Towards the Accurate Automatic Detection of Mesoscale Convective Systems in Remote Sensing Data: From Data Mining to Deep Learning Models and Their Applications
, Alexander Sprygin 4,5, Svyatoslav Elizarov 1, Alexandra Narizhnaya 4, Andrei Shikhov 4,6 and Alexander Chernokulsky 4,7Round 1
Reviewer 1 Report
Review of the paper “Towards the accurate automatic detection of mesoscale convective systems in remote sensing data: from data mining to deep learning models and its applications”
by Krinitsky et al.
The study is dedicated to an attempt to develop a new methodology for determining and classifying mesoscale convective systems in the European territory of Russia. Many different procedures have been proposed, from the selection and definition of events "manually" (with the help of so-called experts) to the use of deep machine learning. In itself, the problem of studying/classifying MCS is very relevant. Such events are rare, but can have a devastating impact on infrastructure, threaten lives, etc.
A great job is done, and I do not see any significant obstacles to its publication, but I will note a few points that, in my opinion, should be discussed.
The most significant remark: in the work a lot of attention is paid to the methodology itself, but significant results of its work are not given: what MCS have been identified, classified? Did you get any database? What is the practical outcome? What should the reader do with this information?
Further minor remarks (with line numbers):
you need to check that the abbreviations are everywhere deciphered in the text.
219 “As we mentioned in Section” -> “As we mention in Section”
374 "??" - 3 and further check the numbering of the figures
Fig 3 why do you need fields” label_uid and track_uid?
Table 3 description move above table (this applies to all tables)
233 “characteristics should persist more than 6 hours” and in Fig. 4 there are much shorter time intervals. Why?
495 "?"
595 "on the contrary"
710 measures
In general, I repeat, the article can be recommended for publication. I would like to ask the editor to check the text for plagiarism (most importantly, for self-plagiarism) as co-authors have many articles on related topics.
I didn't find any problems with English. There are a few typos.
Author Response
Please find the response PDF file attached.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
This manuscript presents (a) the GeoAnnotateAssisted tool for fast and convenient visual identification of MCS in satellite imagery, which is capable of providing AI-generated suggestions of MCS labels; (b) the Dataset of Mesoscale Convective Systems over the European Territory of Russia (DaMesCoS-ETR), which we created using this tool, and (c) the Deep Convolutional Neural Network for the identification of Mesoscale Convective Systems (MesCoSNet), constructed following the RetinaNet architecture, which is capable of identifying MCSs in METEOSAT MSG/SEVIRI data.
The manuscript is well organized but some minor parts need to be revised.
1. Line 110-121, the use of 1,2,3 is not recommended here, as it duplicates the first level heading of the article.
2. Please standardize the format of the images, e.g. font size, font style, etc. especially the Figure12.
3. Line 464, I’m sure it’s an error of “??”
4. Eq.(1) if and yk should leave more distance;
5. 3.1 RetinaNet for the identification of MCS is not very clear to be understand.
And a major revision is suggested.
A minor polishing of the English language is recommended.
Author Response
Please find the response PDF file attached.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
The manuscript is well written and is suggested for publication in Remote Sensing.
