North Atlantic sea level budget revisited

Song, Zhe; Cazenave, Anny; Llovel, William; Storto, Andrea; Bouih, Marie

doi:10.5194/os-22-2903-2026

Articles | Volume 22, issue 5

https://doi.org/10.5194/os-22-2903-2026

© Author(s) 2026. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/os-22-2903-2026

© Author(s) 2026. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 22, issue 5

Research article

|

25 Sep 2026

Research article |

| 25 Sep 2026

North Atlantic sea level budget revisited

Zhe Song, Anny Cazenave, William Llovel, Andrea Storto, and Marie Bouih

Download

Final revised paper (published on 25 Sep 2026)
Supplement to the final revised paper
Preprint (discussion started on 18 Feb 2026)

Interactive discussion

Status: closed

RC1:
'Comment on egusphere-2026-802', Anonymous Referee #1, 21 Apr 2026

attached supplement

Citation: https://doi.org/10.5194/egusphere-2026-802-RC1
- AC1:
  'Reply on RC1', Zhe Song, 11 May 2026
  See below the responses to Reviewer 1. We also attach the tracked revised manuscript called "Song_article_revised_tracked_11May2026" in order that the Reviewer can see our modifications.
  Responses to the Reviewers’ comments
  (in italics, bold)
  
  Reviewer 1
  Review for: “North Atlantic sea level budget revisited” by Zhe Song, Anny Cazenave1, William Llovel, Andrea Storto and Marie Bouih (https://doi.org/10.5194/egusphere-2026-802). This study examines the North Atlantic sea level budget from 2004-2022, comparing satellite altimetry data with estimates of steric and manometric contributions from various sources (ARGO floats, ocean reanalyses, and GRACE). While a complete sea level budget (altimetry balanced by steric and manometric components) was not achieved using GRACE data and any ARGO/reanalysis combinations, it was closed when using full-depth steric data from the CIGAR and ORA5S reanalyses with the CIGAR manometric component. The authors found that the deep steric component of CIGAR significantly contributed to this closure, and highlight potential issues with GRACE measurements, particularly in the Northeastern Atlantic. Even when the overall budget is balanced, significant regional anomalies in sea level trends persist. The paper is generally well-written and the figures are clear. However, the altimetry and GRACE data sets are not sufficiently cited. The numerous analyses presented can make it challenging to discern the main points. A table detailing the mean contributions from each component (altimetry, steric, deep steric, manometric) would improve clarity and facilitate comparison. The conclusions would be strengthened by more robust substantiation.
  A table detailing the mean contributions from each component (altimetry, steric, deep steric, manometric) would improve clarity and facilitate comparison.
  
  Answer: Two tables gathering the trend contribution of each component of the North Atlantic sea level budget have been added in the revised manuscript (new Table 3 for the steric 0-2000m case and new Table 4 for the steric full depth case)
  A key concern is the reliance of the manometric component conclusions on the validity of the CIGAR deep steric data, given the considerable differences observed between CIGAR and ORA5S estimates – potentially indicating differing underlying physical processes.
  
  Answer: In our previous article on the regional sea level budget (Bouih et al., Ocean sciences, 12, 1425-1440, https://doi.org/10.5194/os-21-1425-2025, 2025), we computed the manometric component over 2005-2019 using six different ocean reanalyses (G-CLORS, GLORYS, ORAS5, SODA, CIGAR and FOAM), as well as an ensemble mean of the first five (excluding FOAM because of spurious signal in the Southern Ocean). While some regional differences were observed between the different reanalyses, the CIGAR-based and ensemble mean manometric maps agreed well, including in the North Atlantic. Besides, as shown in the present revised manuscript, comparisons between ‘Altimetry minus Steric’ map (equivalent to the manometric component) and GRACE-based manometric in the North Atlantic clearly shows opposite signs (unlike with CIGAR or ensemble mean manometric), indicating a problem with GRACE.
  
  The observed warming and salinization in the deep western North Atlantic within CIGAR since 2004 should be validated using in-situ data, such as from existing moorings (e.g., see https://www.nature.com/articles/s43247-025-02170-y ).
  
  Answer: Yes indeed, it is what we plan to do, not only using mooring data at 26.5°N but also Deep Argo float measurements since there are a few available in the Northwest Atlantic. However, considering the short delay allowed to revise our manuscript, such an analysis could be achieved before the deadline. We will perform such a work in the near future, comparing in situ data not only with CIGAR but also other available reanalyses.
  We thank Reviewer 1 for the Chomiak et al’s article. The cooling trend observed at 26.5°N in the deep western Atlantic is a long-term trend since 1980. In our study, we show that deep ocean warming emerges around 2016 in the Northwest Atlantic (see for example Cazenave et al., Evidence of increased deep ocean warming from a sea level budget approach, Earth’s Future, 14, e2025EF007403. https://doi.org/10.1029/2025EF007403 2026), thus at the end of the record shown in Chomiak et al.
  
  Considering the potential impact of the derived changes in the manometric component on the Atlantic Meridional Overturning Circulation (AMOC) would enhance the paper's discussion.
  
  Answer: The regional manometric changes are closely inter-connected to the North Atlantic circulation variability. The manometric component mainly reflects redistribution of ocean mass by circulation and, thus, includes the barotropic signature of changes in the gyre circulation and/or
  
  AMOC-driven circulation. The AMOC variability can modify large-scale pressure gradients, western-boundary currents, return flows, etc., which may reflect into the east-west manometric dipole present in the North Atlantic, in the CIGAR reanalysis. However, our present analysis does
  
  not allow establishing causality. It is difficult to conclude whether the changes in manometric sea level will affect the AMOC variability, or this is among the causes of deep western North Atlantic warming and associated halosteric compensation, visible in CIGAR, because of changes
  
  in overturning and ventilation, including a possible AMOC slowdown. Causality and process-oriented analyses are beyond the scope of the present work and will be deepened in a follow-up study.
  --------------------
  Specific points:
  Page 5, line 117: the mascon solution is probably not from GFZ, but rather from GSFC?
  Answer: Corrected.
  
  Page 5, line 128: in the following no individual mascon solutions but only ensemble means are shown and discussed
  Answer : We have revised the wording.
  
  Page 8, line 201-203: how do the global mean values compare between the datasets?
  Answer : We added two tables (new Tables 3 and 4) in which are presented the global mean trends over the study period.
  
  Figure 1: what would be the corresponding ORA5S component?
  Answer : We do not consider ORAS5 manometric component in this study because the corresponding manometric dataset prepared for our previous study (Bouih et al., 2025) by our CNR colleagues (A. Storto and C. Yang) ends in 2019, thus does not cover our whole study time span (2004-2022).
  
  Figures 1, 2, and 3: what is the confidence level for the corresponding trends?
  Answer : Here, our discussion focuses more on the spatial distribution of trends for each dataset. For specific trend value, we compute it in section 3.4.
  
  Page 9/10, lines 224-228: phrasing potentially misleading: larger trends => GRACE show positive trends while CIGAR trends are negative? Most significant differences seem to be rather in the North East Atlantic than near Greenland.
  Answer : We have revised the wording.
  
  Page 13, lines 271-276: Why is it necessary for the deep steric component to align with the upper steric component?
  Answer : The corresponding sentence has been removed
  
  Page 14, lines 292-294: If the reliability of that dataset is questionable, consider excluding it from the study.
  Answer : This is exactly what is written in the text. We discarded the JAMSTEC data.
  
  Figure 4: might be an option to add the corresponding manometric curves
  Answer : We only discuss steric sea level change here. The manometric component curves are presented in Figures 9 and 10.
  
  Figure 6, caption: scatter plots of sea level trends in mm/year?, what does the normalization of the data include?
  Answer : The values presented in the scatter plots (Figure 6) are not the raw sea level trends in mm/yr, but rather the standardized values (Z-scores). The normalization of the data includes the following transformation applied to each spatial field independently:
  
  where x represents the local sea level trend, is the spatial mean of the trends over the North Atlantic region, and is the spatial standard deviation.
  
