Regionalizing the Sea-level Budget With Machine Learning Techniques

Camargo, Carolina M. L.; Riva, Riccardo E. M.; Hermans, Tim H. J.; Schütt, Eike M.; Marcos, Marta; Hernandez-Carrasco, Ismael; Slangen, Aimée B. A.

doi:https://doi.org/10.5194/egusphere-2022-876

Preprints

https://doi.org/10.5194/egusphere-2022-876

Preprints

13 Sep 2022

| 13 Sep 2022

Regionalizing the Sea-level Budget With Machine Learning Techniques

Carolina M. L. Camargo, Riccardo E. M. Riva, Tim H. J. Hermans, Eike M. Schütt, Marta Marcos, Ismael Hernandez-Carrasco, and Aimée B. A. Slangen

Abstract. Attribution of sea-level change to its different drivers is typically done using a sea-level budget (SLB) approach. While the global mean SLB is considered closed, closing the SLB on a finer spatial scale is more complicated due to, for instance, limitations in our observational system and the spatial processes contributing to regional sea-level change. Consequently, the regional SLB has been mainly analysed on a basin-wide scale. Here we investigate the SLB at sub-basin scales, using two machine learning techniques to extract domains of coherent sea-level variability: a neural network approach (Self-Organising Maps) and a network detection approach (δ-MAPS). The extracted domains provide a higher level of spatial detail than entire ocean basins and besides indicating how sea-level variability is connected among different regions. Using these domains we can close the regional SLB world-wide on different spatial scales. Steric variations dominate the temporal sea-level variability and determine a significant part of the total regional change. Sea-level change due to mass transport between ocean and land has a relatively homogeneous contribution to all regions. In highly dynamic regions (e.g., Gulf Stream region) the dynamic mass redistribution is significant. Regions where the SLB cannot be closed highlight processes that are affecting sea level but are not well captured by the observations, such as the influence of western boundary currents. Hence, the use of the SLB approach in combination with machine learning techniques leads to new insights into regional sea-level variability and its drivers.

Received: 02 Sep 2022 – Discussion started: 13 Sep 2022

Download & links

Preprint (PDF, 16434 KB)

Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
Preprint (16434 KB)

Download & links

The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.

Journal article(s) based on this preprint

16 Jan 2023

| Highlight paper

Regionalizing the sea-level budget with machine learning techniques

Carolina M. L. Camargo, Riccardo E. M. Riva, Tim H. J. Hermans, Eike M. Schütt, Marta Marcos, Ismael Hernandez-Carrasco, and Aimée B. A. Slangen

Ocean Sci., 19, 17–41, https://doi.org/10.5194/os-19-17-2023,https://doi.org/10.5194/os-19-17-2023, 2023

Short summary Co-editor-in-chief

Carolina M. L. Camargo et al.

Interactive discussion

Status: closed

CC1:
'Comment on egusphere-2022-876', Paul PUKITE, 13 Sep 2022

"indicating how the ENSO signal is propagated through the Pacific, possibly through coastally trapped waves (Hughes et al., 2019) in the coastal domains"

The ENSO signal shows up throughhout the tropical Pacific sea level simply via the inverse barometer effect. ENSO tracks closely the atmospheric pressure dipole as reealed by the differences between pressure at Darwin and Tahiti (the Southern Oscillation Index). The change is 1 cm for a 1 mBar change in pressure, so that with the SOI extremes showing 14 mBar variation at the Darwin location, this accounts for a 14 cm change in sea-level, roughly matching that shown in the chart below
(sorry for the mangled chart but this comment interface is very primitive)

Citation: https://doi.org/10.5194/egusphere-2022-876-CC1
- AC1: 'Reply on CC1', Carolina M.L. Camargo, 27 Sep 2022
  
