the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Monitoring the regional Ocean Heat Content change over the Atlantic Ocean with the space geodetic approach
Abstract. The estimation of the regional Ocean Heat Content (OHC) is essential for climate analysis and future climate predictions. In this study, we propose a method to estimate and propagate uncertainties in regional OHC changes. The OHC is estimated with space geodetic steric data corrected from salinity variations estimated with in situ measurements. A variance-covariance matrix method is used to propagate uncertainties from space geodetic data to the OHC change. The integrated OHC change over the Atlantic basin is 0.17 W m-2 which represents 21 % of the global OHC trend, with significant trends observed in 52 % of the Atlantic basin. Uncertainties in OHC trends are mainly attributed to manometric sea level change uncertainties. We validate our space geodetic OHC estimates at two test sites, representing the subtropical and subpolar regions of the North Atlantic, highlighting their importance in understanding climate dynamics. Our results show good agreement between space geodetic estimates and in situ measurements in the North Atlantic region. The space geodetic OHC trends reveal a warming pattern in the southern and western parts of the North Atlantic, particularly in the Gulf Stream region, while the northeastern part exhibits cooling trends. Overall, our study provides valuable insights and a new framework to estimate regional OHC change and its uncertainties, contributing to a better understanding of the Earth's climate system and its future projections. The space geodetic OHC change product (version 1.0) is freely available at https://doi.org/10.24400/527896/a01-2022.012 (Magellium/LEGOS, 2022)
This preprint has been withdrawn.
-
Withdrawal notice
This preprint has been withdrawn.
-
Preprint
(6146 KB)
Interactive discussion
Status: closed
-
CC1: 'Comment on essd-2023-236', Armin Koehl, 28 Aug 2023
The manuscript describes a technique to estimate heat content changes from satellite data together with in situ data. The estimate include uncertainty estimation and a comparison to alternative estimates. The product is available to the public, and as such the manuscript is worth to be published. However, I feel that the methodology is not sufficiently complete or well enough described to allow the reader to reproduce the estimation of the ocean heat content (OHC).
The large contribution to the error estimate by manometric contribution is surprising given that this component shows very little contribution to the dynamic sea level change (see for instance Landerer etal (doi.org/10.1175/JPO3013.1) or Couldrey et al (doi.org/10.1007/s00382-020-05471-4)). Dynamic sea level certainly does not contain the important signals related to melting, groundwater change and GIA.
However, this outcome is an argument against this technique and suggest that it may be actually inferior to calculating heat content change directly from temperature data.There is discussion about the different contributions to the uncertainty in the conclusion. Data gaps are reported as the main issue with the technique, while the estimated uncertainties relate mostly to the manometric contribution from the gravity missions. Their procedure was introduced in several ways to be superior to alternative estimates based on for instance temperature data alone, but then their product is evaluated with these kind of products. In the end, I felt a bit lost as how to evaluate the quality and merits of the product and what the true uncertainties are. One merit is the estimate of uncertainty but then I get the impression that these cannot be trusted since the true uncertainty may relate also to contributions not considered. The authors dedicated a substantial part of the conclusion to the discussion of uncertainties, and I believe it will be hard to provide additional details. However as detailed below, there are some places, for instance in the introduction, the comparison to alternative estimates and the presentation of the uncertainies, where some more details or interpretations could be presented.
Details:L 38 What is the role of salinity here?
L68 Why is this uncertainty estimate sufficient ?
Doesn't the contribution of the halosteric signal to the uncertainty need to be considered too? estimating a signal from the difference instead directly from the thermosteric sounds at first a bad idea since the uncertainties add. Add a argument why your approach is still a good idea.L 61-68 Maybe it would be useful to contrast advantages with disadvantage. The full depth estimate of the thermosteric signal assumes that the halosteric signal is known or zero which is not the case. Maybe anticipate some of the caveates discussed later, to provide a more blanced view.
L 78 Please define abbreviation TSSL
eq (2) There needs to be different identifiers for TSSL from temperature and from the geodetic approach.
