Data-Driven Calibration of SWOT’s Systematic Errors: First In-Flight Assessment
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThis manuscript presents the first results of systematic errors calibration for real SWOT data, which is particularly important for SWOT's Inland Water products (HR products). It begin by presenting the uncalibrated systematic-error budget. The implementation of the Level-2 and Level-3 algorithms is then described, followed by an assessment of calibration performance for both Science (Level-2) and CALVAL (Level-2 and Level3) orbits. An initial brief analysis of performance on SWOT HR products is presented.
General concept comments
Despite its importance to the community using HR SWOT data, the manuscript would sometimes be difficult for this community to understand. Acronyms are not always defined. In addition, variable names are not constant (e.g. SLA and SSHA). Units are missing in several figures, and scales are not always the same when results are compared. Section numbers do not follow each other and at least one figure number figure number is duplicated. The form needs to be revised to make it easier to understand and avoid having to "guess" what the authors want to present.
The analysis of the L2 calibration on HR products for the Science orbit is very brief. This analysis could be the subject of a full article (perhaps part 2 of this article). Only the Science orbit (Level2) is presented, which should be clearly mentioned in the introduction and conclusion. Also, it would be more interesting to present results before and after calibration, or to present the importance of this calibration. This information is available in HR products (xovr_cal_c field). This would avoid including other potential errors in the LakeSP product. PLD, geolocation and darkwater errors can affect the results of LakeSP products. It would also be interesting to quantify the quality flag related to calibration (xovr_cal_q), in order to know in what proportion of data the calibration worked well. A spatial and temporal analysis (summer vs. winter) of the calibration would also be relevant. Finally, an analysis of the L3 calibration during the CALVAL orbit would also be relevant. These elements of analysis can at least be mentioned in the discussion or conclusion.
Specific comments
Line 27 : The launch was in December 2022
Line 28 : It is better to identify the fast sampling orbit between April and July 2023, since the satellite was in the commissioning phase between January and March 30, 2023.
Figure 1 : What is the x axis of the middle panels? What are the units? It will be easier to compare data if the left and right panels have the same scale. Do the authors differentiate between SLA and SSHA, since they use both acronyms (which are not defined)?
Figure 2, Figure 6, Figure 10 : The figure is similar to figure 17 in the article by Dibarboure et al, 2022, although the units are not the same (m2/cpkm vs cm2/cpkm). Can you check the units?
Line 109 : total signal
Line 110 : Remove a period at the end of the sentence.
Line 236 : section 2.3 not 3.2
Line 277 bracket is missing [6]
Line 293 : Is the performance of HR products also analyzed with offline reprocessing with a centered time window instead of using the 10 days in the past. This should be specified in section 5.
Pages 13-14 : There is two Figure 9. Figure 9 of page 13 used SSHA. Figure 9 of page 14 need to be larger.
Line 432 : This is section 5 (there is two section 4).
Line 523 : The conclusion becomes section 6.
Author Response
We thank reviewer #1 for the very constructive comments and suggestions. We propose in the following a point-by-point response with mentions to the updated manuscript.
Comment 1 : Despite its importance to the community using HR SWOT data, the manuscript would sometimes be difficult for this community to understand. Acronyms are not always defined. In addition, variable names are not constant (e.g. SLA and SSHA). Units are missing in several figures, and scales are not always the same when results are compared. Section numbers do not follow each other and at least one figure number figure number is duplicated. The form needs to be revised to make it easier to understand and avoid having to "guess" what the authors want to present.
Response 1 : We apologize for the lack of clarity. We checked that all acronyms are now correctly defined, and homogenized the variable names (SSHA is no longer used for instance). Units have been added to the figures (indeed Figure 1 was particularly poor in units) with clear subtitles on each panels to help understanding. We also added missing references to acronyms (such as the CMEMS products). We apologize for the figure numbering. A wrong figure reference was also found. This is now fixed. We hope that the manuscript gains clarity following these changes.
Comment 2 : The analysis of the L2 calibration on HR products for the Science orbit is very brief. This analysis could be the subject of a full article (perhaps part 2 of this article). Only the Science orbit (Level2) is presented, which should be clearly mentioned in the introduction and conclusion.
