the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 18 Oct 2022
A method to enhance the detecting of geostrophic current and its temporal variations with SWOT swath data
Abstract. The Surface Water and Ocean Topography (SWOT) mission, which can map the sea surface height with high spatial and temporal sampling rates simultaneously, has significant potential for detecting mesoscale and submesoscale eddy variations. At present, in the determination of geostrophic current from nadir altimeter or SWOT swath data, the optimal interpolation method is usually used to grid the observations with the space-time covariance function and use a percentage of the signal variance to reduce the long-wavelength error. However, this optimal interpolation method used for nadir altimeters may not be optimal for SWOT as the spatial and temporal characteristics is different. In this study, we propose to first derive the geostrophic currents in each swath from absolute dynamic topography by difference, to reduce the long-wavelength error which is constant along the tracks. And then, based on the temporal characteristics of the signal expected to be detected and high spatial SWOT observations, the spatial covariance function is used only to get the gridded geostrophic currents. The accuracy of the proposed method is verified by one year of simulated data in the Sea of Japan using MITgcm LLC4320 model and the SWOT errors. Compared with the absolute dynamic topography and geostrophic current from LLC4320 model, using the simulated data including errors, the proposed method makes high spatial sampling more effective and can obtain gridded absolute dynamic topography and geostrophic current with better accuracy especially when the number of observations is limited. In terms of the temporal variations of eddy kinetic energy, this method can significantly improve the reconstruction and detected temporal scales of mesoscale eddy variations.
This preprint has been withdrawn.
-
Withdrawal notice
This preprint has been withdrawn.
-
Preprint
(1319 KB)
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2022-1018', Anonymous Referee #1, 12 Dec 2022
A method to enhance the detecting of geostrophic current and its temporal variations with SWOT swath data Shi et al. (2022)
General Comments
The authors present an interesting article detailing methods and improvements for retrieval of geostrophic currents from simulated SWOT observations of the sea surface height, which are expected to contain large magnitude correlated errors.
There seem to be three aspects to the paper: the filtering of swath ADT to suppress long-wavelength (spatially-correlated) errors, testing updated space and time correlation scales in the OI, and comparing the different methods (ADT_STC, UV_STC, and UV_SC) against one another.
The methods presented demonstrate the benefit of filtering the swath ADT (which appears to remove much of the short length-scale noise), and the benefit of calculating the gradient in the ADT fields (to calculate the u/v geostrophic currents) which appears to address the limitation imposed by the large-scale correlated errors in the SWOT ADT.
However, one stated objective of the paper was to demonstrate the need to adjust the spatial and temporal scales in the OI to be more appropriate for SWOT. It was not clear how the spatial and temporal scales used were chosen and whether the two methods (UV_SC and UV_STC) compared the currently used space/time scales with the proposed update, or not. Overall, I found the description of the methods to be somewhat confusing: it was not clear to me how the ADT was produced using the UV_STC method, nor was is clear how the UV_SC and UV_STC methods differ in terms of time-range of inputs and quantity of observations included.
I believe this work is timely and will be of interest to the community, but the description and comparison of the methods (ADT_STC, UV_STC, UV_SC) must first be clarified.
Specific Comments
Line 17: “the long wavelength error (LWE) is constant along tracks”. I don’t think this is true: the LWEs are correlated along track, but are not constant.
Line 44: “spatial resolution of…the resolution of” I think this would more correctly be ”spatial sampling of…the posting of”. The expected feature resolution of SWOT is 15-30km, but the observations will sample much more frequently and the supplied product will have many observation points (every 2km) per resolution element (15-30km).
Line 61-62: “the long wavelength errors” Which of the error components due you include in this definition? Roll and phase only?
Line 62: “constant along track” Same comment as for line 17.
Line 64: “the difference step” I think here you mean that the long-wavelength errors do not affect the short-scale spatial derivatives of the ADT used in the calculation of the geostrophic velocities, but I think you could be more explicit in this at the first mention of the “difference step”.
Line 67: “swath width of 50km” I would suggest to clarify that this is only one side of the full swath.
Figure 1: The rainbow colour-scale used is hard to interpret. I’d suggest a diverging colour-scale (e.g., red to white to blue) to emphasise where the magnitude of the errors is largest.
Figure 1: The range of the colour bar is limited to +/-5cm, but the text notes that the phase and roll errors are at the decimetre level. From my experience of these errors, +/-20cm could be necessary to show the range of the phase/roll errors.
