Review of "Eight years of continuous Rockall Trough transport observations using moorings and gliders"

Recommendation:

Requires major revision before acceptance to "Ocean Science". Suggested improvements and key points are listed below.

Main points requiring revisions:

This study presents an updated methodology for analysis of observational data from Rockall Trough. The new methodology is supposedly better than existing methods, but there is no hard evidence that the new version is indeed an improvement. I recommend adding analyses that quantitatively show why this is the case. To do this, the following approaches come to my mind (but the authors may choose differently):

• One could test the different methods in a numerical model that has sufficient resolution and fidelity in the Rockall Trough (validate the model first - does not need to be GLORYS). In the model world, the actual flow (of volume, heat, and freshwater) is known, and the observing system results can be simulated (in an OSSE type analysis).
• If the biggest improvement is the better time resolution compared to earlier methods, one could also compare spectra of the resulting time series from old/new methods. If the new method is really superior, the shape and power levels in the spectra should demonstrate what kind of aliasing is now avoided, and what the presumably lowered noise floor at high frequencies looks like.

The study does not quantify uncertainties in the resulting time series. Error bars, in the few places where they exist, denote the error of the mean when averaging e.g. several years into an annual cycle, but this says nothing about the uncertainty of the original time series itself. Can we believe the smaller wiggles in figure 5a, or are they instrumental noise? Is the ~0.8 Sv misfit (l. 150) the dominant contribution to uncertainty, or is it the sensor errors that are listed for glider and mooring CTD - how many Sv result from these sensor errors? I recommend including a systematic accounting of the major sources of uncertainty, and adding those to the relevant figures and results. Just copying the manufacturer's sensor specifications and not propagating them into the resulting volume, heat, and freshwater fluxes is, in my opinion, insufficient.

The methodology for heat and freshwater transports is flawed in that underlying data are not available at sufficient spatial resolution, let alone with appropriately co-located measurements. As a bare minimum, there needs to be a validation why the method should still give correct results. This could (again) be done with an OSSE-type numerical simulation. If these validations have been done in some of the referenced work, they need to be summarized here. I did not review the section with heat and freshwater transports at this time.

There is an inconsistency or flaw in the method, in that it uses EOF patterns but use of patterns higher than order one fails to improve the results in the validation step. I suspect this is due to how observational and model data are mixed in the methodology. There are comments about this below, as well as suggested steps to address this.

The overall presentation needs to be polished; I am including detailed comments below. For a publication-ready manuscript, I expect more consistency with labels, abbreviations, etc., across the figures and text.

Improvements to content:

Ll. 47-49:

In the transition between the paragraphs here, it is not clear whether this study solves the shortcomings of the glider observations, or reveals what they were in the first place. Maybe change the wording in the paragrph lines 49 ff. to make this clear.

Ll. 47 and 85:

These state that the glider data are "scattered and sparse in time". It is not very clear what this means, nor what would be considered "good enough". Can this be clarified/quantified somehow? For the introduction, this is probably OK as is, but in the methodology section, I recommend being more quantitative.

Ll. 61 ff.:

The glider CTD sensors are essentially the same as the mooring ones. You are listing what appears to be the factory specifications for sensor accuracy here. For the moorings described later, you list similar accuracies but presumably requiring the tedious calibration procedures referenced there (line 97; McCarthy et al. 2015). Are the glider data processed with similar methods? If not, they will not be as accurate as described.

Ll. 93-95:

It is not clear how the WB1 and WB2 moorings are concatenated. The reference quoted in turn refers to another reference (McCarthy et al., 2015), which also leaves details somewhat open. There is a "correct" way to do this merger, assuming geostrophy and that the current and CTD data are actually available: One can start by integrating the CTD-derived specific volume anomaly upwards from the bottom of the deep mooring. Then, one can "jump" horizontally to the shallower mooring using the geostrophic equation and the currents at the depth of the jump (supposedly some average from nearby current meters on both moorings can be used). Then, continue integrating the CTD-derived data up along the shallower mooring. Is this what was done?

L. 99:

What is meant by correcting the velocity measurements for "sound"? The instruments are sonic current meters, but what corrections need to be made?

Ll. 107 ff.:

Clarify whether the RTADCP will be used, or why it is worth mentioning here. Else, remove.

Ll. 138 ff.:

If the Hilbert EOF analysis is not used, I think this paragraph that refers to it can be removed.