Response 2 : This is right. This paper is a first in-flight assessment and more in-depth assessments of the Level-2 correction will be certainly performed soon. This is why we choose to keep the Level-2 analysis short, although it is covered by Figures 9,10 and 11 including both integrated (RMS maps) and spectral diagnostics. We complemented the discussion about these figures in the manuscript as it was indeed too synthetic (in particular for the spectral description).
As for the performances of Level-2 for the CalVal orbit, we now added clear statements in the introduction and conclusion.. Note that this L2-correction will no longer be implemented in the next reprocessings as the performances do not meet requirements (due to the lack of crossovers, as already anticipated in simulations). The Level-3 correction will be proposed.
Comment 3 : Also, it would be more interesting to present results before and after calibration, or to present the importance of this calibration. This information is available in HR products (xovr_cal_c field). This would avoid including other potential errors in the LakeSP product. PLD, geolocation and darkwater errors can affect the results of LakeSP products. It would also be interesting to quantify the quality flag related to calibration (xovr_cal_q), in order to know in what proportion of data the calibration worked well. A spatial and temporal analysis (summer vs. winter) of the calibration would also be relevant. Finally, an analysis of the L3 calibration during the CALVAL orbit would also be relevant. These elements of analysis can at least be mentioned in the discussion or conclusion.
Response 3 :
This is a good point, we agree that the importance of the calibration was too briefly mentioned in the first version of the manuscript. In particular, the introduction and conclusion were too short on this point, we added a few statements. We also elaborated on the role of the Level-2 calibration in section 3.1 with a new paragraph.
Concerning the quality flag related to the calibration, its prior definition allowed us to improve the mechanism of computing the correction by highlighting unexpected behaviors but not necessarily the quality of the correction (the quality was assessed through cal/val analysis or hydrology validation). The latest version of the flag definition should allow us to use it as required (as a quality flag) and will be used for validating the correction.
The question of temporal analysis (summer vs winter) is also a very good point. It was extensively discussed in the pre-launch paper (e.g. Dibarboure et al. 2022) but not in this first version of post-launch results. At the time when this analysis was performed, we did not have a full year to carry out the analysis. A quick analysis performed this summer confirmed that the variance of the correction signal in the northern hemisphere above 50°N is lower in winter, suggesting that the correction misses a part of the signal. We also verified that arctic crossovers become available during summer. As the paper already contains many figures, we decided to elaborate this point by adding a paragraph in section 5. We also pointed that this seasonality is mostly expected inland over the northern hemisphere, and that future studies will likely characterize the seasonality of performances once several years of data will be available.
Specific comments
Line 27 : The launch was in December 2022
Yes, thanks
Line 28 : It is better to identify the fast sampling orbit between April and July 2023, since the satellite was in the commissioning phase between January and March 30, 2023.
Correct.
Figure 1 : What is the x axis of the middle panels? What are the units? It will be easier to compare data if the left and right panels have the same scale. Do the authors differentiate between SLA and SSHA, since they use both acronyms (which are not defined)?
Thanks for pointing this, the Figure 1 has been changed with all units added, as well as subtitles on each panels to help understanding. The ‘SSHA’ acronym is no longer used in the manuscript and figures (only SLA for ‘Sea Level Anomaly’). We check that in the new version, all acronyms are defined when they first appear in the text.
Figure 2, Figure 6, Figure 10 : The figure is similar to figure 17 in the article by Dibarboure et al, 2022, although the units are not the same (m2/cpkm vs cm2/cpkm). Can you check the units?
Good point, we checked the values and indeed, there is a typo in Dibarboure et al., 2022, the unites were also m2/cpkm instead of cm2/cpkm. We will ask to submit a correction of the figure in Dibarboure et al., 2022 figure 17.
Line 109 : total signal
Thanks, done
Line 110 : Remove a period at the end of the sentence.
Line 236 : section 2.3 not 3.2
Line 277 bracket is missing [6]
All done, thanks (all section numbering has been updated)
Line 293 : Is the performance of HR products also analyzed with offline reprocessing with a centered time window instead of using the 10 days in the past. This should be specified in section 5.
