We would like to thank reviewer 1, Hjálmar Hátún, for the detailed comments on this work as provided in “https://doi.org/10.5194/os-2021-60/RC1”. We will now respond to each comment in turn, with Hátún’s original comments presented in italics. 

Firstly, we consider Hátún’s comments on sea surface height (SSH):

    The discussion on eastward flows in the North Atlantic Current has a long history (which the present study does not appear to acknowledge). Already in 2002, Pingree (2002) linked this to the meridional gradient in sea surface height (SSH), as revealed by satellite altimetry. He discussed increased SSH gradient/transport during NAO+ years (e.g. 1994-1995 and 1999-2000). He (and a large volume of following literature) has associated such interannual changes to the NAO index and variability in the wind stress curl field over the NE Atlantic.

     After this, SSH data (observed and simulated) have been utilized by myself and many others, to discuss this dynamics in a broad and longer context (Hátún and Chafik, 2018, and reference therein). For example is the calculation of the so-called gyre index closely linked to variability mentioned by Pingree (2002), and likely to your eastward transport records based on hydrography. Your analysis on these transports is interesting, and does provide new information/knowledge. Just try to better weave it into the existing volume of knowledge. This includes paying more attention to the interannual fluctuations that you present (Figure 6 and 7). Your study is clearly motivated by identifying drivers behind ecosystem fluctuations along the European continental slope (ECS). We have previously linked these pulses to many ecological aspects in the NE Atlantic (e.g. Hátún et al., 2017, 2016; Jacobsen et al., 2019), and a growing body of evidence shows that this type of variability does also characterize the ECS (Pätsch et al., 2020). You have the evidence, utilize it better.

Hátún is correct in saying we have overlooked SSH gradients as a driver for eastward flow in the North Atlantic. We have also not related our results to the North Atlantic Oscillation (NAO) index, which in hindsight would have been appropriate. We will incorporate SSH studies into the introduction and background materials, as well as attempting to relate our main findings to SSH changes (accessible via GODAS ‘sshg’ files). We will relate our findings back to the NAO index as well. 

Our study is indeed motivated by identifying drivers behind ecosystem fluctuations along the slope and shelf seas. We thank Hátún for drawing our attention to the studies referenced above. We propose to relate our observed hydrographic changes and calculated particle trajectories to the gyre index: this may prove to be a valuable link. We will also introduce a new SSH decadal mean anomaly figure in the style of Figs 2-4, for direct comparison to observations already in the literature. Additionally, we’ll include an estimation of monthly eastward barotropic (SSH-related) transport across the same region as our GODAS estimates (30 °W, 45 – 60 °N), for the time period 1980 – 2019. 

We note the fact that the NAO and poleward flow is out of phase:

    Garcia-Soto et al. (2002) and Pingree (2002) discuss the conditions along the ECS in relation to the relatively narrow slope currents from the south (Bay of Biscay). This topic should be better handled in your work. For example does Pingree (2002) claim that the North Atlantic Current strength and the mentioned poleward flow are out of phase. Weak NAC (aka NAO-) is related to stronger flow of warm and saline waters from Spanish waters – also referred to as Navidad events/years (Garcia-soto, 2002). This seems to be at odd with your perspective (although I follow your argument that NAC waters are being continuously recruited to the slope, north of the Porcupine Bank). This aspect must be better handled.

We agree that the NAO needs to be better handled within our manuscript. This will be done through a more thorough discussion of our results in this context, relating back to the papers highlighted above as well as additional relevant material. 

The more specific suggested figure alterations are, for the most part, wholly appropriate and justified comments. We will comment on each figure (or figure grouping) individually:

    I will below suggest some more specific changes to your work. If you are will to roughly follow this path, I can provide a more detailed review during the next round.

    Fig. 1 does nice illustrate the entrainment of water to the boundary north of Ireland, and no northward bound boundary current south of the Porcupine Bank. Your particle tracking figures, however, suggest near-slope patterns further south. Would velocity quiver maps on a shallower level maybe reveal any influx from the Bay of Biscay?

We are cautious about providing quiver maps at shallower levels. The literature suggests the core of the Slope Current to be approximately 200 – 300m deep (Porter et al, 2018), hence our chosen GODAS depth level of 245m. This will be justified more strongly within the manuscript. The message we were trying to convey was the seasonal and interannual variability of the Slope Current. We are happy to test shallower depth levels and reassess the best level(s) to display here.

