Investigation of North Atlantic Salinity Long-Term Trends Based on Historical Datasets

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

 The salinity of seawater is a key ocean parameter, and the author analyzed the changing characteristics of seawater salinity in the North Atlantic using different datasets. This is a very meaningful job. Regarding the content of this manuscript, I have the following questions that the author needs to answer.

 1. Page 1, line 19. Missing space between “and” and “CO₂”.

 

2. Page 3. Section 2. Data and Methods

Lines 136-137. “The data used in the paper are monthly mean felds of ocean salinity from objective analysis datasets, ……”  

“felds” ? maybe “fields”.

What is the spatial resolution of these datasets? Please provide a detailed introduction to the datasets used in this manuscript.

 3. Section 3. Results

The spatial distribution of the average salinity values in different datasets is basically consistent, but the trend of changes varies significantly. Can the author analyze the reasons for these differences in the manuscript? Which dataset does the author think is the most reliable?

 4. Section 3.3.

Why only conduct a generalized analysis on datasets from 1961-2011? Please explain in the manuscript why the datasets from 1948-2018 were not analyzed?

Will the positive and negative trends of salinity changes in different datasets cancel out in Figure 4, making it difficult to accurately reflect the areas where salinity changes occur?

 5. Section 4. Conclusions

Lines 351-352. “For this 71yr period, median monthly salinity in subtropical latitudes increased by 0.07±0.02 PSU. “

Lines 359-361. “Over this 51yr period, in the vicinity of the Canary upwelling, salinization in the layer 10–400 m is 0.1 PSU and is detected in almost all the datasets used, with the exception of the ESTOC and ORAS4 reanalyses.”

How were the trend values in the above conclusion calculated? I think it would be better to provide specific calculation methods.

 6. Due to the differences in the trend of seawater salinity calculated from different datasets, the author draws the conclusion of this manuscript based on the rule of majority rule. Can the author evaluate the reliability of the dataset used in this paper? Which dataset is the most authoritative?

Comments on the Quality of English Language

None

Author Response

The salinity of seawater is a key ocean parameter, and the author analyzed the changing characteristics of seawater salinity in the North Atlantic using different datasets. This is a very meaningful job. Regarding the content of this manuscript, I have the following questions that the author needs to answer.

The authors express their sincere gratitude to the Reviewer for his careful review of the manuscript and valuable recommendations. The reviewer's comments are given in italics and responses to them are given in regular font.

1. Page 1, line 19. Missing space between “and” and “CO2”.

Tnank you for the remark. We corrected the misprint.

2. Page 3. Section 2. Data and Methods

Lines 136-137. “The data used in the paper are monthly mean felds of ocean salinity from objective analysis datasets, ……” 

“felds” ? maybe “fields”.

Tnank you for the remark. We corrected the misprint.

What is the spatial resolution of these datasets? Please provide a detailed introduction to the datasets used in this manuscript.

Thank you for the comment. We provided additional details on the datasets used, including the time interval covered, the ocean model used, horizontal and vertical resolution, and the data assimilation method.

3. Section 3. Results

The spatial distribution of the average salinity values in different datasets is basically consistent, but the trend of changes varies significantly. Can the author analyze the reasons for these differences in the manuscript? Which dataset does the author think is the most reliable?

Thank you for the remark. We added the section "Discussion", which, as the Introduction, analyzes some of the reasons for the differences in trends. As for the most reliable dataset, this study did not set out to determine the most reliable salinity dataset. Instead, we attempted to clarify the pattern of long-term salinity change using several independent datasets.

4. Section 3.3.

Why only conduct a generalized analysis on datasets from 1961-2011? Please explain in the manuscript why the datasets from 1948-2018 were not analyzed?

The dataset analysis is only performed for 1961–2011 because the greatest number of datasets (8) is available for this period. This number of datasets allows us to make reliable conclusions on salinity trends. The pattern in Figure 4 can be considered as universal because adding another dataset (the ninth) to the analysis can strengthen the conclusions but not change the pattern. For example, if 6 of 8 datasets indicate a trend and the ninth one yields the opposite sign, then 6 of 9 datasets ultimately indicate a trend. In this sense, the choice of the time interval 1961–2011 is not tied to climate events but provides the most complete sample from which the most reliable conclusions can be drawn. The revised version of the article contains a respective clarification.