The XCAL Level-2 data comes from the official product version (PIXC PIC0). This clarification is added at the beginning of section5.
Pages 13-14 : There is two Figure 9. Figure 9 of page 13 used SSHA. Figure 9 of page 14 need to be larger.
This is fixed and Figure 9 has been changed with larger captions (instead of enlarging the whole figure). Thanks for pointing this.
Line 432 : This is section 5 (there is two section 4).
Fixed
Line 523 : The conclusion becomes section 6.
Fixed
########################################################################################################################################################################
We thank reviewer #2 for the very constructive comments and suggestions. We propose in the following a point-by-point response with mentions to the updated manuscript.
The paper presents an outline of SWOT data cal/val for the identification of long-wavelength and systematic errors that affect the KaRIN observations. The methodologies and approaches presented and followed, are those of pre-launch studies, with some modifications for adaptation to the real data.
In general the paper is very interesting and the analysis and results presented are justified, showing the significance of the mission data to oceanographic, geodetic and land applications. There are a few points that need to be addressed in the revised version, as listed below.
1) The introductory section is rather short and the review of the pre-launch methods, followed also in this paper, is not discussed in detail. Also, the related bibliography is rather short and not discussed in detail.
This is right and also pointed by reviewer #1. The introduction was almost entirely re-written, with more details on the history of calibration method development for wide-swath Altimetry (even before the SWOT project, during the first concept studies). Indeed, there is a long history of developments, in particular for the crossover method. We tried to keep the introduction of reasonable size since the manuscript is already long, hoping that the new elements and references discussed are now sufficient.
2) The captions of most Figures are rather large. They should be shortened and the explanatory text moved to the main body of the paper. This applies to Figures 1, 2 3, 6, 7, 9, 10, 11, and 14.
We agree. We managed to shorten the captions of these suggested Figures. Some elements have been added in the text accordingly. Although the caption of Figure 1 is still long, we did not manage to shorten it by more than 1 sentence.
3) The quality of Figure 1, especially the middle panel, is not good and higher resolution should be used. The middle panel is not readable (even at 200% magnification) especially the upper-middle panel.
Thanks, this is fixed. We resized labels and also introduced new ones to help understanding, as well as simplified subtitles.
4) Lines 85-86. What is this factor and how is it computed? Please give more details.
Indeed, the 2-sentence explanation was too short. As now explained in a new paragraph, we generated synthetic 1-D signals following the different spectra, based on an approach detailed in Ubelmann et al., 2014. Then the mask of the SWOT missing data is applied to the synthetic data, followed by the convolution-based data-filling also described in the manuscript (suppressing some energy). Finally, the spectra are computed on the processed synthetic data, and the ratio with the initial spectra represents a loss of energy factor, as a function of wavenumber, due to the missing data segments. This factor (near 2 at short wavelength and 1 at long wavelength) is applied after the computation of the SWOT data spectra.
5) Line 109. "....below the total signal
Done
6) Line 149-150. Why is the residual, after removing the 2nd order polynomial, not shown? It would be valuable, as to the removal of the systematic effects.
This is a good suggestion, we did not show it in the first version as there were already many figures, but we agree this is valuable. This is shown on Figure R1, which is now also in the new version of the manuscript. To construct the figure, we subtracted the successive cross-track polynomial fits and looked at the residual signal in the orbital frame, averaged over 1-day.. The left panel (1st order fit removing Bias only) exhibits the dominant roll signal. The middle panel (2nd order fit removing bias and linear shapes) exhibits the centimetric residuals of quadratic shapes. Finally, the right panel (removing bias, linear and quadratic shapes) confirms that cross-track order 3 signatures are quasi-absent or millimetric.
Figure R1. SLA averaged over 1-day, as a function of the orbital coordinate (in seconds) and the cross-track direction, after removing a zero (biases only), first (bias+linear) and second (bias+linear+quadratic) order fits, from left to right.
This figure also illustrates well the removal of successive systematic errors.