    Figs. 2-4: You can state the association between T and S, and the tight linkage between T and SSH in the text, and only show the density field (Fig. 4). The T-S-density relations are well known between oceanographers, and the T and S figures are not strictly needed. And as suggested below, provide a better figure, which includes a relevant geographic domain, and averages over relevant periods.

Removing Figs. 2 (S) and 3 (T) is a sensible suggestion. We will improve the description of the T and S relationship in the text after removing them, particularly in the discussion. An overview of the entire study area can be provided in an updated version of Fig. 1.

    I would also include altimetry data here. It would (i) validate the chosen in situ hydrography data product, (ii) produce and independent east ward transport record, and (iii) enable you to put your analysis in much better context – and link to the existing literature.

SSH data, from GODAS, will be plotted in the style of Figs 2-4. Please see the previous comments on SSH. We will then link any observations back to the existing literature throughout the discussion.

    I would only show the GODAS-based Hovmöller diagram in Fig. 5. You say that there is mutual agreement between the GODAS-based and the EN4-based. I think there are large differences between them (although basic major feature are comparable). You also describe some limitations with the EN4 dataset (pages 10 and 11). And my impression of the hydrographic signal at ~60°N (which is based on many years of experience and many data sources), is that the GODAS product probably is more reliable for you purpose. Suggestion, skip EN4. It would enable you to produce a clearer figure, and convey a clearer message.

EN4 will be removed in this figure and other figures for better clarity of our findings. Whist both GODAS and EN4 have their limitations, we agree that GODAS is the more appropriate and reliable dataset.

    I would merge Figures 6 and 7 into one two-panel figure with the GODAS-based time series. It is reassuring that the EN4-based series show similar variability, and this could be mentioned with words/correlations.

Current Figs. 6 and 7 will be merged to form a 2-panel plot of GODAS-derived eastward volume transport: a- geostrophic, b- total. The EN4 transports will still be mentioned within our discussion of results.

    It is good to see the transport change in T-S space (Fig. 8). You could, however, zoom in on a narrower TS window, which would enable a better/clearer figure. The TS-transport figure based on ORCA12 (Fig. 9) is actually very different from the GODAS-based figure (Fig. 8). GODAS shows a major decrease around 5-6°C, 35.0 (which must be close to Subpolar Mode Water), which is not reproduce by the ORCA model. Suggestion: Stick to GODAS – skip Fig. 9.

We will remove ORCA12 from this figure and amend the TS window to reflect that: T in range 5 – 18 °C, S in range 34 – 36 PSU. We will still refer back to the differences between GODAS and ORCA12 within the discussion, to create context with our necessary use of the ORCA12 simulation for calculating the particle trajectories that are central to the variable pathway/provenance analysis in Sect. 3.3.

    Fig. 10. Yes the transports are much larger at the northern section (admixture of NAC-derived water, right?), and there is a somewhat worrisome decline in this transport (in line 255, you mention an almost-steady northward transport of 2 Sv after 1995, while I see a continuing decline, also after 1995). While I guess that you already have tested this thoroughly, are you still confident that you capture the entire slope current, with this model extraction? If yes, which current is then presently feeding the Faroe-Shetland Channel?

Our current Fig. 10 will be repurposed to better reflect the changes in northward transport along the shelf edge. We now propose to present 4 northward transport time series: at 51, 54, 56, 58 °N, or alternatively a Hovmuller plot showing northward depth integrated transports over time for the shelf edge 50 – 58 °N. We will test these and decide on a best course of action. Replotting will better reflect the release locations for the particle tracking experiments later in the manuscript. The calculations will be done in the CDFTools package, proving a more accurate estimation of volume transport through the ORCA12 meshgrid. 

    The results from you particle tracking analyses are nice. Keep.

 

    I hope that I have been too frank. I really hope that your work will become a part of the literature, because it is needed. If you (and the editor) think that my suggestions are too drastic, I would be ok with being revoked as an editor. Otherwise, I look forward to read a reworked version J

    Figures (I wrote this, before the text above. You can maybe use it as guidance)

Many of Hátún’s additional comments on the figures are formatting improvements to better utilise the plotting space, or to avoid repetition in the figures and captions. For instance:

    Figure 1: Remove the header “Velocity quiver at 245 m” from each panel, and this common information in the caption. Remove the y-axis information on the right panels, and the x-axis information/labels in the upper panels, enlarge each panel, which removes too much void space between them.

We fully agree with this recommendation, and this change will be implemented in full for the reworked version of the manuscript.

