the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 28 Jun 2023
Drivers of Laptev Sea interannual variability in salinity and temperature
Abstract. Eurasian Rivers provide a quarter of total fresh water to the Arctic, maintaining a persistent fresh layer that covers the surface Arctic Ocean. The Lena River supplies the largest volume of runoff and plays a key role in this system, as runoff outflows into the Laptev Sea as a particularly shallow plume. This freshwater export controls Arctic Ocean stratification, circulation, and basin-wide sea ice area. Previous in-situ and modelling studies suggest that local wind forcing is a primary driver of variability in Laptev sea surface salinity (SSS) with no consensus over the roles of Lena river discharge and sea ice cover in contributing to this variability. Until recently, satellite SSS retrievals were insufficiently accurate for use in the Arctic, due to the low sensitivity of the L-band signal they utilise in cold water and challenges of retrieval near sea ice. However, retreating sea ice cover and continuous progress in satellite product development have significantly improved SSS retrievals, giving satellite SSS data true potential in the Arctic.
This study demonstrates a novel method of using satellite-based SSS, sea surface temperature (SST) data, in-situ observations, and reanalysis products to identify the dominant drivers of interannual variability in Laptev Sea dynamics. Satellite-based SSS is found to agree well with in-situ data in this region (r > 0.8) and provides notable improvements compared to the reanalysis product used in this study (r > 0.7) in capturing patterns and variability observed in in-situ data. The satellite SSS data firmly establishes what has previously been subject to debate due to the limited years and locations analysed with in-situ data: that the zonal wind is the dominant driver of offshore or onshore Lena river plume transport. This finding is affirmed by the strong agreement in SSS pattern in all reanalyses and satellite products used in this study under eastward and westward wind regimes. The pattern of SST also varies with the zonal wind component, and drives spatial variability in sea ice area. The strong correspondence between large scale and local zonal wind dynamics and the key role of SSS and SST variability in driving sea ice and stratification dynamics demonstrates the importance of changes in large-scale atmospheric dynamics for variability in this region as well as for future Arctic sea ice dynamics and freshwater transport.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(5513 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(5513 KB) - Metadata XML
- BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-1403', Céline Heuzé, 21 Jul 2023
Hi Phoebe,
You’re not going to like what comes next. Sorry.
Major comments
I. The objective of this manuscript is unclear
Although both the abstract and the introduction are very long, they are also unfocused, so it is unclear whether the aim of this article is to:
- combine satellite, reanalysis, and in-situ data to study the plume;
- demonstrate that satellite data can be used to look at sea surface salinity in the region;
- or determine which reanalysis product is most adapted for this study.
These different objectives would then result in different structures of the manuscript, with the majority of the product comparisons and validations going to the appendix. It would also affect the time period considered. The choice of in-situ data would then be affected as well; see my next point.
II. The choice of data, especially their resolution, was surprising
The method section did not clarify much, so I am not sure which time period you worked with. That is, you mention that you use UDASH, but that stopped in 2015 and would not really help with SMAP. The manuscript needs an overarching table that says for all types of products over which time period and at which spatio-temporal resolution they are available (not the resolution at which you use them; their native one).
Coupled with the fact that the objective is unclear, I won’t be able to give you a clear direction. But if you want to validate SSS, I would have looked for underway CTD data rather than CTD casts. It will be at approx. 10 m depth, but on most vessels (and especially so on Oden, i.e. for the SWERUS data) the upper 10 m of the CTD casts cannot be used anyway. If you want the upper water column, there might be ITPs nearby, and there should be at least one mooring, but I’m not sure of their time coverage.
The other thing that really surprised me is that you want to investigate a plume dynamics, have a 3-day product available… and downgrade it to monthly resolution. And then regularly show results in the manuscript that you explain with the poor resolution of your monthly product. Use the 3-day version.
And regardless of what you do, you need to say why you do so. Maybe you had a perfectly valid reason for using CTD casts and downgrading everything to one month, but you did not write it so the reader cannot know.
III. Causation is not shown
This is my main issue with your manuscript. You do not demonstrate causation. You produce two composites, and declare that the variable you composited against explains the differences.
Let’s start with the definition of the composites. I agree that on Figure 2, the circulation is different. However on Figure 4, the uncertainty is so large that some of the strongest years could in fact have a value with either sign. See for example 2012 and 2019. As rotation is involved, a metric based on the curl of the wind, or simply on the sea level pressure, may be more effective and robust.
Anyway, the outcome of the composites is that the SSS looks different. But so do the sea ice and the SST, which could both explain the SSS pattern, and even be responsible for the wind differences. Or maybe wind and SSS are both the result of another variable that is not included in your analysis.