  Figure 7/8, caption: map of residual SL trends … im mm/year ?
  Answer : We added the "mm/yr" in the caption for clarity.
  
  Tables 1/2/3: The reported uncertainty may be unrealistically low, considering the uncertainties inherent in the contributing data sources.
  Answer : We only use the coefficient uncertainty of linear regression in our study; it might underestimate the uncertainty value. We will consider a more detailed method to compute the uncertainties in the future.
  
  Figures 9-13: very similar, hard to distinguish
  Answer : We added two tables in the revised version (new Tables 3 and 4) with the trend valuesof each component
  
  Page 31, line 575: steric sea level trends? Page 31, line 583: decreasing trend -> negative trend
  Answer : corrected
  
  Reviewer 2
  The study of the regional sea-level balance is essential for better identifying all the causes of the sea-level rise observed in recent decades. This study, which re-examines the sea-level balance in the North Atlantic over the 2004–2022 period by combining multiple data sources (satellite altimetry and gravimetry, in situ measurements, and ocean reanalyses), will thus enable the scientific community to better understand the imbalance in the sea level budget observed in this region (with significant residuals) by several studies, notably Bouih et al. (2025), which is likely due to an error in estimating the halosteric component in certain Argo float measurements. The authors assessed the impact of each product on the sea-level balance imbalance in the North Atlantic, while accounting for the warming of deep waters below 2,000 m (not sampled by Argo) using estimates derived from ocean reanalyses.
  The authors demonstrated the validity of sea-level measurements obtained via satellite altimetry (within the limits of measurement uncertainties) when they used the CIGAR reanalysis to estimate the barometric component (instead of GRACE) and the steric component of the ocean (instead of Argo data, in order to account for ocean depths greater than 2,000 m). They highlighted a slight decrease in the trend of the North Atlantic halosteric component over the 2010–2016 period, using Argo and CIGAR data for the first 2,000 meters. This decrease could reflect an actual increase in salinity in the region, although salinity measurement errors may have influenced the halosteric component over the entire period. Thus, this study shows that the contribution of deep water (below 2,000 m depth, estimated at 0.62 ± 0.04 mm/year by CIGAR data) to the sea level balance in the North Atlantic is not negligible and must be taken into account to improve the consistency of this balance in the region. Including this factor reduces the residuals of the sea level balance in the North Atlantic by 30% with the GRACE manometric component and by 50% with CIGAR data (and full-depth CIGAR data for the steric component). In the latter case, the average residual trend over the North Atlantic is less than 1 mm/yr, which is of the same order of magnitude as the regional trend error in the gridded altimetry data.
  This study is important because it not only makes a significant contribution to our understanding of the regional sea-level budget in the North Atlantic, but also highlights the need to improve deep-water observations and raises open scientific questions for future research, the answers to which will contribute to a better understanding of the processes involved in studying the regional sea-level budget beyond the North Atlantic.
  This article is very well written and structured, and the results are clear and well analyzed. However, please find attached a list of minor questions to which we would like to receive responses from the authors.
  
  Isn't this extrapolation from 2016 to 2022 a bit rough, given that several studies show an acceleration in the ocean's mass component over the past two decades? Even if the global average trend in ocean mass has since been removed, isn’t there another way to complete the data through 2022?
  
  Answer : The GRD signal is very small, thus a linear extrapolation seems acceptable. Reproducing the Adhikari et al’s study is clearly beyond the scope of the present study.
  
  Give a reference
  
  Answer: This is a very classical procedure
  Why do you use mascons instead of spherical harmonics solutions ?
  
  Answer : In our previous study on the regional sea level budget (Bouih et al., Ocean sciences, 12, 1425-1440, https://doi.org/10.5194/os-21-1425-2025, 2025), we compared the mascon solutions with spherical harmonic solutions and found that the latter are very noisy with the well-known north-south stripes very apparent, especially in the North Atlantic. Thus we prefer to use the mascon solutions.
  
  What might explain the difference in trends between the eastern and western parts of the basin observed by CIGAR? Do you have any ideas?
  
  Answer : This remains an issue to clarify in the future. The transitions seems to be correlated with the bathymetry, in particular the location of the Mid-Atlantic ridge.
  
  Although the objective of the study is to verify the closure of the sea-level budget in the North Atlantic, it would also be interesting for the reader to gain an appreciation of each of the contributions from the regional components of sea level in terms of trend (i.e., mass change and thermosteric and halosteric components down to a depth of 2,000 m) in this region, similar to what you did for the steric component of the deep ocean (which is 0.62 ± 0.04 mm/yr according to CIGAR data).
  
  Answer : Two tables have been added with the trend values of each component of the budget.
  
  Add a “Conclusion” section or make sure to clearly indicate it in the discussion
  
  Answer : The discussion is now strengthened with a few concluding sentences.
  
  Citation: https://doi.org/10.5194/egusphere-2026-802-AC1
RC2:
'Comment on egusphere-2026-802', Anonymous Referee #2, 22 Apr 2026

The study of the regional sea-level balance is essential for better identifying all the causes of the sea-level rise observed in recent decades. This study, which re-examines the sea-level balance in the North Atlantic over the 2004–2022 period by combining multiple data sources (satellite altimetry and gravimetry, in situ measurements, and ocean reanalyses), will thus enable the scientific community to better understand the imbalance in the sea level buget observed in this region (with significant residuals) by several studies, notably Bouih et al. (2025), which is likely due to an error in estimating the halosteric component in certain Argo float measurements. The authors assessed the impact of each product on the sea-level balance imbalance in the North Atlantic, while accounting for the warming of deep waters below 2,000 m (not sampled by Argo) using estimates derived from ocean reanalyses.
The authors demonstrated the validity of sea-level measurements obtained via satellite altimetry (within the limits of measurement uncertainties) when they used the CIGAR reanalysis to estimate the barometric component (instead of GRACE) and the steric component of the ocean (instead of Argo data, in order to account for ocean depths greater than 2,000 m). They highlighted a slight decrease in the trend of the North Atlantic halosteric component over the 2010–2016 period, using Argo and CIGAR data for the first 2,000 meters. This decrease could reflect an actual increase in salinity in the region, although salinity measurement errors may have influenced the halosteric component over the entire period. Thus, this study shows that the contribution of deep water (below 2,000 m depth, estimated at 0.62 ± 0.04 mm/year by CIGAR data) to the sea level balance in the North Atlantic is not negligible and must be taken into account to improve the consistency of this balance in the region. Including this factor reduces the residuals of the sea level balance in the North Atlantic by 30% with the GRACE manometric component and by 50% with CIGAR data (and full-depth CIGAR data for the steric component). In the latter case, the average residual trend over the North Atlantic is less than 1 mm/yr, which is of the same order of magnitude as the regional trend error in the gridded altimetry data.
This study is important because it not only makes a significant contribution to our understanding of the regional sea-level budget in the North Atlantic, but also highlights the need to improve deep-water observations and raises open scientific questions for future research, the answers to which will contribute to a better understanding of the processes involved in studying the regional sea-level budget beyond the North Atlantic.
This article is very well written and structured, and the results are clear and well analyzed. However, please find attached a list of minor questions to which we would like to receive responses from the authors.

