Review of submission os-2021-56

Title:

Refined estimates of water transport through the Aland Sea, Baltic Sea

Authors:

A. Westerlund, E. Miettunen, L. Tuomi, P. Alenius

General:

The study by Westerlund et al. investigates main pathways of water transport through the Aland Sea by means of a high-resolution regional model application. Previous observational and model approaches provided limited understanding of the transport and circulation structure in that area because of the strong seasonality of the regional atmospheric forcing as well as the complex bottom topography which requires high-resolution data coverage both in space and time to adequately capture the main transport characteristics. The aim of the study, as I understand it, is to gain insight into exchange dynamics between the Baltic Proper and the Bothnian Sea at interannual and seasonal scales of the recent past. To this end, the authors constructed a bathymetric representation of the Aland Sea with an unprecedented resolution of 500m horizontally and 200 vertical layers. The applied hindcast simulation with NEMO provided hourly to daily model output for the period 2013-2017.

While the model setup has been evaluated against available station data and seems to perform sufficiently well, my main concern is related to the rather limited use of the comprehensive model output. Substantial parts of the ms are dedicated to the discussion of the model biases and their possible origins. The transport dynamics as being the main focus of the study, by contrast, are presented in a rather descriptive way whithout analyzing and discussing any driving mechanisms or broader context that would finally gain our process understanding of the Aland Sea circulation. Apart from the shown figures, the transport rates are not even quanitfied. In this way, the only new finding of the study seems to be that about 25% of the transport entering the Aland Sea from the south does not follow the main strait at 20.5°E but rather happens through a topographic depression at 19.6°E. Of course, it is valuable to reflect on the model biases. But if the focus on the analysis and discussion of the model results is underrepresented, it conveys a rather defensive and repetitive flavor.

Nevertheless, I do see great potential to use the performed simulation for further analysis that, in my opinion, would substantially increase the depth, relevance and impact of this study. For instance, questions that naturally arise while reading the present version of the ms and could well be addressed, are: What drives the occasional northward turn of the surface flow? Is it exceptional wind conditions? Does the northward surface flow lead to SSH and pressure anomalies in the Aland and Bothnian Sea? Do these anomalies temporally weaken the more steady sub-halocline northward flow? What drives the sub-halocline gyre in the Aland Sea Proper? Why is the surface circulation strongest in summer? Is this related to melt water discharge from land? What are the transport budgets of the individual basins of the Aland Sea? Is the sub-halocline northward transport a continuous steady flow or is part of the inflow returned and exported southward horizontally or temporally or via mixing with the surface flow? Are there regionalized future projections for wind and freshwater discharge available that could be utilized to hypothesize on potential climate change impacts (e.g. those used by Meier et al. 2021, https://doi.org/10.1038/s43247-021-00115-9)? Could one also hypothesize/extrapolate from your results on the water mass exchange to the nutrient supply into the Bothnian Sea?

The conclusions could then follow a more explanatory line, if supported by the model results, such as: Continuous northward transport into the Bothnian Sea is dominated almost entirely by sub-halocline water masses. Northward flow potentially occurs in any area where the bathymetry exceeds the local depth of the halocline. If these areas are wide enough (with respect to the deformation radius?) the Coriolis force aligns the northward flow to the eastern side of the passages, which leads to a shoaling of the halocline at the eastern side and a deepening at the western side. Wind conditions can drive the predominantly southward surface flow of the estuarine-type general circulation towards the north on monthly to seasonal scales, causing significant anomalies in SSH and halocline structure. Further conclusions could be drawn from addressing some of the questions given above.

I therefore would like to encourage the authors to dive deeper into the subject, provide more detailed analysis of the simulation already available, and from this derive more comprehensive and thoughtful conclusions.

Some minor specific comments:

L21: What are these changes in the eutrophication status? Would be helpful information to better understand the context of the study.

Fig.1: Would be helpful to have an additional (small) inlet that shows the entire Baltic Sea and marks the location of the Aland Sea.

L24-35: This paragraph would be better structured if it was split between the Aland Sea and the Archipelago Sea.

L40: Maybe extend the last sentence by: " ... as the Archipelago Sea is too shallow to establish significant sub-halocline fluxes."