Line 182: It’s unclear why a time window of +/-5days is chosen. While the precise choice may be arbitrary, given the timescale of 3.6 days mentioned earlier, some justification seems necessary.
Figure 3: Is the time period used for the figure to match the +/-5day window mentioned before?
Figure 3: The power spectra show that the filtering removes the excess power (and the signal) at scales shorter than ~20km, but there appears to be more small-scale structures present in Fig 3c than in Fig 3a. How is this consistent?
Line 212: The title of this section “Effect of difference” could be improved as the section discusses both the effect of deriving the u/v currents (the difference method) and Gaussian-filtering the ADT. I think it would also be clearer if the “difference method” were explicitly defined somewhere.
Line 213: Is the Gaussian filter applied two-dimensionally, or only in the along-track direction? If 2D, does this introduce artefacts at the swath edge? And if along-track, are there artefacts at the end of tracks?
Line 213: Why was a radius of 14km chosen? You stated that the correlation scale of the model was 27km and the apparent correlation scales of the observations errors is much longer.
Table 1: RRMSE is not widely used, so it would be useful to refer to equation 9 in the table caption.
Line 228: Same question as for line 182.
Line 228: Why was a 0.1degree grid chosen? If the expected limit of feature resolution is ~100km, then you would require 2-3 grid cells per resolution element to retain that resolution.
Figure 4: How was the ADT estimated using the UV_STC method?
Figure 4: The difference fields shown in panels (d) and (e) would be clearer if the colour-scale were centred on zero. This would make it easier to interpret where the bias was very small, and/or whether there is a domain-wide bias.
Line 243: It is not clear to me exactly how the UV_SC and UV_STC methods differ in terms of time-range of inputs and quantity of observations included. Is the UV_SC method equivalent to UV_STC, but with a different time window (+/-5days)?
Line 258: Related to comment on line 243: is the higher correlation between the RMSE and number of observations for UV_STC due to using many fewer observations than in UV_SC?
Technical Corrections/Suggestions
Title: suggest replacing “detecting” with “detection”
Line 15: “is different” -> “are different”
Line 26: “Satellite altimetry” -> “Satellite altimetry has”
Line 27: “Gridding method should be” -> “Gridding methods have been”
Line 41: “interferometric” -> “interferometer”
Line 65: “T/P” Perhaps expand to “TOPEX/Poseidon”
Line 67: “the globe ocean” -> “the whole global ocean”
Line 76: This first sentence could be rephrased to clarify the meaning.
Line 100: “the reality” -> “reality”
Line 106: “the other instrument error is” -> “the other instrument errors are”
Line 107 “those error is” -> “those errors are”
Line 108: “spatially structured” -> “spatial structure”
Line 108: “is constant” -> “are constant”
Line 111: “errors are…cut” -> “errors being...cuts”
Line 135: The omega symbol is used mistakenly for the meridional direction (as well as the angular velocity).
Line 187: “A Global Self-consistent” -> “the Global Self-consistent”
Figure 4: “using ADT_STC and UV_STC method” -> “using the ADT_STC and UV_STC methods”
Figure 4: “The ADT of model” -> “The ADT of the LLC4320 model”
Line 255: “on the overall” -> “overall”
Line 275: “underestimate the gridded results of model” -> “underestimate the magnitude of the gridded model ADT”
Line 280: “then” -> “than”
Figure 7: The panels references need to be updated: a,b,c,d used in figure, a,c,e,g used in caption.
Line 321: “To adapt…” This sentence should be rephrased.
Line 323: “in the swath” -> “of the swath ADT”
Line 337: C3S has not been defined.
Citation: https://doi.org/10.5194/egusphere-2022-1018-RC1 -
RC2: 'Comment on egusphere-2022-1018', Anonymous Referee #2, 28 Dec 2022
## Review
* Shi et al, 2022
* Title: A method to enhance the detecting of geostrophic current and its temporal variations with SWOT swath data
* Submitted to egusphere
* december 2022## Summary
The paper introduces and discusses a method to map geostrophic current from the recently launched SWOT mission dedicated to wide-swath altimetry, using statistical interpolation. The novelty (w.r.t. CMEMS products made with the DUACS system, in particular) is to derive geostrophic currents on the swath before mapping. With nadir altimetry, SSH must be gridded before deriving geostrophic currents. The authors argues that this strategy reduces the consequences of long-wavelength errors in SSH through finite differences, and provides more accurate currents that the DUACS approach.