Will the positive and negative trends of salinity changes in different datasets cancel out in Figure 4, making it difficult to accurately reflect the areas where salinity changes occur?

Yes. Certainly, the presence of positive and negative salinity trends in different datasets makes it difficult to accurately reflect the areas where salinity changes occur. In the context of the method of displaying changes and agreement between datasets, if trends of different signs are obtained in different datasets, they will lead to a discrepancy between the datasets.

To resolve this situation, additional sources of information must be used, and the more the better. For the period 1948–2018, almost the entire North Atlantic will be filled with gray dots. This happens because the salinity trends in the 3 datasets are contradictory, and additional sources of information on salinity changes over such a long period are not available.

For the period 1961–2011, there are enough information sources to obtain reliable conclusions. In a situation where a significant trend is obtained for a separate grid node in one dataset, while there is no significant trend in another dataset, it is necessary to count the number of datasets with such outcomes. Whichever datasets are larger, such a conclusion will be made. This situation also depends on the level of statistical significance of the trend. Result: either there is a reliable trend in the grid node (red), or there is no trend in the grid node (green). In another situation, when significant trends with opposite signs are obtained for a separate grid node, uncertainty arises (these are the gray grid nodes in Figure 4). Again, it is necessary to involve additional sources of information. All "suspicious" grid nodes were analyzed manually after the algorithm ran. For example, for the grid node centered at 49° N, 40° W, 7 of the 8 datasets used show a significant positive trend, and 1 of the 8 datasets (ESTOC) shows a significant negative trend. Here the conclusion is obvious. If any one dataset is removed from the sample (there will be 6 of 7 datasets or 7 of 7 datasets (if ESTOC is removed)), the conclusion made for this grid node will not change.

5. Section 4. Conclusions

Lines 351-352. “For this 71yr period, median monthly salinity in subtropical latitudes increased by 0.07±0.02 PSU. “

Lines 359-361. “Over this 51yr period, in the vicinity of the Canary upwelling, salinization in the layer 10–400 m is 0.1 PSU and is detected in almost all the datasets used, with the exception of the ESTOC and ORAS4 reanalyses.”

How were the trend values in the above conclusion calculated? I think it would be better to provide specific calculation methods.

We agree with the remark. The revised version of the article provides a corresponding clarification.

6. Due to the differences in the trend of seawater salinity calculated from different datasets, the author draws the conclusion of this manuscript based on the rule of majority rule. Can the author evaluate the reliability of the dataset used in this paper? Which dataset is the most authoritative?

This study did not attempt to identify the most reliable salinity dataset. Rather, we attempted to elucidate the general pattern of long-term salinity change using multiple independent datasets. The authority of a particular dataset can be indirectly judged by how closely its salinity trend estimates are to the combined estimates from multiple datasets. This task is complicated by the fact that a particular dataset may well describe consistent salinity trends in one region, but be inconsistent with them in another region. This depends on the ability of the particular dataset to represent local processes and the availability of sufficient salinity data in that region.

Sincerely,

authors

Reviewer 2 Report

Comments and Suggestions for Authors

Dear authors,

   The author uses historical data to analyze the long-term trends and variations in North Atlantic salinity. However, under this thematic framework, many important factors that need to be discussed are overlooked. Here are the suggestions.

1.      The quality and rendering of the image are very poor and need to be redrawn.

2.      Lines 131-134: The authors need to provide a well-justified scientific rationale for why the periods 1948-2018 and 1961-2011 were chosen for comparison.

3.      Section three is based on unclear period classifications, and the author has also not considered any variable parameters that might affect salinity distribution, such as characteristic watersheds and the distribution of salinity in the North Atlantic. Moreover, the author has not conducted further classifications for climate change features, such as the North Atlantic Oscillation and the Arctic Oscillation, which significantly impact salinity distribution. These aspects require detailed discussion.

4.      Additionally, the authors need to consider the impact of pre- and post-salinity satellite launches on salinity observations, which is significant.

Author Response

The author uses historical data to analyze the long-term trends and variations in North Atlantic salinity. However, under this thematic framework, many important factors that need to be discussed are overlooked. Here are the suggestions.

