Statistical Analysis of High–Energy Particle Perturbations in the Radiation Belts Related to Strong Earthquakes Based on the CSES Observations
Round 1
Reviewer 1 Report
(1) Please give a map where the earthquake locations are shown along with their EPZ or make a table with latitude longitude Magnitude depth and EPZ.
(2) Draw a plot of geomagnetic activity indices of DST and KP for the entire period and show the earthquake days.
Please check the grammar and spelling mistakes in the manuscript.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Apart from my comments in the Word document I would like to comment here that the paper does not strictly follow the schema of MDPI papers, consisting on: Introduction, Data Processing, Results, and Discussion.
I admit that sometimes you can differ a bit this organization, but in general lines it is good to follow the recommendations.
In the same line, I would say that in the document, as I already mentioned in the document, there is some parts that I would have included as Results, rather than Data analysis, because they are basicaly an enumeration of the features discovered using the data from CSES satellite. Please revise that and extend the explanation on the data selection and processing, both for earthquakes databases and electron flux data.
Also I would suggest to put some more figures to ilustrate the mathematical way you used to select and count the anomalies.
Finally, I don't know if this is your plan in the end, but I'd prefer the have the figures placed just after they are mentioned in the text, rather than all together at the end of the manuscript.
Comments for author File: Comments.pdf
In general, the paper needs an extensive and detailed review of english language as it contains many errors in words spelling and grammar. Also several sentences must be shortened or simplified in a way that the reader can get the idea faster and more precisely. I've noted a bit of abuse of redundancies, expressing several times the same idea with slightly different words.
In particular, some paragraphs in the Conclusions are almost identical to the ones in the text.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments for author File: Comments.pdf
There are a few typos, please check the text.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 4 Report
The manuscript describes results of a statistical study of the particle precipitation from Van Allen radiation belts before, during, and after major earthquakes (EQ). The study uses data from CSES from 2018. There are 78 cases of major EQs; although some of them (16 cases) are discarded for various reasons, the rest can be effectively utilized to conduct a comprehensive statistical analysis. However, in my view, only a portion of the analysis is shown, and more can be effectively performed to improve the quality of this manuscript.
Major comments:
1. Abstract uses a word choice “we think …” (LN 13). The authors should state what are their conclusions based on their analysis, not just what they “think” which is happening during the major EQs.
2. Section 2 describes all the instruments available in CSES and fails to describe the instruments used in this study. I just figured out it must be HEPP-L based on the figures. Were any data quality control measures considered before using the data sets? No such information is available other than discarding 16 cases for other reasons. How does the precipitation data look like on a typical day (in the absence of EQ), and how does it vary (from typical values) during the major EQs? None of these was presented.
3. First, provide a figure (or at least a table along with geographical locations) to show all the EQ cases that were considered in this study so that the reader can get some idea about the span of the data.
4. In the results, three examples (out of 62 cases) are shown for the evolution of the spatial distribution of electrons. What about other cases? The proper way of addressing this is to estimate the differences in precipitation at these locations (relative to typical quiet day values). Authors should itemize these in the form of either figures or at least in the form of tables to show consistency in the results.
5. Only a small paragraph was written (with a single figure) for the work of time evolution of the electron flux. Again, were the authors able to notice similar differences in other cases as well? If so, show those results to justify the conclusions.
6. Authors concluded that there is an overall 62% of probability for EQ related to electron flux anomalies but this fact is not properly justified by the (described) method of analysis and (shown) results. I even notice that a paragraph from the results section (LN 190 -196) was indeed repeated in the Conclusions (LN 240 - 245).
7. The manuscript contains several statements about previous studies without proper references. For example, statements starting in lines 71, 232, and so on. Moreover, the statement in the LN 232, should not be in the conclusions. If the same team did several works before this manuscript, please state them either in the introduction or results section to support your work (along with the references). I could not see any of such statements in the main body (unless I missed them). Total of 25 references were listed at the end of the manuscript but it seems only 23 of them were used. Again, it is possible I missed them, but I could not track beyond 23.
8. In LN 187, it is stated “electrons move from west to east in the drifting direction”. The authors use only this statement to explain as to why the changes in the precipitation occur in the east or northeast directions. If this drifting is due to E X B, then it depends on direction of E.
Considering the above important points, in my view, the manuscript needs a major revision to justify the conclusions. The manuscript needs substantial improvements in describing methodology, quality control of data, and in better presentation of the analyzed data for both space and time. The authors have to utilize effectively available data sets in a convincing manner (by showing proper figures, tables, along with error estimations) so that the results will justify their conclusions. This is important to improve the quality of this manuscript.
The quality of the English language is adequate enough to understand the manuscript. However, I noticed few typos and moderate editing of language may be required.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
Dear authors.
I've seen a great improvement in the description of the processing methods and the results. Now it's much more clear to the reader how are they performed. The enumeration of steps you showed is a great improvement on that direction.