  Dear Paul Pukite,
  Thank you for your comment.
  The line highlighted in your comment is about the propagation of ENSO signal through the Pacific, possibly by coastally trapped waves (Lines 207-208). The role of coastal and equatorial Kelvin waves and Rossby waves in propagating sea-level signals has been demonstrated in several studies (e.g., Hsieh & Bryan, 1996; Federov and Brown, 2009; Hughes et al., 2019). An example visualization of Rossby waves in sea-level anomalies records can be found in Figure 4 of Chelton et al. (1996). The comment, however, does not question the propagation of ENSO signal, but the impact of ENSO on sea-level change.
  ENSO events are characterized by a larger warming (in case of El Niño) or cooling (La Niña) in the Central Pacific. These variations in ocean heat content have a direct effect on sea surface height due to thermosteric effects (Wang & Picaut, 2004; Domingues et al., 2008). The influence of ENSO on sea level is not only clearly visible on regional steric sea-level maps (e.g., Figure 8 of Camargo et al. (2020)), but also on global mean sea-level curves (e.g., Figure 1 of Boening et al. (2012)). Therefore, the ENSO signal does not appear in sea level only via the inverse barometer effect.
  Kind regards,
  Carolina Camargo, on behalf of the authors
  References:
  Boening, C., Willis, J. K., Landerer, F. W., Nerem, R. S., and Fasullo, J. (2012), The 2011 La Niña: So strong, the oceans fell, Geophys. Res. Lett., 39, L19602, doi:10.1029/2012GL053055.
  Camargo, C. M.L., Riva, R. E. M., Hermans, T. H. J., & Slangen, A. B. A. (2020). Exploring sources of uncertainty in steric sea-level change estimates. Journal of Geophysical Research-Oceans, 125(10).
  Chelton, D. B., & Schlax, M. G. (1996). Global observations of oceanic rossby waves. Science, 272(5259), 234–238.
  Domingues, C., Church, J., White, N. et al. Improved estimates of upper-ocean warming and multi-decadal sea-level rise. Nature 453, 1090–1093 (2008). https://doi-org/10.1038/nature07080
  Fedorov, A. V., & Brown, J. N. (2009). Encyclopedia of ocean sciences. In Equatorial waves (pp. 271–287). essay, Elsevier Ltd. https://doi.org/10.1016/B978-012374473-9.00610-X
  Hsieh, W. W., & Bryan, K. (1996). Redistribution of sea level rise associated with enhanced greenhouse warming: a simple model study. Climate Dynamics, 12(8), 535–544. https://doi.org/10.1007/BF00207937
  Hughes, C. W., Fukumori, I., Griffies, S. M., Huthnance, J. M., Minobe, S., Spence, P., Thompson, K. R., and Wise, A.: Sea Level and the Role of Coastal Trapped Waves in Mediating the Influence of the Open Ocean on the Coast, Surveys in Geophysics, 40, 1467–1492, https://doi.org/10.1007/s10712-019-09535-x, 2019.
  Wang, C., Picaut JoëL, & Wang, C. (2013). Earth's climate. In Understanding enso physics—a review (pp. 21–48). essay, American Geophysical Union : Washington, D. C. https://doi.org/10.1029/147GM02
  
  Citation: https://doi.org/10.5194/egusphere-2022-876-AC1
RC1:
'Comment on egusphere-2022-876', Samantha Royston, 05 Oct 2022

Please see attached PDF report. Overall it is a very well written paper covering an important subject. I feel a few modifications to the manuscript text will improve clarity.

Citation: https://doi.org/10.5194/egusphere-2022-876-RC1
- AC2: 'Reply on RC1', Carolina M.L. Camargo, 11 Nov 2022
  
  Dear Dr. Royston,
  Thank you for your review. We have carefully replied to all your comments. Please see the attached point-by-point reply PDF.
  Kind regards,
  Carolina,
  on behalf of the authors
  
  Citation: https://doi.org/10.5194/egusphere-2022-876-AC2
RC2:
'Comment on egusphere-2022-876', Anonymous Referee #2, 17 Oct 2022

In this manuscript Camargo and colleagues analyze the regional sea level budget (i.e., the sum of individually measured/modelled contributions) to satellite altimetry over the 1993 to 2016 period. They specifically focus on the effect of spatial averaging on the uncertainties in budget closure. For spatial averaging they incorporate an a priori pattern recognition (two different approaches) step, which identifies clusters of homogeneous regions that are then averaged for the budget analysis. They demonstrate that clustering generally improves the budget closure and works significantly better than just using larger blocks. They also demonstrate the importance of the inclusion of an ocean bottom pressure term to the sterodynamic component. Overall, this is a very well written paper using novel approaches with several interesting findings. I therefore have no major reservations regarding the publication of the paper in Ocean Science. Below I provide a couple of comments and suggestions:

General:

I hope I did not overlook anything, but it seems that the authors compare geocentric sea level from satellite altimetry to relative sea level from the budget components, as their budget components also seem to contain crustal components of GRD terms due to contemporary mass change!? To my understanding one must either add those components to satellite altimetry, or only consider the geoid variations in the budget. The term has a substantial contribution to regional sea level according to Frederikse et al. (2017a)

Specific

Line 20: The inverse barometer contribution is missing here

Line 32: or for individual coastline stretches characterized by coherent variability (Frederikse et al., 2016, 2017b; Dangendorf et al., 2021). It has also been closed at a tide gauge level by Wang et al. (2021).

Line 69: I was wondering how the authors treated missing data due to the presence of sea ice at higher latitudes? This might also affect some of the budget misclosures/uncertainties mentioned farther below in the manuscript.

Line 83: How does that compare the deep ocean contribution from Zanna et al. 2019?

Line 85 following: As mentioned as a general comment, the approach seems to be inconsistent with respect to geocentric sea level as measured by satellites.

Line 94: It might be good to provide a little more information here, given that this other paper is still under review. I was also wondering how the estimates differ from those in Frederikse et al. (2020)?

Line 174 following: I am wondering how sensitive the two approaches are to temporal filtering? Former assessments such as Thompson and Merrifield (2014) have focused on decadal scales (which is likely more relevant for trends). Did the authors test sensitivity to smoothing? Also, have the time series been deseasonalized before applying the clustering technique?

Line 207: Or atmospheric teleconnections. Not all of them are connected by coasts

Line 231: does this mean a positive bias?

Line 344: The authors might consider Calafat et al. (2013) & Dangendorf et al., (2014), who initially established that link

Line 349: Southern Hemisphere

References:

Calafat, F. M., Chambers, D. P., & Tsimplis, M. N. (2013). Interâannual to decadal seaâlevel variability in the coastal zones of the Norwegian and Siberian Seas: The role of atmospheric forcing. , (3), 1287-1301.

