Comparison of Terrestrial Water Storage Changes Derived from GRACE/GRACE-FO and Swarm: A Case Study in the Amazon River Basin

Round 1

Reviewer 1 Report

This work compares Swarm and GRACE gravity solutions in spectrum and in the Amazon basin, and conclude that the Swarm solution can detect hydrological signals. The topic is interesting, but similar results and conclusion have been given in previous studies. The manuscript needs to offer something novel to ensure its publication. -- general comments Comparison of water storage between Swarm and GRACE in basins has been made by others, for example, in Lueck et al. 2018, Encarnação et al. 2020. In fact, their work investigated more basins than here. Why do we need a further investigation in a signal basin? What is new in this study? These questions should be clarified in the introduction and conclusion. Besides, I would recommend the authors investigating more basin as the previous study, since results in one basin are less persuasive (and less interesting based on a known conclusion in previous studies). Encarnação et al. 2020: https://essd.copernicus.org/articles/12/1385/2020/ The conclusion on accuracy is drawn based on RMS of a signal month. First, RMS is not straightforwardly associated with accuracy; second, a single month is not representative. More in-depth investigation should be provided. A spectrum combination method is introduced. But how much does it improve the agreement with GRACE? COST-G also applied the algorithm of weighted average, so what is the difference in theory and in result? Make clear in the figure that which version of Swarm solution is used, and whether it has been truncated. The language greatly needs to improve. The manuscript is filled with simple grammar mistakes and colloquial/meaningless expressions. -- specific comments L27. “model degree variance”, the “model” can be omitted. “closer”, compared to what? “10 degree”, should be “degree 10”, make changes throughout the manuscript L35. “annual amplitudes”, should be “annual phases” L43. “had” -> “was” L45, I feel “which” indicates GRACE rather than NASA and DLR, rewrite the sentence. L48, The orbit altitude descends with time. Maybe a reference here is preferable. L49, “which arising from” the sentence makes no sense. A suggestion is “which is influenced by” L50. I feel Reference-2 is not appropriate here. L55. earthquake coseismic change -> coseismic gravity change L74. Remove “can” L75. Two periods at the end of the sentence. L76. “quantify and monitor” L76-79. References for these missions? L87-89. How are Beutle and Ilk’s studies helpful for the conclusion here? L103. “an overview of the research area” L106, title of section 2, why the plural here? Figure 1. The main stream of the Amazon River is not shown. L117, “in this case”? L121-126, grammatically awful, rewrite. L154. This sentence is meaningless. Equation (1), why introduce \delta here? Clm and Slm are better to be subscripts. Equation (1) and (2) give different \sigma_l, it is better to make it clear. L167. A reference for ITG-Grace2018s Equation (4), it should make clear that g_i is only degree dependent (is it?), since p_i is degree dependent. L195, “\beta_1(t)” should be “\beta_1 t” L206. Figure 2 is based on equation-1? L247, “is used to calculated” L251, “any single model” L252, “the Figure.3”, remove “the” L255. “are limited”, too vague. Besides, the conclusion is not supported yet. L263. So here, the GRACE’s C20 has not been replaced yet? Make it clear. L266. “are worse than”, needs more evident and conclusive evidence. “within 15 degree. More than …” how do we get here? Figure 5, typo in the y-axis annotation. Figure 6. Consider using a shorter color range for 900km Fan filter. How about truncating Swarm to degree 12? Explain the white and black color in the colorbar. L296, “noises … has …” L297, “useful signals”? Again ,subject-verb agreement. Table 2. AP is originally in radian, is it converted to degree? If so, two decimals are too much. L315, why use “inversion” here? There is no inversion at all. L321, where are the sudden TWSC decreases? L368, “below 10 degree”, previously it is degree 15.

Author Response

Due to a large amount of response, it is uploaded as an attachment.

Author Response File: Author Response.doc

Reviewer 2 Report

Title: Comparison of Terrestrial Water Storage Changes derived from GRACE/GRACE-FO and Swarm:  Case Study in Amazon River Basin

In this study, the authors use SWARM data to fill the time gap between GRACE and GRACE-FO missions to estimate TWSC over the Amazon River Basin.

The comparison between different SWARM solutions is interesting.  However, the SWARM solution evaluation is not completed, because the analysis was made only for March 2015. Also, the comparisons between GRACE and SWARM over the Amazon River Basin are not satisfactory, especially in the case of signal spatial localization.