There are a lot of additional comments about figures 2-4: 

    Figures 2-4: Remove “ S/T/density decadal mean anomaly, and “205 m” from each title. This information is already in the caption.

The figure titles will be amended and simplified, according to this suggestion.

    You lose out in insight and smear out valuable signal, by strictly averaging over these decades. For example, the 1990s was is contrasting period, with dense waters until the mid-1990s, and much warmer/lighter waters after. The average over these contrasting states is not so meaningful. I am aware of our wish to stay objective, but you have explainable reasons for selecting contrasting periods (e.g. early 1990s and early 2000s), in order to portray spatial hydrographic structures over the North Atlantic. Also pay attention to the (short term) interannual signal (mentioned above).

We agree with this point. We will replot Figs 2-4, with means pre 1997 and post 1997, which will bring them in line with the T-S plots shown in Figs 8 and 9. This will enable a better comparison between the “warm” and “cold” periods/states as observed in previous literature and indeed our own results. SSH will also be plotted here (see previous comments on SSH) in the same style.

    Add some selected isobaths (e.g. 1000m, 2000 m and 3000 m) to these plots. In order to discuss these patterns against previously published key patterns in the North Atlantic (e.g. the spatial sea surface height mode, which is associated with the gyre index), you might what to include a broader meridional window (e.g. 35-65°N, although you only calculate transports over the 45-60°N latitudinal range). Maybe use a bit narrower color ranges, in order to emphasize the obtained patterns.

Isobaths will be added to current figures 2-4 (including the replacement SSH plot). We will select narrower colour bar ranges as suggested. We will plot for 35 – 65 °N, and annotate the figures with the profile used in our calculation of transport at 30 °W, 45-60°N.

Previously in this response, we agreed to limit the plots to just one showing the density anomaly patterns. 

    Figure 5: Maybe use a bit narrower color ranges, in order to emphasize the obtained patterns.

    The obtained Hovmöller diagrams based on GODAS and EN4 are actually rather dissimilar (mentioned above).

A narrower colour range will be trialled, most likely in the range ±0.2 kg m-3.

More minor figure amendments were recommended:

    Figure 11-13

    Just keep the y-axis information on the left panels, and the x-axis info on the upper panels. Enlarge the panels, which would remove the excessive white spaces in these figures.

    Figure 6:

Remove “Geostrophic eastward transport”, from the titles. This info is provided in the caption.

    Figure 7:

Remove “total eastward transport”, from the titles. This info is provided in the caption.

We agree with these recommendations and they will be actioned in full for the amended manuscript.

We once again thank Hatun for his detailed comments. We look forward to presenting a revised manuscript and hope that the amendments and additions satisfy your feedback.

Kind regards, 

Matt Clark,

Lead Author

on behalf of all authors. 

We would like to thank anonymous Reviewer 2, for their comments on this work as provided in “https://doi.org/10.5194/os-2021-60/RC2”. We will now respond to each comment in turn, with Reviewer 2’s original comments presented in italics. For ease of reading, we have retained the headings (in bold) used by the reviewer for clarity.

The data products used

    The ESC region is one of the best observed regions in the sub-polar North Atlantic. While data products such as EN4 and GODAS provide a full 4D overview of the region of interest.

They are also often not great. The authors acknowledge this to some degree, although I find the statements on this quite confusing. In Section 2.1, the authors highlight that GODAS salinity is mostly “synthetic” and “seriously under estimates salinity variability”, but in the discussion in Section 4.1, the EN4 lack of salinity data and gridding methods is flagged as a potential issue. I also find the assumption in lines 107-111 requires further evidence that it is appropriate. 

Line 114-115: The two data products are stated to be independent of each other, but I doubt this is truly the case (e.g. if both incorporate Argo profiles). Particularly the following sentence highlights that these are likely the same four sources (please state here which ones also!).

We acknowledge that, at times, our explanation of the limitations of both the GODAS and EN4 datasets could be stronger and/or clearer. Combining Reviewer 2’s comments with Reviewer 1 (Hátún), we will be removing EN4 data from the figures, but will continue to provide a text-based description of the merits and downsides of EN4. We will provide further justification (line 107 onwards) on why GODAS is an appropriate dataset to use in this region; for the purposes of assessing the hydrography of the North Atlantic basin and determining geostrophic volume transport estimates towards the shelf edge. Whilst GODAS and EN4 do indeed incorporate Argo profiles and other similar data (to be stated as per the recommendation), the ways in which the datasets are assimilated and gridded differ considerably. Clearly this needs further and clearer explanation in the text, backed up with the relevant references. 