## General statements
* The writing is too aproximative for an accurate understanding, especially for non-native English speakers like me, not able to meta-understand the meaning of the text. Errors include wrong or inconsistent tenses, inapropriate vocabulary, missing or inadequate articles before nouns, incomplete sentences, inaccurate wording, etc. A thorough proofreading is essential.
* The text is not at a level appropriate for scientific publication in the present form: many basic information is unecessarily provided (geostrophic equations, gaussian filter equations, etc) and many previous articles on the topic are completely ignored.
* The focus of the paper is not well defined. Is it the issue of SWOT temporal sampling and the related time scales involved in OI? Or a new method to derive gridded products to overcome the long-wavelength error issue? Or the reconstruction of time variations of EKE? None is satisfyingly addressed.
* If the focus is on a new mapping method, standard performance metrics must be adopted. Power Spectrum Densities are common in the SWOT community because SWOT specifications are formulated in terms of spectra. The authors should dive into the large body of litterature on this topic.
* * if the focus is on the removal of long-wavelength errors, diagnostics must be formulated specifically. The along-track and across-track directions musy be separately addressed, given the structure of SWOT errors.
* Perfoming diagnostics with weekly means is inconsistent with the spatial resolution considered in the paper, with SWOT, and with the standard products (DUACS from CMEMS).I think the paper needs more a rewriting than a revision. For this reason I recommend rejection. Below are line-by-line notes that might be useful to the authors for a future submission.
### Abstract
### Introduction
* L30: Esteban-Fernandez et al is not the appropriate reference to support the previous consideration on EKE.
* L39: SWOT is now on orbit. Update tenses!
* L44: The data with resolution 10-70 m are the L1 level, useless for ocean science as is. This mention is misleading for the reader not familiar with SWOT and should be removed.
* L45: the standard is 2 km.
* L45-49 are unclear. The first sentence seems incomplete. The second one is awkward. Do the authors want to say that the small mesoscale features are blurred by SWOT errors?
* L53: what is "the accuracy of the grid"?
* L57: I think that the 2020 paper by Gomez Navarro et al (Remote Sensing, 2020) is a better reference for SWOT denoising.
* L59-65: It is not clear where this discussion leads to. On the one hand, it is suggested that the long-wavelength error is constant along the track, but induces spurious gradients across-track: why? across-track variations of errors are ignored, and the discussion ends on the removal of long-wavelength errors by computing SSH gradients. What about across-track gradients? (I came back to these lines after reading lines 104-114 and figure 1, with even more confusion)
* L65: "the accuracy of the grid". Do you mean "of the gridded product"?
* L82: it should be said how time interpolation is performed if Gaussian covariances are not used anymore. I see it at Line 182, it is too late.### Data and methods
* Lines 104-114 are repeating other papers, or Esteban-Fernandez' SWOT report. It should be acknowledged and kept only if this information is used later to shed light on some results. By the way, where is the "long-wavelength error" cited several times earlier in the text?
* I do not find it relevant to remind the geostrophic expressions, except for introducing notations or specific discussions on expressions. I do not think it is the case.
* Lines 142-145: The unbalanced wave motions are filtered out anyway, by the time averaging.
* Section 2.3: I do not think another description of the OI method is needed. It would be more relevant to cite the DUACS system and emphasize the differences. Also, the very first sentence of the section ("The OI method can be used...") does not clearly state that OI is used in this study.
* Lines 200 and below: I do not think it necessary to recall what a Gaussian filter is.### Results and discussion
* L219-220: Should I understand that the RMSE in velocity is quantitatively compared with the RMSE in ADT?
* L222: I do not think that the simple RRMSE scores allow to conclude that the mesoscale signal of currents is recovered.
* L229: The method to recover ADT from U and V after mapping geostrophic currents has not been described. It is the first time, I think, that the text refers to this process.
* L233: it is stated that the long-wavelength error is constant along-track, but this does not seem consistent with figure 1.
* L243-246: to my opinion, these lines belong to the method section.
* Why performing evaluation on weekly averaged products? The spatial resolution is 0.1, what makes the space and time resolutions fully inconsistent. What is the point of such products? For information, DUACS products are delivered daily at 0.25 resolution.