Ll. 140 ff.:

There is a fundamental inconsistency in the method used here, which needs to be reconciled:

For input parameters, the regression method uses a mix of observational data (from the EB1 mooring) and numerical model data (from the GLORYS analysis at a single nearby but separate location). The observational and numerical data reflect two different "realities" that may be inconsistent with each other. There is no quantitative reasoning supporting the choice made here over other available choices, other than the choice of using only observational data from EB1 (which is shown to be inferior). In order to justify the choice made, I recommend additional analyses:

- Validate that the EOF patterns and magnitudes in the model world reasonably match the observational ones, i.e. recreate figure 3 from GLORYS data alone.

- Validate the model against the existing EB1 and glider observations.

- Use the model transport alone (at full model resolution in space and time), and quantitatively test whether this is inferior to your choice.

- Try to find some combination of input from the model (other than the velocities at the single RTADCP location, but simpler than using everything) that might optimize agreement with your reference data.

Ll. 155 ff.:

I am confused - you are only using the first EOF mode to reconstruct the velocity field, correct? Why does your figure 4b not look like figure 3 (mode 1) then? Shouldn't they be more similar?

Ll. 177-178:

I understand that the mid-basin transport is not the primary focus of this study, but since you are mentioning it here and in figures 6-7: The way I read the reference, the mid-basin transport is calculated using an assumption of no motion at ~1800 m. This is not consistent with how I read figure 1(b), in that there is more "red" at depth than "blue". I can only assume that there is a substantial amount of variability at that depth. In order to quantify the error from this reference level assumption, can you provide the time series of the velocity (averaged between WB2 and EB1) at 1800 m from the 17 ADCP sections from figure 1(b)? Multiplying this with the water depth and the section width will show you the error in terms of volume transport. I would not be surprised if that error were as large as your entire signal in figure 5. If I am doing the math right, 5 mm/s velocity variability will translate to 1 Sv error, but please double-check and provide the actual number.

Ll. 184 ff.:

I think there is a flaw in how the heat and freshwater transports are derived, in that the underlying temperature and salinity observations do not provide data at locations where it is needed. The way I understand the explanations below equations 4 and 5, the temperature and salinity profiles are from the moorings at the western and eastern edges (and an average of these). However, at least for the mid-basin transports, the section is much longer than typical mesoscale length scales (a case made obvious by figure 1). In order for equations 4 and 5 to hold, both the velocity field v as well as Theta and S profiles must be known at some sort of eddy-resolving resolution in situ. I would not have a lot of confidence in the outcome of these equations unless the methodology has been validated somehow (e.g. in a high-resolution numerical model that reasonably depicts the mesoscale eddy field inside Rockall Trough, where you can then compute the ground truth from equations 4 and 5 using the full model field and compare it with a version that resembles sparse observations). If this has already been done in one of the references, it should be explained here, together with some quantitative uncertainty estimate.

Ll. 188 ff., eqn. (5):

I think the sign of the salinities is wrong. As it stands, high salinity would give you higher Qf.

Ll. 205-207:

If you discuss the undercurrent here, you should refer to it in figure 4 and also include a panel in figure 4 that shows the glider data for comparison. Reg. the overestimate: The error bars overlap, so you could also say that the measurements agree within the given uncertainty, couldn't you?

Ll. 208 ff.:

There needs to be an explanation of the bias correction, why it is needed, and what it improves. Perhaps a reference to a publication plus a one-sentence summary is sufficient, but the way it stands, it sounds as if the bias "correction" actually makes agreement with observations here worse.

Ll. 257-258:

The sentence here uses the words "better" and "robust". I don't think the manuscript in its present state actually demonstrates that the new data are "better", although there are good reasons to believe this is true. It should, however, be demonstrated. As for "robust", I am not sure what that is supposed to mean - it refers to the seasonal cycle, but figure 5 shows that one can basically draw a straight line through the plot at about 1 Sv and be within all the error bars. If anything, this shows that the data cannot determine the presence of an annual cycle with certainty, doesn't it?

Ll. 269-271:

This sentence claims that the ESC is "disproportionately important" for something. When I look at figure 6, I find this to be completely untrue on two accounts: One, the heat and freshwater contributions are fairly proportional to the volume transport, and two, the EW contribution is not very important (instead, the total is dominated by the mid-basin transport).

Improvements to text (readability/appearance/typos):

Abbreviations and acronyms:

The amount of abbreviations is overwhelming. Some are not used consistently throughout the text (e.g. RTWB1 vs. WB1). Some are never spelled out (geogr. names from fig. 1). To make things a bit easier, be sure to:

- spell out each abbreviation at its first occurrence and additionally in each figure caption if it occurs inside a figure,

- avoid inconsistencies,

- consider adding a list of all abbreviations in the supplemental materials.

Ll. 17/18:

Remove "Fu et al." reference if unpublished, else update here and add proper citation to the reference list.