The lack of consideration for the forcing mechanisms

    The paper is highly descriptive of what is going on, but lacks to place this into the context of the forcing mechanisms. For example, there is no consideration for the positioning of the sub-polar front in the North Atlantic, there is also no consideration for the wind-forcing of the circulation of the wider SPNA and how this influences “recruitment” into the ESC or otherwise. Especially given the discussion on zonal current variability, I find these quite major omissions in the analysis. The discussion is more a continued description of the results presented, rather than any contextualisation in terms of previous work and/or forcing mechanisms. Section 4.2 is more speculative on implications, and a repeat of what has already been stated in the introductions.

We agree that the discussion needs to better compare our results with previous published works and the likely forcing mechanisms. This should include analysis of the sub-polar front, NAO index and surface winds. Specifically, we will discuss and further emphasize:

•    how changing wind stress curl at basin scale forces changes in the strength and configuration of the subpolar gyre, of consequence for barotropic inflow to the Slope Current

•    how changing wind stress in the vicinity of the European shelf break forces changes in the Slope Current via Ekman transport towards the shelf break 

•    how changing buoyancy forcing at basin scale forces changes in baroclinic inflow to the Slope Current

The reductionist statements on salinity and lack of error estimates

    The authors rely on the data products to provide accurate baroclinic transport, but based on potentially erroneous salinity data. There is little quantification of salinity error or overall error analysis, it is therefore difficult to know whether this really is of no significance to the results presented.

We will further emphasize that density variability is dominated by temperature variability in the subpolar North Atlantic. Repeating the thermal wind analysis with climatological salinity, we can evidence that uncertainty in the GODAS salinity data does not substantially affect our results or conclusions.

Lack of a general figure with key circulation features and locations

    The paper lacks visual cues of the lines/boxes etc used, as well as a figure that highlights of the focus area of the study sits within the Sub-Polar North Atlantic (SPNA). Even for someone with expertise in the region, it is difficult at times to follow which transect has been used or across which box particles have been quantified. None of the figures show the “analysis region” (line 165) in full, for example.

We will annotate figures accordingly (for example: our shelf edge transects can be added to our amended Fig. 1) and enhance the text appropriately, to clarify the choice and use of selected transects and sub-regions, throughout.

Further Comments (some, but not all, minor)

    Line 34—36: Johnson and colleagues find that the changes in the water mass properties and nutrients concentrations at the Extended Ellett Line are related to changes in properties of the circulation. To my mind, this is not the same as changes in concentrations in upstream flows, as the authors state.

We are familiar with this previous finding, and we will clarify that changes in property transport, consequential for the shelf edge and shelf sea of northwest Europe, are the combined consequence of changes in properties and flows (as everywhere!); we will emphasize that properties and flows may be correlated, as evidenced in our analysis by a weaker, warmer Slope Current in the early 2000s, while acknowledging that property changes may be the dominant factor at locations where volume transport is relatively steady (such as EEL).

    Line 46-47: Suggest rewrite for clarity “The Gulf Stream flows between these two and eventually …”

    Line 51: “However, not all of the water follows this pathway.” (missing “of”).

The minor revisions on lines 46-47 and 51 as recommended above will be implemented in full to improve clarity and readability.

    Line 52-56: This description neglects some of the other exchanges in the northern North

Sea, particularly the Norwegian Trench inflow. The authors have spent great length emphasising the importance of the ESC to the marine ecosystem of the continental shelf, so a correct description here is warranted.

We will endeavour to cover the Norwegian Trench inflow in our introduction, referring back to relevant literature.

    Line 67-69: This reduction in temperature was also accompanied with some very strong reductions in salinity. The region of the ESC was at its freshest for more than 120 years.

Please see Holliday et al., 2020.

Thanks to the reviewer for highlighting this paper. We will highlight declining salinity in the ESC and wider SPG region.

    Line 131-132: This is solely the N-S directed component of the ESC volume transport.

There is no further indication of this assumption in the paper.

We are not sure what the reviewer is referring to here. Equation 1 (on line 129) is simply the eastward velocity change over depth, from the thermal wind relationship. The equation was used to calculate a geostrophic velocity, which is then integrated over depth to obtain the eastward volume transport; therefore indicating transport flowing towards the shelf edge, as a measure of the inflow towards the shelf break, where geostrophic flow must turn to the north. It isn’t until Fig. 10 where we introduce the concept of northward-directed ESC volume transport. Captions for Figs 6 – 10 clearly state the direction of transport being plotted, as well as the location of the transport sections (whether geostrophic or total volume transport).