7) For the polynomial model, why was a 2nd order model used? Why not 3rd and/or 4th. Please provide a quantative and qualitative assessment. The discussion in lines 147-153 is not enough and some more analysis and e.g. statistics, magnitudes of higher order effects (or even their PSDs) would be nice to shown.
Indeed, the manuscript was not clear enough on this point, especially before we show the new Figure R1. Now with the figure (and added explanation), it is clear that no 3rd or 4th order signatures are present. And this is actually what was expected in simulation during the pre-launch studies (to described roll, phase, timing and baseline dilation errors, cf the Esteban-Fernandez SWOT error document referenced in the text). This is why here, we applied that 2nd order model, but verified that no significant orbital signal was present after its application (Figure R1). This statement is now better detailed in the text, and we hope this is now clear enough for the reader.
8) Figure 4 is poor in resolution and the lettering/numbering in the axes/titles is not visible. Maybe this figure can be split in two which would be re-drawn and enlarged. As they are now, not much can be seen.
This is fixed. (we considered splitting, but given the large number of figures, we chose to improve the resolution and fix unaligned panels).
Figure 5 was also fixed.
9) Line 175. Rephrase as (see Figure 9 in [4])
Yes, done
10)Line 207. ".....at the orbital frequency"
Yes, done.
11) Lines 225-226. Why is the k^-2 not indicated with a properly written mathematical symbol?
Yes, we now mention “a -2 power law” instead.
12) Lines 249. Omit The in front of Table 1.
Partially fixed, but will be rearranged in the article format version.
13) Line 293. derives instead of derive
Is now referred to ‘The performances’
14) Lines 361. ....at all wavelengths.
Done
15) I really appreciate the discussion in sections 3 and 4 which is quite complete and instructing in this new mission.
Thanks
16) Please note that the section entitled "Inland assessment of the calibration" is also numbered 4.
Fixed (all sections/figures and new figures numbers have been updated)
17) The left panel of Figure 14 is not of good quality and should be redrawn. Also the map inlets are not visible.
Updated
As already mentioned, the paper is well written and it is very instructive in view of this new mission and its assessment. Therefore, I would recommend its publication after minor corrections.
We thank the reviewer again for his very useful comments and suggestions.
Author Response File: Author Response.docx
Reviewer 2 Report
Comments and Suggestions for AuthorsThe paper presents an outline of SWOT data cal/val for the identification of long-wavelength and systematic errors that affect the KaRIN observations. The methodologies and approaches presented and followed, are those of pre-launch studies, with some modifications for adaptation to the real data.
In general the paper is very interesting and the analysis and results presented are justified, showing the significance of the mission data to oceanographic, geodetic and land applications. There are a few points that need to be addressed in the revised version, as listed below.
1) The introductory section is rather short and the review of the pre-launch methods, followed also in this paper, is not discussed in detail. Also, the related bibliography is rather short and not discussed in detail.
2) The captions of most Figures are rather large. They should be shortened and the explanatory text moved to the main body of the paper. This applies to Figures 1, 2 3, 6, 7, 9, 10, 11, and 14.
3) The quality of Figure 1, especially the middle panel, is not good and higher resolution should be used. The middle panel is not readable (even at 200% magnification) especially the upper-middle panel.
4) Lines 85-86. What is this factor and how is it computed? Please give more details.
5) Line 109. "....below the total signal
6) Line 149-150. Why is the residual, after removing the 2nd order polynomial, not shown? It would be valuable, as to the removal of the systematic effects.
7) For the polynomial model, why was a 2nd order model used? Why not 3rd and/or 4th. Please provide a quantative and qualitative assessment. The discussion in lines 147-153 is not enough and some more analysis and e.g. statistics, magnitudes of higher order effects (or even their PSDs) would be nice to shown.
8) Figure 4 is poor in resolution and the lettering/numbering in the axes/titles is not visible. Maybe this figure can be split in two which would be re-drawn and enlarged. As they are now, not much can be seen.
9) Line 175. Rephrase as (see Figure 9 in [4])
10)Line 207. ".....at the orbital frequency"
11) Lines 225-226. Why is the k^-2 not indicated with a properly written mathematical symbol?
