Exploiting the Combined GRACE/GRACE-FO Solutions to Determine Gravimetric Excitations of Polar Motion

Round 1

Reviewer 1 Report

This paper presents in a clear and rigorous way an original use of space gravimetry measurements obtained by the GRACE and GRACE-FO missions, consisting in estimating from them the equatorial components of the gravity excitation of the pole motion. The authors first recall the simple linear relationship between these components and the degree-2 order-1 spherical harmonic coefficients of geopotential. These coefficients are determined monthly from GRACE and GRACE-FO measurements by various analysis centres around the world. Since these coefficients are corrected for non-tidal atmospheric and oceanic effects, their temporal variations reflect the effects of continental hydrology and the cryosphere on the gravimetric excitation of polar motion. The study of temporal variations in this gravimetric excitation should therefore be a means of investigating the hydrology and cryosphere, provided that the sensitivity of the GRACE/GRACE-FO measurements is fine enough.

To assess this sensitivity, the authors propose to systematically compare the so-called GAO reference geodetic time series with optimal combinations of GRACE/GRACE-FO solutions provided by different analysis centres. The optimal combinations are assumed to be noise-reduced by using the generalized TCH method. The comparison of the series concerns 3 statistical indicators (zero-lag correlation coefficient, standard deviation, root mean square difference) represented graphically on Taylor diagrams.

The authors then discuss the sensitivities obtained from the combinations of solutions made according to the periods of the observed signals (annual, semi-annual, terannual).

I found the explanations of the methodology very clear and the subsequent interpretations rather convincing. I appreciated the use of Taylor diagrams as a visual and synthetic tool for comparing time series. The bibliography is also very well provided. The references in the bibliography help to substantiate the claims made.

In the printed attached version of the paper, I just highlighted two minor typos on page 8 (equation 4) and page 17 (caption of figure 9).

If it were necessary to improve a detail of the article, I would suggest to explain more precisely the mention which accompanies the Taylor diagrams by recalling the definitions of the "critical value of the correlation coefficient " and the "standard error of the difference between correlation coefficients".

This in no way compromises the quality of the article and I would like to congratulate them for their remarkable work.

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

Comments:

The authors present a new combination of GRACE c21 and s21 data by TCH method, and then compare it with observed geodetic data and other similar data. The results are generally obvious, but the discussion and some details should be further given for publish. I recommend to add some contents such as:

1)     Give the reference and discussion on how to convert observed PM to excitation, due to different method will give different results.

2)     Give some details on AAM and OAM, such as pressure levels, integral method, boussinesq correction of OAM etc. and especially noise level for both data. Due to GAO is acquired by subtract AAM and OAM from geodetic observation.

3)     The TCH method is often used for clock error estimation which take each clock as an independent observation. Anyway, GRACE data given by different centers use the same data source but with some different processing method. This is very different with the case of clock. i.e., the different GRACE products are highly correlated. Thus, how to consider such correlated GRACE data, are TCH still suitable here? Why? Even TCH can be used here, what’s the true meaning for the combination? Does that can give a better result?

4)     The SLR results for c21 and s21 is regarded as the most precise data in the current case, it should be added or at least discussed.

5)     The Discussion part of the paper only show the result, but have less info to discuss the errors in GAO, AAM, OAM, GRACE, SLR data set etc.. I hope there are deep discussion for the questions that I presented here at least.

6)     The TCH method should cite the original paper, your equations are not clear here.

 Other comments:

L48, 1,4,5,…12, it’s better to be 1,3, 5-12. Others are similar to it.

L197-198: how to convert observed PM to excitation should be pointed out, such as references.

Fig.2, It’s better to replot fig2a as the form of fig2b., and the two figures can be arrange in one line.

Why does eq.(4) have N data and eq.(5) only have N-1 data?

Eq. (6) where is M? N-1 solution centres? Or M centres?

Table 1, we can see that for almost all the case, the weight of CSR is almost the lowest (CSR was chosen as the basis). Why? If so, why not chose COST-G which contains SLR result that we believe it’s the best results? And you say that TCH method is free of the basis time series chosen, is that right? Bold font for highest weight is more obvious, but not italicized.

L534-535, “they visibly underestimate the amplitudes observed for GAO.”It should NOT say “underestimate” because you here only compare the two, but not estimate. Otherwise, if GAO is not totally correct (AAM and OAM may have large error too), we can not draw such conclusion too.

Author Response

Please see the attachment.

Author Response File: Author Response.docx