Ll. 61-62:

Remove "PSU".

L. 64: Change "Avaraged" to "Averaged"

L. 105: Change "gab" to "gap"

GLORYS references: When I search for "GLORYS" in the manuscript, I see inconsistent occurrences of GLORYS12, GLORYS21, GLORYS2, followed by equally inconsistent version numbers 1, 2, or 12. Please determine the exact version number used, and correct this throughout the manuscript. Why not just call it "GLORYS" everywhere and reference the full name once and once only with the dataset citation?

Ll. 158-159: I found the reference to the future equations confusing. This sentence can perhaps be removed.

L. 250:

I am wondering if "Conclusions" would be a more appropriate title than "Discussion" here.

L. 272:

Change "targetted" to "targeted"

Suggestions for figures and captions:

Use consistent labels and spelling across figures:

Figures 1(b), 3, 4, 6, S1 all have depth as vertical axis, but sometimes it counts positive up, sometimes down, and the axis label is spelled differently almost every time.

Ditto for 1(a), 1(b), 3, 4, S1 for longitude.

Ditto for time in 2, 7, 8, S2.

Figure 1:

Add (a) to top panel.

Increase size of panel (b) such that it is roughly as wide as panel (a).

Increase font size of panel (b) such that it roughly matches that of panel (a).

In caption, start the panel (b) part with verbiage that says what is shown (e.g., "cross-section view of Rockall Trough section").

In caption, be consistent with abbreviations - use either RTWB1 or WB1, ditto for ...2.

Panel (b) shows things that are not explained in the caption - explain or remove these: two dashed lines with red dots on top, one solid line with blue dot on top. I assume these are the same as in figure 4, but this needs clarification.

Use the same green symbols on panels (a) and (b), i.e. either circles on both panels, or triangles on both.

Either explain all abbreviations in the caption (include the geographic names of panel (a)), or point to a list (potentially in the supplemental materials) where they are explained.

In caption, change "focusses" to "focuses".

Figure 2:

Add cruise IDs to potential list of acronyms, and refer to this list in the caption.

In caption, change "CTD sensors (d-f)." to "CTD sensors (c-f)."

Figure 3:

The figure caption here refers to "meridional velocity", whereas figure 4 mentions "across section velocity". These two things are almost identical, but not 100%. Please confirm that each caption correctly describes the quantity shown, or correct as appropriate.

Make the axes limits identical to those from figure 4, and add the same vertical lines/symbols for orientation.

Figure 4:

Use the same symbols and colors as figure 1 (a) and (b) for the mooring etc. locations.

Figure 5:

The figure seems to show volume transport, but the caption calls the data "velocities depth-averaged...". Correct the caption such that it calls out the correct physical quantity.

Add the time periods for the blue and green curves in the legend (why just the orange and black?).

Figure 8:

The font size is inconsistently large compared to the other figures. Reduce font size to make it look similar to the others.

The paper presents an 8-year time series of volume, heat, and freshwater transports through the Rockall Trough in the eastern subpolar North Atlantic. The estimates utilize glider and mooring measurements, together with data from an ocean reanalysis product. These transports are critical in inducing hydrographic changes downstream within the subpolar gyre as well as further north in the Nordic Seas, which makes accurate estimates highly desirable. In particular, this analysis emphasizes a new method that it introduces for the wedge transport estimates, along with the new freshwater/heat transport estimates. However, I have two main concerns about both: it is hard to see a clear improvement from the new method compared to the old method, so it would require thorough validation; the seemingly arbitrary definitions of freshwater/heat transport would limit the usefulness of the estimates, e.g., when comparing them to other existing estimates in the region. I recommend a moderate/major revision with more detailed comments outlined as follows.

Main comments:

1. The authors proposed a new method for reconstructing the velocities in the eastern wedge for the upper 1000 meters, which had been previously reconstructed primarily using velocity data from an ocean reanalysis product. However, the robustness of these new glider-based estimates requires careful and more thorough validation. The first main issue is related to the velocity derived from the glider paths. The authors acknowledge this issue (lines 84-85) but unfortunately did not address the associated uncertainty and its subsequent impact. A related issue concerns the comparisons with various other estimates, which are briefly discussed in section 3.1. As this new method is poised to be a key improvement over previous estimates, a thorough discussion and stronger evidence of its superiority are needed.