I recommend the rejection of this manuscript, but encourage the authors to address the following points and submit an improved version.

• Main suggestions:

If SWARM model has a spatial resolution of 1300 – 2000 km. Please, could the authors justify that the spatial resolution of Swarm is enough to study TWSC of the Amazon basin?  

 In section 4:

Concerning the “precision evaluation”, the authors claim that the accuracy of all the Swarm models is comparable to that of GRACE below degree 10. However, the study was made for only one month (March 2015). Please could the authors clarify this point? What happens in another time period?

 Please could the authors clarify the conclusion of figure 5? What is the meaning of the fluctuations of the correlation values?

Line 262 “C20 term is negative value, due to lower precision of C20 coefficient in GRACE and Swarm models”. A comparison with C20 of SLR could be useful.  

The authors recommended truncated to degree 15 the SWARM SH coefficients. However, their analysis was made only for March 2015.

Line 303 “According to the global EWH distribution of the Swarm and GRACE, the results of the two models are the same in some areas with stronger signal, for example, Amazon River basin, Congo River basin and Ganges River basin.”  To justify this claim, could the author show the differences between GRACE and SWARM solution? Looking at figure 6, for example, at the Amazon River basin, the localization of the maximum signal is not the same for GRACE and SWARM.  Also, SWARM solution has a stronger negative signal over the Orinoco than GRACE. Is the 900 km filter used? Are the Swarm coefficients truncated to degree 15?

Figure 8 shows that GRACE and SWARM TWS are not correlated, then how could they be consistent?

In Figure 9, the resulted linear trends are not consistent between GRACE and SWARM. Could the authors explain the conclusions? Line 338 “In the most region, the results of two models are the consistent”

A figure of annual amplitudes differences could be useful.  Amplitudes seem to be consistent, but not the spatial localization of annual signals. Could the authors explain the phase differences?

Figure 8, 9 and 10 don’t show that GRACE and SWARM are in good agreement.

Line 390.  The authors said  ” And in order to ensure the balance between weakening the stripe errors and retaining the real signals, it’s suggest to use 600km Fan filtering to process the Swarm data” But if we look at figure 6, the SWARM model is very noisy. Please could the authors explain their choice?

 

Other suggestions

Abstract:

Line 35: “annual amplitudes”?? or annual phases??

Line 277 please, could the authors explain briefly why Fan filter is better than that of Gaussian filter?

English must be revised.

Author Response

Due to a large amount of response, it is uploaded as an attachment.

Author Response File: Author Response.doc

Round 2

Reviewer 1 Report

This is my second round of review. The manuscript has been greatly improved and I am satisfied with most of it. I only have some concerns about the comparison of GRACE and Swarm solutions.

As the manuscript shows, the C20 term of Swarm is quite bad. I feel it should also be replaced by the SLR solution (maybe it has been done, but should be written out).

The process of geo-center terms and GIA correction of Swarm solutions is not described. For a fair comparison (in maps and trends), the Swarm solutions should follow the GRACE's processing steps in these aspects.

If I understand correctly, GRACE is truncated to degree 60 and applied with a Gaussian 300 km filter. I wonder if it is also truncated to degree 17 and smoothed by a 700 km filter (the same as Swarm), will the agreement be improved? 

 

Author Response

Because there is much response to the reviewer, they are uploaded as an attachment.

Author Response File: Author Response.doc

Reviewer 2 Report

The authors have addressed all my suggestions. Therefore, I recommend this manuscript be accepted after minor revisions.

Minor suggestions and corrections

Title of Section 3.3 could be changed by for example “scale factor estimated from GLDAS”.

The subsections of section 3.4 are not properly labelled.

Line 382 Please, could the authors quantify how much GRACE annual signal is larger than the Swarm annual signal?

Line 404 Please, could the authors change this sentence? “As the degree increases, real signals detected are less and less and the noise values are larger and larger. Therefore, the gap between the degree variances of the time-variable gravity signals of these two kinds of models becomes bigger and bigger” It is not clear.

Line 433 “The comparison shows a certain similarity in terms of annual changes (Figure.9 and 10)”could the authors specify the similarities? Amplitudes, localization?

Line 445 But the long term trends of SWARM and GRACE are not consistent, then what is the conclusion?

Line 450  “The results show that Swarm data can fill the data  gap to some extent.” Some extent??? I think the authors need to be more specific about the conclusions.

Author Response

Because there is much response to the reviewer, they are uploaded as an attachment.

Author Response File: Author Response.doc