What you are really showing is that the hypothesis that wind drives SSS is not incompatible with the observations. But as your analyses currently are, you do not demonstrate the causality. One option is therefore to just rephrase everything, removing all mentions of ”the wind drives” and saying what I just wrote. But that’s rather underwhelming a result.
Instead, you could utilise the 3-day product in its full 3-day glory. For each point, or for the overall region, do a lagged (temporal) correlation analysis of the relationship between wind (available at daily resolution; downgrade to 3-day) and the 3-day SSS from BEC. Then see which variable drives the other, based on these values. I personally would push the analysis and perform the same calculation also with the SST and the sea ice, and have an overall result matrix that shows for each pair of variables for which lag the correlation is maximal and what that correlation value is.
Less major comments, in order of appearance
Salinity unit: You meant ”psu”or really”pss”? Oceanographers use the absolute salinity in g/kg now.
Subsection 2.1.1: not detailed enough. Either there or in the introduction, you need to be more specific about which satellite measures at which band, has which footprint, repeat time, etc. Something similar to the first paragraph of section 2.1 of these people: https://tc.copernicus.org/articles/12/921/2018/tc-12-921-2018.pdf , but for all sensors (at least SMOS and SMAP, and then explanations about how the different products combine them).
Subsection 2.2.1: also not detailed enough; what do you mean by ”correlation” and ”RMSD”? I assume that you took all points available, regardless of location and time, and basically did a regression? Given that the plume is both time and space dependent, as shown on your figure, I would recommend you verify the temporal and spatial accuracy separately. You may need more points for this, agreed, but see Major comment II.
Section 3.1: You do not show the RMSD. See my previous comment anyway, but that could be added to the table -which could be shortened once the manuscript is more focused, see Major comment I.
Sea ice: Line 377 onwards you give statistics of the sea ice area, without specifying over which region. Overall, and based on the figures shown in the manuscript, the area does not seem to matter as much as the southern / eastern extent. I would rather use such extent, if doing the correlations suggested above. If not, then do not even quantify it; your maps are very clear.
Runoff: Line 405, the discussion starts with an analysis of the correlation between runoff and SSS. It is not specified, but I assume that the runoff values are published elsewhere? If so, reproduce them here, and do a proper (lagged) correlation analysis.
Arctic Oscillation: Same comment, no information about where the AOI comes from and the correlation analysis is not shown.
Citation: https://doi.org/10.5194/egusphere-2023-1403-RC1 - AC1: 'Reply on RC1', Phoebe Hudson, 11 Aug 2023
-
RC2: 'Comment on egusphere-2023-1403', Anonymous Referee #2, 04 Aug 2023
Dear coauthors,
Your paper offers valuable insights into the drivers of Laptev Sea dynamics and interannual variability in salinity and temperature. It provides evidence that the salinity and temperature signatures agree in different reanalyses and satellite products under different wind regimes. Addressing the major comments proposed will improve the clarity and robustness of the research, enhancing its contribution to the scientific community.
Major Comments:
The focus of the paper:
The stated objective of the paper is to determine the drivers of the interannual variability of the Laptev Sea dynamics. However, a significant portion of the paper is devoted to the validation/intercomparison of satellite Sea Surface Salinity (SSS) products. While the effort to validate and compare these products is commendable, it appears to dominate the narrative, diverting attention from the primary objective of identifying the drivers of Laptev Sea dynamics.
To address this concern, I see two options:
a) Dividing the current study into two separate papers, each with a distinct focus and stating clearly the objectives (Validation and Intercomparison of Satellite SSS Products / Drivers of Laptev Sea Interannual Variability in Salinity and Temperature).
b) Focus on the primary objective during the narrative and if you feel that the validation/intercomparison of satellite SSS products is an essential part of the methodology, it would be beneficial to include at least one example figure showcasing the different products. Additionally, providing information on the p-value of your correlations, bias, and spectral analysis to assess the effective resolution of the products will improve the methodological rigor and transparency of the study.
Methods:
- In subsection 2.2.2, it is not clear why the analysis is not performed with all four satellite products. To ensure the robustness of the study, it would be beneficial to explain the reasons for the exclusion, if any, of certain products in the analysis.
- The temporal resolution of satellite SSS products is a critical factor when studying the dynamics of a region like the Laptev Sea, where rapid changes can occur over short time scales. If you choose not to use 3-day or available 8-day satellite Sea Surface Salinity (SSS) products, providing a clear and well-justified argument for this decision is crucial.