Dangendorf, S., Calafat, F. M., Arns, A., Wahl, T., Haigh, I. D., & Jensen, J. (2014). Mean sea level variability in the North Sea: Processes and implications. , (10).

Frederikse, T., Riva, R., Kleinherenbrink, M., Wada, Y., van den Broeke, M., & Marzeion, B. (2016). Closing the sea level budget on a regional scale: Trends and variability on the Northwestern European continental shelf. , (20), 10-864.

Frederikse, T., Riva, R. E., & King, M. A. (2017a). Ocean bottom deformation due to presentâday mass redistribution and its impact on sea level observations. , (24), 12-306.

Frederikse, T., Simon, K., Katsman, C. A., & Riva, R. (2017b). The seaâlevel budget along the N orthwest A tlantic coast: GIA, mass changes, and largeâscale ocean dynamics. , (7), 5486-5501.

Zanna, L., Khatiwala, S., Gregory, J. M., Ison, J., & Heimbach, P. (2019). Global reconstruction of historical ocean heat storage and transport. , (4), 1126-1131.

Citation: https://doi.org/10.5194/egusphere-2022-876-RC2
- AC3: 'Reply on RC2', Carolina M.L. Camargo, 11 Nov 2022
  
  Dear Reviewer,
  Thank you for your review. We have carefully replied to all your comments. Please see the attached point-by-point reply PDF.
  Kind regards,
  Carolina,
  on behalf of the authors
  
  Citation: https://doi.org/10.5194/egusphere-2022-876-AC3
RC3:
'Comment on egusphere-2022-876', Anonymous Referee #3, 24 Oct 2022
Review of “Regionalizing the Sea-level Budget With Machine Learning Techniques” by Camargo et al.

General comments:

Global sea level budgets are examined using two machine learning techniques. Through identifying regions of similar sea level variability, the authors examined sea level budget in different basins of the world oceans. It is found that for most of the ocean regions, sea level variation can be explained using steric height changes and mass transport between ocean and land. But for some highly dynamic regions, the sea level budget closure may be affected by the mass redistribution associated with strong western boundary currents. All these make sense to this Reviewer. This is an excellent example of SOM application in oceanography and climate research community. I would like to recommend the manuscript be accepted after some minor revision. Specific comments are listed as follows.

Specific comments:

Pioneer work on SOM analysis of sea level variability should be properly mentioned. These include the first time SOM analysis of the satellite altimetry data (Liu et al., 2008), and the dual-SOM applications including the regionalizing of sea level variability in the Gulf of Mexico (Liu et al., 2016). It would be good to add the following information to the paragraph explaining the SOM (L138 - L156) or the Introduction part (L44-45):

“SOM has been used to extract patterns of sea level variability from satellite altimetry data (Liu et al., 2008; Weisberg and Liu, 2017, Nickerson et al., 2021). Dual-SOM application has been proposed to analyse sea level data, one focused on the characteristic spatial patterns, and the other focused on the characteristic time series, using sea level in the Gulf of Mexico as an example (Liu et al., 2016). The latter resulted in regionalizing the sea level variability, and is pursued here in this study to analyse global sea level data.”

L361-L362 indicate the challenges of sea level budget in coastal regions. This is true, as coastal ocean dynamics of sea level (e.g., Liu et al., 2007) are more complicated than that of deep ocean, and key dynamics may not be properly represented in the global data. It would be good to add a sentence to L364 about the sea level budget issues for coastal regions:

“Note that sea level budget for coastal regions is more challenging (Liu et al., 2007) with some of the dominant coastal ocean dynamics are not properly represented in the global data sets.”

Throughout the manuscript, “sea-level” should be changed to “sea level” ––– no hyphen.

The abbreviations of “sea level change” and “sea level budget” are not necessary at all. They do not save much space in text, rather they may cause inconveniences to readers, as readers may need to go back to search what they stand for, particularly for the case of many other acronyms are used later.

L23, it would be good to provide an example, Chambers et al. (2014), for this sentence.

L90, GRD is not defined.

Line 139, it would be good to insert a sentence to mention the powerfulness of the SOM technique: “It was demonstrated to be more powerful than conventional feature extraction methods (e.g., Liu and Weisberg, 2005).

References:

Chambers, D.P., A. Cazenave, N. Champollion, H. Dieng, W. Llovel, R. Forsberg, K. von Schuckmann, Y. Wada (2014), Evaluation of the global mean sea level budget between 1993 and 2014, , 315-333, Springer, Cham.

Liu, Y., and R.H. Weisberg (2005), Patterns of ocean current variability on the West Florida Shelf using the self-organizing map, 110, C06003, http://dx.doi.org/10.1029/2004JC002786.

Liu, Y., and R.H. Weisberg (2007), Ocean currents and sea surface heights estimated across the West Florida Shelf, , 37(6), 1697-1713, http://dx.doi.org/10.1175/JPO3083.1.

Liu, Y., R.H. Weisberg, and Y. Yuan (2008), Patterns of upper layer circulation variability in the South China Sea from satellite altimetry using the self-organizing map. , 27(Supp.), 129-144.