Citation: https://doi.org/10.5194/egusphere-2026-802-RC2
- AC3:
  'Reply on RC2', Zhe Song, 11 May 2026
  See below the responses to Reviewer 2. We also attach the tracked revised manuscript called "Song_article_revised_tracked_11May2026" in order that the Reviewer can see our modifications.
  Responses to the Reviewers’ comments
  (in italics, bold)
  
  Reviewer 1
  Review for: “North Atlantic sea level budget revisited” by Zhe Song, Anny Cazenave1, William Llovel, Andrea Storto and Marie Bouih (https://doi.org/10.5194/egusphere-2026-802). This study examines the North Atlantic sea level budget from 2004-2022, comparing satellite altimetry data with estimates of steric and manometric contributions from various sources (ARGO floats, ocean reanalyses, and GRACE). While a complete sea level budget (altimetry balanced by steric and manometric components) was not achieved using GRACE data and any ARGO/reanalysis combinations, it was closed when using full-depth steric data from the CIGAR and ORA5S reanalyses with the CIGAR manometric component. The authors found that the deep steric component of CIGAR significantly contributed to this closure, and highlight potential issues with GRACE measurements, particularly in the Northeastern Atlantic. Even when the overall budget is balanced, significant regional anomalies in sea level trends persist. The paper is generally well-written and the figures are clear. However, the altimetry and GRACE data sets are not sufficiently cited. The numerous analyses presented can make it challenging to discern the main points. A table detailing the mean contributions from each component (altimetry, steric, deep steric, manometric) would improve clarity and facilitate comparison. The conclusions would be strengthened by more robust substantiation.
  A table detailing the mean contributions from each component (altimetry, steric, deep steric, manometric) would improve clarity and facilitate comparison.
  
  Answer: Two tables gathering the trend contribution of each component of the North Atlantic sea level budget have been added in the revised manuscript (new Table 3 for the steric 0-2000m case and new Table 4 for the steric full depth case)
  A key concern is the reliance of the manometric component conclusions on the validity of the CIGAR deep steric data, given the considerable differences observed between CIGAR and ORA5S estimates – potentially indicating differing underlying physical processes.
  
  Answer: In our previous article on the regional sea level budget (Bouih et al., Ocean sciences, 12, 1425-1440, https://doi.org/10.5194/os-21-1425-2025, 2025), we computed the manometric component over 2005-2019 using six different ocean reanalyses (G-CLORS, GLORYS, ORAS5, SODA, CIGAR and FOAM), as well as an ensemble mean of the first five (excluding FOAM because of spurious signal in the Southern Ocean). While some regional differences were observed between the different reanalyses, the CIGAR-based and ensemble mean manometric maps agreed well, including in the North Atlantic. Besides, as shown in the present revised manuscript, comparisons between ‘Altimetry minus Steric’ map (equivalent to the manometric component) and GRACE-based manometric in the North Atlantic clearly shows opposite signs (unlike with CIGAR or ensemble mean manometric), indicating a problem with GRACE.
  
  The observed warming and salinization in the deep western North Atlantic within CIGAR since 2004 should be validated using in-situ data, such as from existing moorings (e.g., see https://www.nature.com/articles/s43247-025-02170-y ).
  
  Answer: Yes indeed, it is what we plan to do, not only using mooring data at 26.5°N but also Deep Argo float measurements since there are a few available in the Northwest Atlantic. However, considering the short delay allowed to revise our manuscript, such an analysis could be achieved before the deadline. We will perform such a work in the near future, comparing in situ data not only with CIGAR but also other available reanalyses.
  We thank Reviewer 1 for the Chomiak et al’s article. The cooling trend observed at 26.5°N in the deep western Atlantic is a long-term trend since 1980. In our study, we show that deep ocean warming emerges around 2016 in the Northwest Atlantic (see for example Cazenave et al., Evidence of increased deep ocean warming from a sea level budget approach, Earth’s Future, 14, e2025EF007403. https://doi.org/10.1029/2025EF007403 2026), thus at the end of the record shown in Chomiak et al.
  
  Considering the potential impact of the derived changes in the manometric component on the Atlantic Meridional Overturning Circulation (AMOC) would enhance the paper's discussion.
  
  Answer: The regional manometric changes are closely inter-connected to the North Atlantic circulation variability. The manometric component mainly reflects redistribution of ocean mass by circulation and, thus, includes the barotropic signature of changes in the gyre circulation and/or
  
  AMOC-driven circulation. The AMOC variability can modify large-scale pressure gradients, western-boundary currents, return flows, etc., which may reflect into the east-west manometric dipole present in the North Atlantic, in the CIGAR reanalysis. However, our present analysis does
  
  not allow establishing causality. It is difficult to conclude whether the changes in manometric sea level will affect the AMOC variability, or this is among the causes of deep western North Atlantic warming and associated halosteric compensation, visible in CIGAR, because of changes
  
  in overturning and ventilation, including a possible AMOC slowdown. Causality and process-oriented analyses are beyond the scope of the present work and will be deepened in a follow-up study.
  --------------------
  Specific points:
  Page 5, line 117: the mascon solution is probably not from GFZ, but rather from GSFC?
  Answer: Corrected.
  
  Page 5, line 128: in the following no individual mascon solutions but only ensemble means are shown and discussed
  Answer : We have revised the wording.
  
  Page 8, line 201-203: how do the global mean values compare between the datasets?
  Answer : We added two tables (new Tables 3 and 4) in which are presented the global mean trends over the study period.
  
  Figure 1: what would be the corresponding ORA5S component?
  Answer : We do not consider ORAS5 manometric component in this study because the corresponding manometric dataset prepared for our previous study (Bouih et al., 2025) by our CNR colleagues (A. Storto and C. Yang) ends in 2019, thus does not cover our whole study time span (2004-2022).
  
  Figures 1, 2, and 3: what is the confidence level for the corresponding trends?
  Answer : Here, our discussion focuses more on the spatial distribution of trends for each dataset. For specific trend value, we compute it in section 3.4.
  
  Page 9/10, lines 224-228: phrasing potentially misleading: larger trends => GRACE show positive trends while CIGAR trends are negative? Most significant differences seem to be rather in the North East Atlantic than near Greenland.
  Answer : We have revised the wording.
  
  Page 13, lines 271-276: Why is it necessary for the deep steric component to align with the upper steric component?
  Answer : The corresponding sentence has been removed
  
  Page 14, lines 292-294: If the reliability of that dataset is questionable, consider excluding it from the study.
  Answer : This is exactly what is written in the text. We discarded the JAMSTEC data.
  
  Figure 4: might be an option to add the corresponding manometric curves
  Answer : We only discuss steric sea level change here. The manometric component curves are presented in Figures 9 and 10.
  
  Figure 6, caption: scatter plots of sea level trends in mm/year?, what does the normalization of the data include?
  Answer : The values presented in the scatter plots (Figure 6) are not the raw sea level trends in mm/yr, but rather the standardized values (Z-scores). The normalization of the data includes the following transformation applied to each spatial field independently:
  
  where x represents the local sea level trend, is the spatial mean of the trends over the North Atlantic region, and is the spatial standard deviation.
  
  Figure 7/8, caption: map of residual SL trends … im mm/year ?
  Answer : We added the "mm/yr" in the caption for clarity.
  
  Tables 1/2/3: The reported uncertainty may be unrealistically low, considering the uncertainties inherent in the contributing data sources.
  Answer : We only use the coefficient uncertainty of linear regression in our study; it might underestimate the uncertainty value. We will consider a more detailed method to compute the uncertainties in the future.
  