The authors thank the Reviewer for his helpful comments and valuable suggestions that contributed to improving the manuscript. The reviewer's comments are given in italics below, and responses to them are given in regular font.

1. The quality and rendering of the image are very poor and need to be redrawn.

Thank you for the comment. In the revised version of the paper, the figures have been changed.

2. Lines 131-134: The authors need to provide a well-justified scientific rationale for why the periods 1948-2018 and 1961-2011 were chosen for comparison.

Thank you for your comment. The revised version of the paper provides the rationale for choosing the periods under consideration. The choice of the first period (1948–2018) is related to the need to analyze the longest-term changes in the North Atlantic salinity based on available historical data. The choice of the second period (1961–2011) is due to the need to compare the longest time series based on the largest number of datasets. Our goal was to determine the general pattern of the long-term change in the North Atlantic salinity. Therefore, in this task, we were interested in the longest and most accessible data on salinity.

3. Section three is based on unclear period classifications, and the author has also not considered any variable parameters that might affect salinity distribution, such as characteristic watersheds and the distribution of salinity in the North Atlantic. Moreover, the author has not conducted further classifications for climate change features, such as the North Atlantic Oscillation and the Arctic Oscillation, which significantly impact salinity distribution. These aspects require detailed discussion.

We agree with the remark. Since we are interested in long-term trends in North Atlantic salinity obtained on the basis of several arrays of historical data, the article is devoted only to this issue. As the Reviewer correctly notes, the analysis of further classifications of climate change features is not carried out in the scope of this paper. We do this in order not to overload the presentation. Although the Reviewer is right that climate signals have a significant impact on the salinity distribution, we believe that these aspects require separate detailed consideration.

4. Additionally, the authors need to consider the impact of pre- and post-salinity satellite launches on salinity observations, which is significant.

Thank you for the remark. The launch of satellites allows one to obtain unique data on the state of the ocean surface. Based on satellite platforms, it has become possible to study changes in salinity on the ocean surface. However, estimates of salinity on the ocean surface from Aquarius have only been available since 2011. This time interval is not yet sufficient to analyze long-term changes in salinity. In addition, satellites provide global high-resolution observations of the ocean surface. It is also of interest to study the evolution of salinity of subsurface waters.

Sincerely,

authors

Reviewer 3 Report

Comments and Suggestions for Authors

Long term changes in North Atlantic salinity are examined using a number of existing reanalysis datasets.  This is carried out using quantile regression, and some general trends in broad areas, and more localised trends are identified.  Overall this is an interesting paper with some useful analysis and results. There are some general points and also some minor errors to be considered.

In the introduction there could be more discussion of the datasets, how they were generated, and any issues with this in terms of evaluating salinity over time (especially given the variation in measurement methods over the time period in question, and how this might influence or be corrected by the reanalysis products used).

In the conclusions, some discussion of why particular datasets might give different results would be good (but the fact that a number of different sets, with different methods, give similar results does give some confidence in the robustness of the results).

Some minor errors, by line number:

19 and CO2 [space]

100 great than that for the period [I think this is what you mean]

122 Most reanalysis datasets

136 fields

153 . Applicability

166 possible using

240 are mostly not pronounced

Comments on the Quality of English Language

minor corrections, see above

Author Response

Long term changes in North Atlantic salinity are examined using a number of existing reanalysis datasets.  This is carried out using quantile regression, and some general trends in broad areas, and more localised trends are identified.  Overall this is an interesting paper with some useful analysis and results. There are some general points and also some minor errors to be considered.

The authors express their deep gratitude to the Reviewer for useful comments and recommendations that contributed to improving the manuscript. The Reviewer's comments are given in italics below, and responses to them are given in regular font.

In the introduction there could be more discussion of the datasets, how they were generated, and any issues with this in terms of evaluating salinity over time (especially given the variation in measurement methods over the time period in question, and how this might influence or be corrected by the reanalysis products used).

Thank you for the remark. This issue is partially covered in the penultimate paragraph of the Introduction section, which is why this section turned out to be extensive. In the revised version of the article, the first paragraph of the section "Data and Methods" has been changed. A brief description of the datasets used has been added to this paragraph.

In the conclusions, some discussion of why particular datasets might give different results would be good (but the fact that a number of different sets, with different methods, give similar results does give some confidence in the robustness of the results).