* L273-281: The analysis of the reason why UV\_SC performs better than UV\_STC seems far too simplistic to me. At these latitudes, such a small 5x5 domain is not visited by SWOT for at least 5 days, perhaps more. Depending of this, UV\_SC might performs better with a 10-day window, but worse with a 8-day window, of UV\_STC with Lt=5 instead of 3.6 . A sensitivity analysis is missing, especially because SWOT is at the core of the work.### Conclusions
* L321: incomplete sentence.
* L325: "using meridional (V) geostrophic current only as input may improve the
accuracy.": where is this coming from? I did not see any experiment showing this in the paper.Citation: https://doi.org/10.5194/egusphere-2022-1018-RC2
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2022-1018', Anonymous Referee #1, 12 Dec 2022
A method to enhance the detecting of geostrophic current and its temporal variations with SWOT swath data Shi et al. (2022)
General Comments
The authors present an interesting article detailing methods and improvements for retrieval of geostrophic currents from simulated SWOT observations of the sea surface height, which are expected to contain large magnitude correlated errors.
There seem to be three aspects to the paper: the filtering of swath ADT to suppress long-wavelength (spatially-correlated) errors, testing updated space and time correlation scales in the OI, and comparing the different methods (ADT_STC, UV_STC, and UV_SC) against one another.
The methods presented demonstrate the benefit of filtering the swath ADT (which appears to remove much of the short length-scale noise), and the benefit of calculating the gradient in the ADT fields (to calculate the u/v geostrophic currents) which appears to address the limitation imposed by the large-scale correlated errors in the SWOT ADT.
However, one stated objective of the paper was to demonstrate the need to adjust the spatial and temporal scales in the OI to be more appropriate for SWOT. It was not clear how the spatial and temporal scales used were chosen and whether the two methods (UV_SC and UV_STC) compared the currently used space/time scales with the proposed update, or not. Overall, I found the description of the methods to be somewhat confusing: it was not clear to me how the ADT was produced using the UV_STC method, nor was is clear how the UV_SC and UV_STC methods differ in terms of time-range of inputs and quantity of observations included.
I believe this work is timely and will be of interest to the community, but the description and comparison of the methods (ADT_STC, UV_STC, UV_SC) must first be clarified.
Specific Comments
Line 17: “the long wavelength error (LWE) is constant along tracks”. I don’t think this is true: the LWEs are correlated along track, but are not constant.
Line 44: “spatial resolution of…the resolution of” I think this would more correctly be ”spatial sampling of…the posting of”. The expected feature resolution of SWOT is 15-30km, but the observations will sample much more frequently and the supplied product will have many observation points (every 2km) per resolution element (15-30km).
Line 61-62: “the long wavelength errors” Which of the error components due you include in this definition? Roll and phase only?
Line 62: “constant along track” Same comment as for line 17.
Line 64: “the difference step” I think here you mean that the long-wavelength errors do not affect the short-scale spatial derivatives of the ADT used in the calculation of the geostrophic velocities, but I think you could be more explicit in this at the first mention of the “difference step”.
Line 67: “swath width of 50km” I would suggest to clarify that this is only one side of the full swath.
Figure 1: The rainbow colour-scale used is hard to interpret. I’d suggest a diverging colour-scale (e.g., red to white to blue) to emphasise where the magnitude of the errors is largest.
Figure 1: The range of the colour bar is limited to +/-5cm, but the text notes that the phase and roll errors are at the decimetre level. From my experience of these errors, +/-20cm could be necessary to show the range of the phase/roll errors.
Line 182: It’s unclear why a time window of +/-5days is chosen. While the precise choice may be arbitrary, given the timescale of 3.6 days mentioned earlier, some justification seems necessary.
Figure 3: Is the time period used for the figure to match the +/-5day window mentioned before?
Figure 3: The power spectra show that the filtering removes the excess power (and the signal) at scales shorter than ~20km, but there appears to be more small-scale structures present in Fig 3c than in Fig 3a. How is this consistent?
Line 212: The title of this section “Effect of difference” could be improved as the section discusses both the effect of deriving the u/v currents (the difference method) and Gaussian-filtering the ADT. I think it would also be clearer if the “difference method” were explicitly defined somewhere.
Line 213: Is the Gaussian filter applied two-dimensionally, or only in the along-track direction? If 2D, does this introduce artefacts at the swath edge? And if along-track, are there artefacts at the end of tracks?