12) Lines 249. Omit The in front of Table 1.
13) Line 293. derives instead of derive
14) Lines 361. ....at all wavelengths.
15) I really appreciate the discussion in sections 3 and 4 which is quite complete and instructing in this new mission.
16) Please note that the section entitled "Inland assessment of the calibration" is also numbered 4.
17) The left panel of Figure 14 is not of good quality and should be redrawn. Also the map inlets are not visible.
As already mentioned, the paper is well written and it is very instructive in view of this new mission and its assessment. Therefore, I would recommend its publication after minor corrections.
Author Response
Response to reviewers
We thank reviewer #1 for the very constructive comments and suggestions. We propose in the following a point-by-point response with mentions to the updated manuscript.
Comment 1 : Despite its importance to the community using HR SWOT data, the manuscript would sometimes be difficult for this community to understand. Acronyms are not always defined. In addition, variable names are not constant (e.g. SLA and SSHA). Units are missing in several figures, and scales are not always the same when results are compared. Section numbers do not follow each other and at least one figure number figure number is duplicated. The form needs to be revised to make it easier to understand and avoid having to "guess" what the authors want to present.
Response 1 : We apologize for the lack of clarity. We checked that all acronyms are now correctly defined, and homogenized the variable names (SSHA is no longer used for instance). Units have been added to the figures (indeed Figure 1 was particularly poor in units) with clear subtitles on each panels to help understanding. We also added missing references to acronyms (such as the CMEMS products). We apologize for the figure numbering. A wrong figure reference was also found. This is now fixed. We hope that the manuscript gains clarity following these changes.
Comment 2 : The analysis of the L2 calibration on HR products for the Science orbit is very brief. This analysis could be the subject of a full article (perhaps part 2 of this article). Only the Science orbit (Level2) is presented, which should be clearly mentioned in the introduction and conclusion.
Response 2 : This is right. This paper is a first in-flight assessment and more in-depth assessments of the Level-2 correction will be certainly performed soon. This is why we choose to keep the Level-2 analysis short, although it is covered by Figures 9,10 and 11 including both integrated (RMS maps) and spectral diagnostics. We complemented the discussion about these figures in the manuscript as it was indeed too synthetic (in particular for the spectral description).
As for the performances of Level-2 for the CalVal orbit, we now added clear statements in the introduction and conclusion.. Note that this L2-correction will no longer be implemented in the next reprocessings as the performances do not meet requirements (due to the lack of crossovers, as already anticipated in simulations). The Level-3 correction will be proposed.
Comment 3 : Also, it would be more interesting to present results before and after calibration, or to present the importance of this calibration. This information is available in HR products (xovr_cal_c field). This would avoid including other potential errors in the LakeSP product. PLD, geolocation and darkwater errors can affect the results of LakeSP products. It would also be interesting to quantify the quality flag related to calibration (xovr_cal_q), in order to know in what proportion of data the calibration worked well. A spatial and temporal analysis (summer vs. winter) of the calibration would also be relevant. Finally, an analysis of the L3 calibration during the CALVAL orbit would also be relevant. These elements of analysis can at least be mentioned in the discussion or conclusion.
Response 3 :
This is a good point, we agree that the importance of the calibration was too briefly mentioned in the first version of the manuscript. In particular, the introduction and conclusion were too short on this point, we added a few statements. We also elaborated on the role of the Level-2 calibration in section 3.1 with a new paragraph.
Concerning the quality flag related to the calibration, its prior definition allowed us to improve the mechanism of computing the correction by highlighting unexpected behaviors but not necessarily the quality of the correction (the quality was assessed through cal/val analysis or hydrology validation). The latest version of the flag definition should allow us to use it as required (as a quality flag) and will be used for validating the correction.