Thank you for the remark. The revised version of the paper has a section "Discussion". As the reviewer correctly notes, if different datasets with different methods yield similar results, this instills confidence in the estimates obtained. This is the ideology on which the presented article is based. It is difficult to trust unverified estimates obtained from only one dataset. Therefore, in this article, we adapted the method of Tebaldi et al. 2011, which was originally intended for verifying climate models from the CMIP6 project. Ideally, it would be better to take the data of any objective analysis as a reference array and compare all other ocean reanalyses with it. However, the results of the article show that the data of the three objective analyses used slightly differ in representing such a climate signal as a long-term trend. Therefore, we analyze the data of objective ocean analyses on a par with the data of ocean reanalyses. This attempt allowed us to obtain small-area regions with salinization, but these regions are recorded for most of the considered datasets. We would like to share this highly reliable result with our readers in this article.

Some minor errors, by line number:

19 and CO2 [space]

100 great than that for the period [I think this is what you mean]

122 Most reanalysis datasets

136 fields

153 . Applicability

166 possible using

240 are mostly not pronounced

Thank you for the remarks. We corrected these items.

Sincerely,

authors

Reviewer 4 Report

Comments and Suggestions for Authors

Dear authors,

 

General Comment: This study employed multiple long-term datasets to quantify long-term salinity trends in the northern part of the Atlantic. And also explores the extent to which salinity trends can be reliably determined using existing available datasets. 

The introduction is too long, 2 full pages introduction. It contains some unnecessary details, kindly remove them all and reduce it to a sizeable length.

 

 

Specific comments:

Line 1: Can you add a background introductory sentence to your abstract before talking about the methodology and results?

Line 16-17: Change the word “as” to give the sentence a meaning

Line 216: 236. Bring this section up right before Fig. 1

Line 277-305:  Take it up right before Fig. 3

Line 323-340:  Take it up right before Fig. 4

 

The analysis from the different datasets used shows some differences, what are the causes of the differences? Are these differences reasonable enough to affect long-term climate projections?

 

A new section “Discussion” is necessary, to link your results to past studies

 

Comments on the Quality of English Language

minor

Author Response

The authors express gratitude to the Reviewer for a number of useful comments, which undoubtedly contributed to the improvement of the manuscript. The Reviewer's comments are given in italics below, and the responses to them are given in regular font.

General Comment: This study employed multiple long-term datasets to quantify long-term salinity trends in the northern part of the Atlantic. And also explores the extent to which salinity trends can be reliably determined using existing available datasets.

The introduction is too long, 2 full pages introduction. It contains some unnecessary details, kindly remove them all and reduce it to a sizeable length.

We agree with the remark. By now, many studies have been conducted on the long-term variability of salinity in the North Atlantic. This explains the extensive Introduction section in the present article. One of the important points that we wanted to show in the Introduction is the coincidence of the amplitude of the intense decadal variability of salinity and the amplitude of the long-term salinity trend. This factor and the dependence of the analysis of linear trends on the length of time series lead to difficulties in analyzing and assessing the long-term evolution of salinity. However, the Introduction section in the article turned out to be overloaded and difficult to understand. Therefore, as the reviewer correctly points out, we removed some unnecessary, in our opinion, details.

 Specific comments:

Line 1: Can you add a background introductory sentence to your abstract before talking about the methodology and results?

We agree with the remark. The revised version of the article contains a corresponding clarification.

 Line 16-17: Change the word “as” to give the sentence a meaning

We agree with the remark. The first sentence of the article has been changed in the revised version.

Line 216: 236. Bring this section up right before Fig. 1

Line 277-305:  Take it up right before Fig. 3

Line 323-340:  Take it up right before Fig. 4

The authors do not fully understand the reason of this action. In any case, this is the subject of the journal's rules and format. We hope that the Editors will adjust the figures' positions due to the required standards.

The analysis from the different datasets used shows some differences, what are the causes of the differences? Are these differences reasonable enough to affect long-term climate projections?

The reasons for the differences in ocean salinity trend estimates based on the different datasets used are partially listed in the penultimate paragraph of the section "Introduction". The impact of these differences on long-term climate projections is not the subject of this paper. However, it is safe to say that if a climate model poorly represents historical changes in the parameter under study, then the climate projections obtained on its basis have low reliability. Therefore, it is obvious that it is necessary to know the exact historical changes in the parameter under study. The North Atlantic salinity trend estimates obtained from several datasets in our paper have high reliability and will be useful in verifying climate model data in the historical period.