- The paper mentions a validation/intercomparison of satellite SSS products. To strengthen this aspect, I suggest including an example figure showing the different products for comparison. Additionally, information on the p-values of your correlations, bias, and spectral analysis to determine the effective resolution of the different products should be included.
- The decision to use the median of the products for analysis should be justified. It might be more appropriate to use the product that best aligns with in-situ information, has a higher spatiotemporal resolution, or realistically agrees with the expected dynamics of the area. If the median approach is retained, the reasoning behind this choice should be elaborated.
Results:
- The results section lacks concrete analysis and tangible results to support the discussed relationships with the Arctic Oscillation Index, and river runoff.
Discussion:
- The discussion/conclusion emphasizes that wind is the dominant driver of offshore or onshore Lena River plume transport. To strengthen this claim, the study should include additional evidence from the analysis correlating composites and other drivers, for example, the river runoff, ice melting, etc.
- Line 415: your claim that because GLORYS12V1, which doesn't include interannually varying river runoff, replicates the SSS pattern well as compared to satellite SSS, suggests that variability in river runoff is not a significant contributor to the interannual variability in GLORYS12V1 SSS. However, GLORYS12V1 utilizes in-situ SSS data, which is how it reproduces the SSS pattern. This does not negate the potential influence of river runoff on interannual SSS variability. The absence of river runoff does not imply that it has no impact on SSS dynamics. Moreover, the correlation between GLORYS12V1 and satellite SSS patterns does not necessarily indicate causation, river runoff could influence Laptev SSS variability, if you make this argument, the paper should conduct a more comprehensive analysis that explicitly investigates the impact of river runoff on interannual SSS variability.
- In Section 4.1, there is a discussion about correlating composites to river runoff, but no results are shown to support the argument. The analysis should be included to provide tangible evidence for the discussion. Additionally, the use of the BEC SSS should be addressed if you want to compare your results to the study of the product as in Umbert et al. 2021, who uses this product. The absence of BEC SSS figures and the correlation against river runoff data should be explained to ensure a coherent argument.
Figures:
- Figure 1, it is unclear why the wind over plots are not the mean for September, similar to the salinity over plots. Providing an explanation for this difference would enhance the figure's clarity and interpretation.
- In Figure 3, I strongly suggest to include the other two satellite SSS products
Minor comments:
Introduction
I suggest including a reference to Umbert et al. 2021 as it also uses SMOS SSS to characterize the Lena River plume in the Laptev Sea, which could provide valuable context and potential links between the two studies.
Methods
In line 280, it is mentioned that the median product is generated using GLORYS12V1, LOCEAN SMOS, and both SMAP products. However, it seems there might be a discrepancy, as it was previously stated that there were four satellite products. This inconsistency needs clarification.
Results
Section 3.3 is missing, but it is referred to in the text as "3.2 3". The authors should correct this discrepancy and make sure the section numbers are accurate.
In Table 1, it is puzzling that the median product has more observations than any of the individual products. The authors should address this discrepancy and provide an explanation for the data variations to ensure the table's accuracy and consistency.
Citation: https://doi.org/10.5194/egusphere-2023-1403-RC2 - AC2: 'Reply on RC2', Phoebe Hudson, 22 Aug 2023
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-1403', Céline Heuzé, 21 Jul 2023
Hi Phoebe,
You’re not going to like what comes next. Sorry.
Major comments
I. The objective of this manuscript is unclear
Although both the abstract and the introduction are very long, they are also unfocused, so it is unclear whether the aim of this article is to:
- combine satellite, reanalysis, and in-situ data to study the plume;
- demonstrate that satellite data can be used to look at sea surface salinity in the region;
- or determine which reanalysis product is most adapted for this study.
These different objectives would then result in different structures of the manuscript, with the majority of the product comparisons and validations going to the appendix. It would also affect the time period considered. The choice of in-situ data would then be affected as well; see my next point.
II. The choice of data, especially their resolution, was surprising
The method section did not clarify much, so I am not sure which time period you worked with. That is, you mention that you use UDASH, but that stopped in 2015 and would not really help with SMAP. The manuscript needs an overarching table that says for all types of products over which time period and at which spatio-temporal resolution they are available (not the resolution at which you use them; their native one).
Coupled with the fact that the objective is unclear, I won’t be able to give you a clear direction. But if you want to validate SSS, I would have looked for underway CTD data rather than CTD casts. It will be at approx. 10 m depth, but on most vessels (and especially so on Oden, i.e. for the SWERUS data) the upper 10 m of the CTD casts cannot be used anyway. If you want the upper water column, there might be ITPs nearby, and there should be at least one mooring, but I’m not sure of their time coverage.