  Figures 9-13: very similar, hard to distinguish
  Answer : We added two tables in the revised version (new Tables 3 and 4) with the trend valuesof each component
  
  Page 31, line 575: steric sea level trends? Page 31, line 583: decreasing trend -> negative trend
  Answer : corrected
  
  Reviewer 2
  The study of the regional sea-level balance is essential for better identifying all the causes of the sea-level rise observed in recent decades. This study, which re-examines the sea-level balance in the North Atlantic over the 2004–2022 period by combining multiple data sources (satellite altimetry and gravimetry, in situ measurements, and ocean reanalyses), will thus enable the scientific community to better understand the imbalance in the sea level budget observed in this region (with significant residuals) by several studies, notably Bouih et al. (2025), which is likely due to an error in estimating the halosteric component in certain Argo float measurements. The authors assessed the impact of each product on the sea-level balance imbalance in the North Atlantic, while accounting for the warming of deep waters below 2,000 m (not sampled by Argo) using estimates derived from ocean reanalyses.
  The authors demonstrated the validity of sea-level measurements obtained via satellite altimetry (within the limits of measurement uncertainties) when they used the CIGAR reanalysis to estimate the barometric component (instead of GRACE) and the steric component of the ocean (instead of Argo data, in order to account for ocean depths greater than 2,000 m). They highlighted a slight decrease in the trend of the North Atlantic halosteric component over the 2010–2016 period, using Argo and CIGAR data for the first 2,000 meters. This decrease could reflect an actual increase in salinity in the region, although salinity measurement errors may have influenced the halosteric component over the entire period. Thus, this study shows that the contribution of deep water (below 2,000 m depth, estimated at 0.62 ± 0.04 mm/year by CIGAR data) to the sea level balance in the North Atlantic is not negligible and must be taken into account to improve the consistency of this balance in the region. Including this factor reduces the residuals of the sea level balance in the North Atlantic by 30% with the GRACE manometric component and by 50% with CIGAR data (and full-depth CIGAR data for the steric component). In the latter case, the average residual trend over the North Atlantic is less than 1 mm/yr, which is of the same order of magnitude as the regional trend error in the gridded altimetry data.
  This study is important because it not only makes a significant contribution to our understanding of the regional sea-level budget in the North Atlantic, but also highlights the need to improve deep-water observations and raises open scientific questions for future research, the answers to which will contribute to a better understanding of the processes involved in studying the regional sea-level budget beyond the North Atlantic.
  This article is very well written and structured, and the results are clear and well analyzed. However, please find attached a list of minor questions to which we would like to receive responses from the authors.
  
  Isn't this extrapolation from 2016 to 2022 a bit rough, given that several studies show an acceleration in the ocean's mass component over the past two decades? Even if the global average trend in ocean mass has since been removed, isn’t there another way to complete the data through 2022?
  
  Answer : The GRD signal is very small, thus a linear extrapolation seems acceptable. Reproducing the Adhikari et al’s study is clearly beyond the scope of the present study.
  
  Give a reference
  
  Answer: This is a very classical procedure
  Why do you use mascons instead of spherical harmonics solutions ?
  
  Answer : In our previous study on the regional sea level budget (Bouih et al., Ocean sciences, 12, 1425-1440, https://doi.org/10.5194/os-21-1425-2025, 2025), we compared the mascon solutions with spherical harmonic solutions and found that the latter are very noisy with the well-known north-south stripes very apparent, especially in the North Atlantic. Thus we prefer to use the mascon solutions.
  
  What might explain the difference in trends between the eastern and western parts of the basin observed by CIGAR? Do you have any ideas?
  
  Answer : This remains an issue to clarify in the future. The transitions seems to be correlated with the bathymetry, in particular the location of the Mid-Atlantic ridge.
  
  Although the objective of the study is to verify the closure of the sea-level budget in the North Atlantic, it would also be interesting for the reader to gain an appreciation of each of the contributions from the regional components of sea level in terms of trend (i.e., mass change and thermosteric and halosteric components down to a depth of 2,000 m) in this region, similar to what you did for the steric component of the deep ocean (which is 0.62 ± 0.04 mm/yr according to CIGAR data).
  
  Answer : Two tables have been added with the trend values of each component of the budget.
  
  Add a “Conclusion” section or make sure to clearly indicate it in the discussion
  
  Answer : The discussion is now strengthened with a few concluding sentences.
  
  Citation: https://doi.org/10.5194/egusphere-2026-802-AC3
AC2:
'Reply on RC2', Zhe Song, 11 May 2026
See below the responses to Reviewer 2. We also attach the tracked revised manuscript called "Song_article_revised_tracked_11May2026" in order that the Reviewer can see our modifications.
Responses to the Reviewers’ comments
(in italics, bold)

Reviewer 1
Review for: “North Atlantic sea level budget revisited” by Zhe Song, Anny Cazenave1, William Llovel, Andrea Storto and Marie Bouih (https://doi.org/10.5194/egusphere-2026-802). This study examines the North Atlantic sea level budget from 2004-2022, comparing satellite altimetry data with estimates of steric and manometric contributions from various sources (ARGO floats, ocean reanalyses, and GRACE). While a complete sea level budget (altimetry balanced by steric and manometric components) was not achieved using GRACE data and any ARGO/reanalysis combinations, it was closed when using full-depth steric data from the CIGAR and ORA5S reanalyses with the CIGAR manometric component. The authors found that the deep steric component of CIGAR significantly contributed to this closure, and highlight potential issues with GRACE measurements, particularly in the Northeastern Atlantic. Even when the overall budget is balanced, significant regional anomalies in sea level trends persist. The paper is generally well-written and the figures are clear. However, the altimetry and GRACE data sets are not sufficiently cited. The numerous analyses presented can make it challenging to discern the main points. A table detailing the mean contributions from each component (altimetry, steric, deep steric, manometric) would improve clarity and facilitate comparison. The conclusions would be strengthened by more robust substantiation.
A table detailing the mean contributions from each component (altimetry, steric, deep steric, manometric) would improve clarity and facilitate comparison.

Answer: Two tables gathering the trend contribution of each component of the North Atlantic sea level budget have been added in the revised manuscript (new Table 3 for the steric 0-2000m case and new Table 4 for the steric full depth case)
A key concern is the reliance of the manometric component conclusions on the validity of the CIGAR deep steric data, given the considerable differences observed between CIGAR and ORA5S estimates – potentially indicating differing underlying physical processes.

Answer: In our previous article on the regional sea level budget (Bouih et al., Ocean sciences, 12, 1425-1440, https://doi.org/10.5194/os-21-1425-2025, 2025), we computed the manometric component over 2005-2019 using six different ocean reanalyses (G-CLORS, GLORYS, ORAS5, SODA, CIGAR and FOAM), as well as an ensemble mean of the first five (excluding FOAM because of spurious signal in the Southern Ocean). While some regional differences were observed between the different reanalyses, the CIGAR-based and ensemble mean manometric maps agreed well, including in the North Atlantic. Besides, as shown in the present revised manuscript, comparisons between ‘Altimetry minus Steric’ map (equivalent to the manometric component) and GRACE-based manometric in the North Atlantic clearly shows opposite signs (unlike with CIGAR or ensemble mean manometric), indicating a problem with GRACE.

The observed warming and salinization in the deep western North Atlantic within CIGAR since 2004 should be validated using in-situ data, such as from existing moorings (e.g., see https://www.nature.com/articles/s43247-025-02170-y ).

Answer: Yes indeed, it is what we plan to do, not only using mooring data at 26.5°N but also Deep Argo float measurements since there are a few available in the Northwest Atlantic. However, considering the short delay allowed to revise our manuscript, such an analysis could be achieved before the deadline. We will perform such a work in the near future, comparing in situ data not only with CIGAR but also other available reanalyses.
We thank Reviewer 1 for the Chomiak et al’s article. The cooling trend observed at 26.5°N in the deep western Atlantic is a long-term trend since 1980. In our study, we show that deep ocean warming emerges around 2016 in the Northwest Atlantic (see for example Cazenave et al., Evidence of increased deep ocean warming from a sea level budget approach, Earth’s Future, 14, e2025EF007403. https://doi.org/10.1029/2025EF007403 2026), thus at the end of the record shown in Chomiak et al.

Considering the potential impact of the derived changes in the manometric component on the Atlantic Meridional Overturning Circulation (AMOC) would enhance the paper's discussion.

