9f: The large spring chlorophyll discrepancies in the northwestern domain deserve additional explanation.
Figure 10f: Similarly, the surface nitrate differences in the northwestern domain warrant discussion.
Discussion
Lines 809-810: “Integrated biological metrics… (Table 1)” Agreement in integrated carbon stocks alone is not sufficient to demonstrate skill in zooplankton dynamics and potential fisheries applications.
Line 926: “Potential mechanisms underlying this bias include inadequate top-down control from light-dependent grazing by microzooplankton”. This further motivates inclusion of zooplankton evaluation, if observational datasets are available.
Refrences
Dueri, S., Bopp, L. and Maury, O., 2014. Projecting the impacts of climate change on skipjack tuna abundance and spatial distribution. Global change biology, 20(3), pp.742-753.
Friedland KD, Stock C, Drinkwater KF, Link JS, Leaf RT, Shank BV, Rose JM, Pilskaln CH, Fogarty MJ. Pathways between primary production and fisheries yields of large marine ecosystems. PloS one. 2012 Jan 20;7(1):e28945.
Kearney, K.A., Stock, C., Aydin, K. and Sarmiento, J.L., 2012. Coupling planktonic ecosystem and fisheries food web models for a pelagic ecosystem: Description and validation for the subarctic Pacific. Ecological Modelling, 237, pp.43-62.
Stock, C.A., John, J.G., Rykaczewski, R.R., Asch, R.G., Cheung, W.W.L., Dunne, J.P., Friedland, K.D., Lam, V.W.Y., Sarmiento, J.L., & Watson, R.A. (2017). Reconciling fisheries catch and ocean productivity. PNAS, 114, E1441-E1449.