L46/47: It is stated that there are 'no clearly defined water masses' in the Aland Sea but in the next sentence, 'the existence of a deep water type' is mentioned. Isn´t this contradictory?

L55: The unit [g kg-1] is used to refer to a change in salintiy. The cited study by Palosuo, however, dates back to the year 1964, where probably [psu] was used.

L62-66: This paragraph can be condensed to: "... More recently, numerical modelling allows us to investigate intra- and inter-annual variability with much richer detail than we could with observations alone."

L82: "... in this topographically complex and irregular area."

L83: "... information about the bottom topography and related dynamics of different exchange pathways."

L88: "... represents realistic bathymetric features in the investigated area."

L89/90: The aim of the study is not well outlined. I assume, the aim is at least to provide a detailled understanding of the present-day water mass exchange dynamics in the Aland Sea. And maybe some more such as to draw conclusions on the nutrient fluxes in the area or generally to provide information for the development of a science-based marine management strategy?

L96: You may refer to the flushing times of the Aland Sea to justify the comparatively short spinup time of 6-7 months.

L110: Why does the use of a sea ice model with thermodynamic formulation reduce computational demands?

L110-114: Would you expect that the water mass exchange is different during years/winters with large ice cover?

L116-119: The calculation of volume transports does not have to be explained. These sentences could be condensed to: "We analyze volume transports across several transects to investigate the pathways of water exchange more closely."

L128-133: Suggest to condense this paragraph to: "To mitigate artificial interpolation issues we checked and edited ... to ensure that it accurately represents the coastline and depth variations in the 0.25 NM resolution model domain."

L134-140: Suggest to delte this paragraph as it does not contain important information.

L141: Why is it advantagous or necessary to smooth the steepest bathymetry gradients?

L182: What are these processes? Would be helpful to name a few examples: "... other processes such as ..."

Fig.3: Have you compared the station data also with the ocean reanalysis product you use to drive the model? As you are mentioning in L457, might be that the S biases are related to the boundary conditions.

L209ff: Suggest to delete the first sentence and start with "We evaluate modelled current magnitudes and ...". Also suggest to delete L211-213 (While time ... ADCP was located.)

L261: Why can the stronger surface flow at the western side be expected? Would be interesting to elaborate more on the dynamics.

L264: Why is the persistency of surface currents lower in autumn? Due to stronger or more frequent south-westerly wind conditions?

L267: What driving mechanism turned the surface flow northward during winter 2013/2014?

Fig.8: Why is the surface circulation strongest in summer? Due to melt water discharge and export?

L296: At least give correlation coefficients to support this statement.

L316: "... through the Northern Aland Sill."?

L326: What are 'the issues discoverd by Tuomi et al. and Miettunen et al.'?

general comments:

The authors configured a numerical model to investigate currents and resulting volume transports in the Åland Sea located between the Baltic proper and the Gulf of Bothnia in the northern Baltic Sea. This work is based on former model configurations with the aim to improve the performance of the the simulations in that challenging region with the focus on volume transports in unprecedented detail.

Large parts of the manuscript are dedicated to model description and validation even outside the validation section in the results chapter. Also the results sections on currents and transports descripe the simulations more or less in comparison with data and general knowlege in large parts and, consequently, do not contain much new facts. However, some new insight about seasonal current patterns, the surface transport and the pathways of deep water is given. Nevertheless, also the discussion focuses on the plausibility of the results, possible causes for deviations from existing findings, and ways to improve the model setup. Motivated by e.g. the importance of nutrient fluxes through the Åland Sea for the eutrophication status of the Gulf of Bothnia, unfortunately, neither implications of the findings on this nor the role of any relevant physical phenomena or driver for them and implications from that are discussed in detail and the manuscript remains more technical in this way. For this reason, the manuscript just has to be seen as a basework to develop modelling skills for the investigated region as the authors note themselves.

