Consistent picture of the horizontal circulation of the Atlantic Ocean over three decades

Cainzos et al

This is a detailed study of inverse model calculations of the major current systems (upper, intermediate and deep) throughout the Atlantic basin using 3 decades of hydrographic data of latitudinal sections from 45S up to 58N. Extensive comparisons of the results against past literature studies is presented. The final conclusion is that, at least from these data, there is no evidence of statistically significant changes in the current systems between decades. A nice set of figures are presented with the major current transports identified by longitude where they cross each section along with an extensive table of the inverse model mass, heat and freshwater transports.  The paper which represents a substantial set of work and will be a useful future reference should be published

There are however several pieces of missing information from the paper, and also the text presentation needs to be improved in places. Main things that need further discussion  are;

• Details of when the actual sections used were taken. Currently we are simply told if section was taken in each decade. A table of the sections used with their woce/go-ship identifications and the dates could be used
• More explicit statements about how the uncertainties from the inverse solution currents are derived and what they represent. What has been assumed about uncertainties in the input data? The uncertainties in the output are presented as if these really represent the uncertainties in a decadal mean current. But key assumptions have been made in particular that the current systems are in steady state because they are only being sampled once? and these latitude samples are all at different times.  These are well known issues with steady inverse models but the caveats need to be clearly stated.  I suggest that an additional paragraph relevant to these issues can be added after L155 stating important caveats to interpreting results
• Heat and Freshwater transports are given for each separate current component. It should be stated how these are defined because these are not formally separable from the mass transport.
• There is virtually no proper discussion of water mass transformation between sections? Lines 397 are randomly introduced but a little more clear discussion would be useful. I appreciate that this is not the main subject of the paper but since the water masses and currents are being defined by their density classes and the transport strengths are changing as currents cross different latitudes it would be useful to be given some of this information where large transformations between classes are being derived by the inverse analysis. A standard way to do this would be to add some up/down arrows on figure 7,8 or 9 to show conversions between upper, intermediate and deep waters?  

The current schematic figures 7,8,9  are very useful but are not referenced in text where the current details are discussed. I suggest Figs 7,8,9 are moved up to be introduced as summary results near the start  and then references should be included to these figures in the text at the same time and the transport sections are discussed as the geographic referencing would greatly help understand the connections between sections.

More minor point by point issues are listed below;

Line 87 assumed  level of no motion.  This is not “known”

L95  It is not a “fact” that currents are close to a steady time mean. It is an “assumption”.   This assumption is a very large caveat on the uncertainty values

L122 Does the reference depth vary with longitude?

L175 Are quoted MC transports the average over 3 decades?

L244 The longitudinal extent and widths quoted for many currents (eg. BeC here) are not so easy to decide upon from the transport sections. Often the main currents seem to occupy much smaller widths.  This often leads to very different widths being identified for different decades.  A clearer statement about how these ranges are defined would be helpful. L360, L507, L564 other examples. In L633 for example there seem to be large + and – values which cancel out and a very small value is given in text over a large extent?

L259-262  The paragraph structure is very unclear. Some are very long some very short (as here) Some more thought about what should be in each paragraph is needed throughout to help readability

L337  750m is not very shallow??

L365-367 Notice the very large uncertainties given in other literature relative to tiny uncertainties you are giving. This is clear example of where the main point 2 above needs more careful discussion

  L416 low relative to what?

L464 The section figures always use 55N as the average latitude but the text always uses 58N. Perhaps be consistent?

L488-91 repetative of Brearley result

L512 Sometimes hugely different values are obtained in previous literature. These are cases where some suggested explanation would be useful?

L540 East of DWBC?

L544  Still worth marking where you have taken data from on section

L654 In summary be very clear what surface intermediate and deep mean

L680 “Errors” often used when you mean “uncertainties”

L686 recuperation => recovery

L690 upper => upward

L720  Care to comment on the importance or not of the FW transport into the Atlantic from the south vis a vis AMOC stability?

Are you inverse solution transports available along sections as available datasets?? Along with density transformations presumably?

There are small grammatical/english issues through the manuscript which could be corrected by a more careful reading

Review of “Consistent picture of the horizontal circulation of the Atlantic Ocean over three decades” by Cainzos et al.

The authors use an inverse model applied to hydrographic sections from 45°S to 60°N in the Atlantic and repeated over 3 decades to estimate the circulation and the volume, heat and freshwater transports associated with the main currents. In addition to quantifying these quantities from North to South over three different decades, a major conclusion of the study is that the circulation has not changed significantly over the three decades. The analysis required considerable effort. It is relevant and has clear figures and, in general, adequate reference is made to previous work.