 A new section “Discussion” is necessary, to link your results to past studies

We agree with the remark. A new section, "Discussion", has been added to the revised version of the article.

Sincerely,

authors

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The author made modification according to the review comments and also answered relevant questions.

One suggestion:

  Please provide the address and date of  data acquisition.

Author Response

The author made modification according to the review comments and also answered relevant questions.

We thank the Reviewer for contributing in improvement of the paper.

One suggestion:

  Please provide the address and date of  data acquisition.

Thank you for your notice. We added this information into the Data Availability Statement (prior to the bibliography).

Reviewer 2 Report

Comments and Suggestions for Authors

Dear Authors,

Thanks for your reply, but it still doesn't address my suggestion.

Author Response

Dear Authors,

Thanks for your reply, but it still doesn't address my suggestion.

Dear Reviewer,

thank you for your contribution in revision of our paper. We are rewriting our responses to your previous comments with corrections. Additions are marked with green.

The author uses historical data to analyze the long-term trends and variations in North Atlantic salinity. However, under this thematic framework, many important factors that need to be discussed are overlooked. Here are the suggestions.

The authors thank the Reviewer for his helpful comments and valuable suggestions that contributed to improving the manuscript. The reviewer's comments are given in italics below, and responses to them are given in regular font.

1. The quality and rendering of the image are very poor and need to be redrawn.

Thank you for the comment. In the revised version of the paper, the figures have been changed.

2. Lines 131-134: The authors need to provide a well-justified scientific rationale for why the periods 1948-2018 and 1961-2011 were chosen for comparison.

Thank you for your comment. The revised version of the paper provides the rationale for choosing the periods under consideration. The choice of the first period (1948–2018) is related to the need to analyze the longest-term changes in the North Atlantic salinity based on available historical data. The choice of the second period (1961–2011) is due to the need to compare the longest time series based on the largest number of datasets. Our goal was to determine the general pattern of the long-term change in the North Atlantic salinity. Therefore, in this task, we were interested in the longest and most accessible data on salinity.

3. Section three is based on unclear period classifications, and the author has also not considered any variable parameters that might affect salinity distribution, such as characteristic watersheds and the distribution of salinity in the North Atlantic. Moreover, the author has not conducted further classifications for climate change features, such as the North Atlantic Oscillation and the Arctic Oscillation, which significantly impact salinity distribution. These aspects require detailed discussion.

We agree with the remark. Since we are interested in long-term trends in North Atlantic salinity obtained on the basis of several arrays of historical data, the article is devoted only to this issue. As the Reviewer correctly notes, the analysis of further classifications of climate change features is not carried out in the scope of this paper. We do this in order not to overload the presentation. Although the Reviewer is right that climate signals have a significant impact on the salinity distribution, we believe that these aspects require separate detailed consideration. However, the reviewer is right that climate signals have a significant impact on the salinity distribution. Therefore, we added links to the Discussion section to articles documenting the analysis of the North Atlantic Oscillation variability role on the salinity distribution in the North Atlantic.

4. Additionally, the authors need to consider the impact of pre- and post-salinity satellite launches on salinity observations, which is significant.

Thank you for the remark. The launch of satellites allows one to obtain unique data on the state of the ocean surface. Based on satellite platforms, it has become possible to study changes in salinity on the ocean surface. However, estimates of salinity on the ocean surface from Aquarius have only been available since 2011. However, estimates of salinity on the ocean surface are available only since 2010 (Vinogradova et al. 2019). This time interval is not yet sufficient still insufficient to analyze long-term changes in salinity. In addition, satellites provide global high-resolution observations of the ocean surface. It is also of interest to study the evolution of salinity of subsurface waters.  The corresponding clarification has been made in the revised text of the article. In addition, satellites provide global observations of the ocean surface with high resolution. This determines the high potential for using satellite products to study short-term variability in ocean surface salinity.

Sincerely,

authors

 

Reviewer 4 Report

Comments and Suggestions for Authors

Good luck

Author Response

The authors thank the Reviewer for helpful recommendations, which made it possible to improve the paper significantly.