Answer: The regional manometric changes are closely inter-connected to the North Atlantic circulation variability. The manometric component mainly reflects redistribution of ocean mass by circulation and, thus, includes the barotropic signature of changes in the gyre circulation and/or

AMOC-driven circulation. The AMOC variability can modify large-scale pressure gradients, western-boundary currents, return flows, etc., which may reflect into the east-west manometric dipole present in the North Atlantic, in the CIGAR reanalysis. However, our present analysis does

not allow establishing causality. It is difficult to conclude whether the changes in manometric sea level will affect the AMOC variability, or this is among the causes of deep western North Atlantic warming and associated halosteric compensation, visible in CIGAR, because of changes

in overturning and ventilation, including a possible AMOC slowdown. Causality and process-oriented analyses are beyond the scope of the present work and will be deepened in a follow-up study.
--------------------
Specific points:
Page 5, line 117: the mascon solution is probably not from GFZ, but rather from GSFC?
Answer: Corrected.

Page 5, line 128: in the following no individual mascon solutions but only ensemble means are shown and discussed
Answer : We have revised the wording.

Page 8, line 201-203: how do the global mean values compare between the datasets?
Answer : We added two tables (new Tables 3 and 4) in which are presented the global mean trends over the study period.

Figure 1: what would be the corresponding ORA5S component?
Answer : We do not consider ORAS5 manometric component in this study because the corresponding manometric dataset prepared for our previous study (Bouih et al., 2025) by our CNR colleagues (A. Storto and C. Yang) ends in 2019, thus does not cover our whole study time span (2004-2022).

Figures 1, 2, and 3: what is the confidence level for the corresponding trends?
Answer : Here, our discussion focuses more on the spatial distribution of trends for each dataset. For specific trend value, we compute it in section 3.4.

Page 9/10, lines 224-228: phrasing potentially misleading: larger trends => GRACE show positive trends while CIGAR trends are negative? Most significant differences seem to be rather in the North East Atlantic than near Greenland.
Answer : We have revised the wording.

Page 13, lines 271-276: Why is it necessary for the deep steric component to align with the upper steric component?
Answer : The corresponding sentence has been removed

Page 14, lines 292-294: If the reliability of that dataset is questionable, consider excluding it from the study.
Answer : This is exactly what is written in the text. We discarded the JAMSTEC data.

Figure 4: might be an option to add the corresponding manometric curves
Answer : We only discuss steric sea level change here. The manometric component curves are presented in Figures 9 and 10.

Figure 6, caption: scatter plots of sea level trends in mm/year?, what does the normalization of the data include?
Answer : The values presented in the scatter plots (Figure 6) are not the raw sea level trends in mm/yr, but rather the standardized values (Z-scores). The normalization of the data includes the following transformation applied to each spatial field independently:

where x represents the local sea level trend, is the spatial mean of the trends over the North Atlantic region, and is the spatial standard deviation.

Figure 7/8, caption: map of residual SL trends … im mm/year ?
Answer : We added the "mm/yr" in the caption for clarity.

Tables 1/2/3: The reported uncertainty may be unrealistically low, considering the uncertainties inherent in the contributing data sources.
Answer : We only use the coefficient uncertainty of linear regression in our study; it might underestimate the uncertainty value. We will consider a more detailed method to compute the uncertainties in the future.

Figures 9-13: very similar, hard to distinguish
Answer : We added two tables in the revised version (new Tables 3 and 4) with the trend valuesof each component

Page 31, line 575: steric sea level trends? Page 31, line 583: decreasing trend -> negative trend
Answer : corrected

Reviewer 2
The study of the regional sea-level balance is essential for better identifying all the causes of the sea-level rise observed in recent decades. This study, which re-examines the sea-level balance in the North Atlantic over the 2004–2022 period by combining multiple data sources (satellite altimetry and gravimetry, in situ measurements, and ocean reanalyses), will thus enable the scientific community to better understand the imbalance in the sea level budget observed in this region (with significant residuals) by several studies, notably Bouih et al. (2025), which is likely due to an error in estimating the halosteric component in certain Argo float measurements. The authors assessed the impact of each product on the sea-level balance imbalance in the North Atlantic, while accounting for the warming of deep waters below 2,000 m (not sampled by Argo) using estimates derived from ocean reanalyses.
The authors demonstrated the validity of sea-level measurements obtained via satellite altimetry (within the limits of measurement uncertainties) when they used the CIGAR reanalysis to estimate the barometric component (instead of GRACE) and the steric component of the ocean (instead of Argo data, in order to account for ocean depths greater than 2,000 m). They highlighted a slight decrease in the trend of the North Atlantic halosteric component over the 2010–2016 period, using Argo and CIGAR data for the first 2,000 meters. This decrease could reflect an actual increase in salinity in the region, although salinity measurement errors may have influenced the halosteric component over the entire period. Thus, this study shows that the contribution of deep water (below 2,000 m depth, estimated at 0.62 ± 0.04 mm/year by CIGAR data) to the sea level balance in the North Atlantic is not negligible and must be taken into account to improve the consistency of this balance in the region. Including this factor reduces the residuals of the sea level balance in the North Atlantic by 30% with the GRACE manometric component and by 50% with CIGAR data (and full-depth CIGAR data for the steric component). In the latter case, the average residual trend over the North Atlantic is less than 1 mm/yr, which is of the same order of magnitude as the regional trend error in the gridded altimetry data.
This study is important because it not only makes a significant contribution to our understanding of the regional sea-level budget in the North Atlantic, but also highlights the need to improve deep-water observations and raises open scientific questions for future research, the answers to which will contribute to a better understanding of the processes involved in studying the regional sea-level budget beyond the North Atlantic.
This article is very well written and structured, and the results are clear and well analyzed. However, please find attached a list of minor questions to which we would like to receive responses from the authors.

Isn't this extrapolation from 2016 to 2022 a bit rough, given that several studies show an acceleration in the ocean's mass component over the past two decades? Even if the global average trend in ocean mass has since been removed, isn’t there another way to complete the data through 2022?

Answer : The GRD signal is very small, thus a linear extrapolation seems acceptable. Reproducing the Adhikari et al’s study is clearly beyond the scope of the present study.

Give a reference

Answer: This is a very classical procedure
Why do you use mascons instead of spherical harmonics solutions ?

Answer : In our previous study on the regional sea level budget (Bouih et al., Ocean sciences, 12, 1425-1440, https://doi.org/10.5194/os-21-1425-2025, 2025), we compared the mascon solutions with spherical harmonic solutions and found that the latter are very noisy with the well-known north-south stripes very apparent, especially in the North Atlantic. Thus we prefer to use the mascon solutions.

What might explain the difference in trends between the eastern and western parts of the basin observed by CIGAR? Do you have any ideas?

Answer : This remains an issue to clarify in the future. The transitions seems to be correlated with the bathymetry, in particular the location of the Mid-Atlantic ridge.

Although the objective of the study is to verify the closure of the sea-level budget in the North Atlantic, it would also be interesting for the reader to gain an appreciation of each of the contributions from the regional components of sea level in terms of trend (i.e., mass change and thermosteric and halosteric components down to a depth of 2,000 m) in this region, similar to what you did for the steric component of the deep ocean (which is 0.62 ± 0.04 mm/yr according to CIGAR data).

Answer : Two tables have been added with the trend values of each component of the budget.