2.The overall results and the analysis are comparatively limited. Much of the focus is on the volume transport and is similar to the previous publication based on shorter records. The authors indicated that the heat and freshwater transport estimates are presented for the first time here; if so, I believe they deserve a more in-depth analysis. Some questions could be readily addressed by the data, e.g., what is the relative contribution of velocity versus T/S variations to the transport variability? Which subregion dominates the T/S transport variability? Caution is needed regarding the definition of the heat and freshwater transports, as this may introduce uncertainties and make it difficult to interpret the results when comparing them to other estimates in the region. I would suggest the authors clarify what they are presenting and include sufficient information on how to contextualize the presented estimates with those in previous publications.

Other comments:

Line 10: This is a commendable goal, but the framework's validity requires more thorough demonstration.

Line 77: Add more details on the common section?

Figure 1: Please label all moorings in the main and inset plots.  For the caption: is ‘RTWB1’ the same as ‘WB1’?  Same question for ‘RTWB2’ and ‘WB2’.

Figure 2 and the related text: The notation ‘WB 1/2' is confusing. Please clarify if it referes to a single mooring location or a composite of WB1 and WB2.

Section 2.3:  Please elaborate on why the supplementary datasets are necessary and how they are specifically used in the analysis.

Line 117: Please provide clear definition for ‘western wedge’, ‘mid-basin’ and ‘eastern wedge’. Their spatial extents are not clear.

Line 140: Please label ‘RTADCP’ in the relevant figures.

Line 151:  What is ‘glider data at EB1’?  If velocity data from gliders are used at EB1, I would suggest additional tests to validate the reconstruction.  For instance, compare a reconstruction using: (a) mooring velocity at EB1 + first EOF, (b) velocity from GLORYS at the ADCP location + first EOF.  These two tests would better quantify importance of the EB1 and ADCP velocities for the eastern wedge transport.

Line 160 (Figure 4): To directly assess the impact from the new method, please include a comparison of the transport time series from the new and old methods.

Line 174: The description of the velocity reconstruction is confusing. Please elaborate with more details.  Specifically, clarify what ‘WB1/2’ refers to and where the ‘WB1/2 position’ is, both in the text and on the figures.

Line 185: How was the reference temperature determined? Elaborate on how to the choice of the reference influences the interpretation of the resulting temperature transport.

Line 189: Similar for the choice of salinity reference (Sref). A discussion on the sensitivity of the freshwater transport results to different Sref values is needed to understand the robustness of the estimates.

Line 219:  The reconstruction relied on a reanalysis product in addition to the mooring data. Please clarify.

Figure 5: The labels are confusing. Please specify what each line represents in the caption. Also, provide more details on how the estimates were obtained. For example, what is the difference between ‘old’ and ‘old resampled’?

Lines 203-204:  Be cautious with phrases like ‘differ notably’.  Specify what the numbers after plus/minus represent (e.g., standard errors).  And discuss the statistical significance of the mean differences.

Line 206: The analysis should consider the error bars (uncertainty) for both estimates when comparing them.

Line 212:  How does the gliders' irregular sampling affect the mean transport estimate? The glider-based transport is used as a benchmark in section 4.1, but its own robustness should be evaluated more thoroughly earlier in the paper.

Figure 6: The black line is also dashed – does it represent the old method? Please clarify in the caption.

Line 223: The heat and freshwater transport values are highly sensitive to the arbitrary choice of the respective reference, especially given the non-zero volume transport across the section.  This major caveat should be emphasized.

Figure 7bdf: what do the vertical color bars represent?  The mean value appears to be incorrectly placed in panel f – please verify.

Line 240:  The analysis would be significantly enhanced by investigating the potential causes of the observed differences.  For example, what is the relative contribution by T/S vs velocity changes? How is it related to volume transport and T/S properties in the specific wedges and mid-basin? 

Line 248: How do property changes in the region subsequently affect the corresponding transports across the Trough?  This would provide a strong linkage to the broader analysis.

Line 253:  It would be valuable to contextualize these transport estimates within the broader understanding of subpolar volume, heat, and freshwater transports.

Line 256: While the new method is a key novelty, its necessity and efficacy have not been thoroughly validated. A more rigorous comparison with the old method and a clear demonstration of its improvement are necessary.

Line 267: It would be great to include an analysis for assessing the effect of salinity changes.

Line 270: Again, my suggestion is to separate and quantify the impact of volume transport variability in different parts of the section on the overall property transport variability.

Line 272: Consider adding a discussion on how the presented results are related to the other transport estimates within the subpolar region.