L 87-88 Since the expansion coefficient is strongly temperature dependent, it is important to state the reference state assumed for the expansion coefficient. Assuming is 0C and 35 PSU would be a very bad choice, because coefficients near the reference state will be very different from the actual coefficient in a substantial part of the ocean. I think (2) makes only sense for a linear density equation, because expansion coefficient may be very different for a whole range from 0 to the actual temperature, while the changes may be mostly be around the actual temperature and very different from 0.
l 31 Not clear how the changing temperature is used to calculate halosteric sea level.
l 143 How is merging done?
l 153-154 Some extra details would be helpful since the brief introduction above gave the impression that this may not have been done in the best way,
Section 3.3.3 Also for the validation products it would be useful to know how the salinity data has been used.
L 199 Please specify in what way the method was extended. Check the grammar of the sentence.
l 201-202 Not clear how the variance covariance matrices are defined. Please provide a definition, is covariance only in time? What do the two arguments t stand for?
l 208, eq (5) Not clear which terms are in that budget. What is i indicating?
l 210 Not clear how this works, this is dispersion in time?
Maybe also a comment on how independent the members of the ensemble are and how that affects the estimate.l 211 How is the additional uncertainty computed?
eq (6) Not clear to me over which dimension the cov is calculated that the result depends on i,j,lat, lon and t. And why there are two t arguments.
eq (9) why to the power of t, where does it come from? Or is is a superscipt, maybe transposed with the period indicating the dot product ?
l 235-239, eq(10) not clear where the uncertainty is formulated or is \hat(beta) the uncertainty. It is described as estimator of beta. What is X in (10)?
Figure 1 What is the time period? April 2002 to Dec 2020 for all estimates here?
l 249 What is the expectation here. The hypothesis from the Introduction is that the present estimate is the better product than NOC or ISAS21. How does one use this comparison to validate this assumption? I can believe that OVIDE is the better yet sparse product that can be used to validate your product, if you show that your product has smaller rms to OVIDE than ISAS21 you could make a point, otherwise one could point to greater detail but also larger noise in your product. I am missing here some conclusions for this comparison.
I only take that it is an alternative, but that's not how you started out.l 288-289 Isn't the given reason and the pattern the same thing?
l 297 Where is the value 3 Wm2 shown? Maybe it would be better to show the actual value of the contributions to the uncertainty rather then the percent due to MAN.
l 307-308 Doesn't the evaluation assume ISAS21 to be the better product, contradicting the premise of the manuscript ?
l 317-318 Maybe useful to report also the correlation between ISAS21 and OVIDE
Fig.5 Add longitude range from the text here.
L 353 Isn't 15 W/m2 only for the subtropical gyre while negative values apply for the subpolar gyre? Maybe you want to say order 15W/m2?
L 360-361 I think nothing has shown so far. The information comes in the next sentence. Maybe you want to reformulate these sentences and discuss Figs E1/F1 a little more.
Table 1 not clear what time_covar is.
Citation: https://doi.org/10.5194/essd-2023-236-CC1 -
RC1: 'Comment on essd-2023-236', Anonymous Referee #1, 08 Dec 2023
Review of
Monitoring the regional Ocean Heat Content change over the Atlantic Ocean with the space geodetic approach
by
Rousseau, V., et al.
Summary: This manuscripts aims to present a novel data set of the regional ocean heat content change over the Atlantic Ocean derived using a synergistic approach that combines the space geodetic approach with in-situ observations. One of the key novelties is the attempt to handle regional uncertainty contribution in an improved way and to provide ocean heat content trends and their uncertainties along with the ocean heat content itself. The resulting data are intercompared for two highy dynamic regions in terms of ocean heat content change with independent estimates of the ocean heat content based on in-situ and ARGO float measurements.
General comments:
GC1: I have to admit I am not sure whether ESSD is the correct journal to publish this manuscript. The overwhelming focus of the manuscript is on geophysics, the current OHC conditions and its changes. The manuscript appears not to be sufficiently well tailored to be published in a data set journal. This begins already with the title, continues with an abstract which highlights geophysics rather than issues of the data set and its "evaluation", and continues with a too my opinion insufficient focus on describing the limitations of the input data and the methodological approach. The conclusion is far too long for a manuscript that aims to present and publish a data set - which ESSD is for.