Line 213: Why was a radius of 14km chosen? You stated that the correlation scale of the model was 27km and the apparent correlation scales of the observations errors is much longer.
Table 1: RRMSE is not widely used, so it would be useful to refer to equation 9 in the table caption.
Line 228: Same question as for line 182.
Line 228: Why was a 0.1degree grid chosen? If the expected limit of feature resolution is ~100km, then you would require 2-3 grid cells per resolution element to retain that resolution.
Figure 4: How was the ADT estimated using the UV_STC method?
Figure 4: The difference fields shown in panels (d) and (e) would be clearer if the colour-scale were centred on zero. This would make it easier to interpret where the bias was very small, and/or whether there is a domain-wide bias.
Line 243: It is not clear to me exactly how the UV_SC and UV_STC methods differ in terms of time-range of inputs and quantity of observations included. Is the UV_SC method equivalent to UV_STC, but with a different time window (+/-5days)?
Line 258: Related to comment on line 243: is the higher correlation between the RMSE and number of observations for UV_STC due to using many fewer observations than in UV_SC?
Technical Corrections/Suggestions
Title: suggest replacing “detecting” with “detection”
Line 15: “is different” -> “are different”
Line 26: “Satellite altimetry” -> “Satellite altimetry has”
Line 27: “Gridding method should be” -> “Gridding methods have been”
Line 41: “interferometric” -> “interferometer”
Line 65: “T/P” Perhaps expand to “TOPEX/Poseidon”
Line 67: “the globe ocean” -> “the whole global ocean”
Line 76: This first sentence could be rephrased to clarify the meaning.
Line 100: “the reality” -> “reality”
Line 106: “the other instrument error is” -> “the other instrument errors are”
Line 107 “those error is” -> “those errors are”
Line 108: “spatially structured” -> “spatial structure”
Line 108: “is constant” -> “are constant”
Line 111: “errors are…cut” -> “errors being...cuts”
Line 135: The omega symbol is used mistakenly for the meridional direction (as well as the angular velocity).
Line 187: “A Global Self-consistent” -> “the Global Self-consistent”
Figure 4: “using ADT_STC and UV_STC method” -> “using the ADT_STC and UV_STC methods”
Figure 4: “The ADT of model” -> “The ADT of the LLC4320 model”
Line 255: “on the overall” -> “overall”
Line 275: “underestimate the gridded results of model” -> “underestimate the magnitude of the gridded model ADT”
Line 280: “then” -> “than”
Figure 7: The panels references need to be updated: a,b,c,d used in figure, a,c,e,g used in caption.
Line 321: “To adapt…” This sentence should be rephrased.
Line 323: “in the swath” -> “of the swath ADT”
Line 337: C3S has not been defined.
Citation: https://doi.org/10.5194/egusphere-2022-1018-RC1 -
RC2: 'Comment on egusphere-2022-1018', Anonymous Referee #2, 28 Dec 2022
## Review
* Shi et al, 2022
* Title: A method to enhance the detecting of geostrophic current and its temporal variations with SWOT swath data
* Submitted to egusphere
* december 2022## Summary
The paper introduces and discusses a method to map geostrophic current from the recently launched SWOT mission dedicated to wide-swath altimetry, using statistical interpolation. The novelty (w.r.t. CMEMS products made with the DUACS system, in particular) is to derive geostrophic currents on the swath before mapping. With nadir altimetry, SSH must be gridded before deriving geostrophic currents. The authors argues that this strategy reduces the consequences of long-wavelength errors in SSH through finite differences, and provides more accurate currents that the DUACS approach.
## General statements
* The writing is too aproximative for an accurate understanding, especially for non-native English speakers like me, not able to meta-understand the meaning of the text. Errors include wrong or inconsistent tenses, inapropriate vocabulary, missing or inadequate articles before nouns, incomplete sentences, inaccurate wording, etc. A thorough proofreading is essential.
* The text is not at a level appropriate for scientific publication in the present form: many basic information is unecessarily provided (geostrophic equations, gaussian filter equations, etc) and many previous articles on the topic are completely ignored.
* The focus of the paper is not well defined. Is it the issue of SWOT temporal sampling and the related time scales involved in OI? Or a new method to derive gridded products to overcome the long-wavelength error issue? Or the reconstruction of time variations of EKE? None is satisfyingly addressed.