However, I would still make two comments on some of the answers you gave me in the letter:
About the periods of time:
Your response: In the data processing process, each case is divided into 0.1-0.3 MeV/0.3-3MeV, day-side/ night-side, and 5 time periods. In other words, there are 20 spatial distribution maps for each case.
To simplify the work, , we divide each month into 6 segments: 1-5; 6-10; 11-15; 16-20; 21-25; 26-30. For example, in Figure 1, if an earthquake occurs on the 19th, we consider the interval from 16-20 as "at the time of EQ," and 21-25 as the five days after the EQ. I acknowledge that this approach may introduce some errors, but the spatial distribution maps themselves are already smoothed averages over a five-day period. I believe that this approach has a minimal impact on the results.
My comment: As you say, accounting for these five-day smoothed average, theese errors may be reduced, even though I'm not completey sure. I woud have prefered using a simple algorithm that divided the time in real 5-day intervals from the EQ day, not these "fixed" 5-day intervals. But in any case, if this was your procedure, I think you should explain it better in the processing section.
My former comment on Lines 121 and 143.
In both sections: “Spatial” and “temporal” analyses, I’m missing more on the “data processing”, on how the initial data is mathematically processed to get the results. Then section seems to be more a “Results” section than a “Data processing” section. Particularizing: how did you decided if a perturbation is a perturbation. Did you used any mathematical function or threshold values? Is it simply a visual check? This is very important in terms of reproducibility and also to increase the reliability of the statistical results in the next sections.
Response: Thanks for the advice. To begin with, I've included my updated table in the new version, and this table is based on all the plots I made. The stats for Figure 6 and Figure 7 also come from this table.
We put our basis of evaluation and data processing methods in "Data selection and analysis" in Line 121-133
When it comes to the time distribution plots, those are calculated using a specific algorithm. But for the spatial distribution plots, there’s a visual judgment involved which is thought to be at a high significance
My comment: I think the table is a piece of very valuable information. And about the last comment you made me, this is one of the major corrections I was lacking in the initial version. Now it is mostly corrected, but I think that if the selection of anomalies on the map was made by visual inspection it should be mentioned in the text. This approach may not be wrong if the criteria used to visually detect the anomalies are well established. But for me, it has two important drawbacks:
1. It is difficult to be completely objective in some cases. The decision on "what is an anomaly" and "where it is located from the epicenter" can be slightly affected.
2. It is not scalable. Now, having 62 valid earthquakes, I understand you have checked 1240 maps. (5 intervals, day/night, and two energy intervals). These are a lot of cases, which in the other hand, may help smoothing the possible errors induced by the visual seection.
I think a computational approach for future studies must be considered. Nowadays technology gives many opportunities to process these problems in several ways. And you could increase the amount of data processed and the quality of the results.
However, and again, if this is the method used and you are confident with the criteria followed in the visual inspection, I suggest remarking it on the the manuscript, detailing the criteria, and discussing some ways to automatize this task as future work in the conclusions.
These are all my comments I wanted to mention in this second revision of the manuscript. Overall I think that the quality increased, and it could be ready to be published after pulishing the comments I made above.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 4 Report
N/A
Author Response
Dear reviewers, editor:
Thank you for your great advice and comments which are very helpful to improve our manuscript. we have improved the description of our data processing section:
We have modified this section in Line 140-151:
” The revisiting period of the CSES satellite is 5 days, that is, every 5 days the satellite can complete a global Earth coverage. Therefore, we use a set of 5 days data to map the smoothed flux of energetic particles near the epicentre. In addition, we divided the energy range into 0.1-0.3 Mev and 0.3-3.0 MeV, and splitting the pixels in $1^{\circ}\times1^{\circ}$. We also separated the night-side and day-side data in the statistics. To simplify data processing, each month is divided into 6 sets: from the 1st to the 5th, the 6th to the 10th, and so on until the 26th to the 30th. The set in which the EQ date falls is regarded as at the time of the EQ (refer to Figure 2-4(d)). The forward sets are: 5 days before the EQ (refer to Figure 2-4(c)), 10 days before the EQ (refer to Figure 2-4(b)), and 15 days before the EQ (refer to Figure 2-4(a)); and the last set is the 5 days after the EQ (refer to Figure 2-4(e)). The criterion for anomaly identification is from the second part in the article. We have marked the anomalies with red arrows in the figures. ”
We added this section in Line 130-132:
“ We can identify anomalies in the high-energy particle flux figures easily with visual interpretation. ”
For future work, we have added this discussion section in Line 289-293:
“ We can also make a conjecture if we can try to predict EQs from the flux of energetic particles in the radiation belts. Of course, while we are confident in these results, visual interpretation is not objective enough and inefficient in practical applications. We will also develop image recognition algorithms to identify anomalies using artificial intelligence. “
