Forest Fire Prediction Based on Long- and Short-Term Time-Series Network

Round 1

Reviewer 1 Report

Remarks

The authors propose a forest fire prediction model based on a long-term and short-term time series network (LSTNet) to improve the accuracy of forest fire forecasts. They consider eight factors that influence forest fires, and the data were obtained through remote sensing satellites and GIS. Pearson analysis and multicollinearity test were used to filter and validate the forest fire factors. The proposed prediction model has higher accuracy and stronger robustness than traditional machine learning models. 

 

The article is well-written and structured. Moreover, it deals with an interesting topic. The description of the model in section 2.4 could be improved to make the proposed approach clearer.

 

 

Questions and suggestions 

• On page 2, the acronym NDVI is used, and it is only introduced later on page 3.
• On page 2, the acronym SVM should be introduced.
• The authors should insert space afetr the legend of Table 1 and after Table 2.
• Before the brackets with citations should insert a space. For example, on page 5, it appears “whether[“ and “ ocurrence[“.
• On page 6, instead of “ranges from [-1,1],” it should be written “ranges in the interval [-1,1]”.
• On page 6, it is not necessary to introduce R in “R (Red)” because the authors always use “Red”.
• On page 7, it should correct “a value between [-1,1]”. If you use between, you cannot present the interval.
• On page 7: the caption of figure 2 is on the next page.
• On page 9, the acronym CNN should be introduced.
• On page 10, the acronym RNN should be introduced.
• On page 5, Figure 5 has some inaccuracies: some arrows need to be included, there is a very short arrow, and a + sign needs to be included. The operation X used in the diagram could be confusing with X_t. Equations (11) and (12) use another notation for the product.
• On page 11, equation (18), is the upper limit of the summation correct?
• On page 10: use lowercase and uppercase w_k. It should be equalized. 
• On page 13, there are typos in the first sentence of section 3.3. It should be rewritten.

Author Response

Please see the attachment.

Author Response File: Author Response.doc

Reviewer 2 Report

Dear authors!

Thank you for your article. I think this article is in the scope of Forests journal, but some disadvantages were found in current version of the manuscript.

Abstract

I think you should follow to the common structure of the abstract: Introduction, Background, Methods, Results and Discussion and Conclusion.

Introduction

I think you must provide separate section Background with literature review. At present time I found only 35 cited references. It is too small quantity for research article with such wide topic like forest fire danger prediction. Please, completely rework this section with additional references. I estimate this quantity about 50-60 unique references with works around the world. It must be noted, that Forests is an international journal and you must provide global background for your research.

Materials and Methods

I think you must extend the Table 1 and consider forest fire factors with ignition sources like human activity and lightnings over the controlled territory. As you knew, forest fire is multistage phenomenon that can be described as follow: inert heating, drying, pyrolysis, ignition, flame combustion and afterburning. You must consider ignition source when you are developing forest fire danger prediction methods or models.
Please, revise this section.

The same situation with table 2.

Results

Please, explain figures 8. How did you compute Ignition probability without ignition sources? Please, revise this part of your manuscript.

Discussion

This section is too small. PLease, make a journey through the cited references in Background to contast and compare your results with published works in Discussion. You should completely rework this section.

Conclusion

I think current conclusion is too general. Please, mark out 3-4 key finding and formulate the corresponding conclusions in this section.

References

This section is too small. Please, revise according Background suggestions.

Author Response

Please see the attachment.

Author Response File: Author Response.doc

Round 2

Reviewer 2 Report

Dear authors!

Thank you for your revision.

I agree with your clarifications and explanations.

But I still have to suggest some improvements of your manuscript.

I think you have to include human and lightning activity to the Table 1 and Table 2 with status "Not taking into account". And write extensive explanations of limitations.

Also you should extensively rework conclusion with brief description of further researches, taking into account, human and lightning activity factors.