    Line 133: Has no reference velocity or assumption of Level of No Motion been applied?

Further in the paper, there is discussion of the baroclinic and barotropic components of transport. I think this could be clarified here in the methodology.

    Line 135: Why not state x 10-6. The e-notation seems a relic of the coding.

Equation 2 (on line 135) states that the calculation is performed with -1000m as our reference/no motion level. This can be made clearer in the accompanying text. The “e” notation is from the coding and it is sensible to replace this with “10 to power” notation.

    Line 150-167: I found the description of the particle tracking methodology could be better: it is very detailed about some things (e.g. reference the initial positions file), but lacked details on other. Are particles released from all grid cells? Or only grid cells following the continental shelf edge? Later in the paper there is also mention of particles crossing certain transects. I would recommend some major rewrite of this section to ensure transparency and repeatability.

As we outline (lines 155-159), ‘Particles were released proportional to the northwards transport at the shelf edge, between 50 –  60 °N; as such the exact number of particles released varied with each experiment. Releases were designed to target the “core” of the Slope Current:  cells were defined as an active release location if there was northward transport present and that the bathymetry was between 200 – 250 m, taking care to exclude other shallow areas such as the Rockall Bank.’ More specifically, particles are allocated at all grid cells within this definition, on sub-grids in proportion to local speed. By our definition, these grid cells do follow the shelf edge. Regarding the ‘census’ of particles crossing the 30W transect, this can be included in Figs. 11-13, and we will more clearly explain how we define ‘crossing’, in order to obtain the histograms in Fig. 14-15.

    Line 204-205: This is quite a narrow temperature range. Please consider justification or broadening.

Within the text, we have already justified this temperature range in line 204 – “based on the mean decadal temperature anomalies presented in Figure 3”. This will be improved with further justification.

    Line 209-211: Is this difference in transports an artefact of the referencing or the fact salinity is poorly constrained in the reanalysis?

In response to Reviewer 1, we are removing the ORCA12 time series, so in the revised manuscript we will not address these differences.

    Line 231-247 (and probably throughout): Practical Salinity is a unitless quantity. It should be used as such. Therefore text should say “Practical salinity in the range 34.25-36 …).

That being said, oceanographers agreed to adopt the TEOS-10 convention, and the authors already use the Gibbs Python functions, so Absolute Salinity should be used.

We confirm that salinity as provided by GODAS is in units kg/kg, akin to absolute salinity (units g/kg); we will clarify this in the revised manuscript.

    Line 250-252: Recommend to plot these transect locations on a map. If not on a general overview map, at least on one of the pre-ceding figures.

This is a good idea. We will adapt Fig. 1 to show the profiles used for calculating northward transport at the shelf edge. Likewise, for the eastward transports at 30 °W, we will add the profiles to Figs. 2-4.

    Line 250-288 (Section 3.3): This section is very descriptive, but lacks interpretation (here or later in the discussion) on how this relates to what is already known of the region’s circulation. I was unclear what the authors consider the novel finding from this analysis. 

We will revisit this text to provide more quantitative perspective, and to emphasize our key finding that the inflows feeding the Slope Current are systematically different when we consider years of ‘cold/strong’ and ‘warm/weak’ Slope Current transport. In the latter years, we find that a shift towards reduced inflows from more subtropical latitudes. We consider this to be a novel finding, via a Lagrangian methodology that is more intuitive than observations of transport time series at separate (e.g., upstream, downstream) locations.

    Figure 1: A continuous colour bar is not helpful to the reader. I would suggest using fewer, more discrete intervals in the colour scheme. The quiver is also quite difficult to see and may need to be scaled up. Which months do the authors consider winter/summer? [I note the colour bar does improve in future figures]

    Figure 2/3: The decadal distinctions are a human reflection of the calendar, rather than a reflection of the physical ocean climate. The “warm/cold phases” are not specifically associated with the changes from the 90s to the 00s – for example with major changes happening mid-1990s. The authors should consider using a more objective way of combing years into more meaningful “warm/cold” or “strong/weak” composites.

Figs 2-4 will be revised to show the shift between the warm and cool periods of pre and post 1997 respectively. This will then better align with the T-S figures presented in Figs 8-9, and also will better link back to the observed temperature shift in the previous literature and our own study.

Once again, we thank the reviewer for their comments and suggestions.

Kind regards, 

Matt Clark,

Lead Author

on behalf of all authors.