Add a “Conclusion” section or make sure to clearly indicate it in the discussion

Answer : The discussion is now strengthened with a few concluding sentences.
Citation: https://doi.org/10.5194/egusphere-2026-802-AC2

Peer review completion

AR – Author's response | RR – Referee report | ED – Editor decision | EF – Editorial file upload

AR by Zhe Song on behalf of the Authors (11 May 2026) Author's response Author's tracked changes Manuscript

ED: Referee Nomination & Report Request started (12 May 2026) by Matjaz Licer

RR by Anonymous Referee #3 (07 Jun 2026)

RR by Anonymous Referee #4 (28 Jun 2026)

Suggestions for revision or reasons for rejection

The manuscript entitled “North Atlantic Sea Level Budget Revisited” addresses the important question of closing the regional sea level budget in the North Atlantic during the 2004–2022 period. The study builds upon and extends the work of Bouih et al. (2025). The authors identify the significant contributions of deep-ocean steric sea level changes below 2000 m and the manometric sea level derived from the CIGAR reanalysis. They also highlight potential errors in the GRACE and GRACE-FO datasets. Overall, this study makes a significant contribution to improving our understanding of the contemporary sea level budget in the North Atlantic.
The manuscript is generally well written and clearly presented. I therefore recommend minor revisions to clarify several statements and improve readability for the benefit of the readers. The detailed comments are below:

1. Line 61-63, the sentence may be rephased to be :”At the global scale, the two main contributors to the global mean sea level rise are ocean mass (known as barystatic sea level, Gregory et al., 2019) and thermosteric change.”
2. Line 96, typo. Should be “Section 4”.
3. Section 2.1.4. Bouih et al. (2025) evaluated six ocean reanalysis products. In this study, only CIGAR and ORAS5 were selected. Please provide a clearer justification for choosing ORAS5 over the other products considered by Bouih et al. (2025), such as GLORYS, C-GLORS, SODA, and FOAM. Can you summarize the major differences between CIGAR and ORAS5 in terms of the vertical physics and data-assimilation schemes since you indicated that they may have important implications for the results presented in Figure 3.
4. Section 2.2.1. You derived CIGAR manometric sea level. Why did not you compute ORAS5 manometric sea level? Please include some explanation.
5. Line 200. “this least-squares fit” should be “the linear least-squares fit”.
6. Line 210, “the law of error propagation, considering all terms of the sea level budget”. Can you define the residuals and explain explicitly how the error is calculated? For example ,the residual = altimeter sea level - manometric sea level -steric sea level. Assuming the uncertainties of these three terms are independent, the uncertainty of the residual is calculated by adding the individual uncertainties in quadrature.
7. Line 228. “with the most significant difference observed near Greenland.” To me, the most significant difference in Figure 1c is in the northeast Atlantic near Spain and France.
8. Line 236 and 241, 251. Typo. “Figure 3a-c” should be Figure 2a-c.
9. Line 275-276. “triggered by vertical physics and, to a lesser extent, vertical error correlations embedded in the 3D-Var scheme”. Could you elaborate more on this?
10. Line306-307, can you cite some literatures to backup your suspicion “Given the unrealistically stationary thermosteric signal observed in ORAS5 below 2000m”. What is the current observational consensus regarding deep-ocean warming below 2000 m depth? Please provide a brief summary and relevant references.
11. Line 336. “Figure 6a, b shows the scatter plots between SIO gridpoint steric trend” This statement is inconsistent with the caption of Figure 6. Please double check and clarify if Figure 6 shows trends or normalized trends.
12. Lines 339,344, 345, 350. The correlation needs to be quantified using correlation coefficients and tests of statistical significance.
13. Figure 6. Please clarify what you plotted in Figure 6. What is the unit of x and y axis? Do you normalize the trend? If so, how do you normalize the trend?
14. Line 617-618. Maybe rephrase the sentence to be “What is the impact of using EN4 (without applying a salinity drift correction) in the assimilation procedure on the CIGAR steric component estimates?”

Referee Report: PDF

Hide

RR by Anonymous Referee #5 (28 Jun 2026)

Suggestions for revision or reasons for rejection

The manuscript examines the North Atlantic sea level budget for 2004-2022 time span, comparing estimates of manometric and steric contributions from GRACE, Argo floats and ocean reanalysis with satellite altimetry data. The authors find that deep ocean steric contribution is important for closure of the North Atlantic sea level budget, that the main contribution to the North Atlantic residual trend comes from errors on the regional components, rather than from the errors in the global mean trends, discover a trend in the Argo-based (SIO dataset) halosteric component from 2012, which according to them suggest a real physical increase in salinity of the North Atlantic, and conclude that errors in GRACE data are the main cause for the difference between the manometric components from GRACE and CIGAR reanalysis and one of the culprits for the inability to close the North Atlantic sea level budget with manometric contribution from GRACE.

Overall, I find this manuscript to be promising and scientifically relevant, and I believe it has the potential to make a valuable contribution to the field. However, some of the conclusions are not fully justified by the presented results and the manuscript would benefit from some improvements in its structure and focus. I believe the manuscript is worthy of publication, but only after substantial revisions addressing the issues listed below.

Main issues

1. The conclusion that GRACE errors are solely responsible for the difference between GRACE and CIGAR-based manometric component and the reason why North Atlantic sea level budget cannot be closed with GRACE data might be true, but the results presented in the manuscript are not enough to support it. The authors mainly base it on the comparison between the total sea level from altimetry, steric sea level from Argo (SIO dataset, upper 2000 m only), and the manometric component from GRACE and CIGAR reanalysis. They treat the sum of the manometric and upper ocean steric component as equivalent to altimetry and conclude that the manometric component that provides a better match is better. However, they also show that the deep ocean steric component is important in North Atlantic and cannot be ignored, which makes this comparison and their conclusion flawed. Furthermore, the CIGAR-manometric component is based on a reanalysis that assimilates Argo data (from EN4), and the authors themselves demonstrate that SIO and EN4 are very similar, so CIGAR and SIO cannot be considered entirely independent. The authors should account for the deep ocean contribution into this comparison and add discussions about other possible causes for the differences to strengthen their conclusion.

2. Some references to datasets are missing from the manuscript. What is the satellite altimetry dataset the authors use? I can guess which one it is, but that dataset has a name (DT2021 is just the version) and a DOI, which the authors do not provide. The link the authors provide is not to the dataset but to the data store landing page. GRACE datasets also do not have references. All datasets included in this study should be clearly named and referenced.

3. In the Data section the authors present three Argo datasets (SIO, EN4 and JAMSTEC) and two reanalysis products (CIGAR and ORAS5), but then throughout the manuscript they pick and choose which of them are used for different comparisons without giving reasons why. They also use GRACE mascon solutions and not spherical harmonics without explaining why. Previous reviewers already asked about some of these choices and the authors responded in the rebuttal but none of these answers were included into the revised version of the manuscript. Apart from the question already asked in the first review, I would like to know why do authors use only the SIO dataset for the steric component in Sect. 3.1 and 3.3? They explain why they exclude JAMSTEC in Sect. 3.2 (which is also not enough as it should be explained before the first time they exclude datasets), but they never explain why they do not use EN4. It often makes sense to focus on specific datasets, but then that decision needs to be explained, otherwise readers might wonder what would the result be with the other dataset and why it was not shown.

4. Sect. 3.5, which is the most important in the manuscript, as it is showing attempts to close the sea level budget using different datasets, is not well organized and it contains too much irrelevant information, while not showing important information in a clear way. Four figures (Figs. 9, 10, 12, 13) show a side by side comparison of the sea level budget with the steric component from SIO and CIGAR, which has very minor differences. At the same time, the difference with and without the inclusion of the deep ocean, and the difference between GRACE and CIGAR manometric, which are relevant for the main conclusions of the manuscript, can only be analysed by comparing multiple figures, which is rather hard. I would strongly advise the authors to reduce the number of figures in this section by merging Figs. 9 and 10, as well as Figs. 12 and 13, after picking one of the steric components. Figures with the other steric component can be shown in the appendix for completeness.

5. The structure of the manuscript is overall not bad but it could definitely use some improvements. Originally, the authors first presented individual components (Sect. 3.1-3.2), and then compared their combinations (Sect. 3.3-3.5), which is a good structure, but in the revised manuscript they added a comparison with altimetry minus steric component into Sect. 3.1, which does not make sense. I stronly advise to remove this comparison and go back to the original structure. As the conclusions about GRACE vs. CIGAR manometric component cannot be drawn by comparing them to altimetry minus SIO dataset, the authors should either add the deep ocean steric component in Sect. 3.3 and include multiple steric products, as they do elsewhere, or completely remove this section and move their discussion about the benefits of CIGAR manometric over GRACE to (currently) Sect. 3.4 and 3.5. It has more impact if shown that CIGAR manometric works better when combined with all steric component products than if that conclusion is based on only one product.