GC2: While the data set does already exist it is not yet in a format that satisfies CF- and other conventions sufficiently well. See my specific comments to the data set at the end of my review.
GC3: While the manuscript is partly motivated with telling the reader that you will come up with regional uncertainty estimates I do not find this motivation is convincingly followed throughout the manuscript. Neither are the uncertainties actually shown in the manuscript nor is a comprehensive effort invested to understand the results - particular when it comes to the different contributions to the uncertainty derived. Whether the uncertainties provided are considered of credible magnitude (and why) is not laid out in a convincing way.
GC4: From a technical view-point I find what is sold as "validation" at best an intercomparison between the product itself with some other products that have inherent uncertainties and limitations of use which I did not find discussed in a convincing manner. Here the manuscript has some room for improvement.Specific comments:
L39-41: Do quantitative estimates exist of the amount of heat uptake that is neglected by using the Argo network of profiling floats?L50: It would be good to tell the reader which elements the manometric sea level change has and which of these elements are of particular relevance for your work. Since my understanding of manometric sea level change includes - among others - also fast fluctuating atmospheric loading, and precipitation components as well as boundary current influences - which potentially cannot be resolved (well) with spaceborne gravimetric methods, it would be good to know which are the main processes you are considering here.
L50-52: Does this approach also work in the presence of sea ice? How? Is it as accurate there as it is over open water?
And: How close can one get to the coasts with this method?L83: While I understand that the IEEH coefficient is an integrated quantity, I have difficulties to understand how it takes into account pressure, salinity, water composition, and temperature dependent variations into account; heat expansion and salt contraction coefficients do have these dependencies.
L124/125: These ocean mass changes near Greenland and Antarctica, what is their assumed main cause? Is it freshwater input and hence salinity change? Or is this a correction with respect to the gravitational forces of these two bodies on the adjacent oceans? Please add this piece of information.
L125: While temporal and spatial resolution for SLA data is given, it is not provided here. I assume the spatial resolution of the MAN data is considerably coarser. Please provide the missing information.
L149/150: "In this present study" refers to this paper here, right? Why were these uncertainties not propagated?
L162-167: Isn't this a very crude over-simplification? I mean, whereever the regional (or local) IEEH is smaller than the global one you say that the regional (or local) uncertainty in IEEH is smaller than the global uncertainty whereas whereever the regional IEEH is larger than the global one the uncertainty in IEEH is larger than the global uncertainty. Does this make sense? Aren't the smaller coefficients not often prone to larger uncertainties? I find this not yet convincing and suggest that you provide more details as to why you think this is a viable approximation.
L172-173: "over the depths 0-2000m" --> If the NOC data set is limited to these depths, how are you then able to evaluate a fully vertically integrated product for places with water depths exceeding 2000m?
L183: "relying on ... bias" --> Not sufficiently clear. Is this additive bias the OHC contribution from depths below 2000 m then, those depths the NOC Argo OI data set does not cover?
I invite the authors to provide information about the location and water depth of this evaluation site.L225-228: "Moreover, ..." --> I was wondering whether this issue applies to your data set and how critical it is that there exists no quantification of this source of error. Does it have the potential to jeopardize your uncertainty estimates by potentially having a much larger uncertainty contribution than thought so far?
L248-258: "SSL change ... OVIDE section" --> I have a few comments to this pararaph and the associated figures.
- The first thing is that I am stumbling about the term "validation". What I see in the figures are comparisons of two products and not comparisons of a product with independent high accuracy in-situ measurements which would eventually warrant usage of the term "validation". Modifying the wording might help.
- Second, showing two maps side by side is good but showing a difference map of the trends, therein also highlighting which grid cells are actually giving statistically significant trends, would be of advantage. You could also provide a scatterplot and/or a heat map of the differences on the trends.
- The time series in Fig. 2c I find interesting in that sense that i) the few ship-based observations seem to match well most of the time but that ii) the other two data sets show clearly phases where they are in very good agreement but also many phases where the agreement is not that convincing. I was wondering whether you could comment on that.L268: Ok, you tested the choice of the time reference but within which temporal limits?
