Seasonal prediction of ocean biogeochemistry in the Northeast U.S. Large Marine Ecosystem using a regional ocean-biogeochemistry model

Ross, Andrew; Stock, Charles; Koul, Vimal; Jia, Liwei; Johnson, Nathaniel; McHugh, Colleen

doi:10.5194/egusphere-2026-5262

Preprints

https://doi.org/10.5194/egusphere-2026-5262

Preprints

16 Sep 2026

| 16 Sep 2026

Status: this preprint is open for discussion and under review for Ocean Science (OS).

Seasonal prediction of ocean biogeochemistry in the Northeast U.S. Large Marine Ecosystem using a regional ocean-biogeochemistry model

Andrew Ross, Charles Stock, Vimal Koul, Liwei Jia, Nathaniel Johnson, and Colleen McHugh

Abstract. We couple an ocean biogeochemical model to a regional ocean model of the Northwest Atlantic and assess the potential skill of bottom dissolved oxygen, aragonite saturation state, and pH forecasts along the Northeast U.S. Continental Shelf using a reanalysis-forced model run as the true biogeochemical state. Across 30 years of retrospective forecasts initialized four times per year, the model has substantial potential skill for all three variables, with skill generally highest in the deep Gulf of Maine and lowest in the shallower Mid-Atlantic Bight and Georges Bank. Forecasts initialized in January and April benefit from the insulation of the bottom from the unpredictable atmosphere by stratification and tend to have greater skill at longer lead times. Using two case studies, we show how predictable variations in ocean temperature, salinity, and currents are associated with predictable variability in bottom biogeochemistry in both the Mid-Atlantic Bight and Gulf of Maine. The forecast model evaluated in this study has been developed as an operational prototype for NOAA's Changing Ecosystems and Fisheries Initiative, with the goal of applying the forecasts to improve the yield and sustainability of living marine resources.

Received: 31 Aug 2026 – Discussion started: 16 Sep 2026

Competing interests: One of the coauthors serves as 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.

Download & links

Preprint (PDF, 3656 KB)

Supplement (473 KB)

Download & links

Andrew Ross, Charles Stock, Vimal Koul, Liwei Jia, Nathaniel Johnson, and Colleen McHugh

Status: open (until 11 Nov 2026)

Post a comment Subscribe to comment alert

RC1:
'Comment on egusphere-2026-5262', Lachlan Phillips, 15 Oct 2026 reply
General Comments
This manuscript details a biogeochemical model (MOM6-COBALT-NWA12) coupled to a regional ocean model of the Northwest Atlantic and assesses the potential forecast skill for bottom dissolved oxygen, pH and aragonite saturation state across the Northeast U.S. Continental Shelf. The authors use a nudged analysis simulation, with temperature and salinity weakly nudged towards GLORYS12 to initialise 30 years of 10-member hindcast ensembles started four times per year. They then evaluate these forecasts against the same analysis as reanalysis-referenced potential skill. The authors report substantial potential skill, highest in the Gulf of Maine and from winter initialisations and demonstrate the sources of that skill with two case studies.
The paper is well written and makes a useful contribution to operational biogeochemical forecasting with a focus on fisheries management. The manuscript was a pleasure to read, and I consider it suitable for publication in Ocean Science after some minor revisions. My main concern is that the Discussion section does not draw closely on some of the interesting points raised in the Results. This is detailed below in the specific comments followed by some minor technical corrections.
Specific Comments
The Discussion does not draw closely on the Results. The discussion is sensible and clear but reads largely as a general list of future improvements rather than an interpretation of the specific results. As a result, the authors leave some very interesting findings unaddressed that I feel would significantly benefit the manuscript. I suggest expanding the discussion around the sources and seasonal windows of predictability the results reveal and framing the proposed improvements in those terms. In particular:
A key result of the paper is the dependence of skill on the initialisation season, with forecasts initialised in January retaining skill several months longer than forecasts initialised in October (particularly for bottom oxygen). The authors attribute this to stratification isolating the bottom from the atmospheric forcing with skill dropping off in autumn as less predictable surface conditions are mixed downwards [L279-283]. This is restated in the Discussion [L432-434] but it’s not explored further beyond a general case for improved data assimilation and initial conditions. I would love to read more discussion of what this seasonal dependence means for the forecast system and which forecasts would benefit most from improved initialisation.

Another interesting result that is stated but not expanded upon in the discussion is that oxygen skill depends mainly on initialisation, while pH and aragonite saturation skill depend more on region [L284-285]. I think this point warrants further discussion, and I would like to see the authors' take on this. One possibility is their different air-sea equilibrium timescales, with bottom O2 anomalies reset by winter ventilation while carbonate anomalies persist through winter mixing as water-mass memory.

The Discussion highlights that the high skill in the Gulf of Maine is consistent with predictable advection [L434-436], but the lower skill in the Mid-Atlantic Bight receives only indirect comment through river discharge [L500-503], and the abrupt drop in skill for November conditions there [L290-292] is not discussed. I would be curious to see the authors’ interpretation of these patterns.

In Georges Bank and in other region/lead combinations, skill is high but often not significantly better than persistence [L289-290; Fig. 7]. This is important for an operational system as it highlights where the dynamical forecast adds value and where it does not. A brief discussion of where and why the model outperforms persistence would strengthen the case for the forecast system.

The two case studies are highlights of the manuscript, but I think they demonstrate more than the authors draw from them in the discussion. They are cited as evidence for the importance of initial conditions [L441-443], but they appear to illustrate two distinct sources of predictability: short-lead evolution of the Gulf Stream mesoscale features in the Mid-Atlantic, and multi-month persistence of initial water-mass anomalies in the Gulf of Maine. These operate on different lead times and therefore require different modelling approaches (good initial ocean state for the former, and large enough ensembles for the latter), and drawing out the distinction could improve the Discussion as a whole.