The question of temporal analysis (summer vs winter) is also a very good point. It was extensively discussed in the pre-launch paper (e.g. Dibarboure et al. 2022) but not in this first version of post-launch results. At the time when this analysis was performed, we did not have a full year to carry out the analysis. A quick analysis performed this summer confirmed that the variance of the correction signal in the northern hemisphere above 50°N is lower in winter, suggesting that the correction misses a part of the signal. We also verified that arctic crossovers become available during summer. As the paper already contains many figures, we decided to elaborate this point by adding a paragraph in section 5. We also pointed that this seasonality is mostly expected inland over the northern hemisphere, and that future studies will likely characterize the seasonality of performances once several years of data will be available.
Specific comments
Line 27 : The launch was in December 2022
Yes, thanks
Line 28 : It is better to identify the fast sampling orbit between April and July 2023, since the satellite was in the commissioning phase between January and March 30, 2023.
Correct.
Figure 1 : What is the x axis of the middle panels? What are the units? It will be easier to compare data if the left and right panels have the same scale. Do the authors differentiate between SLA and SSHA, since they use both acronyms (which are not defined)?
Thanks for pointing this, the Figure 1 has been changed with all units added, as well as subtitles on each panels to help understanding. The ‘SSHA’ acronym is no longer used in the manuscript and figures (only SLA for ‘Sea Level Anomaly’). We check that in the new version, all acronyms are defined when they first appear in the text.
Figure 2, Figure 6, Figure 10 : The figure is similar to figure 17 in the article by Dibarboure et al, 2022, although the units are not the same (m2/cpkm vs cm2/cpkm). Can you check the units?
Good point, we checked the values and indeed, there is a typo in Dibarboure et al., 2022, the unites were also m2/cpkm instead of cm2/cpkm. We will ask to submit a correction of the figure in Dibarboure et al., 2022 figure 17.
Line 109 : total signal
Thanks, done
Line 110 : Remove a period at the end of the sentence.
Line 236 : section 2.3 not 3.2
Line 277 bracket is missing [6]
All done, thanks (all section numbering has been updated)
Line 293 : Is the performance of HR products also analyzed with offline reprocessing with a centered time window instead of using the 10 days in the past. This should be specified in section 5.
The XCAL Level-2 data comes from the official product version (PIXC PIC0). This clarification is added at the beginning of section5.
Pages 13-14 : There is two Figure 9. Figure 9 of page 13 used SSHA. Figure 9 of page 14 need to be larger.
This is fixed and Figure 9 has been changed with larger captions (instead of enlarging the whole figure). Thanks for pointing this.
Line 432 : This is section 5 (there is two section 4).
Fixed
Line 523 : The conclusion becomes section 6.
Fixed
########################################################################################################################################################################
We thank reviewer #2 for the very constructive comments and suggestions. We propose in the following a point-by-point response with mentions to the updated manuscript.
The paper presents an outline of SWOT data cal/val for the identification of long-wavelength and systematic errors that affect the KaRIN observations. The methodologies and approaches presented and followed, are those of pre-launch studies, with some modifications for adaptation to the real data.
In general the paper is very interesting and the analysis and results presented are justified, showing the significance of the mission data to oceanographic, geodetic and land applications. There are a few points that need to be addressed in the revised version, as listed below.
1) The introductory section is rather short and the review of the pre-launch methods, followed also in this paper, is not discussed in detail. Also, the related bibliography is rather short and not discussed in detail.
This is right and also pointed by reviewer #1. The introduction was almost entirely re-written, with more details on the history of calibration method development for wide-swath Altimetry (even before the SWOT project, during the first concept studies). Indeed, there is a long history of developments, in particular for the crossover method. We tried to keep the introduction of reasonable size since the manuscript is already long, hoping that the new elements and references discussed are now sufficient.
2) The captions of most Figures are rather large. They should be shortened and the explanatory text moved to the main body of the paper. This applies to Figures 1, 2 3, 6, 7, 9, 10, 11, and 14.
We agree. We managed to shorten the captions of these suggested Figures. Some elements have been added in the text accordingly. Although the caption of Figure 1 is still long, we did not manage to shorten it by more than 1 sentence.
3) The quality of Figure 1, especially the middle panel, is not good and higher resolution should be used. The middle panel is not readable (even at 200% magnification) especially the upper-middle panel.
Thanks, this is fixed. We resized labels and also introduced new ones to help understanding, as well as simplified subtitles.
4) Lines 85-86. What is this factor and how is it computed? Please give more details.
Indeed, the 2-sentence explanation was too short. As now explained in a new paragraph, we generated synthetic 1-D signals following the different spectra, based on an approach detailed in Ubelmann et al., 2014. Then the mask of the SWOT missing data is applied to the synthetic data, followed by the convolution-based data-filling also described in the manuscript (suppressing some energy). Finally, the spectra are computed on the processed synthetic data, and the ratio with the initial spectra represents a loss of energy factor, as a function of wavenumber, due to the missing data segments. This factor (near 2 at short wavelength and 1 at long wavelength) is applied after the computation of the SWOT data spectra.
5) Line 109. "....below the total signal
Done
6) Line 149-150. Why is the residual, after removing the 2nd order polynomial, not shown? It would be valuable, as to the removal of the systematic effects.
This is a good suggestion, we did not show it in the first version as there were already many figures, but we agree this is valuable. This is shown on Figure R1, which is now also in the new version of the manuscript. To construct the figure, we subtracted the successive cross-track polynomial fits and looked at the residual signal in the orbital frame, averaged over 1-day.. The left panel (1st order fit removing Bias only) exhibits the dominant roll signal. The middle panel (2nd order fit removing bias and linear shapes) exhibits the centimetric residuals of quadratic shapes. Finally, the right panel (removing bias, linear and quadratic shapes) confirms that cross-track order 3 signatures are quasi-absent or millimetric.
Figure R1. SLA averaged over 1-day, as a function of the orbital coordinate (in seconds) and the cross-track direction, after removing a zero (biases only), first (bias+linear) and second (bias+linear+quadratic) order fits, from left to right.
This figure also illustrates well the removal of successive systematic errors.
7) For the polynomial model, why was a 2nd order model used? Why not 3rd and/or 4th. Please provide a quantative and qualitative assessment. The discussion in lines 147-153 is not enough and some more analysis and e.g. statistics, magnitudes of higher order effects (or even their PSDs) would be nice to shown.
Indeed, the manuscript was not clear enough on this point, especially before we show the new Figure R1. Now with the figure (and added explanation), it is clear that no 3rd or 4th order signatures are present. And this is actually what was expected in simulation during the pre-launch studies (to described roll, phase, timing and baseline dilation errors, cf the Esteban-Fernandez SWOT error document referenced in the text). This is why here, we applied that 2nd order model, but verified that no significant orbital signal was present after its application (Figure R1). This statement is now better detailed in the text, and we hope this is now clear enough for the reader.
8) Figure 4 is poor in resolution and the lettering/numbering in the axes/titles is not visible. Maybe this figure can be split in two which would be re-drawn and enlarged. As they are now, not much can be seen.
This is fixed. (we considered splitting, but given the large number of figures, we chose to improve the resolution and fix unaligned panels).
Figure 5 was also fixed.
9) Line 175. Rephrase as (see Figure 9 in [4])
Yes, done
10)Line 207. ".....at the orbital frequency"
Yes, done.
11) Lines 225-226. Why is the k^-2 not indicated with a properly written mathematical symbol?
Yes, we now mention “a -2 power law” instead.
12) Lines 249. Omit The in front of Table 1.
Partially fixed, but will be rearranged in the article format version.
13) Line 293. derives instead of derive
Is now referred to ‘The performances’
14) Lines 361. ....at all wavelengths.
Done
15) I really appreciate the discussion in sections 3 and 4 which is quite complete and instructing in this new mission.
Thanks
16) Please note that the section entitled "Inland assessment of the calibration" is also numbered 4.
Fixed (all sections/figures and new figures numbers have been updated)
17) The left panel of Figure 14 is not of good quality and should be redrawn. Also the map inlets are not visible.
Updated
As already mentioned, the paper is well written and it is very instructive in view of this new mission and its assessment. Therefore, I would recommend its publication after minor corrections.
We thank the reviewer again for his very useful comments and suggestions.
Author Response File: Author Response.docx