Minor issues

49 "by a factor of 2" This has been corrected in the conclusion in the last revision, but not here.

122-125 "Additionally, the GAD product derived from AOD1B models (Dobslaw et al., 2017; Flechtner et al., 2014), which represents non-tidal atmospheric and oceanic mass redistribution, is added back over the ocean areas to restore the ocean bottom pressure, which combines the effects of ocean mass and atmospheric loading." Why use AOD1B RL06, when there is a newer version released in 2022: https://doi.org/10.1029/2022JB024360? Especially for a paper that in the title includes word "revisited" it is important to use more recent information than the previous papers covering the same topic. If there is a reason for not using the newest release, e.g., this part of processing is already included into the mascon solution, that should be noted in the text. You should also specify which parts of the processing was performed by you and which is part of the dataset.

129 "In this study, we use the ensemble mean of the three mascon solutions." How different are the mascon solutions and would perhaps individual solutions achieve a better result? It would strengthen the conclusions if you can demonstrate that you considered differences in GRACE products.

Sect. 2.2.1 Computation of the manometric component: The explanation of the term "manometric" should come before explaining the methods how to obtain it.

Sect. 2.2.2 Post processing of the data: It is not entirely clear whether post-processing of the data, particularly the interpolation to a 1 degree grid, was performed before or after computing the manometric component. Presumably before, to be able to combine different datasets, but then it would also make more sense to have this section before the previous one.

Figs. 1 and 5: They are essentially the same, use the same products and the only difference is whether you subtract manometric or steric component from altimetry. It is redundant to show both.

Sect 3.2 Just a detail, but since Sect 3.2 is split into 3 parts (Argo products, reanalysis products and timeseries for both) why do reanalysis products and time series get their own sections (3.2.1 and 3.2.2), and Argo products do not? Perhaps there should be a Sect 3.2.3 here.

Fig. 3: It would be nice if it was written on the figure that the two left columns are the upper 2000 m and the below 2000 m contributions. Same for other figures with many subplots, it is hard to analyse them while looking at very long lists of letters in the caption describing what each subplot represents.

328-332 "From Figure 5, we clearly observe that the spatial trend pattern of SIO is more consistent with the “Altimetry minus CIGAR manometric” estimate, as both exhibit positive signals in the western North Atlantic. In contrast, the “Altimetry minus GRACE manometric” estimate displays a prominent positive signal in the eastern North Atlantic which is absent in the SIO data." Altimetry - GRACE also has a positive signal in the western North Atlantic, that is not what makes it differ from SIO, it is its positive signal in the east.

379-381 "As shown in Figure 7a,e, the resulting residuals implicitly include a deep ocean steric contribution, which is not accounted for since the steric data are limited here to the upper 2000 m." This sentence is not clear. 1. How is that seen from the figure? Is it specifically in a and e, and not in others? Why?

418 "For all cases except the ORAS5 deep ocean case," It might be worth explicitly stating that ORAS5 does not have much signal in the deep ocean, as shown earlier.

425-426 " when using CIGAR for both the steric (full depth and manometric component)." This is phrased wrong, manometric is not part of steric contribution, which this sentence implies. Ending bracket is probably in the wrong place.

429-431 "Notably, the sea level budget is effectively nearly closed within error bars when combining CIGAR manometric and CIGAR steric data." This may be true, but cannot really be seen from the discussed Fig. 8, as it does not include error bars. It is also hardly surprising when most components are from the same dataset and dependent on each other.

455-458 "Since the CIGAR manometric component does not fully account for the external sea water mass addition to the
ocean due to land ice melt and terrestrial waters, we adopt GRACE data to represent the total ocean mass manometric sea level change (i.e., global mean trend accounted for)." I do not understand this. What is different here compared to using GRACE vs CIGAR manometric component in previous sections?

476-478 "For SIO, a slight halosteric decrease is observed over 2012-2015, with a linear trend of -1.10±0.94 mm/yr during the such period, then it increases to -3.20±0.9 mm/yr over 2016-2022." It seems rather stable in the 2012-2015 time span, same as CIGAR, and they both have a negative trend from 2016. The 2015-2015 "trend" is probably just a result of the peak in 2013 that is more prominent in SIO than in CIGAR, but calculation of trend over 3 years is not very meaningful.

606-608 "(3) As of 2012, a negative trend is observed in the SIO halosteric component which suggests a physically real increase in salinity of the North Atlantic. Again, this needs further investigation." It really only starts in 2016, in all Argo datasets.

625-626 "The North Atlantic sea level budget is closed within data uncertainties when using the CIGAR reanalysis for both the manometric and steric components." Isn't that at least partially a consequence of the way reanalysis is produced? It is supposed to be consistent with itself.

635-638 "Why GRACE manometric data give less good results in the Northeast Atlantic than the CIGAR-based manometric sea level? Which inaccurate correction applied to GRACE is involved? Is it related to the GIA or the geocenter corrections applied to GRACE?" At least some of these questions could be answered in this paper by comparing the results of different GRACE products.

Hide

ED: Reconsider after major revisions (30 Jun 2026) by Matjaz Licer

AR by Zhe Song on behalf of the Authors (22 Jul 2026) Author's response Author's tracked changes Manuscript

ED: Referee Nomination & Report Request started (27 Jul 2026) by Matjaz Licer

RR by Anonymous Referee #4 (04 Aug 2026)

Suggestions for revision or reasons for rejection

The revised manuscript entitled “North Atlantic sea level budget revisited” has been significantly improved. In particular, the globally averaged trend was consistently removed throughout the analysis. I therefore recommend minor revisions. My detailed comments are provided below.
1. Line 37, “This was observed but not in this study, …” should be “In this study, …”
2. Line 48, “ With manometric, the budget is closed within data uncertainties.” I found this sentence is not clear. It may be revised to be “The budget can be closed within the data uncertainties when CIGAR manometric and full steric sea levels are included.”
3. Line 131, change “withdraw” to “remove”
4. Figure 2. Panels e and f are not discussed. The main difference shown in panels e and f is in subtropical gyre between 20-40 degree.
5. Line 198-199. “Regarding the trend uncertainty, it is expressed as the standard error of the least-squares fit.” Since a three-month moving-average filter was applied to the time series, the effective number of degrees of freedom is reduced when estimating the standard error of the least-squares fit. Was this effect taken into account? If not, the trend uncertainties may be underestimated. In that case, the uncertainty estimates reported in Table 1 should be revised, and the associated errors may be larger than those currently reported.
6. Line 366. “… allows to almost close regional sea level budget.” Are you referring to Table 1? If so, please explicitly cited the table.
7. Line 372 “We can thus conclude that the closure of the sea level budget holds within the data uncertainties in this case.” This statement is not consistent with previous statement in Line 366. It may be revised to be “We therefore conclude that, although a small residual remains, the sea-level budget closure is achieved within the data uncertainties for this case.”

Hide

RR by Anonymous Referee #5 (12 Aug 2026)

Suggestions for revision or reasons for rejection

The manuscript examines the North Atlantic sea level budget for 2004-2022 time span, comparing estimates of manometric and steric contributions from GRACE, Argo floats and ocean reanalysis with satellite altimetry data. The authors find that using the manometric component from the CIGAR reanalysis instead of GRACE helps with closure of the North Atlantic sea level budget and that using deep ocean steric contribution reduces the North Atlantic sea level budget residuals by 30. This is the third round of review overall, and the second time I have reviewed the manuscript.

The manuscript could potentially have scientific significance and its quality has improved substantially since the previous round of review, but there are still several issues that were not addressed properly or at all, as well as some new ones I discovered. If this was in an earlier stage of the review process, I would recommend another revision, however, as it is already the third review and the reviewers are specifically asked to assess whether the manuscript is acceptable for final publication, I have to say that no, it is definitely not.