* If the focus is on a new mapping method, standard performance metrics must be adopted. Power Spectrum Densities are common in the SWOT community because SWOT specifications are formulated in terms of spectra. The authors should dive into the large body of litterature on this topic.
* * if the focus is on the removal of long-wavelength errors, diagnostics must be formulated specifically. The along-track and across-track directions musy be separately addressed, given the structure of SWOT errors.
* Perfoming diagnostics with weekly means is inconsistent with the spatial resolution considered in the paper, with SWOT, and with the standard products (DUACS from CMEMS).I think the paper needs more a rewriting than a revision. For this reason I recommend rejection. Below are line-by-line notes that might be useful to the authors for a future submission.
### Abstract
### Introduction
* L30: Esteban-Fernandez et al is not the appropriate reference to support the previous consideration on EKE.
* L39: SWOT is now on orbit. Update tenses!
* L44: The data with resolution 10-70 m are the L1 level, useless for ocean science as is. This mention is misleading for the reader not familiar with SWOT and should be removed.
* L45: the standard is 2 km.
* L45-49 are unclear. The first sentence seems incomplete. The second one is awkward. Do the authors want to say that the small mesoscale features are blurred by SWOT errors?
* L53: what is "the accuracy of the grid"?
* L57: I think that the 2020 paper by Gomez Navarro et al (Remote Sensing, 2020) is a better reference for SWOT denoising.
* L59-65: It is not clear where this discussion leads to. On the one hand, it is suggested that the long-wavelength error is constant along the track, but induces spurious gradients across-track: why? across-track variations of errors are ignored, and the discussion ends on the removal of long-wavelength errors by computing SSH gradients. What about across-track gradients? (I came back to these lines after reading lines 104-114 and figure 1, with even more confusion)
* L65: "the accuracy of the grid". Do you mean "of the gridded product"?
* L82: it should be said how time interpolation is performed if Gaussian covariances are not used anymore. I see it at Line 182, it is too late.### Data and methods
* Lines 104-114 are repeating other papers, or Esteban-Fernandez' SWOT report. It should be acknowledged and kept only if this information is used later to shed light on some results. By the way, where is the "long-wavelength error" cited several times earlier in the text?
* I do not find it relevant to remind the geostrophic expressions, except for introducing notations or specific discussions on expressions. I do not think it is the case.
* Lines 142-145: The unbalanced wave motions are filtered out anyway, by the time averaging.
* Section 2.3: I do not think another description of the OI method is needed. It would be more relevant to cite the DUACS system and emphasize the differences. Also, the very first sentence of the section ("The OI method can be used...") does not clearly state that OI is used in this study.
* Lines 200 and below: I do not think it necessary to recall what a Gaussian filter is.### Results and discussion
* L219-220: Should I understand that the RMSE in velocity is quantitatively compared with the RMSE in ADT?
* L222: I do not think that the simple RRMSE scores allow to conclude that the mesoscale signal of currents is recovered.
* L229: The method to recover ADT from U and V after mapping geostrophic currents has not been described. It is the first time, I think, that the text refers to this process.
* L233: it is stated that the long-wavelength error is constant along-track, but this does not seem consistent with figure 1.
* L243-246: to my opinion, these lines belong to the method section.
* Why performing evaluation on weekly averaged products? The spatial resolution is 0.1, what makes the space and time resolutions fully inconsistent. What is the point of such products? For information, DUACS products are delivered daily at 0.25 resolution.
* L273-281: The analysis of the reason why UV\_SC performs better than UV\_STC seems far too simplistic to me. At these latitudes, such a small 5x5 domain is not visited by SWOT for at least 5 days, perhaps more. Depending of this, UV\_SC might performs better with a 10-day window, but worse with a 8-day window, of UV\_STC with Lt=5 instead of 3.6 . A sensitivity analysis is missing, especially because SWOT is at the core of the work.### Conclusions
* L321: incomplete sentence.
* L325: "using meridional (V) geostrophic current only as input may improve the
accuracy.": where is this coming from? I did not see any experiment showing this in the paper.Citation: https://doi.org/10.5194/egusphere-2022-1018-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 357 | 140 | 30 | 527 | 20 | 20 |
- HTML: 357
- PDF: 140
- XML: 30
- Total: 527
- BibTeX: 20
- EndNote: 20
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1