Liu, Y., R.H., Weisberg, S. Vignudelli, and G.T. Mitchum (2016), Patterns of the Loop Current system and regions of sea surface height variability in the eastern Gulf of Mexico revealed by the self-organizing maps, 121, 2347-2366, http://dx.doi.org/10.1002/2015JC011493.

Weisberg, R.H., and Y. Liu (2017), On the Loop Current penetration into the Gulf of Mexico, , 122, 9679-9694, http://dx.doi.org/10.1002/2017JC013330.

Nickerson, A.K., R.H. Weisberg, and Y. Liu (2022), On the evolution of the Gulf of Mexico Loop Current through its penetrative, ring shedding and retracted states, , 69(11), 4058-4077, https://doi.org/10.1016/j.asr.2022.03.039
Citation: https://doi.org/10.5194/egusphere-2022-876-RC3
- AC4: 'Reply on RC3', Carolina M.L. Camargo, 11 Nov 2022
  
  Dear Reviewer,
  Thank you for your review. We have carefully replied to all your comments. Please see the attached point-by-point reply PDF.
  Kind regards,
  Carolina,
  on behalf of the authors
  
  Citation: https://doi.org/10.5194/egusphere-2022-876-AC4
RC4:
'Comment on egusphere-2022-876', Anonymous Referee #4, 03 Nov 2022

The authors offer a new perspective on regional sea level budget (SLB) closure by focusing on two machine learning (ML) algorithms. They show how self organizing maps (SOM) and d-MAPS allow to identify spatially coherent regions and (i) close the SLB in most of those regions and (ii) further reduce the undertainties otherwise present when using the gridded data alone. Importantly, by focusing on two ML tools they are able to demonstrate the robustness of their conclusions under the models architecture considered.

This is a good, well written paper and it further demonstrate the benefit of focusing on coherent patterns rather than gridded data. I reccomend publication after some minor revisions.

(a) Section 2.1. Is terrestrial water storage included in the sea level budget? If not, can the authors give an explanation on why it was not considered?

(b) Figure 1. In a recent paper, Wang et al. 2021 closed the sea level budget at tide gauges locations. In Figure 1 of Wang et al. 2021, the authors show different components of sea level (SL) randing from stereodynamic SL to GIA, Glaciers etc. Can the authors explain possible differences between their Figure 1 and the one in Wang et al. 2021. In any case the authors should at least cite that work.

(c) Section 2.3.

(c.1) It would be clearer if SOM and d-Maps are introduced in two different subsections or paragraphs (Section 2.3.1 and 2.3.2)

(c.2) I understand that in both SOM and d-Maps, domains are identified after removing the seasonal cycle and trends. This is reasonable. It is my understanding though that after the domain identification step, the time series considered in each domain are averaged time series with seasonality and trends included. Is this correct? This should be clearly stated in the manuscript and it is missing at the momen (my bad in case I missed it).

(d) Section 4.1. The budget is closed in 77/92 d-Maps domains. I wonder if the remaining 15 domains where the budget is not closed are mainly found in the Southern Ocean. In that region I see lots of very small regions which could be just noise. If that is the case I suggest the author to add this in the paper.

(e) Figure 4. "Sea level budget trends (mm/yr) for (a) d-MAPS ad (b) SOM" I think it should be the opposite: Sea level budget trends (mm/yr) for (a) SOM and (b) d-MAPS.

J. Wang, J. A. Church, X. Zhang, J. M. Gregory, L. Zanna, and X. Chen. Evaluation of the local sea- level budget at tide gauges since 1958. Geophysical Research Letters, 48:e2021GL094502, 2021. doi: https://doi.org/10.1029/2021GL094502.

Citation: https://doi.org/10.5194/egusphere-2022-876-RC4
- AC5: 'Reply on RC4', Carolina M.L. Camargo, 11 Nov 2022
  
  Dear Reviewer,
  Thank you for your review. We have carefully replied to all your comments. Please see the attached point-by-point reply PDF.
  Kind regards,
  Carolina,
  on behalf of the authors
  
  Citation: https://doi.org/10.5194/egusphere-2022-876-AC5

Interactive discussion

Status: closed

CC1:
'Comment on egusphere-2022-876', Paul PUKITE, 13 Sep 2022

"indicating how the ENSO signal is propagated through the Pacific, possibly through coastally trapped waves (Hughes et al., 2019) in the coastal domains"

The ENSO signal shows up throughhout the tropical Pacific sea level simply via the inverse barometer effect. ENSO tracks closely the atmospheric pressure dipole as reealed by the differences between pressure at Darwin and Tahiti (the Southern Oscillation Index). The change is 1 cm for a 1 mBar change in pressure, so that with the SOI extremes showing 14 mBar variation at the Darwin location, this accounts for a 14 cm change in sea-level, roughly matching that shown in the chart below
(sorry for the mangled chart but this comment interface is very primitive)

Citation: https://doi.org/10.5194/egusphere-2022-876-CC1
- AC1: 'Reply on CC1', Carolina M.L. Camargo, 27 Sep 2022
  