The other thing that really surprised me is that you want to investigate a plume dynamics, have a 3-day product available… and downgrade it to monthly resolution. And then regularly show results in the manuscript that you explain with the poor resolution of your monthly product. Use the 3-day version.
And regardless of what you do, you need to say why you do so. Maybe you had a perfectly valid reason for using CTD casts and downgrading everything to one month, but you did not write it so the reader cannot know.
III. Causation is not shown
This is my main issue with your manuscript. You do not demonstrate causation. You produce two composites, and declare that the variable you composited against explains the differences.
Let’s start with the definition of the composites. I agree that on Figure 2, the circulation is different. However on Figure 4, the uncertainty is so large that some of the strongest years could in fact have a value with either sign. See for example 2012 and 2019. As rotation is involved, a metric based on the curl of the wind, or simply on the sea level pressure, may be more effective and robust.
Anyway, the outcome of the composites is that the SSS looks different. But so do the sea ice and the SST, which could both explain the SSS pattern, and even be responsible for the wind differences. Or maybe wind and SSS are both the result of another variable that is not included in your analysis.
What you are really showing is that the hypothesis that wind drives SSS is not incompatible with the observations. But as your analyses currently are, you do not demonstrate the causality. One option is therefore to just rephrase everything, removing all mentions of ”the wind drives” and saying what I just wrote. But that’s rather underwhelming a result.
Instead, you could utilise the 3-day product in its full 3-day glory. For each point, or for the overall region, do a lagged (temporal) correlation analysis of the relationship between wind (available at daily resolution; downgrade to 3-day) and the 3-day SSS from BEC. Then see which variable drives the other, based on these values. I personally would push the analysis and perform the same calculation also with the SST and the sea ice, and have an overall result matrix that shows for each pair of variables for which lag the correlation is maximal and what that correlation value is.
Less major comments, in order of appearance
Salinity unit: You meant ”psu”or really”pss”? Oceanographers use the absolute salinity in g/kg now.
Subsection 2.1.1: not detailed enough. Either there or in the introduction, you need to be more specific about which satellite measures at which band, has which footprint, repeat time, etc. Something similar to the first paragraph of section 2.1 of these people: https://tc.copernicus.org/articles/12/921/2018/tc-12-921-2018.pdf , but for all sensors (at least SMOS and SMAP, and then explanations about how the different products combine them).
Subsection 2.2.1: also not detailed enough; what do you mean by ”correlation” and ”RMSD”? I assume that you took all points available, regardless of location and time, and basically did a regression? Given that the plume is both time and space dependent, as shown on your figure, I would recommend you verify the temporal and spatial accuracy separately. You may need more points for this, agreed, but see Major comment II.
Section 3.1: You do not show the RMSD. See my previous comment anyway, but that could be added to the table -which could be shortened once the manuscript is more focused, see Major comment I.
Sea ice: Line 377 onwards you give statistics of the sea ice area, without specifying over which region. Overall, and based on the figures shown in the manuscript, the area does not seem to matter as much as the southern / eastern extent. I would rather use such extent, if doing the correlations suggested above. If not, then do not even quantify it; your maps are very clear.
Runoff: Line 405, the discussion starts with an analysis of the correlation between runoff and SSS. It is not specified, but I assume that the runoff values are published elsewhere? If so, reproduce them here, and do a proper (lagged) correlation analysis.
Arctic Oscillation: Same comment, no information about where the AOI comes from and the correlation analysis is not shown.
Citation: https://doi.org/10.5194/egusphere-2023-1403-RC1 - AC1: 'Reply on RC1', Phoebe Hudson, 11 Aug 2023
-
RC2: 'Comment on egusphere-2023-1403', Anonymous Referee #2, 04 Aug 2023
Dear coauthors,
Your paper offers valuable insights into the drivers of Laptev Sea dynamics and interannual variability in salinity and temperature. It provides evidence that the salinity and temperature signatures agree in different reanalyses and satellite products under different wind regimes. Addressing the major comments proposed will improve the clarity and robustness of the research, enhancing its contribution to the scientific community.
Major Comments:
The focus of the paper:
The stated objective of the paper is to determine the drivers of the interannual variability of the Laptev Sea dynamics. However, a significant portion of the paper is devoted to the validation/intercomparison of satellite Sea Surface Salinity (SSS) products. While the effort to validate and compare these products is commendable, it appears to dominate the narrative, diverting attention from the primary objective of identifying the drivers of Laptev Sea dynamics.
