Review Reports - A Method for Recognition of Sudden Commencements of Geomagnetic Storms Using Digital Differentiating Filters

Round 1

Reviewer 1 Report

This manuscript presents a new method for the detection of sudden commencements, which are characteristic abrupt changes of the magnetic field that precede magnetic storms. SC detection is important because it allows to predict magnetic storms. Various techniques have been used in the past, all with their advantages and caveats. Here, an optimized estimate of the maximum amplitude of the first derivative is used as a SC indicator. The performance of the detection mechanism has been tested with real data from 16 magnetic observatories for the year 2015. The correct detection probability is 75% if the probability of false positives is limited to 5%. This is a good performance. Therefore, the manuscript deserves publication.

The presentation style needs to be improved, especially for the mathematical part, which is difficult to read. I have proposed several improvements in the attached commented PDF.

As far as the content is concerned, there are only few major points that needs to be addressed for a more in-depth discussion:

1) You mention the possibility to improve the detection method by "optimizing its parameters". However, these parameters are already gained from an optimization routine: can you substantiate how a further optimization could be performed?

2) Further improvement of this and other SC detection methods requires to understand what triggers a false positive, since avoiding false positives requires to increase the detection threshold, thus also increasing the probability of not detecting a real event, as you show in Fig. 3. Therefore, it would be very helpful is a figure depicting a false positive is added and the reason for this detection would be discussed. For instance, the first thing that comes to my mind is an isolated peak (e.g. by noise), which gives a large maximum derivative without having the expected characteristics of a SC.

Comments for author File: Comments.pdf

Author Response

Response to Review 1

The authors - Victor Getmanov, Roman Sidorov , Alexei Gvishiani thank the reviewer for the work done on reviewing our manuscript «A method for recognition of sudden commencement geomagnetic storms predictors using digital differentiating filters». We agree with all the comments. Changes have been made in the revised text of the article.

In the revised text, the style of presentation has been improved based on the proposed comments contained in the PDF file. Additionally, the title of the article and the name of the filter have been changed.
On the "optimization". On page 6 the penultimate phrase has been changed and clarified. «The method can be improved due to optimal сhoice of the quantity of the analyzed magnetic observatories and increasing the temporal interval of observations.
On p. 6 on Figure 3 the calculation results are displayed, that show both the probabilities of correct SC recognition and the false recognition probabilities. Therefore, in our opinion, adding a separate figure demonstrating the false recognition probabilities is redundant.

On behalf of co-authors – Prof. V.G. Getmanov

Reviewer 2 Report

This paper proposes a method to be applied on magnetic data coming from observatories at ground (but maybe it could be also applied on satellite data in the future?) to detect Sudden Commencement events. Authors designed a decision-making rule to recognize these events. They used this methodology in a single case study (1-day of ASC observatory data) and in 16 datasets coming from 16 observatories during 1-yr to estimate probabilities of real and false recognitions.

I think that the proposal is interesting and the exposition of the methods and results is clear (but please consider my comments in the annotated PDF). I think that after some moderate revisions this paper could be accepted to publication.

General comments:
1. This paper could be interesting for geomagnetic community and this is the reason why I think that it should be highlighted in the title (maybe replacing events by geomagnetic storms/events?).

2. In my opinion, some concepts such as derivative filter, noise, etc should be briefly introduced at the Introduction section. Obviously most of them are very well known for the community but I think that it could be useful.

3. Please, revise with calm the notation (indices, variable names, etc) used along the paper to be consistent (I have indicated some inconsistencies in the annotated PDF but maybe I missed something), or clarify when necessary.

Specific comments/suggestions:

I have provided an annotated PDF of the manuscript where you can find the specific comments and suggestions. Also some typos that I detected.

Some important remarks are:

1. About the optimization method: would it be possible to test the robustness of your optimization method in comparison with other methods (see annotated PDF)?

2. Section 2.3: the concept of real SC recognitions should be revised

3. Authors show just an example about their method on a single case study (ASC). I'm wondering if it could be possible to extend the analysis to other case studies. What happens if you use another magnetic observatory? Are the results consistent? What's the change with respect to the optimal values analysis? I suggest adding at least another example of observatory (in the opposite hemisphere in middle latitudes) to check the results

4. In Section 3.2: Why didn't authors use the 156 observatories of INTERMAGNET? Computational reasons?

5. I think that it could be useful to include, for the optimal threshold, a confusion matrix of the results in Figure 3 (see annotated PDF)

6. It would be a good exercise to compare these results with the effectiveness and productivity of other methods of SC recognitions mentioned in the Introduction

Comments for author File: Comments.pdf

Author Response

Responses to general comments:

1.In the revised text, we have changed the title: «events» has been changed to «geomagnetic storms», and the filter name has been changed.

In the introduction, the brief information is presented on the proposed method, that is related to the estimation of derivatives for discrete geomagnetic data. On Page 2, after the third paragraph, an explanation has been added: «The proposed approach to solving the formulated problem of SC recognition consists in using differentiating filters for geomagnetic data. The structure of such filters is selected based on the problem of optimal use of data from a set of magnetic observatories for a certain time interval.
The corrections have been made for the denotations according to the reviewer’s remarks.

Responses to particular comments:

The proposed optimization method is reliable enough because a combination of the least squares method that leads to solving the linear equation system, and the discrete sorting method. This remark is introduced in the text of the manuscript - page4 at the end ofSection2.2.
Section 2.3. - regarding the proposed concept of the probability of correct SC recognition. Of course, this part of the article is not ideal; instead of revising, the authors propose to interpret the proposed concept as an approximate one. The first paragraph of Section 2.3 was replaced, the word “optimal” was removed, and the sentence “Consider an approximate scheme for calculating the indicated probabilities” was added.
In order to make the data from different observatories identical in a certain way, the normalization of observations was applied - part of the questions on item 3 of the review is removed. Please look at the list of observatories showing that the observatories on all continents worldwide are considered - the other part of the questions of item 3 of the review is removed.
The authors did not use data from all INTERMAGNET observatories due to technical and organizational reasons.
Of course, the value of the optimal recognition threshold depends on the statistical characteristics of the initial geomagnetic observatory data. The calculation of the optimal threshold is associated with the solution of a rather complex problem of digital processing of these observations, which is much beyond the scope of this article.
In the publications [4-9], solutions are given for various formulations of SC recognition problems. The main feature of the solutions presented is that observations from one observatory are used to obtain them. The proposed SC recognition here is based on a whole set of observatories. This approach is potentially more effective. In this regard, in paragraph 4 of the article with a discussion, the last paragraph is placed, which repeats the content of paragraph 6 of the answer to the review.

On behalf of the co-authors - prof. Getmanov V.G.

Round 2

Reviewer 2 Report

I would like to thank you very much to the authors for addressing most of my comments and suggestions of the first review. I think that this paper will be ready to be published after minor revision.

Check the name changes with respect to the previous version of the paper: derivative filter by differentiating filter and generalized derivative by maximum absolute/amplitude derivative (decide absolute or amplitude, please).
Some notation should be clarified (indicated in the annotated PDF).
Include units when necessary (indicated in the annotated PDF in the most of cases but please, check this carefully along the manuscript).

Comments for author File: Comments.pdf

Author Response

Dear Reviewer,

Thank you very much for your comments.

We've checked the text and fixed the terms.
We decided to use the "maximum amplitude derivative" term as the most relevant.

Notations have been clarified, units have been added where necessary.

Please see the corrected version of the manuscript and the PDF with the marked-up changes.

On behalf of co-authors
Dr. R.V. Sidorov

Author Response File: Author Response.pdf