L297: If I understood you correctly then one of the main innovations of this paper is the derivation of OHC trend uncertainties and you invested a lot of space in the paper describing how you computed it. I strongly recommend to not only show a map of the MAN contribution in the supplementary figure D1 but to also show a map of the OHC trend uncertainties alongside Fig. 3.
Browsing through the data I also became aware of the distribution of both, the trend and its uncertainty. Based on this I recommend to also show a map of the relative uncertainty in the trend because as written on can get the impression that you are talking about 3 W m-2 uncertainty max associated also with the maximum trends, i.e. about 15 Wm-2, hence 20% relative uncertainty at max. But the truth is different and I bet there is a considerable area where the relative uncertainty in the trend is 50% or more.L298/299: "with a contribution ..." --> I can see that you would be after further studies that could determine which contribution in the MAN uncertainties play the largest role. However, do you understand why this MAN contribution is so dominant? Almost 100% in the west of the Atlantic basin is something. Why is this? Readers might appreciate some comments on this in the written text.
Figure 4: Also in this case a reader might appreciate to see a difference map and/or a scatterplot or heat map of the differences between the two products shown in panels a) and b). I repeat my comment about the colors of the OVIDE section in panels a) and b) and the snapshots in panel c). The unit of the trends lack the reference time period.
L365/366: "continuous measurements of the full ocean" seems a bit weird in light of an altimeter which essentially senses along-track with a km-scale footprint with 14 overpasses per day distributed over the full globe. This is far from being "continuous" or "full". I suggest to rephrase accordingly.
L385/386: Uncertainties of the IEEH have only been estimated at global scale, true, but you suggested how to regionalize them. This approach, however, has not been evaluated further in this work and should be among the next steps to pursue.
Table 1:
- The variable name is "ohc" - aka Ocean heat content but the description is Ocean heat content CHANGE ... If it is a change than it possibly also involves a time scale but this is not reflected in the units.- The units of ohc_trend and ohc_trend_uncertainty lack the reference time dimension.
Figure E1: I note the trends derived from the two different products are identical - unlike in Figure 4. You might want to comment on that in the text. Possibly this is how it should be. It might also make sense to stress that in case of the time mean IEEH obviously the error envelope increases considerably.
Typos / Editoral comments:
L103: The "and S&-MF very soon" is not relevant for this paper and should be deleted.L118: Typo: ensenble --> ensemble
Figure 2: Just a small editoral hint: The cruise-based snapshots in panel c) could be aligned more directly with the OVIDE sections in panels a) and b) if both would be in the same color. Currently you use green AND red. You could pick one.
L311: I am not sure I understand where this ~-2-5 belongs to. Is this a difference? Is the sign negative or positive? Not clear.
L318: Something is "overestimated" by a negative value ... appears strange to me, please check.
L385, typo: "futur" --> "future"
Comments to the netCDF file- I recommend to add the attribute missing_value to all variables that are not coordinates, i.e. not the time and latitude/longitude variables
- It is rather unusual to have a _FillValue for the time and latitude/longitude coordinates. Are there really missing values? Usually one has a complete coverage of the region of interest with grid cells and only the value of the respective geophysical variable is missing and gets a _FillValue and a missing_value
- I recommend to not use NaN as the entry for _FillValue (and/or missing value) because it is not a numerical value; as far as I know the _FillValue or missing_value should be an element from within the range of values that is allowed by the respective data type, i.e. for example -127b for datatype byte and -32767 for datatype short integer.
- You need to check all your standard names for compatibility with the CF-convention standard name tables. I tried to find "ocean_heat_content" for instance but it does not exist.
- Units like W/m² should read W m-2
- The attribute name "unit" has a "string " prepended. This needs to be removed.
- I did not check the netCDF file using available CF-convention compatibility checkers but am sure - in light of the list of global attributes seen - that there is room for improvement and complementing of the global attributes as well.
- I note that the time axis in the data is not consistenly pointing at the same day of the month of which you are showing the mean value. Currently it is from the 15th (Feb.) to the 18th (March) with a yet not sufficiently well motivated switch between the 16th and the 17th for the other months. I recommend to correct that.
- Why is there a shift from more positive values in the first two thirds of the record to more negative values later on for variable ohc_var_covar_matrix?
- What explains the meridional banding in the values of the same variable (e.g. in 2009 and 2011)?
Citation: https://doi.org/10.5194/essd-2023-236-RC1 -
EC1: 'Comment on essd-2023-236', Dagmar Hainbucher, 21 Feb 2024
As topic editor, I would like to point out to the authors that I see a considerable effort to revise the paper. Both reviewers see considerable shortcomings not only in the methods used but also in the credibility of the outcomes. In parts, I think the paper has to be completely rewritten to meet the demands of the reviewers and the journal. After the revision, a further review process would be carried out by the two reviewers.
Citation: https://doi.org/10.5194/essd-2023-236-EC1
Interactive discussion
Status: closed
-
CC1: 'Comment on essd-2023-236', Armin Koehl, 28 Aug 2023
The manuscript describes a technique to estimate heat content changes from satellite data together with in situ data. The estimate include uncertainty estimation and a comparison to alternative estimates. The product is available to the public, and as such the manuscript is worth to be published. However, I feel that the methodology is not sufficiently complete or well enough described to allow the reader to reproduce the estimation of the ocean heat content (OHC).
The large contribution to the error estimate by manometric contribution is surprising given that this component shows very little contribution to the dynamic sea level change (see for instance Landerer etal (doi.org/10.1175/JPO3013.1) or Couldrey et al (doi.org/10.1007/s00382-020-05471-4)). Dynamic sea level certainly does not contain the important signals related to melting, groundwater change and GIA.
However, this outcome is an argument against this technique and suggest that it may be actually inferior to calculating heat content change directly from temperature data.There is discussion about the different contributions to the uncertainty in the conclusion. Data gaps are reported as the main issue with the technique, while the estimated uncertainties relate mostly to the manometric contribution from the gravity missions. Their procedure was introduced in several ways to be superior to alternative estimates based on for instance temperature data alone, but then their product is evaluated with these kind of products. In the end, I felt a bit lost as how to evaluate the quality and merits of the product and what the true uncertainties are. One merit is the estimate of uncertainty but then I get the impression that these cannot be trusted since the true uncertainty may relate also to contributions not considered. The authors dedicated a substantial part of the conclusion to the discussion of uncertainties, and I believe it will be hard to provide additional details. However as detailed below, there are some places, for instance in the introduction, the comparison to alternative estimates and the presentation of the uncertainies, where some more details or interpretations could be presented.
Details:L 38 What is the role of salinity here?
L68 Why is this uncertainty estimate sufficient ?
Doesn't the contribution of the halosteric signal to the uncertainty need to be considered too? estimating a signal from the difference instead directly from the thermosteric sounds at first a bad idea since the uncertainties add. Add a argument why your approach is still a good idea.L 61-68 Maybe it would be useful to contrast advantages with disadvantage. The full depth estimate of the thermosteric signal assumes that the halosteric signal is known or zero which is not the case. Maybe anticipate some of the caveates discussed later, to provide a more blanced view.
L 78 Please define abbreviation TSSL
eq (2) There needs to be different identifiers for TSSL from temperature and from the geodetic approach.
L 87-88 Since the expansion coefficient is strongly temperature dependent, it is important to state the reference state assumed for the expansion coefficient. Assuming is 0C and 35 PSU would be a very bad choice, because coefficients near the reference state will be very different from the actual coefficient in a substantial part of the ocean. I think (2) makes only sense for a linear density equation, because expansion coefficient may be very different for a whole range from 0 to the actual temperature, while the changes may be mostly be around the actual temperature and very different from 0.
l 31 Not clear how the changing temperature is used to calculate halosteric sea level.
l 143 How is merging done?
l 153-154 Some extra details would be helpful since the brief introduction above gave the impression that this may not have been done in the best way,
Section 3.3.3 Also for the validation products it would be useful to know how the salinity data has been used.
L 199 Please specify in what way the method was extended. Check the grammar of the sentence.
l 201-202 Not clear how the variance covariance matrices are defined. Please provide a definition, is covariance only in time? What do the two arguments t stand for?
l 208, eq (5) Not clear which terms are in that budget. What is i indicating?
l 210 Not clear how this works, this is dispersion in time?
Maybe also a comment on how independent the members of the ensemble are and how that affects the estimate.l 211 How is the additional uncertainty computed?
eq (6) Not clear to me over which dimension the cov is calculated that the result depends on i,j,lat, lon and t. And why there are two t arguments.
eq (9) why to the power of t, where does it come from? Or is is a superscipt, maybe transposed with the period indicating the dot product ?
l 235-239, eq(10) not clear where the uncertainty is formulated or is \hat(beta) the uncertainty. It is described as estimator of beta. What is X in (10)?
Figure 1 What is the time period? April 2002 to Dec 2020 for all estimates here?
l 249 What is the expectation here. The hypothesis from the Introduction is that the present estimate is the better product than NOC or ISAS21. How does one use this comparison to validate this assumption? I can believe that OVIDE is the better yet sparse product that can be used to validate your product, if you show that your product has smaller rms to OVIDE than ISAS21 you could make a point, otherwise one could point to greater detail but also larger noise in your product. I am missing here some conclusions for this comparison.
I only take that it is an alternative, but that's not how you started out.l 288-289 Isn't the given reason and the pattern the same thing?
l 297 Where is the value 3 Wm2 shown? Maybe it would be better to show the actual value of the contributions to the uncertainty rather then the percent due to MAN.
l 307-308 Doesn't the evaluation assume ISAS21 to be the better product, contradicting the premise of the manuscript ?
l 317-318 Maybe useful to report also the correlation between ISAS21 and OVIDE
Fig.5 Add longitude range from the text here.
L 353 Isn't 15 W/m2 only for the subtropical gyre while negative values apply for the subpolar gyre? Maybe you want to say order 15W/m2?
L 360-361 I think nothing has shown so far. The information comes in the next sentence. Maybe you want to reformulate these sentences and discuss Figs E1/F1 a little more.
Table 1 not clear what time_covar is.
Citation: https://doi.org/10.5194/essd-2023-236-CC1 -
RC1: 'Comment on essd-2023-236', Anonymous Referee #1, 08 Dec 2023
Review of
Monitoring the regional Ocean Heat Content change over the Atlantic Ocean with the space geodetic approach
by
Rousseau, V., et al.
Summary: This manuscripts aims to present a novel data set of the regional ocean heat content change over the Atlantic Ocean derived using a synergistic approach that combines the space geodetic approach with in-situ observations. One of the key novelties is the attempt to handle regional uncertainty contribution in an improved way and to provide ocean heat content trends and their uncertainties along with the ocean heat content itself. The resulting data are intercompared for two highy dynamic regions in terms of ocean heat content change with independent estimates of the ocean heat content based on in-situ and ARGO float measurements.
General comments:
GC1: I have to admit I am not sure whether ESSD is the correct journal to publish this manuscript. The overwhelming focus of the manuscript is on geophysics, the current OHC conditions and its changes. The manuscript appears not to be sufficiently well tailored to be published in a data set journal. This begins already with the title, continues with an abstract which highlights geophysics rather than issues of the data set and its "evaluation", and continues with a too my opinion insufficient focus on describing the limitations of the input data and the methodological approach. The conclusion is far too long for a manuscript that aims to present and publish a data set - which ESSD is for.
GC2: While the data set does already exist it is not yet in a format that satisfies CF- and other conventions sufficiently well. See my specific comments to the data set at the end of my review.
GC3: While the manuscript is partly motivated with telling the reader that you will come up with regional uncertainty estimates I do not find this motivation is convincingly followed throughout the manuscript. Neither are the uncertainties actually shown in the manuscript nor is a comprehensive effort invested to understand the results - particular when it comes to the different contributions to the uncertainty derived. Whether the uncertainties provided are considered of credible magnitude (and why) is not laid out in a convincing way.
GC4: From a technical view-point I find what is sold as "validation" at best an intercomparison between the product itself with some other products that have inherent uncertainties and limitations of use which I did not find discussed in a convincing manner. Here the manuscript has some room for improvement.Specific comments:
L39-41: Do quantitative estimates exist of the amount of heat uptake that is neglected by using the Argo network of profiling floats?L50: It would be good to tell the reader which elements the manometric sea level change has and which of these elements are of particular relevance for your work. Since my understanding of manometric sea level change includes - among others - also fast fluctuating atmospheric loading, and precipitation components as well as boundary current influences - which potentially cannot be resolved (well) with spaceborne gravimetric methods, it would be good to know which are the main processes you are considering here.
