Enhancing Cyclone Intensity Prediction for Smart Cities Using a Deep-Learning Approach for Accurate Prediction

Round 1

Reviewer 1 Report

In general, the manuscript describes a novel study. However, major improvement/modifications required in the manuscript before it can be accepted for publication. Some major comments are:

1. The title must include the words "prediction of cyclone". Also, the words "Rainwater Harvesting and Smart Irrigation for Sus-tainable Water Management in Smart Cities" should be removed, as I do not see any result related to rainwater harvesting or smart irrigation. Rainwater harvesting (RWH) can be effective, even without cyclone. I really dont understand why the authors trying to relate RWH with cyclone?

2. Last 4 lines of abstract, first 9 lines of 'Introduction', two dot points on Page 3, Sections 3.3 and 3.4 are totally irrelevant and MUST be excluded. I do not see any relation with the cyclone study.

3. Figure 3: split between Pages 6 & 7

4. Second paragraph of the 'Introduction' directly starts with the cyclone studies. Authors should cite some papers, which used similar techniques for other hydrological phenomena. Some relevant studies are mentioned below. Authors should cite these before jumping into cyclone matter.

i) Abed, M., Imteaz, M., Ahmed, A.N. and Huang, Y.F. (2022) Modelling Monthly Pan Evaporation Utilising Random Forest and Deep Learning Algorithms, Scientific Reports, Vol. 12(1):13132, DOI: 10.1038/s41598-022-17263-3.

ii) Oad, S., Imteaz, M.A. and Mekanik, F. (2023) Artificial Neural Network (ANN) based long-term streamflow forecasting models using climate indices for three tributaries of Goulburn River, Australia. Climate, Vol. 11, No. 152. DOI: 10.3390/cli11070152

iii) Ghamariadyan, M. and Imteaz, M.A. (2021) Prediction of seasonal rainfall with one-year lead time using climate indices: A Wavelet Neural Network scheme, Water Resources Management, Vol. 35, pp. 5347–5365. DOI: 10.1007/s11269-021-03007-x.

 

 

 

5. Very is minimum is presented on the accuracy of the model (only Figure 6). Authors should present some more results on the model accuracy.

6. Some tables and figures are duplication of results (i.e. Table 3-Fig 6, Table 4-Fig 7, Table 5-Fig 8, Table 6-Fig 9). Please keep either the figures or the corresponding tables. No need to show both.

7. Last line of page 17: "compared to existing Tong et al. [1] and Na et al" should be mentioned as "compared to the results by Tong et al. and Na et al."

8. English needs to be improved. Second para of Conclusion mentions "Cyclone is extreme weather systems that improve over the tropical ocean." no meaning?

 

English needs to be improved. Some comments made, however a thorough checking is necessary.

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 2 Report

The manuscript of the authors from India, Egypt and Saudi Arabia is concerned with " Enhancing Rainwater Harvesting and Smart Irrigation for Sustainable Water Management in Smart Cities using Deep Learning Approach for Accurate Prediction".

Thank you for the opportunity to review such useful and actual research study.

The manuscript presents the integration of rainwater harvesting plus smart irrigation in smart cities that offers several benefits and it helps reduce the demand for freshwater sources. By using rainwater for non-potable purposes and optimizing irrigation practices, cities can reduce their dependence on traditional water sources.

This submitted article could be with the aim and scope of the MDPI journal Atmosphere.

• Abstract & introduction: These two parts are focused on the paper's main aim and the new contributions of authors to the state of the art. The abstract with keywords very effectively summarizes the manuscript.

The key objective for the authors is to predict the intensity of the cyclones, as well as to improve the accuracy of cyclone intensity prediction. To address this challenge, they proposed a new technique „Linear Support Vector Regressive Gradient Descent Jaccardized Deep Multilayer Perceptive Classifier“.

• Materials & methods: Based on the existing research, the authors developed of the very interesting and novel experimental technique called Linear Support Vector Regressive Gradient Descent Jaccardized Deep Multilayer Perceptive Classifier (LEGEMP) to improve the accuracy of cyclone intensity prediction by using of deep learning approaches.

It is clear how all of the data were obtained. This section gives readers enough information so that they can use the study for other areas.   

• Results & discussion: The data are well-controlled and robust and results are well-presented with relevant and current tables, figures, and references.

Results section were obtained and the methods used to analyze the data are progressive and scientific sound. With some adaptation, this described methodology of could be used more generally.

I hope that authors will continue with such interesting and inspirative research.

I have no comments for improving of manuscript before publishing.

 

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 3 Report

The title needs to reflect the content of the paper - rainwater harvesting (RWH) is at the top of the title, but none of the outputs are applied to it. The last 2 words "accurate prediction" isnt qualified, accurate prediction of what?

Upon reading the abstract, I found that the paper was about predicting the intensity of cyclones, RWH is mentioned at the very end, but feels like an afterthought.

The introduction begins with RWH and until the sentence ending “sustainable water future” was reasonable. However, suddenly “cyclone intensity” was mentioned, which could have been applied to RWH, but wasn’t and was followed by a sentence including “tragedy management” which didn’t really fit. The list of models needed context, ie applying back to the rest of the paper to follow. The first 3 bullet points of the research contributions are actually the methodology, which is then repeated in the methodology section, so needs to be removed and replaced with 2 or 3 aims. The last 2 bullet points don’t appear to be relevant to the rest of the paper,and repeat some of the first paragraph of the paper and so need to be removed.

Section 2 is very list-like and could do with being rewritten more coherently.