  Dear Paul Pukite,
  Thank you for your comment.
  The line highlighted in your comment is about the propagation of ENSO signal through the Pacific, possibly by coastally trapped waves (Lines 207-208). The role of coastal and equatorial Kelvin waves and Rossby waves in propagating sea-level signals has been demonstrated in several studies (e.g., Hsieh & Bryan, 1996; Federov and Brown, 2009; Hughes et al., 2019). An example visualization of Rossby waves in sea-level anomalies records can be found in Figure 4 of Chelton et al. (1996). The comment, however, does not question the propagation of ENSO signal, but the impact of ENSO on sea-level change.
  ENSO events are characterized by a larger warming (in case of El Niño) or cooling (La Niña) in the Central Pacific. These variations in ocean heat content have a direct effect on sea surface height due to thermosteric effects (Wang & Picaut, 2004; Domingues et al., 2008). The influence of ENSO on sea level is not only clearly visible on regional steric sea-level maps (e.g., Figure 8 of Camargo et al. (2020)), but also on global mean sea-level curves (e.g., Figure 1 of Boening et al. (2012)). Therefore, the ENSO signal does not appear in sea level only via the inverse barometer effect.
  Kind regards,
  Carolina Camargo, on behalf of the authors
  References:
  Boening, C., Willis, J. K., Landerer, F. W., Nerem, R. S., and Fasullo, J. (2012), The 2011 La Niña: So strong, the oceans fell, Geophys. Res. Lett., 39, L19602, doi:10.1029/2012GL053055.
  Camargo, C. M.L., Riva, R. E. M., Hermans, T. H. J., & Slangen, A. B. A. (2020). Exploring sources of uncertainty in steric sea-level change estimates. Journal of Geophysical Research-Oceans, 125(10).
  Chelton, D. B., & Schlax, M. G. (1996). Global observations of oceanic rossby waves. Science, 272(5259), 234–238.
  Domingues, C., Church, J., White, N. et al. Improved estimates of upper-ocean warming and multi-decadal sea-level rise. Nature 453, 1090–1093 (2008). https://doi-org/10.1038/nature07080
  Fedorov, A. V., & Brown, J. N. (2009). Encyclopedia of ocean sciences. In Equatorial waves (pp. 271–287). essay, Elsevier Ltd. https://doi.org/10.1016/B978-012374473-9.00610-X
  Hsieh, W. W., & Bryan, K. (1996). Redistribution of sea level rise associated with enhanced greenhouse warming: a simple model study. Climate Dynamics, 12(8), 535–544. https://doi.org/10.1007/BF00207937
  Hughes, C. W., Fukumori, I., Griffies, S. M., Huthnance, J. M., Minobe, S., Spence, P., Thompson, K. R., and Wise, A.: Sea Level and the Role of Coastal Trapped Waves in Mediating the Influence of the Open Ocean on the Coast, Surveys in Geophysics, 40, 1467–1492, https://doi.org/10.1007/s10712-019-09535-x, 2019.
  Wang, C., Picaut JoëL, & Wang, C. (2013). Earth's climate. In Understanding enso physics—a review (pp. 21–48). essay, American Geophysical Union : Washington, D. C. https://doi.org/10.1029/147GM02
  
  Citation: https://doi.org/10.5194/egusphere-2022-876-AC1
RC1:
'Comment on egusphere-2022-876', Samantha Royston, 05 Oct 2022

Please see attached PDF report. Overall it is a very well written paper covering an important subject. I feel a few modifications to the manuscript text will improve clarity.

Citation: https://doi.org/10.5194/egusphere-2022-876-RC1
- AC2: 'Reply on RC1', Carolina M.L. Camargo, 11 Nov 2022
  
  Dear Dr. Royston,
  Thank you for your review. We have carefully replied to all your comments. Please see the attached point-by-point reply PDF.
  Kind regards,
  Carolina,
  on behalf of the authors
  
  Citation: https://doi.org/10.5194/egusphere-2022-876-AC2
RC2:
'Comment on egusphere-2022-876', Anonymous Referee #2, 17 Oct 2022

In this manuscript Camargo and colleagues analyze the regional sea level budget (i.e., the sum of individually measured/modelled contributions) to satellite altimetry over the 1993 to 2016 period. They specifically focus on the effect of spatial averaging on the uncertainties in budget closure. For spatial averaging they incorporate an a priori pattern recognition (two different approaches) step, which identifies clusters of homogeneous regions that are then averaged for the budget analysis. They demonstrate that clustering generally improves the budget closure and works significantly better than just using larger blocks. They also demonstrate the importance of the inclusion of an ocean bottom pressure term to the sterodynamic component. Overall, this is a very well written paper using novel approaches with several interesting findings. I therefore have no major reservations regarding the publication of the paper in Ocean Science. Below I provide a couple of comments and suggestions:

General:

I hope I did not overlook anything, but it seems that the authors compare geocentric sea level from satellite altimetry to relative sea level from the budget components, as their budget components also seem to contain crustal components of GRD terms due to contemporary mass change!? To my understanding one must either add those components to satellite altimetry, or only consider the geoid variations in the budget. The term has a substantial contribution to regional sea level according to Frederikse et al. (2017a)

Specific

Line 20: The inverse barometer contribution is missing here

Line 32: or for individual coastline stretches characterized by coherent variability (Frederikse et al., 2016, 2017b; Dangendorf et al., 2021). It has also been closed at a tide gauge level by Wang et al. (2021).