Technical Corrections
The authors clearly state that by “skill” they are referring specifically to “reanalysis-referenced potential skill” [L185-190]. The shorthand “skill” is used throughout the document, which is perfectly fine, however I would suggest retaining the term “potential” in the Conclusions section as has been done in the Abstract so this remains clear to quick readers of the manuscript.

Nudging temperature and salinity without adjusting velocity may affect mixing and therefore biogeochemistry. A sentence on whether this has been checked could be useful.

L323 “In both the analysis simulation, observations (Section 2.5), and forecasts…” drop the word both (referring to three things).

Fig. 6 caption: The phrase “note that this line is repeated on the x-axis” is a little unclear. I took this to mean the climatology is repeated across the two-year x-axis but I had to read this twice.

Incorrect cross-reference. L213 says observations are used to evaluate select forecasts “in the Discussion (Section 4)”, but that evaluation is in Section 3.3.

Reply
Citation: https://doi.org/10.5194/egusphere-2026-5262-RC1
RC2: 'Comment on egusphere-2026-5262', Tianning Wu, 26 Oct 2026 reply

General Comments

This manuscript investigated the seasonal forecast skill of MOM6-COBALT-NWA12 in the Northeast U.S. Large Marine Ecosystem (LME). The authors developed and evaluated 30 years of nudged analysis simulation and retrospective forecasts. The study provides a useful foundation for regional biogeochemical forecasting, with potential values for regional stakeholders. The manuscript is clearly written and well suited to Ocean Science. I recommend publication after minor revision. My main suggestion is to expand the discussion of several interesting findings that have been briefly introduced. Further interpretation of these findings would strengthen the connection between forecast skill and the underlying physical and biogeochemical processes. Specific comments are listed below.

Specific Comments

1. In lines 232-248 and Fig. 2, the authors suggest that the contrasting biogeochemical responses during the 2012 and 2015-2016 marine heatwaves may reflect differences in their physical drivers, including atmospheric forcing and advection of Gulf Stream water. I would suggest further elaboration of how these mechanisms influence biogeochemical conditions in the Northeast U.S. LME.

2. In Section 3.1, regional differences in model/observation comparison may reflect both model errors and uneven observational sampling. I suggest including a map showing the locations of the point observations and indicating where the larger discrepancies occured. It is helpful to discuss how representative the observations are of each region and whether the regional statistics are influenced by sampling distribution or model error.

3. Lines 266-275 and Fig. 6 show forecast drift in several variables and regions, particularly in aragonite saturation state across all four regions. Although the authors described how this drift is corrected, its origins need further discussion. It would be useful to examine whether the biogeochemical drift also accompanies drift in physical conditions or only reflects drift in the biogeochemical model. This would help clarify the rationale and limitations for correcting drift statistically.

4. In Section 3.3.1, the authors mentioned that ensemble mean smooths Gulf Stream meanders and rings, particularly at long lead times in the forecasts initialized in April. I suggest showing members with contrasting Gulf Stream meander and ring evolution and evaluating their corresponding biogeochemical forecast skills. This would help evaluate whether more accurate Gulf Stream circulation can contribute to better biogeochemical predictions on the shelf. It would also assess how uncertainty in Gulf Stream forecast contributes to biogeochemical forecast uncertainty, providing useful context for developing and interpreting probabilistic forecasts from ensembles.

Reply

Citation: https://doi.org/10.5194/egusphere-2026-5262-RC2

Andrew Ross, Charles Stock, Vimal Koul, Liwei Jia, Nathaniel Johnson, and Colleen McHugh

Supplement

https://doi.org/10.5194/egusphere-2026-5262-supplement

Data sets

Model source code for "Seasonal prediction of ocean biogeochemistry in the Northeast U.S. Large Marine Ecosystem using a regional ocean-biogeochemistry model'" A. Ross et al. https://doi.org/10.5281/zenodo.22210571

Model code and software

Model output for "Seasonal prediction of ocean biogeochemistry in the Northeast U.S. Large Marine Ecosystem using a regional ocean-biogeochemistry model'" A. Ross et al. https://doi.org/10.5281/zenodo.22211043

Andrew Ross, Charles Stock, Vimal Koul, Liwei Jia, Nathaniel Johnson, and Colleen McHugh

Viewed

Total article views: 453 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	Supplement	BibTeX	EndNote
264	139	50	453	54	64	51

HTML: 264
PDF: 139
XML: 50
Total: 453
Supplement: 54
BibTeX: 64
EndNote: 51

Views and downloads (calculated since 16 Sep 2026)

Month	HTML	PDF	XML	Total
Sep 2026	182	73	30	285
Oct 2026	82	66	20	168

Cumulative views and downloads (calculated since 16 Sep 2026)

Month	HTML	PDF	XML	Total
Sep 2026	182	73	30	285
Oct 2026	82	66	20	168

Viewed (geographical distribution)

Total article views: 453 (including HTML, PDF, and XML) Thereof 453 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 27 Oct 2026

Short summary

This paper examines the accuracy of forecasts of ocean oxygen and acidity, both critical for the healthy functioning of fish and marine ecosystems, along the continental shelf off the Northeast U.S. coastline. The forecasts are produced using a novel numerical model able to resolve fine oceanographic details. Comparison with a reference dataset shows that the model often accurately predicts oxygen and acidity up to several months in advance.


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