L50-52: Does this approach also work in the presence of sea ice? How? Is it as accurate there as it is over open water?
And: How close can one get to the coasts with this method?L83: While I understand that the IEEH coefficient is an integrated quantity, I have difficulties to understand how it takes into account pressure, salinity, water composition, and temperature dependent variations into account; heat expansion and salt contraction coefficients do have these dependencies.
L124/125: These ocean mass changes near Greenland and Antarctica, what is their assumed main cause? Is it freshwater input and hence salinity change? Or is this a correction with respect to the gravitational forces of these two bodies on the adjacent oceans? Please add this piece of information.
L125: While temporal and spatial resolution for SLA data is given, it is not provided here. I assume the spatial resolution of the MAN data is considerably coarser. Please provide the missing information.
L149/150: "In this present study" refers to this paper here, right? Why were these uncertainties not propagated?
L162-167: Isn't this a very crude over-simplification? I mean, whereever the regional (or local) IEEH is smaller than the global one you say that the regional (or local) uncertainty in IEEH is smaller than the global uncertainty whereas whereever the regional IEEH is larger than the global one the uncertainty in IEEH is larger than the global uncertainty. Does this make sense? Aren't the smaller coefficients not often prone to larger uncertainties? I find this not yet convincing and suggest that you provide more details as to why you think this is a viable approximation.
L172-173: "over the depths 0-2000m" --> If the NOC data set is limited to these depths, how are you then able to evaluate a fully vertically integrated product for places with water depths exceeding 2000m?
L183: "relying on ... bias" --> Not sufficiently clear. Is this additive bias the OHC contribution from depths below 2000 m then, those depths the NOC Argo OI data set does not cover?
I invite the authors to provide information about the location and water depth of this evaluation site.L225-228: "Moreover, ..." --> I was wondering whether this issue applies to your data set and how critical it is that there exists no quantification of this source of error. Does it have the potential to jeopardize your uncertainty estimates by potentially having a much larger uncertainty contribution than thought so far?
L248-258: "SSL change ... OVIDE section" --> I have a few comments to this pararaph and the associated figures.
- The first thing is that I am stumbling about the term "validation". What I see in the figures are comparisons of two products and not comparisons of a product with independent high accuracy in-situ measurements which would eventually warrant usage of the term "validation". Modifying the wording might help.
- Second, showing two maps side by side is good but showing a difference map of the trends, therein also highlighting which grid cells are actually giving statistically significant trends, would be of advantage. You could also provide a scatterplot and/or a heat map of the differences on the trends.
- The time series in Fig. 2c I find interesting in that sense that i) the few ship-based observations seem to match well most of the time but that ii) the other two data sets show clearly phases where they are in very good agreement but also many phases where the agreement is not that convincing. I was wondering whether you could comment on that.L268: Ok, you tested the choice of the time reference but within which temporal limits?
L297: If I understood you correctly then one of the main innovations of this paper is the derivation of OHC trend uncertainties and you invested a lot of space in the paper describing how you computed it. I strongly recommend to not only show a map of the MAN contribution in the supplementary figure D1 but to also show a map of the OHC trend uncertainties alongside Fig. 3.
Browsing through the data I also became aware of the distribution of both, the trend and its uncertainty. Based on this I recommend to also show a map of the relative uncertainty in the trend because as written on can get the impression that you are talking about 3 W m-2 uncertainty max associated also with the maximum trends, i.e. about 15 Wm-2, hence 20% relative uncertainty at max. But the truth is different and I bet there is a considerable area where the relative uncertainty in the trend is 50% or more.L298/299: "with a contribution ..." --> I can see that you would be after further studies that could determine which contribution in the MAN uncertainties play the largest role. However, do you understand why this MAN contribution is so dominant? Almost 100% in the west of the Atlantic basin is something. Why is this? Readers might appreciate some comments on this in the written text.
Figure 4: Also in this case a reader might appreciate to see a difference map and/or a scatterplot or heat map of the differences between the two products shown in panels a) and b). I repeat my comment about the colors of the OVIDE section in panels a) and b) and the snapshots in panel c). The unit of the trends lack the reference time period.
L365/366: "continuous measurements of the full ocean" seems a bit weird in light of an altimeter which essentially senses along-track with a km-scale footprint with 14 overpasses per day distributed over the full globe. This is far from being "continuous" or "full". I suggest to rephrase accordingly.
L385/386: Uncertainties of the IEEH have only been estimated at global scale, true, but you suggested how to regionalize them. This approach, however, has not been evaluated further in this work and should be among the next steps to pursue.
Table 1:
- The variable name is "ohc" - aka Ocean heat content but the description is Ocean heat content CHANGE ... If it is a change than it possibly also involves a time scale but this is not reflected in the units.- The units of ohc_trend and ohc_trend_uncertainty lack the reference time dimension.
Figure E1: I note the trends derived from the two different products are identical - unlike in Figure 4. You might want to comment on that in the text. Possibly this is how it should be. It might also make sense to stress that in case of the time mean IEEH obviously the error envelope increases considerably.
Typos / Editoral comments:
L103: The "and S&-MF very soon" is not relevant for this paper and should be deleted.L118: Typo: ensenble --> ensemble
Figure 2: Just a small editoral hint: The cruise-based snapshots in panel c) could be aligned more directly with the OVIDE sections in panels a) and b) if both would be in the same color. Currently you use green AND red. You could pick one.
L311: I am not sure I understand where this ~-2-5 belongs to. Is this a difference? Is the sign negative or positive? Not clear.
L318: Something is "overestimated" by a negative value ... appears strange to me, please check.
L385, typo: "futur" --> "future"
Comments to the netCDF file- I recommend to add the attribute missing_value to all variables that are not coordinates, i.e. not the time and latitude/longitude variables
- It is rather unusual to have a _FillValue for the time and latitude/longitude coordinates. Are there really missing values? Usually one has a complete coverage of the region of interest with grid cells and only the value of the respective geophysical variable is missing and gets a _FillValue and a missing_value
- I recommend to not use NaN as the entry for _FillValue (and/or missing value) because it is not a numerical value; as far as I know the _FillValue or missing_value should be an element from within the range of values that is allowed by the respective data type, i.e. for example -127b for datatype byte and -32767 for datatype short integer.
- You need to check all your standard names for compatibility with the CF-convention standard name tables. I tried to find "ocean_heat_content" for instance but it does not exist.
- Units like W/m² should read W m-2
- The attribute name "unit" has a "string " prepended. This needs to be removed.
- I did not check the netCDF file using available CF-convention compatibility checkers but am sure - in light of the list of global attributes seen - that there is room for improvement and complementing of the global attributes as well.
- I note that the time axis in the data is not consistenly pointing at the same day of the month of which you are showing the mean value. Currently it is from the 15th (Feb.) to the 18th (March) with a yet not sufficiently well motivated switch between the 16th and the 17th for the other months. I recommend to correct that.
- Why is there a shift from more positive values in the first two thirds of the record to more negative values later on for variable ohc_var_covar_matrix?
- What explains the meridional banding in the values of the same variable (e.g. in 2009 and 2011)?
Citation: https://doi.org/10.5194/essd-2023-236-RC1 -
EC1: 'Comment on essd-2023-236', Dagmar Hainbucher, 21 Feb 2024
As topic editor, I would like to point out to the authors that I see a considerable effort to revise the paper. Both reviewers see considerable shortcomings not only in the methods used but also in the credibility of the outcomes. In parts, I think the paper has to be completely rewritten to meet the demands of the reviewers and the journal. After the revision, a further review process would be carried out by the two reviewers.
Citation: https://doi.org/10.5194/essd-2023-236-EC1
Data sets
Atlantic OHC from space: Heat content change over the Atlantic Ocean by space geodetic approach Magellium/LEGOS https://doi.org/10.24400/527896/A01-2022.012
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
628 | 263 | 54 | 945 | 54 | 58 |
- HTML: 628
- PDF: 263
- XML: 54
- Total: 945
- BibTeX: 54
- EndNote: 58
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1