Review of the manuscript egusphere-2025-3167.pdf

General comments

The manuscript, entitled “Eight years of continuous Rockall Trough transport observations using moorings and gliders”, presents new estimates of the European Slope Current (ESC) and North Atlantic Current (NAC) volume transports through the Rockall Trough (RT), adding two years to the existing time series. For the first time, heat and freshwater transports are also presented. The new data set includes temporal overlap of mooring and glider data from the ESC. The authors have developed a new methodology, where mooring and glider data are combined, allowing for better estimates of the ESC transports. The new methodology gives a better representation of the Eastern Wedge covering the ESC, with the undercurrent now visible in the data. Interestingly, they also find, that the variability of the ESC seems independent from the NAC variability. Over the eight years of observations, the total RT transport does not have a significant trend. The manuscript highlights the benefits of combining various data sets to get a better representation and understanding of physical processes in the ocean.

Generally, the manuscript is within the scope of OS and the language is fairly good. But the lack of consistency in e.g. figures and abbreviations gives the impression that the manuscript is written in a hurry and that the authors have not spent enough time on polishing the manuscript. Also, the discussion is relatively short and appears more like a summary of the results section, while the results are discussed already in that section.

Thus, major revisions are needed before publication in Ocean Science.

Specific comments

Abstract

Line 8-9: Here you say, that you have produced, for the first time, a continuous ESC transport time series, but as I understand, the volume transport is an update of Fraser et al., 2022. On the other hand, heat and freshwater transports for ESC and RT are presented for the first time. I suggest that you highlight that in the abstract.  

Introduction

Line 15-16: The listing of impacted regions could be rearranged in order of appearance downstream of RT, i.e. Arctic last.

Line 29: “together with”

Data and Methods

In your description, you use both “glider” and “gliders”. I suggest being consistent.

Line 59: Cross-hatched region is not visible on a print out. See also comments to Figure 1 below.

Line 61: Should it be “an SBE41”?

Line 91-95: See comments to Figure 1 below.

Line 107-108: This sentence gives the impression that no data exist from the RTADCP. Please modify.

Line 159: Please add, that Eq. 3-5 are given in section 3.4.

Line 185 and 189: How did you select the reference values? Please clarify in the text.

Results

I suggest that you add a table listing the various transport values given in the text and maybe also correlation coefficients. A table gives a better overview and it is easier for the reader to compare the different branches.

Line 200: Please specify Figure 5a.

Line 203: Please add Sv to the transport values.

Line 224: Replace “small” with “close to zero”

Line 225: “an increasing trend”? Are these trends significant?

Discussion

As mentioned above, the discussion is relatively short and is more like a summary of the Results section. Must be rewritten into a proper discussion section.

Line 265: “controlled by”

Figures and legends

Figure 1: This figure needs some updates. Firstly, the resolution in both panels is too low.

The map must be marked with an a). The green circles are not very visible and it is hard to see, that there actually are three of them. The speed on the map seems okay, but what are the arrows based on? I do not have comments on the swirls here and there, but the arrow located at ~12°W; ~64°N is flowing in the wrong direction! See e.g. Hansen et al., (2023, Fig 14) or Orvik and Niiler (2002, Fig 1 and 3b).

Hansen et al., 2023 (https://doi.org/10.5194/os-19-1225-2023); Orvik and Niiler 2002 (https://doi.org/10.1029/2002GL015002)

The hydrographic section in b) is from Frazer et al. 2022 (their Fig 1.b in a good resolution), but the resolution here is too low to see details in the figure. Houpert et al., 2020 have a similar figure (their Fig 2.a) and they nicely define the WW, MB and EB areas on top of the section. Together with the limit at depth, this would give a much better illustration of which area the three transport estimates are calculated for. As the figure is now, the cross-hatching is hard to see, especially on print, and the different symbols on the surface are not described. Please update the figure and legend accordingly.

Figure 2: Fine resolution. The names on the top are cruise id’s? Please clarify in the legend.

Figure 3: Please use “°W” for consistency with other figures.

Figure 4: Blue square marks the RTADCP – right? This information should also be added to Figure 1b.

Figure 5: The figure legend is insufficient. Reference to a) and b) is missing. The black dotted lines and the vertical bars (in b) are not described in the legend. Moving on to figures 6, 7 and 8, you here use the colors blue, orange and green for MB, EW and WW – but they are not consistent in all figures. Please select one color for each region and use them consistently. Please do not re-use these colors in Figure 5 (except for the EW color). What is the temporal resolution in a)?

Figure 6: Please repeat the description of the numbers in the panels.

Figure 7: Please add more details to the legend and include ref to all panels. The red lines (solid and dotted) in b, d and f are not very visible. Please modify.

Figure 8: Please be consistent in the use of colors and avoid to use orange for temperature in a) and blue for WB1/2 in b). You should probably also avoid to use red and green in the same figure.

2^nd line in legend should read “(b) isolating”.

Supplementary

The title is different from the manuscript title.