The language is good and clear. The content has some repetions from the introduction throughout the intermediate chapters all the way to the conclusions. The figures are of good quality and easy to understand.

specific comments:

line 102: I think it would be good to briefly explain the physical meaning of the Samagorinsky parameter.

line 107: Maybe some examples from literature should be given here.

line 163: Is there some literature proving this and the opposite for Föglö Degerby?

line 188: I agree that timing of the events seems good for which Figure 2 is appropriate. However, absolute differences are in the order of 0.1 m over most of the time shown. This would be much easier to see in a difference plot which is by definition much more appropriate to show differences. For me the following questions arise here: Is 0.1 m difference really quite small or what is to be called small here and why? And what is about the reference levels? Are they compareable at all? Are they determined in a comparable way? Can a target difference of 0 be expected from this point of view? Depending on your answers to these questions, it may be better to skip SSH differences as a mean for validtion, explain why a difference of 0 cannot be expected (mainly different reference levels I would assume) and focus validation on the timing and magnitude of the shown variations. Else, it should be explained why 0.1 m can be considered as small here and what implications these differences have on the dynamics and transports.

line 198: Similar problem than before with SSH: What is the expected accuracy and why? Which accuracy is needed for the planned investigation to be reasonable?

line 206: Why (see comment on line 198))? Which implications does this have for the investigations?

line 238 and before: How large is the error or accuracy of the numbers given in this paragraph and that one before?

Figure 3: It is a bit a pitty that it is not possible for all profiles to unambiguously assign the corresponding NEMO profile to the monitoring profile. I thought about indicating corresponding profiles from monitoring and model by colour, but for this the needed 12 to 32 different colours for all profiles is definitely too much and not manageable.

Figure 5: General problem with these plots: They are not normalised to phase space. But here this seems not to be very relevant for the discussed findings.

line 266: A good measure to show that would be to present the magnitudes of the vector means normalised to the means of the vector magnitudes. The effect itself is trivial and simply geometric.

lines 341/343/348/355: The mesoscale is defined by the first baroclinic Rossby radius. It would be good to give a number of its absolute size in this region. I suspect it to be quite small in this region and the model showing only mesoscale structures. Is the model capable to simulate submesoscale structures reasonably at all?

line 457: This also raises the question how much bigger the model domain should be in relation to the investigated area. Would the model results be more robust against disturbances or biases in the boundary data if the model domain is somewhat larger than the investigated area? Why is the relation between both choosen as it is here?

technical corrections:

line 41: Witting (1908) is missing in the references.

line 45:  Granquist (1938) is missing in the references.

line 54: Is F64 correct? In Figure 1 station F69 is located in Lågskär Deep. If F64 is correct it should be emphasised somehow that a station quite far away was used to draw this conclusion. Suggestion: ‚He also concluded from data from station F64, although it is located in the Åland Sea proper, that at Lågskär Deep …‘

line 78: Suggestion: Change ‚big depth gradients‘ to ‚large depth gradients‘.

lines 206/407/434: Suggestion: Replace ‚sensible‘ by ‚reasonable‘.

line 268: Suggestion: ‚In the other winters‘ instead of ‚Of the other winters‘

line 334: Suggestion: Better write ‚are plausible‘ than ‚were plausible‘. Check used times like this in the whole section! I would suggest to use present time to descripe your findings in general in the whole manuscript, especially also in section 3.

line 352: Suggestion: Put ‚at our disposal‘ to the end of the sentence for better readability.

line 361: Suggestion: ‚are‘ instead of ‚to be‘ to make it a correct sentence.

line 372: Suggestion: ‚direction of the mean seasonal current‘ sounds more correct.

line 386: Although the Knudsen theorem is commonly quite known, maybe is would be good to add the reference here.

line 387: ‚note‘ instead of ‚notes‘

line 422: Suggestion: Ommit ‚than‘, just ‚less ideal‘ reads better and maybe is more correct.

line 426: Suggestion: ‚a minor issue‘ instead ‚less of an issue‘.

line 485: ‚flows‘ instead ‚flow‘

General comments:

The authors have adapted a high-resolution model, which has already been used in previous studies, to the Åland Sea area in order to investigate in unprecedented detail the water transports between the Åland Sea and the Bothnian Sea in the north and the Baltic Proper in the south. Previous modelling studies as well as observational data lack detail, leaving open questions about, for example, the changing eutrophication status of the Bothnian Sea. This study now aims - at least to my reading - to fill these gaps in understanding (at least partially).