Line 69: I was wondering how the authors treated missing data due to the presence of sea ice at higher latitudes? This might also affect some of the budget misclosures/uncertainties mentioned farther below in the manuscript.

Line 83: How does that compare the deep ocean contribution from Zanna et al. 2019?

Line 85 following: As mentioned as a general comment, the approach seems to be inconsistent with respect to geocentric sea level as measured by satellites.

Line 94: It might be good to provide a little more information here, given that this other paper is still under review. I was also wondering how the estimates differ from those in Frederikse et al. (2020)?

Line 174 following: I am wondering how sensitive the two approaches are to temporal filtering? Former assessments such as Thompson and Merrifield (2014) have focused on decadal scales (which is likely more relevant for trends). Did the authors test sensitivity to smoothing? Also, have the time series been deseasonalized before applying the clustering technique?

Line 207: Or atmospheric teleconnections. Not all of them are connected by coasts

Line 231: does this mean a positive bias?

Line 344: The authors might consider Calafat et al. (2013) & Dangendorf et al., (2014), who initially established that link

Line 349: Southern Hemisphere

References:

Calafat, F. M., Chambers, D. P., & Tsimplis, M. N. (2013). Interâannual to decadal seaâlevel variability in the coastal zones of the Norwegian and Siberian Seas: The role of atmospheric forcing. , (3), 1287-1301.

Dangendorf, S., Calafat, F. M., Arns, A., Wahl, T., Haigh, I. D., & Jensen, J. (2014). Mean sea level variability in the North Sea: Processes and implications. , (10).

Frederikse, T., Riva, R., Kleinherenbrink, M., Wada, Y., van den Broeke, M., & Marzeion, B. (2016). Closing the sea level budget on a regional scale: Trends and variability on the Northwestern European continental shelf. , (20), 10-864.

Frederikse, T., Riva, R. E., & King, M. A. (2017a). Ocean bottom deformation due to presentâday mass redistribution and its impact on sea level observations. , (24), 12-306.

Frederikse, T., Simon, K., Katsman, C. A., & Riva, R. (2017b). The seaâlevel budget along the N orthwest A tlantic coast: GIA, mass changes, and largeâscale ocean dynamics. , (7), 5486-5501.

Zanna, L., Khatiwala, S., Gregory, J. M., Ison, J., & Heimbach, P. (2019). Global reconstruction of historical ocean heat storage and transport. , (4), 1126-1131.

Citation: https://doi.org/10.5194/egusphere-2022-876-RC2
- AC3: 'Reply on RC2', Carolina M.L. Camargo, 11 Nov 2022
  
  Dear Reviewer,
  Thank you for your review. We have carefully replied to all your comments. Please see the attached point-by-point reply PDF.
  Kind regards,
  Carolina,
  on behalf of the authors
  
  Citation: https://doi.org/10.5194/egusphere-2022-876-AC3
RC3:
'Comment on egusphere-2022-876', Anonymous Referee #3, 24 Oct 2022
Review of “Regionalizing the Sea-level Budget With Machine Learning Techniques” by Camargo et al.

General comments:

Global sea level budgets are examined using two machine learning techniques. Through identifying regions of similar sea level variability, the authors examined sea level budget in different basins of the world oceans. It is found that for most of the ocean regions, sea level variation can be explained using steric height changes and mass transport between ocean and land. But for some highly dynamic regions, the sea level budget closure may be affected by the mass redistribution associated with strong western boundary currents. All these make sense to this Reviewer. This is an excellent example of SOM application in oceanography and climate research community. I would like to recommend the manuscript be accepted after some minor revision. Specific comments are listed as follows.

Specific comments:

Pioneer work on SOM analysis of sea level variability should be properly mentioned. These include the first time SOM analysis of the satellite altimetry data (Liu et al., 2008), and the dual-SOM applications including the regionalizing of sea level variability in the Gulf of Mexico (Liu et al., 2016). It would be good to add the following information to the paragraph explaining the SOM (L138 - L156) or the Introduction part (L44-45):

“SOM has been used to extract patterns of sea level variability from satellite altimetry data (Liu et al., 2008; Weisberg and Liu, 2017, Nickerson et al., 2021). Dual-SOM application has been proposed to analyse sea level data, one focused on the characteristic spatial patterns, and the other focused on the characteristic time series, using sea level in the Gulf of Mexico as an example (Liu et al., 2016). The latter resulted in regionalizing the sea level variability, and is pursued here in this study to analyse global sea level data.”

L361-L362 indicate the challenges of sea level budget in coastal regions. This is true, as coastal ocean dynamics of sea level (e.g., Liu et al., 2007) are more complicated than that of deep ocean, and key dynamics may not be properly represented in the global data. It would be good to add a sentence to L364 about the sea level budget issues for coastal regions:

“Note that sea level budget for coastal regions is more challenging (Liu et al., 2007) with some of the dominant coastal ocean dynamics are not properly represented in the global data sets.”

Throughout the manuscript, “sea-level” should be changed to “sea level” ––– no hyphen.

