AutoST-Net: A Spatiotemporal Feature-Driven Approach for Accurate Forest Fire Spread Prediction from Remote Sensing Data

Round 1

Reviewer 1 Report

I reviewed the manuscript "AutoST-Net: A Spatiotemporal Feature-driven Approach for Accurate Forest Fire Spread Prediction from Remote Sensing Data" by Chen et al. The topic is interesting, especially in the context of climate change, and the manuscript is well-written. I have some comments that need to be addressed so that the quality of the manuscript will be improved.

1- Figure 1, Please revise Figure 1, and present the study area by adding the point of fire events (seven forest fire areas). Please also add the coordinates to the figure.

2- Table 2, What is Geopotential height? Please add relevant explanations, along with its association with forest fires. How much was the accuracy of Himawari-8 fire data over the study area? Please explain in detail. How the KBDI was calculated?

3- Line 162; How did you resample the coarse resolution data to 2 km?

4- Figure 2; Please improve the quality of the presentation, and add suitable legends for sub-images. Please enlarge each sub-image for a better presentation.

5- Line 189; Please justify what was the intention of considering a 1-hour temporal resolution. Please elaborate more about choosing the closest available data for clarification.

6-  Table 2; Are the comparisons made using the test dataset (two distinct forest fires)? If not, please include a similar table and compare the performances based on the test dataset. If yes, please separate the results for each test dataset for clarification. Moreover, did you consider hyperparameter tuning for other algorithms? how many epochs were used for model training? For a fair comparison, all models should be tuned according to their specifications. 

7- Figure 7; Please add results from the two test datasets for better visual comparison comparison.

8- Table 4, The current comparison has limited information, as it is not clear which parameter has the most influence. Therefore, further analysis is required to solve this limitation and make this section more exhaustive. you can add the feature importance of each parameter for clarification.

Author Response

Dear Reviewer,

First and foremost, I would like to express my heartfelt appreciation for the time and effort you have dedicated to reviewing my manuscript. Your valuable comments are of great importance to me, and I have taken careful consideration and applied your suggestions to enhance the quality of my manuscript.

Regarding the revisions you suggested, I have meticulously gone through them and made detailed modifications. Please see the attachment.

I look forward to your next review and wish you well.

Sincerely,

Xuexue Chen

Author Response File: Author Response.pdf

Reviewer 2 Report

Dear authors!

At present time your article has some disadvantages.

Introduction

Introduction section is totally unacceptable at present time. I found only 20 cited works in Introduction. It is too small. Please, fully rework this section. Firstly, You must extend review on forest fire spread simulation taking into account different models such as statistical, physical-based, geometrical, graph, probabilistic, empirical. etc in context of your research with ANN.  

More over, You must provide review of sattellites that can be used for purposes of forest fire spread simulation. It is necessary too proovide at least information on price and availability of cosmic images, orbital information, period of sensing for definite territory, region of monitoring.

Also you should add sentence "The purpose of the paper is ..." at the end of the Introduction. Also you can add 2-4 Objectives.

Quantity of suggested citation in Introduction is totally poor. I think total number of cited works in Introduction must reach at least 50-60 original woorks published by scientist from the USA, Russian Federation, Canada, South Europe.

Data

Please, provide precise and full set of information on sattellite Himawari-8. Also give information on products used for detecting the hot spots over the land. Moreover, clarify period of monitoring for study area bt Himawari-8 sattellite and its orbital positioning. Also you must support possibility to use images with spatial resolution on the scale of 2 km for the purposes of predicting  forest fire spread. You need too provide supporting calculations.

Figure 2 is to small. Please, increase pictures to 3-4 times. At present time text is hard to read.

Methodology

Figure 3 - Please, increase size of your text.

Please, support possibility to use images with 32x32 pixels for the purposes of predicting forest fire spread.

Please, provide scheme and description of artificial neural network used in your work with mathematical description.

Please, clarify mathematical matter of procedures Encoder and Decoder.

Figure 4. Please, provide detailed scheme of this mechanism taking into account mathematical basis.

The same situatioon with Figures 5 and 6.

Please, provide information on implemented program realization of your algorithm. What program language and rapid application development software were used in you work?

Please, clarify the modes of work for your computational software. Is this automative or manual mode?

What software was used for visualization of results?

Experiment & Results

Table 2. Please, provide description of all models used in comparative analysis? 

Provide execution time of your software and compared models.

Clarify what mode provided by your algorithm. Is this real-time, retrospective modes or your software can work in advance the real time?

Figure 7. Please, increase the picture. Add information on real fire incident used for this computations.

Discussion

Please, add Discussion section.

Compare and contrast your obtained results with works cited in Introduction. Not only with neural networks. If you can not provide quantative comparison You will should provide qualitative or descriptive analysis.

Please, specify you results in context of advantages and disadvantages.

Make description of limitations of your research.

Provide comprehensive description of future researches.

Conclusion

Please, rework the Conclusion section.

Provide, numbered set of 3-5 key findings with corresponding conclusions.

Clarify possibilities of prectical usage of your developments on the territoty of whole state. What kind of attepmts must be performed for practical usage in forest service?

References

Please, completely rework this list.

Add new cited works according to my remarks concern the Introduction section.

No comments.

Author Response

Dear Reviewer,

First and foremost, I would like to express my heartfelt appreciation for the time and effort you have dedicated to reviewing my manuscript. Your valuable comments are of great importance to me, and I have taken careful consideration and applied your suggestions to enhance the quality of my manuscript.

Regarding the revisions you suggested, I have meticulously gone through them and made detailed modifications. Please see the attachment.

I look forward to your next review and wish you well.

Sincerely,

Xuexue Chen

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

I appreciate the authors for considering all comments. I find the responses convincing, and the manuscript has improved in clarity according to the comments.

Please include the high-resolution figures before possible finalization. The legend in Figure 2 is not observable.

Author Response

Dear Reviewer,

First and foremost, I would like to express my heartfelt appreciation for the time and effort you have dedicated to reviewing my manuscript. Your valuable comments are of great importance to me, and I have taken careful consideration and applied your suggestions to enhance the quality of my manuscript.

Regarding the revisions you suggested, I have meticulously gone through them and made detailed modifications. Please see the attachment.

Best wishes for you.

Sincerely,

Xuexue Chen

Author Response File: Author Response.pdf

Reviewer 2 Report

Dear authors!

Thank you for your revision.

I accepted the majority of your improvements.

Only one remark: You should provide citations in Discussion section.

No comments

Author Response

Dear Reviewer,

First and foremost, I would like to express my heartfelt appreciation for the time and effort you have dedicated to reviewing my manuscript. Your valuable comments are of great importance to me, and I have taken careful consideration and applied your suggestions to enhance the quality of my manuscript.

Regarding the revisions you suggested, I have meticulously gone through them and made detailed modifications. Please see the attachment.

Best wishes for you.

Sincerely,

Xuexue Chen

Author Response File: Author Response.pdf