I do, however, have a number of comments that need to be adequately addressed by the authors before the manuscript can be considered for publication.

*** Major Comments ***

Comparison of the results of the inverse model with previously published studies shows that the inverse model gives satisfactory results in most regions except at subpolar latitudes where the circulation is more barotropic. In the North Atlantic subpolar gyre, the transports of the western boundary currents are strongly underestimated by the inverse model. Along 45°S, the Zapiola anticyclone is not resolved. This is a well-known defect of this type of inverse model based on an a priori state with a zero velocity reference level selected between two water masses and which fails in barotropic current regions where the zero velocity reference level is a poor approximation of reality. This defect can only be remedied by the use of direct current measurements. The problem is well described in Alvarez et al (Journal of Marine Research, 2002). In the subpolar gyre, the transport estimates subsequently produced by Lherminier et al. (Journal of Geophysical Research 2007 and Deep Sea Research, 2010) or Holliday et al. (Journal of Geophysical Research 2018) have all combined direct current measurements and hydrographic data. This is also the case for the transport estimate at 45°S proposed by Saunders & Bacon (Journal of Physical Oceanography, 1995). The consequence is that the circulation at the northern and southern boundaries of the inverse model is incorrect, with artefacts related to the poor estimation of the transport of barotropic currents (e.g. a southward transport of DSOW along the western side of the Reykjanes Ridge). The authors need to remedy this problem. There is also the question of the impact of this problem on the estimation of transport at other latitudes.

The results of the inverse model are interpreted as being representative of a decade. While this assumption may be acceptable for integrated coast-to-coast quantities for which the mesoscale contributions may have been averaged, I do not believe that this can be the case for the estimation of quantities associated with currents of limited regional extent. The authors are aware of this and acknowledge for instance that their results may be biased by the seasonal cycle, as for their discussion of the Canary Current transport. Additionally, the inverse model is forced by an Ekman transport calculated for the time of the cruise, which seems to contradict a search for a solution representative for a decade. Shouldn't a decade-averaged Ekman transport have been used? The authors need to find a way to present their results without interpreting the variability of individual currents as decadal variability (or spend more time convincing the reader that their solution has been derived in such a way that daily to interannual variability has been filtered from their results). In this context, the significance of the errors on te transports determined by the inverse model should be clarified. Should they be interpreted as an estimate of sub-decadal variability and if so why?

A major point of discussion in the manuscript is the transport of the different currents. The way in which the geographical boundaries (horizontal and vertical) of the different currents are defined is not made explicit, although this is a crucial point for the interpretation of the results. The reader needs to know whether from one latitude to another the geographical boundaries of the currents have been chosen in a relevant way to allow a meaningful comparison. Should recirculations be included in the transport estimate of a current (this would seem to be the best option to allow a comparison of transport across latitudes without a bias being introduced by the existence or not of recirculations)? In some places, the choice of the geographical limits of the currents goes against the definitions usually chosen in the literature, as is the case for the North Atlantic Current (NAC) for which the definition chosen in this manuscript only includes a small part of the NAC compared to the choice made by Holliday et al. (2018) for example, making it difficult to compare it with 48°N where the whole NAC is taken into account. The authors should explain the criteria used to define the geographic boundaries of the different currents and how they ensure the meridional consistency of the different transport estimates.

The summary figures presented at the end of the manuscript include only part of the meridional transports through the different hydrological sections. The diagram presented in Figure 7 do not conserve volume. For example, summing all the volume transports through the OSNAP-E section, the residual volume transport is -12.8 Sv, which is greater than any of the individual contributions. Major circulation elements are missing! This should be corrected so that an approximate volume balance be achieved in the circulation sketch of Figure 7. This also poses a problem when interpreting the freshwater and heat fluxes of the individual currents, which must be interpreted at zero volume transport.

The manuscript presents weighted temperature and salinity transports for each of the currents. No conclusions are drawn from these values. It would seem that if the manuscript were to include an analysis in terms of water mass it would be more appropriate for it to be used to determine the geographical limits of the currents.

*** more minor comments ***

Lines 16 to 18 : Referring to Figure 2, I find a net upper ocean transport through 45°S of 2 Sv for the upper ocean. Noting that 61 Sv enters the South Atlantic Ocean in the upper layer, without noting this close to zero net (coast-to-coast) transport is misleading as it neglects outflowing waters. This illustrate the problem of solely considering the transports of individual currents without placing them in a wider context. The same remark can be made for the 55 Sv exported southwards through 45°S. What is the significance of this number given that the net coast-to-coast transport is much lower (Figures 5 and 6)?

Line 102 : what about heat transport ?

Line 109 : I’m not sure what you mean by “geographical structure”.