The methodology starts by defining cyclones, but not in a way useful to their application to RWH – contextualising them in terms of their destructive capability, not their potential for RWH, all a bit confusing. Various words are used, but not defined eg “attributes”; “features”, “3 major processes” are followed by 2: attribute selection and classification. The last 2 sentences above Figure 1 and the first one below it are repeats. In figure 1 the box “selected significant” seems incomplete.

Formula 1 needs a key – what do the lower case numbers, _m1 and _m2 signify? Figure 3 and its description needs to be placed far earlier to avoid this kind of confusion.

The whole of page 6 is highly repetitive, with Y=+1 indicating relevant “features” and Y=-1 indicating irrelevant features described 3 times, and the Herfindahl model description repeated from para 3 Section 3, and also bullet point 2 in section 1. On the bottom of page 6 an Algorithm 1 is mentioned – the only time it is mentioned, and it isn’t explained.

Page 7 – give ANN in full

Page 8 – “perceptrons” or “neurons” are mentioned, but not defined. A “synapse” is defined, but never used again. Figure 2 is incorrectly referred to. Section 3.3 is very repetitive, and also doesn’t 1fit – it isn’t a methodology, it is descriptive. The same can be said of section 3.4 until the paragraph starting “Figure 4”. Unless these 2 sections are properly applied to the rest of the paper, they should be removed. The rest of section 3.4 carries on from section 3.2. Here a neuron is also a node, or it was a perceptron at the top of the page. Please just use one term, as it gets very confusing.

Page 10 – Figure 5 mentions “neuron”, the paragraph beneath “perceptron” – are the same thing or not? Bottom of the page “algorithm 2” is mentioned in the first sentence of the paragraph which is incomplete.

Page 11 – “neuron” is used….

Page 12 – section 4, needs to justify comparison against 2 models that had already been criticised in the introduction – the fact that the developed model outperformed the 2 existing ones therefore comes as no surprise. Once again there are undefined terms used – what is the difference between a feature and an attribute? The Bay of Bengal and its data are mentioned twice in section 4 and again in section 5.1. The list of features/ attributes are given in section 4, again in section 5.1 and yet again at the bottom of page 16. Section 5.1 refers to “instances” of which each one consists of “parameters or attributes or features” – all this repetitive terminology really needs to be tidied up so the reader understands what is going on.

Page 13 – Tables 1 and 2 are not referred to. The whole of the paragraph at the top of Page 13 is a repeat of elsewhere and needs removing.

Tables 3, 4, 5, and 6 need removing as they simply repeat figures 6, 7, 8, and 9.

Page 15 – in the paragraph below Fig 7, “3 methods” are mentioned, but only 2 given – the same is true at the top of page 17.

Page 17 – the paragraph under Table 6 needs to be rewritten far more clearly and with less repetition. What is “after-reason improvement”?

Page 18 – The conclusions mostly don’t conclude, they are also repetitive, describing once again the construction of the model, and trying to squeeze RWH and irrigation in where they don’t really fit.

 

As a whole the paper is highly repetitive and hence hard to follow. The developed model needs to be firmly applied to RWH – how much can be harvested during certain cyclones (can it?) how can it be distributed around the smart city? If you accurately predict the intensity, wind speed etc of a cyclone, how does that help RWH? The formulae are not really applied clearly to anything, and it would be useful for them to be introduced by something like: “in order to” do something, rather than the formula describes rate of error, or similarity… Personally, I think it would be far better to take all mention of RWH and smart cities out as they don’t add anything – apply the model instead to hazards, the ability to notify people of impeding cyclones – should they be evacuated etc? Or allow it to become a modelling paper, which I think would be of interest on its own merits to the scientific community and well worth publishing – certainly in a journal such as “Atmosphere”.

The English needs to be far more clearly expressed, it is hard to follow, over complex and highly repetitive. Often, sentence structure is split by figures or tables, which also leads to repetition. 

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 4 Report

The work deals with an interesting topic and seems to be very current. Research on alternative water resources is obviously necessary, especially in dry regions. Sustainable exploitation of natural resources and their protection is of key importance for smart development. To achieve this, it is necessary to implement alternative sources of water in all areas of economy. This study presents a novel method called LEGEMP developed for improved cyclone intensity prediction, containing two diverse stages: feature selection and classification. This research is very important because applying deep learning algorithms to rainwater harvesting systems, it is possible to optimize the collection, storage, and distribution of rainwater, ensuring its efficient utilization. The outcomes of this study can be beneficial for designing local strategies of water management.

The paper falls within the scope of Atmosphere journal.

The paper is well-organized, containing all of the expected components. The methodology is effective in attaining the object of this work. In the Introduction, the Authors provided a brief research background discussing the current state of prediction models and rainwater harvesting systems. Discussion and Conclusions are supported by the results presented in this paper.

The text requires an editorial correction. There are no line numbers in the text, which makes it impossible to accurately indicate errors. There are a lot of unnecessary spaces in the text or they are missing between words and punctuation marks. Tables and figures should be formatted in accordance with the journal's guidelines. Figure 3 is divided into two pages. 

There are descriptions of the algorithms on pages 7 and 11. This should be included as a figure or table and cited in the text of the article.

Author Response

Please see the attachment

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Authors' have addressed all the comments issued earlier. The manuscript is acceptable.

Author Response

Please see the attachment

Author Response File: Author Response.doc

Reviewer 3 Report

Comments for author File: Comments.pdf

There are still issues of repetition, but this has been much improved. 

Author Response

Please see the attachment

Author Response File: Author Response.pdf