The abbreviations of “sea level change” and “sea level budget” are not necessary at all. They do not save much space in text, rather they may cause inconveniences to readers, as readers may need to go back to search what they stand for, particularly for the case of many other acronyms are used later.

L23, it would be good to provide an example, Chambers et al. (2014), for this sentence.

L90, GRD is not defined.

Line 139, it would be good to insert a sentence to mention the powerfulness of the SOM technique: “It was demonstrated to be more powerful than conventional feature extraction methods (e.g., Liu and Weisberg, 2005).

References:

Chambers, D.P., A. Cazenave, N. Champollion, H. Dieng, W. Llovel, R. Forsberg, K. von Schuckmann, Y. Wada (2014), Evaluation of the global mean sea level budget between 1993 and 2014, , 315-333, Springer, Cham.

Liu, Y., and R.H. Weisberg (2005), Patterns of ocean current variability on the West Florida Shelf using the self-organizing map, 110, C06003, http://dx.doi.org/10.1029/2004JC002786.

Liu, Y., and R.H. Weisberg (2007), Ocean currents and sea surface heights estimated across the West Florida Shelf, , 37(6), 1697-1713, http://dx.doi.org/10.1175/JPO3083.1.

Liu, Y., R.H. Weisberg, and Y. Yuan (2008), Patterns of upper layer circulation variability in the South China Sea from satellite altimetry using the self-organizing map. , 27(Supp.), 129-144.

Liu, Y., R.H., Weisberg, S. Vignudelli, and G.T. Mitchum (2016), Patterns of the Loop Current system and regions of sea surface height variability in the eastern Gulf of Mexico revealed by the self-organizing maps, 121, 2347-2366, http://dx.doi.org/10.1002/2015JC011493.

Weisberg, R.H., and Y. Liu (2017), On the Loop Current penetration into the Gulf of Mexico, , 122, 9679-9694, http://dx.doi.org/10.1002/2017JC013330.

Nickerson, A.K., R.H. Weisberg, and Y. Liu (2022), On the evolution of the Gulf of Mexico Loop Current through its penetrative, ring shedding and retracted states, , 69(11), 4058-4077, https://doi.org/10.1016/j.asr.2022.03.039
Citation: https://doi.org/10.5194/egusphere-2022-876-RC3
- AC4: 'Reply on RC3', Carolina M.L. Camargo, 11 Nov 2022
  
  Dear Reviewer,
  Thank you for your review. We have carefully replied to all your comments. Please see the attached point-by-point reply PDF.
  Kind regards,
  Carolina,
  on behalf of the authors
  
  Citation: https://doi.org/10.5194/egusphere-2022-876-AC4
RC4:
'Comment on egusphere-2022-876', Anonymous Referee #4, 03 Nov 2022

The authors offer a new perspective on regional sea level budget (SLB) closure by focusing on two machine learning (ML) algorithms. They show how self organizing maps (SOM) and d-MAPS allow to identify spatially coherent regions and (i) close the SLB in most of those regions and (ii) further reduce the undertainties otherwise present when using the gridded data alone. Importantly, by focusing on two ML tools they are able to demonstrate the robustness of their conclusions under the models architecture considered.

This is a good, well written paper and it further demonstrate the benefit of focusing on coherent patterns rather than gridded data. I reccomend publication after some minor revisions.

(a) Section 2.1. Is terrestrial water storage included in the sea level budget? If not, can the authors give an explanation on why it was not considered?

(b) Figure 1. In a recent paper, Wang et al. 2021 closed the sea level budget at tide gauges locations. In Figure 1 of Wang et al. 2021, the authors show different components of sea level (SL) randing from stereodynamic SL to GIA, Glaciers etc. Can the authors explain possible differences between their Figure 1 and the one in Wang et al. 2021. In any case the authors should at least cite that work.

(c) Section 2.3.

(c.1) It would be clearer if SOM and d-Maps are introduced in two different subsections or paragraphs (Section 2.3.1 and 2.3.2)

(c.2) I understand that in both SOM and d-Maps, domains are identified after removing the seasonal cycle and trends. This is reasonable. It is my understanding though that after the domain identification step, the time series considered in each domain are averaged time series with seasonality and trends included. Is this correct? This should be clearly stated in the manuscript and it is missing at the momen (my bad in case I missed it).

(d) Section 4.1. The budget is closed in 77/92 d-Maps domains. I wonder if the remaining 15 domains where the budget is not closed are mainly found in the Southern Ocean. In that region I see lots of very small regions which could be just noise. If that is the case I suggest the author to add this in the paper.

(e) Figure 4. "Sea level budget trends (mm/yr) for (a) d-MAPS ad (b) SOM" I think it should be the opposite: Sea level budget trends (mm/yr) for (a) SOM and (b) d-MAPS.

J. Wang, J. A. Church, X. Zhang, J. M. Gregory, L. Zanna, and X. Chen. Evaluation of the local sea- level budget at tide gauges since 1958. Geophysical Research Letters, 48:e2021GL094502, 2021. doi: https://doi.org/10.1029/2021GL094502.