A major problem is that the authors have not adequately addressed the reviewers' comments in the revision and rebuttal. Except for one comment, the 'final author reply to the editor' does not quote the relevant revised or newly added text in the manuscript, despite the OS's instructions for submission of revised manuscript clearly stating that it should. Having to search the manuscript for changes corresponding to each comment makes it unnecessarily difficult and time-consuming to assess whether the comments have been adequately addressed. Nonetheless, there are multiple instances of the authors responding only to part of a question, responding to a question in the rebuttal but not in the manuscript or claiming that they modified the manuscript to address it, but there are no such changes in the manuscript. These are not necessarily all issues present in the manuscript, as the structure of the rebuttal does not allow me to systematically verify all changes within the available review time, but they are sufficient for me to make a recommendation. Below is the list of all issues from the last review that the authors did not address properly, as well as of some new issues I detected. I recommend the authors to take their time to fully address them before re-submitting their manuscript.

Comments related to previous review

- Part of comment 1: "Furthermore, the CIGAR-manometric component is based on a reanalysis that assimilates Argo data (from EN4), and the authors themselves demonstrate that SIO and EN4 are very similar, so CIGAR and SIO cannot be considered entirely independent."

- Comment 2: "Some references to datasets are missing from the manuscript. What is the satellite altimetry dataset the authors use? I can guess which one it is, but that dataset has a name (DT2021 is just the version) and a DOI, which the authors do not provide. The link the authors provide is not to the dataset but to the data store landing page. GRACE datasets also do not have references. All datasets included in this study should be clearly named and referenced." - Authors claim in the rebuttal that they modified the Data section to provide clear references for all datasets and products used in their study, but they did not. In fact, there are no changes in Sect. 2.1.1 and 2.1.3 of the revised manuscript nor the track changes file, which describe satellite altimetry and Argo datasets. There is also, e.g., no DOI 10.1007/s00190-019-01252-y, which belongs to the article that accompanies the monthly GRACE GSFC mascon solutions.

- Part of comment 3: "They also use GRACE mascon solutions and not spherical harmonics without explaining why." - The authors explain this in another comment, but completely ignore it here. More importantly, they do not implement this answer into the manuscript.

- Comment 7: "Why use AOD1B RL06, when there is a newer version released in 2022: https://doi.org/10.1029/2022JB024360? Especially for a paper that in the title includes word "revisited" it is important to use more recent information than the previous papers covering the same topic. If there is a reason for not using the newest release, e.g., this part of processing is already included into the mascon solution, that should be noted in the text. You should also specify which parts of the processing was performed by you and which is part of the dataset." - The authors explain the processing steps in the rebuttal and claim that the additional processing step they performed is already described in the manuscript, but the manuscript does not differentiate which processing steps are a part of the downloaded mascon solution and which are performed by the authors.

- Comment 8: "How different are the mascon solutions and would perhaps individual solutions achieve a better result? It would strengthen the conclusions if you can demonstrate that you considered differences in GRACE products." - The authors reply in the rebuttal, but do not implement the answer into the manuscript. If a decision is made based on the results of another published study, that should be explicitly written in the manuscript, including a reference to that study.

- Comment 10: "Sect. 2.2.2 Post processing of the data: It is not entirely clear whether post-processing of the data, particularly the interpolation to a 1 degree grid, was performed before or after computing the manometric component. Presumably before, to be able to combine different datasets, but then it would also make more sense to have this section before the previous one." - The authors claim in the rebuttal that they revised the post-processing section to clarify the order of operations, but there are no such changes in said section in the revised manuscript, nor in the track changes file.

- Part of comment 18: "It is also hardly surprising when most components are from the same dataset and dependent on each other." - The authors respond to the rest of this comment, but ignore this particular statement.

- Comment 22: "Isn't that at least partially a consequence of the way reanalysis is produced? It is supposed to be consistent with itself." - (1) The answer in the rebuttal does not answer my concern. What does "may be" independent mean? Are they confirmed to be independent, do you not know, do you assume they are? Do you not compute the manometric component by subtracting steric from sterodynamic? How are they possibly independent? (2) This should be discussed in the manuscript, not just in the rebuttal. Generally, the answer to reviewers' questions should be implemented into the manuscript, or the authors should explain in the rebuttal why they do not consider it appropriate to include it. Simply answering the question in the rebuttal is not sufficient; if a reviewer wanted to know the answer, so will some readers.

Additional comments (major)

- Some sentences do not make sense in the revised version, perhaps parts of the text were lost while editing. E.g., " This was observed but not in this study, we revisit the North Atlantic sea level budget, using satellite altimetry, GRACE..." (L37-38); "With manometric, the budget is closed within data uncertainties." (L48); "... the EN4 curve (and to a little extent also SIO and JAMSTEC) shows a strong decrease as of 2016, possibly linked to the salinity drift in EN4 and JAMSTEC data sets, to together with a potential physical salinity increase." (L251-253). The authors should check the revised manuscript carefully to avoid such issues.
- Since both GRACE mascons and the satellite altimetry dataset are anomalies relative to a specific time period (2004.0-2009.999 for JPL mascons, possibly 1993-2012 for altimetry, if I am guessing correctly which dataset is used) how are different reference periods handled in this study when comparing products? This needs to be included in the manuscript.
- Sect. 2.2.1 (1) The explanation of how the manometric component is computed from the reanalysis is not clear enough. You explain subtracting local steric effects but do not explain how you obtain these local steric effects. You should include that part. (2) Using a manometric component directly from a GRACE dataset is not really a method of computing the manometric component, so this section should be limited to specifically calculation of the manometric component from other components. (3) Unrelated to the manometric component, you should include some text about the calculation of the steric component in general, not just in the Argo data section. Is the steric component from the reanalysis calculated in the same way?
- L193-195 "For the GRACE-based manometric component, uncertainties can be derived from the dispersion of the data sets with respect to the ensemble mean." Those are not the only uncertainties of GRACE data. Do mascon solutions not have uncertainty estimates? I am pretty sure they do. Why do you not use them?
- L271-272 "These maps are supposed to represent the manometric component, plus the deep ocean contribution." Is the deep ocean contribution not removed? The previous sentence states it is "altimetry minus steric full depth". Also, "supposed to represent" is a strange and too informal phrasing. Do they represent or are they supposed to represent but actually do not?

Additional suggestions

- Section titles are generally too long, they should be labels to distinguish the sections, not summaries. Also, Sect. 3.4.1 should not exist, as there is no Sect. 3.4.2; sub-levels should only be used if the level is split into multiple parts.
- Introduction could use some improvements. The context of introducing the Bouih et al. (2025) study should precede the details given about it, and it should be written more clearly what are the differences between it and your study and why.
- L378-387 This long list of questions at the end of the paper is very unusual for a scientific research article. It would be more appropriate to synthesize the main implications and limitations of the study, rather than presenting a large number of questions that are left unanswered. In particular, the questions pertaining to something that has been at least partially addressed in the manuscript (GIA correction) should not be repeated as fully open questions, while some other should also be discussed earlier.

Hide

ED: Reconsider after major revisions (17 Aug 2026) by Matjaz Licer

AR by Zhe Song on behalf of the Authors (03 Sep 2026) Author's response Author's tracked changes Manuscript

ED: Publish as is (04 Sep 2026) by Matjaz Licer

AR by Zhe Song on behalf of the Authors (14 Sep 2026) Manuscript

Short summary

The objective of this study is to understand why the North Atlantic sea level budget is not closed over the Argo and Gravity Recovery and Climate Experiment (GRACE) era. We show that using an ocean reanalysis for the ocean mass component instead of GRACE, and accounting for deep ocean warming allows quasi-closure of the North Atlantic sea level budget within data uncertainties.