Anomaly Detection of Sensor Arrays of Underwater Methane Remote Sensing by Explainable Sparse Spatio-Temporal Transformer

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

First of all, about the title that it is different between the text and the supplementary material. In the text is Anomaly Detection of Sensor Arrays of Underwater Methane Remote Sensing by Explainable Sparse Spatio-Temporal Transformer, while in the supplementary is Anomaly Detection of Methane Sensor Arrays under Industrial Conditions by Explainable Sparse Spatio-Temporal Transformer. I invite the authors to decide which is the preferred.

In any case in the title of the manuscript the term Remote Sensing is included, this term normally indicates methodologies using Earth observation approach, please clarify to avoid misunderstandings (I suggest to remove).

The main drawback is the description of the used methodology and the results is almost completely obscure, in particular 2. Theoretical Fundamentals and 3. Experiment and Results. The suggestion is to reconsider the text, rewrite completely to be more easily understandable by the readers, including the not expert ones in the matter. This helps to increase the clarity and then to enlarge the potential audience. Consequently abstract and conclusions have to be reconsider too.

In line of this, for instance line 21 explain the sentence   The data input into the explainable sparse block could decrease the time complexity to O(n), and particularly O(n); lines 27-29 explain The F1 score of the model put forward and anomaly accuracy is 0.92, which is superior to other reconstructed models such as convolutional_autoencoder (0.81) and long-short term memory_autoencoder (0.66). Explain carefully the parameters and the formulas introduced in the text.

The acronyms have to be explained the first time they are cited.

Appendix 1 and 2 (lines 143-144), I suggest to include both in the text, but they are obscure and difficult to read, therefore rewrite better. The references cited there are unclear and uncomplete. The content of the appendix 3 seems not discussed, please clarify.

The figure captions have to be self-explanatory, many figures have captions not sufficient to describe the figure(s) itself (e.g., figs. 1, 2, etc.). The tables have the same problems, explain carefully all the parameters included.

Moreover, the tables (? and to be numbered?) between the lines 141-142, 193-194, 265-266 are not explained at all.

REFERENCES

The crosscheck between refs in the text and in the list is difficult. Please use references easily findable, avoiding as much as possible citations of proceedings of conferences and workshops.

The authors have to make an effort to increase the references cited with a more international panorama, most of them at the moment are frmn Chinese lietarure,

Lines 115-116-117 Clark et al. (2019) is not cited and the number [45] does not correspond; lines 121-122 same but it is indicated [46?].

The refs [45] and [46] in the References chapter are not mentioned in the text. The total refs in the list are [48], while in the text is cited a ref [49] Kitaev et al. (2020) not reported in the list.

Minor comments

line 15 The increasing occurrences --> The increasing discovery

line 21  faults --> failures

 

Comments on the Quality of English Language

In general, it is not clear and it is very difficult to understand. There are also mistakes in the words.

Author Response

1

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

Manuscript Title: Anomaly Detection of Sensor Arrays of Underwater Methane Remote Sensing by Explainable Sparse Spatio-Temporal Trans- 3 former 

This paper takes a case study applying an explainable sparse spatio-temporal transformer for detecting the Anomaly of the underwater methane remote sensing sensor arrays.  I think that the paper may be acceptable provided extensive modification as below:

(1) the introduction part seem to be general, it should focus on the development and existing problems about Interpretable Explainable Sparse Spatio-Temporal Transformer and its application inAnomaly Detection, this aspect can be more comprehensive.

(2) You stated the public dataset was adopted,I think it is necessary to provide detailed supplementary explanations on the sources, characteristics, and other properties of the data;

(3)  How to evaluate the credibility and reliability of the underwater methane remote sensing sensor arrays? What kind of evaluation indicators are there?

(4) What are the reasons that lead to differences in the performance of dropout operation?

(5) How to to estimate and strengthen the visualization of variable-importance？It is best for the paper to better reflect the interpretability of the model.

 

(6) Have any interference signals in your experiment? How to avoid abnormal values caused by interference signals in the sensor array to  affect the real detection of methane anomalies?