After a clearly structured description of the methods and used data, the results chapter is unfortunately no longer clearly ordered, as the model validation is not limited to the model validation part, but comprises large parts of the rest of the chapter. Accordingly, this chapter unfortunately contains only a little new findings on water exchange in the Åland Sea. In the discussion section that follows, the focus is then on possible mainly model-related reasons for the deviation of the model results from measurements, again on validation against already known findings from older studies and on potential further improvements to the model setup. In conclusion, it does not seem so clear whether the focus of this study is on the water transport through the Åland Sea itself or on the demonstration that the model can reproduce this transport very well.

All in all, the manuscript reads very well to me as a comprehensive model introduction or validation paper, describing a model that works in my opinion very well in the challenging region of the Åland Sea, and is certainly suitable for numerous and diverse follow-up studies in the region, as the authors themselves write in the outlook. In contrast, the analysis of water transport in the Åland Sea is unfortunately somewhat superficial. For example, no reasons for exceptionally occurring northward mean seasonal currents are discussed, and the knowledge gap on the change in the eutrophication status of the Bothnian Sea mentioned in the introduction is not even addressed in the discussion.

I would therefore recommend the authors to carry out more in-depth and somewhat more comprehensive analyses on water transport on the basis of the already available data used for this study (possibly including the meteorological data) or to shift the focus/objective of this study more towards the applicability/quality of the chosen model setup.

specific comments:

line 21: Could the water transports through the Åland Sea explain the changes in eutrophication status of Bothnian Sea? If so, why? Are there changes in transports? It would be good to refer to this again in the discussion section.

line 103: Why is the Smagorinsky parameter “rn_csmc” the only parameter explicitly mentioned here? What makes the selection of this parameter (as opposed to other parameters) so important? This selection would have to be explained in much more detail. Alternatively, one could list all selected parameters as an appendix or similar and not mention it here at all.

line 109-114: I think thermodynamic formulation means an ice model without ice drift. What effect on water transports can be expected from ice drift (both in mild and in harsher winters) that it is explicitly stated that only mild winters occurred in the modelling period?

line 162/163: Considering the location of Turku and Forsmark, it is easy to imagine that these gauges are not representative for the main study area. However, this should be explained in more detail (e.g. also by plotting the position of the gauges in Fig. 1).

Fig. 2: The time series obviously shows a high correlation, but there also seems to be a relatively large bias, so that I would like to see the numbers of the statistical parameters in addition to the figures. Can this bias be explained by a bias in the boundary conditions only or are there other reasons?

line 193 ff.: In contrast to the thermocline, the halocline is only mentioned in this subchapter in connection with a salinity bias above it. Since the halocline plays a very important role in the further course of the analysis, I would recommend describing the of the modelled halocline quality (location and gradient compared to the observations) in more detail. Due to the large number of individual profiles, this cannot be seen directly from Fig. 3 in my opinion.

line 260f.: Why is it to be expected that the current velocities on the western side are stronger than on the eastern side. I would like some words of explanation.

line 267-269: What is the reason for this special characteristic of the winter 2013/14 (and also of the winter 2016/17)? Were there exceptional wind conditions?

line 326 f.: It would be good to explicitly mention here the issues that have been discovered in previous publications.

line 360:  Why did not use a slightly larger model area if you suspect errors in your main study area due to conditions at the (relatively) near open model boundary?

line 363 f.: I would have given the numbers of missing data in earlier chapters and not just in the discussion.

line 371: Same as line 267-269: What is/could be the reason for these two “special” cases?

technical corrections::

line 54: Is station F64 really meant here? According to Figure 1, station F69 (and not F64) is positioned in Lågskär deep, so in this context F69 is probably meant.

line 78: “big depth gradients” should be changed to “large depth gradients”

line 206/407: I would prefer the use of "reasonable" instead of "sensible"

line 234: Figure 6 is mentioned in the text before Figure 5 - I would recommend to change the numbering of these figures.

line 238: The overestimation of the U-component should be between 0.006 and 0.034 (and not 0.34) m s-1.

line 372: It must be “direction of the mean seasonal current”