Line 118 : 9 should be )

Lines 130-132 : I recommend computing the Ekman transport for each pair of stations so that each Ekman transport value is ascribed to the local density (and not a section averaged density what can create bias when densities are outcropping along the section).

Line 145 : In your manuscript on the anthropogenic carbon, you mention conservation of biogeochemical properties and here only volume and salinity. This is confusing. Are you using a different inverse solution. If yes this is not reasonable and I strongly recommend that you present here an inverse model solution that is consistent with previous publications.

Lines 149-151 : This is a repeat.

Equations (4) : Some notations are not defined. What is the value of the reference salinity ?

Lines 174-175 and 185 : These numbers are also compatible with those of Maamaatualahutapu et al. (Journal of Marine Research, 1998).

Lines 200-201 : The recirculation immediately east of the Brazil Current is of the same amplitude as the Brazil Current itself. How should this structure be interpreted?  Could this be an eddy? Which quantity is more suitable for comparison at another latitude, the transport of the southern branch or the sum of the transport of the southern and northern branches?  This point deserves discussion and the choice made needs to be justified.

Lines 216-217 : the observed variability, determined from snapshots, cannot be attributed with certainty to decadal variability.

Line 218 : This is a pattern consistent … a reference is needed to support the statement.

Line 236 : “appear to decrease among the first decades” … I think that  you do not have the data set to ascribe the variability of regional currents to the decadal time scale. I stop at this point to comment in detail on the subject, but the whole manuscript needs to be revised to take this point into account.

Line 244 : You should explain why your western BeC boundary (2.9E) is different from that chosen by Saunders & King (1995) (10.5E) and why this boundary is a better choice.

Lines 272-274 : Could you indicate how, in your circulation scheme, the SEC at 30°S connects to the BeC at 45°S?

Line 313 : How do Piecuch's results (2020) compare on a ten-year average with your results?

Line 349 : it is more than twice as large and not ‘slightly’ higher.

Lines 364-355: Above you mention that your transports have a better agreement with the fall estimates found in the literature and now it seems that it is with the spring ones. Could you clarify this point?

Line 375 : The North Atlantic Current is a current that extends from the surface to the bottom. It should be made clear that you are only studying the upper part of it (and justify this).

Line 385 : In Meinen and Watts (2000), the North Atlantic Current extends down to the bottom, this might explain the difference with your estimate, which is limited to the upper layer.

Line 394 : Please justify why you are using limits for the NAC that are so different from those used in OSNAP (Holliday et al. 2018) ?

Line 402 : Houpert et al (2018) only considered the Rockall Trough branch of the NAC.

Lines 408-413 : You may be interested in the article by Petit et al (Journal of Geophysical Research, 2019) which provides an updated and slightly different view of the Irminger Current and East Reykjanes Ridge Current.

Line 416 : The low values for the ERRC transport can be explained by the inability of your inverse model to reproduce the circulation in a barotropic current such as the ERRC.

Lines 434-436 :  See above statement that also applies to the EGC transport estimation.

Line 471 : Garcia-Ibanez et al. are discussing contributions from source water masses. This cannot be directly compared to your definition of ISOW which includes all mixing components.

Line 464-475 : ISOW transport is clearly overestimated. This could be an artefact due to the underestimation of southward transport in the EGC and ERRC.  The model compensates by increasing the southward transport in regions where the geostrophic current shear is larger.

Line 478-479 : Holliday et al. (2018) find 3Sv northward for the deep current transport in the eastern part of the Irminger Sea. Your results contradict both previous transport estimates derived with the same data and the water mass properties that indicate LSW and ISOW at this location.

Line 481 : DSOW should have transport weighted temperature lower than that of ISOW. Note that the temperature range you are mentioning for DSOW is that of LSW. LSW is missing from your analyses even though it is  a major component of the DWBC.

Line 531 : Your upper limit is too deep. You are missing part of the LSW, which is a major component of the DWBC.

Line 540 : I’m confused, do you mean “East of the DWBC” ? Recirculation is only evident for 2000-2009.

Line 551 : Schott et al. (2005) used both hydrography and ADCP velocity measurements.

Line 622 : It might be useful to mention the blocking of northward flowing AABW by the Walvis Ridge in the eastern basin.

Line 659 : I thought the SEC was mainly fed by the South Atlantic Current.

Line 673 : Diapycnal upwelling between layers could be added to Figure 7. I note that there is overlap in density between the upper and deep layers. This is for instance the case at 45°S with the upper layer extending from the surface to 27.84, and the deep layer extending from 27.58 to 28.10.  Are the transports in the overlapping densities counted twice ? Could you please clarify this point ? My advice would be to avoid overlaps in density between layers.