Citation: https://doi.org/10.5194/egusphere-2022-876-RC4
- AC5: 'Reply on RC4', Carolina M.L. Camargo, 11 Nov 2022
  
  Dear Reviewer,
  Thank you for your review. We have carefully replied to all your comments. Please see the attached point-by-point reply PDF.
  Kind regards,
  Carolina,
  on behalf of the authors
  
  Citation: https://doi.org/10.5194/egusphere-2022-876-AC5

Peer review completion

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

AR by Carolina M.L. Camargo on behalf of the Authors (01 Dec 2022) Author's response Author's tracked changes Manuscript

ED: Publish subject to technical corrections (08 Dec 2022) by John M. Huthnance

Dear Authors
Thank-you for your revised manuscript addressing the reviewer comments. I am now asking that you consider the “Technical Corrections” below and then upload your manuscript to the Copernicus/OS editorial system for publication (no more intervention by myself). There will be copy-editing and you should check that this retains your intended meaning. Thank-you for publishing in Ocean Science.
Yours sincerely
John Huthnance (editor)

Technical Corrections (and other comments)
“sea level”; I agree with your criterion for when to use a hyphen.
Reviewer-suggested references. You should only include these where you consider that they benefit the manuscript.
Lines 6-7. “provide a higher level of spatial detail than entire ocean basins” is clumsy and “higher level” is not really appropriate. Maybe “provide detail within ocean basins”?
Line 41. “higher spatial resolution budget” –> “budget with finer spatial resolution”?
Line 86. “up to” –> “down to”?
Line 93. Omit first “the”
Lines 93-94. It is not clear from the description (i) that this GRD component of sea level includes a change of level simply because more water in the ocean implies more volume; “Deformation . . response of the Earth to . .” rather ignores the volumetric contribution to sea level. This is of course corrected in lines 98-99 but it would help if the description (i) were clearer.
Lines 143-144. Better “. . Contrary to r, nRMSE closer to 0 indicates better performance.”?
Lines 190-191. “and potentially . . activity”. This second half of the sentence lacks a verb (to match “are” in the first half) and therefore has unclear meaning.
Lines 282-283. “. . from 0.81 . . . MAPS to 0.98 . .”?
Line 335. “being” –> “is” but perhaps better still “. . surprising; the seasonal cycle is predominantly steric.”
Line 346. “reduces” –> “decreases”
Figure 5. In response to Reviewer 1 line-by-line comment 1, you say “On Figure 5, the whiskers (not error bars) show the full distribution, while the box shows the quantiles of the data.” I think this information needs to be in the caption. Do you mean “quartiles”?
Line 380. “. . different depending on whether . .”
Lines 458-459. “. . regions surrounding Indonesia and Japan . .”
Figure B5 caption line 2. “Starts” –> “Stars”
Figure B7 caption line 4. “less” –> “fewer”

Hide

AR by Carolina M.L. Camargo on behalf of the Authors (09 Dec 2022) Manuscript

Journal article(s) based on this preprint

16 Jan 2023

| Highlight paper

Regionalizing the sea-level budget with machine learning techniques

Carolina M. L. Camargo, Riccardo E. M. Riva, Tim H. J. Hermans, Eike M. Schütt, Marta Marcos, Ismael Hernandez-Carrasco, and Aimée B. A. Slangen

Ocean Sci., 19, 17–41, https://doi.org/10.5194/os-19-17-2023,https://doi.org/10.5194/os-19-17-2023, 2023

Short summary Co-editor-in-chief

Carolina M. L. Camargo et al.

Viewed

Total article views: 722 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
501	195	26	722	2	2

HTML: 501
PDF: 195
XML: 26
Total: 722
BibTeX: 2
EndNote: 2

Views and downloads (calculated since 13 Sep 2022)

Month	HTML	PDF	XML	Total
Sep 2022	183	66	10	259
Oct 2022	133	46	6	185
Nov 2022	123	37	10	170
Dec 2022	39	30	0	69
Jan 2023	23	16	0	39
Feb 2023	0
Mar 2023	0

Cumulative views and downloads (calculated since 13 Sep 2022)

Month	HTML	PDF	XML	Total
Sep 2022	183	66	10	259
Oct 2022	133	46	6	185
Nov 2022	123	37	10	170
Dec 2022	39	30	0	69
Jan 2023	23	16	0	39
Feb 2023	0
Mar 2023	0

Viewed (geographical distribution)

Total article views: 678 (including HTML, PDF, and XML) Thereof 678 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 25 Mar 2023

Download

The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.

Preprint (16434 KB)
Metadata XML

Closing the sea level budget (between observed rise and the sum of its causes) has been a challenge, is an ongoing effort and has primarily concerned the global mean. Here, the authors use machine learning to identify sub-areas with similar trends to close the sea level budget on a regional level, with much reduced errors compared with 1-degree grid points.

Short summary

Sea-level change is mainly caused by variations in the ocean’s temperature and salinity and land ice melting. Here, we quantify the contribution of the different drivers to the regional sea-level change. We apply machine learning techniques to identify regions that have similar sea-level variability. These regions show how the different drivers contribute differently to sea-level change depending on the region, highlight how large-scale ocean circulation controls regional sea-level change.


Total:	0
HTML:	0
PDF:	0
XML:	0