To address this concern, I see two options:
a) Dividing the current study into two separate papers, each with a distinct focus and stating clearly the objectives (Validation and Intercomparison of Satellite SSS Products / Drivers of Laptev Sea Interannual Variability in Salinity and Temperature).
b) Focus on the primary objective during the narrative and if you feel that the validation/intercomparison of satellite SSS products is an essential part of the methodology, it would be beneficial to include at least one example figure showcasing the different products. Additionally, providing information on the p-value of your correlations, bias, and spectral analysis to assess the effective resolution of the products will improve the methodological rigor and transparency of the study.
Methods:
- In subsection 2.2.2, it is not clear why the analysis is not performed with all four satellite products. To ensure the robustness of the study, it would be beneficial to explain the reasons for the exclusion, if any, of certain products in the analysis.
- The temporal resolution of satellite SSS products is a critical factor when studying the dynamics of a region like the Laptev Sea, where rapid changes can occur over short time scales. If you choose not to use 3-day or available 8-day satellite Sea Surface Salinity (SSS) products, providing a clear and well-justified argument for this decision is crucial.
- The paper mentions a validation/intercomparison of satellite SSS products. To strengthen this aspect, I suggest including an example figure showing the different products for comparison. Additionally, information on the p-values of your correlations, bias, and spectral analysis to determine the effective resolution of the different products should be included.
- The decision to use the median of the products for analysis should be justified. It might be more appropriate to use the product that best aligns with in-situ information, has a higher spatiotemporal resolution, or realistically agrees with the expected dynamics of the area. If the median approach is retained, the reasoning behind this choice should be elaborated.
Results:
- The results section lacks concrete analysis and tangible results to support the discussed relationships with the Arctic Oscillation Index, and river runoff.
Discussion:
- The discussion/conclusion emphasizes that wind is the dominant driver of offshore or onshore Lena River plume transport. To strengthen this claim, the study should include additional evidence from the analysis correlating composites and other drivers, for example, the river runoff, ice melting, etc.
- Line 415: your claim that because GLORYS12V1, which doesn't include interannually varying river runoff, replicates the SSS pattern well as compared to satellite SSS, suggests that variability in river runoff is not a significant contributor to the interannual variability in GLORYS12V1 SSS. However, GLORYS12V1 utilizes in-situ SSS data, which is how it reproduces the SSS pattern. This does not negate the potential influence of river runoff on interannual SSS variability. The absence of river runoff does not imply that it has no impact on SSS dynamics. Moreover, the correlation between GLORYS12V1 and satellite SSS patterns does not necessarily indicate causation, river runoff could influence Laptev SSS variability, if you make this argument, the paper should conduct a more comprehensive analysis that explicitly investigates the impact of river runoff on interannual SSS variability.
- In Section 4.1, there is a discussion about correlating composites to river runoff, but no results are shown to support the argument. The analysis should be included to provide tangible evidence for the discussion. Additionally, the use of the BEC SSS should be addressed if you want to compare your results to the study of the product as in Umbert et al. 2021, who uses this product. The absence of BEC SSS figures and the correlation against river runoff data should be explained to ensure a coherent argument.
Figures:
- Figure 1, it is unclear why the wind over plots are not the mean for September, similar to the salinity over plots. Providing an explanation for this difference would enhance the figure's clarity and interpretation.
- In Figure 3, I strongly suggest to include the other two satellite SSS products
Minor comments:
Introduction
I suggest including a reference to Umbert et al. 2021 as it also uses SMOS SSS to characterize the Lena River plume in the Laptev Sea, which could provide valuable context and potential links between the two studies.
Methods
In line 280, it is mentioned that the median product is generated using GLORYS12V1, LOCEAN SMOS, and both SMAP products. However, it seems there might be a discrepancy, as it was previously stated that there were four satellite products. This inconsistency needs clarification.
Results
Section 3.3 is missing, but it is referred to in the text as "3.2 3". The authors should correct this discrepancy and make sure the section numbers are accurate.
In Table 1, it is puzzling that the median product has more observations than any of the individual products. The authors should address this discrepancy and provide an explanation for the data variations to ensure the table's accuracy and consistency.
Citation: https://doi.org/10.5194/egusphere-2023-1403-RC2 - AC2: 'Reply on RC2', Phoebe Hudson, 22 Aug 2023
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 298 | 150 | 32 | 480 | 17 | 22 |
- HTML: 298
- PDF: 150
- XML: 32
- Total: 480
- BibTeX: 17
- EndNote: 22
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Phoebe Alice Hudson
Adrien Martin
Simon Josey
Alice Marzocchi
Athanasios Angeloudis
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(5513 KB) - Metadata XML