Comments on the Quality of English Language

the writing style, capitalization and spelling errors and occasionally inappropriate, such as repetition and missing words, There are also some statements and conclusions which are either commonplace or insufficiently supported by the presented data or reasoning

Author Response

1

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

Please consider my comments and suggestions. I am not convinced that the paper is ready for publication at the moment. Please work on the discussion section, in particular.

Comments for author File: Comments.pdf

Comments on the Quality of English Language

The language of the paper is good. However, I would suggest it is proofread to correct a few grammatical and language errors.

Author Response

1

Author Response File: Author Response.docx

Reviewer 4 Report

Comments and Suggestions for Authors

Currently, underwater remote sensing of methane is of great importance. The authors of the presented article claim that it contains an explainable spatiotemporal transformer model that was proposed for methane-sensor array anomaly detection in the underwater methane remote sensing applications. A signal processing algorithm is proposed that may be of some interest to specialists.

The article can be published after correction and addition of the necessary information:

1.      Line 33. Methane hydrate is also called combustible ice which seems like ice and combusts easily when it comes in contact with an open flame [1-3].

For a scientific paper, this presentation of methane hydrate looks unscientific. Here it is necessary to define gas hydrate, to say where methane gas hydrate actually is and in what quantity, how much methane is contained in its unit volume, and only then can we add that it looks like snow and at the same time burns if it is set on fire.

2.      Line 34. Combustible ice mining techniques of developed countries, such as the United States, Japan, and Germany, have always been in the forefront of the world.

It is appropriate to say here that back in 1969, in Russia (Messoyakha field, Nankai), methane production from gas hydrate began for the first time in the world (10.1016/j.jngse.2012.08.002).

3.      Line 35. Its energy generated by combustion is as ten times more as natural gas, gasoline, and coal, making it possible for low-carbon society.

It is not clear where this statement comes from and what it is about? Link required.

4.      Line 39. Therefore, underwater remote sensing of methane is of great importance for the exploitation of the combustible ice and protecting the environment [11].

A completely incomprehensible conclusion, since, for example, there are no gas hydrates in the underwater gas pipeline. Therefore, it is better to remove the phrase “the exploitation of the combustible ice and” from the sentence. At the same time, you can add a link to the importance of remote recording of powerful natural methane emissions, for example those that occur on the Arctic shelf (10.1126/science.118222)

5.      A separate paragraph should be added to the introduction, in which it is necessary to give a short overview of existing remote methods for detecting underwater methane emissions (for example, acoustic: 10.1098/rsta.2014.0451), the states of gas hydrates in the near-surface layer of bottom sediments (for example, 10.3390/geosciences12090315) and other technologies with advising links. Compare the methods with each other, describing the advantages and disadvantages.

6.      In a separate paragraph, it is necessary to indicate with references which methane leakage sensors will be discussed in the work, why they are taken as a basis and what advantages they have compared to other sensors.

7.      Line 267. 3.1. Experiment Setup.

It is necessary to describe Experimental setup using the gas sensor array in more detail. It is not clear from the above text whether such a system can work in natural conditions or only in the laboratory. Which system should be used in real conditions? How exactly does it work (describe in as much detail as possible). How many such systems are needed to organize remote monitoring of the water area? Do similar systems currently operate and where?

8.      Line 268. The public dataset was adopted in this paper.

The meaning of this proposal is not entirely clear. Needs to be rewritten.

Author Response

1

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The author had made some improvements to the paper. I still have a few suggestions.

Firstly, I feel it is inappropriate to place the discussion section after the conclusion. Additionally, I feel that this discussion paragraph should be more detailed，the conclusion section is too short, For some other parts the language seems to be more concise;

Secondly, What is the reason why your revised manuscriptuniformly deleted "remote"? Based on this, is the paper within the scope of the journal? In fact, there is a significant gap between your content and my understanding of remote sensing.

Comments on the Quality of English Language

Language can be improved and refined.

Author Response

1

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

I do not see much improvement in the discussion. The discussion is extremely important to this paper as it will give it a novelty. Please can you spend more time illustrating how your results are adding to the pool of knowledge?

Comments for author File: Comments.pdf

Comments on the Quality of English Language

The English language has improved, though there are a few errors here and there. Once the final manuscript is ready, I suggest the authors get a native speaker to edit the language.

Author Response

1

Author Response File: Author Response.docx