Review Reports

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This manuscript describes an algorithm used to predict medium and long term LOD and DUT1. The paper is tidy and well written. Myself not being an absolute expert in this area (more geodesy in general) makes it very hard to assess the importance of this work. I strongly believe that the authors must improve their placement of this work into existing knowledge and what it means. This also leads to my second major comment, that the authors basically only use one restricted prediction period. Why not extend the prediction to 2024 and assess the overall suitability of this prediction method?

Major comments:

1.      Introduction: the authors try to place their work into the overall geodesy framework. Though they are using inappropriate literature as references. Please find better (review) papers or books when referring to the various space geodetic techniques, or mentioning ‘various excitation sources of Earth rotation’. The literature review needs to be imporved.

2.      The paper lacks the bigger picture and leading the reader through it by referring to general understanding. I.e., when mentioning the C04 series; yes, the reader might know it. But please help them along by repeating some of its main characteristics. Such as, this is a combined data product, with daily resolution etc. The authors further use a lot of abbreviations. This makes the paper difficult to read.

3.      What is the purpose of this paper? Did you take part in the EOPCC? If not, why not? Is your aim to do the same as contributors did during the EOPCC?

4.      When doing dUT1. Did you actually train your data on dUT1 values or LOD values? Or both? Why do you expect different results when looking at dUT1 and LOD? If you only convert (by using the time derivative), why would you expect the results to be different? Please make this clear.

5.      Lines 334-341: do I understand it correctly, that final data was not available? Why? We have 2024 now, surely all the finals are available by now.

6.      Section 5.3: discussion of prediction errors: I do not understand why the authors are only using one period for verification, ending in 2022. We now have the year 2024, so almost two more years where this method could be applied to. What is the aim of this work? Is the intention to use it for better improved long term predictions? If so, then the authors have to prove that it is working, also in non-La Nina/El Nino years. This would allow some quality assessment for general purposes and not only for one year. What is the idea for the future? Will the training period (currently 30 years) be ever growing? Or does it make sense to restrict to a shorter (but possibly more precise) training period?

 

Overall, I struggle to place the importance of this work. I believe this is good work, but strongly recommend to work on the bigger picture and direct outcomes and applications and future plans before I can recommend publication.

If I want to be very cynical (and clearly I am no expert in this field), I could say: well, the authors took 30 years of training data and then developed a method that allows them to predict one additional year of LOD/DUT1 quite well or better than what was published on the IERS predictions. But is it any good for the future and other time periods. I am sure that I am wrong with this assessment, but please convince me in the manuscript.

Minor comments:

1.      What resolution is your input data series? 3-hourly, daily?

2.      I am not sure I understand the difference between base series and training series. Please clarify.

3.      Figure 10, caption. Please provide more information here.

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Reasonable methods are used  to predict the LOD and UTC-UT1 difference. The manuscript is well written.

A minor comment:

In sections 3.2.1. "Fitting results" it is mentioned that periods of 121.75 days, 182.62 days, 365.24 days, 9.3 years, and 18.6 years remain after removing the tidal signals. Is there a reference supporting this statement, or could this be shown in a Figure of the spectrum?

The prediction is up to 360 days, while the periods chosen for the WSL model include multi-year (9.3 and 18.6 years). Therefore I would question to use "long-term" prediction in the title and abstract and at many places in the text, this would be more than one year. The work is mainly about "medium-term" (~30-360 days) predictions.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

(1) A brief summary (one short paragraph) outlining the aim of the paper
and its main contributions;

Paper deals with determination of the Length of Day (LOD) and the Universal Time (UT1), which play very important role
in satellite positioning, deep space exploration, and in GNSS and other satellite techniques in general. To predict
LOD and UT1 authors apply the least squares fitting combined with auto-regressive models and Polynomial Curve Fitting as well.

Authors show, that their method is well-suited for medium- and long-term prediction of the LOD and UT1-UTC.
Significant improvements in prediction accuracy were obtained for periods ranging from 90 to 360 days, where the
average accuracy improved over 20% compared to IERS Bulletin A. The largest prediction accuracy increase
was 49.5% and 59.2%. Impressive!


(2) Broad comments highlighting areas of strength and weakness. These
comments should be specific enough for authors to be able to respond;

To validate the methodology presented in this paper, authors processed a huge amount of data
at weekly intervals within the time span from 2021.09 to 2022.12. The predictions were compared with those from the
International Earth Rotation Service.

Only remark on Title: Too long, big number of abbreviations.  

 
(3) Specific comments referring to line numbers, tables, or figures.

Tables are fine.
Figures are fine. However, I have some doubts about Figures 16-19. As a patterns, they look nice, however from technical
point of view are useless. By my head, they could be omited, making paper shorter.

(4) Are the references appropriate?

Presented references are very appropriate.


(5) Overall conclusion

Scientific soundness is high, practical findings are very promising.
Paper could be printed in "Remote sensing" journal.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

In my initial review I raised serious concerns regarding (a) the repeatability of this study as well as (b) how useful the presented algorithm might be for the future. The current manuscript does not answer this to a sufficient degree which I consider absolutely essential for publication in a scientific journal of this level. 

The presented work as it is now is a prediction algorithm tuned to one specific test period. There is no mention on how this work can/will be used for future predictions. This is of no use. My requests to test the developed methods to another time period (and I again disagree with the authors: we have 2024, and your training period ends in 2022, there is enough data to check your algorithm) was not followed. On could also shorten the training period and choose another test period in previous times.

Consequently I cannot see how this work is of interest and use to the community. 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf