Predictive Modeling of Forest Fires in Yunnan Province: An Integration of ARIMA and Stepwise Regression Analysis

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This article compares two forest fire prediction models based on ARIMA and Stepwise regression models.

Although the article presents significant results, some suggestions must be taken into account to improve the article:

 1. The summary must be reviewed and written again in order to show the relevance of the problem, methodological aspects, and results. For example, you should not include equations in the summary, as the variables are unknown. It is difficult for the reviewer to understand the summary.

2. The introduction needs to clarify why to use the ARIMA and Stepwise regression methods for comparison. Why were they chosen?

3. In general, both in the summary and in the introduction, the problem could be more straightforward.

4. In lines 106 to 113, the data set will be used to build the models presented. However, the article did not cover a section on selecting variables associated with forest fire prediction. This part is essential in model building.

5. “Section 3.1 Stepwise Regression Model” was significantly reduced, with insufficient justification to associate with the results.

6. The results are quite consistent, showing fit metrics and comparisons. I just missed an assessment of the uncertainties associated with these estimates. Did the authors think about the possibility of developing uncertainty analysis associated with forecast models?

 

Author Response

1. The summary must be reviewed and written again in order to show the relevance of the problem, methodological aspects, and results. For example, you should not include equations in the summary, as the variables are unknown. It is difficult for the reviewer to understand the summary.

Thank you for your feedback. The summary will be revised to enhance clarity on the relevance of the research problem, methodological aspects, and results. We will ensure that equations are omitted and that the summary is comprehensible without prior knowledge of the variables.

2. The introduction needs to clarify why to use the ARIMA and Stepwise regression methods for comparison. Why were they chosen?

In the introduction, we will clarify our rationale for choosing ARIMA and Stepwise regression methods. Specifically, we will discuss their respective strengths in modeling and predicting forest fires, and the theoretical and empirical considerations that led to their selection.

3. In general, both in the summary and in the introduction, the problem could be more straightforward.

We will streamline the narrative in both the summary and the introduction to directly address the research problem, making the purpose and objectives of the study more straightforward.

4. In lines 106 to 113, the data set will be used to build the models presented. However, the article did not cover a section on selecting variables associated with forest fire prediction. This part is essential in model building.

We acknowledge the omission and will include a section dedicated to the variable selection process for forest fire prediction, detailing the criteria and methodology used in determining the variables for our models.

5. “Section 3.1 Stepwise Regression Model” was significantly reduced, with insufficient justification to associate with the results.

The section "3.1 Stepwise Regression Model" will be expanded to provide a more comprehensive justification for its use and to better associate the methodology with the results presented in the study.

6. The results are quite consistent, showing fit metrics and comparisons. I just missed an assessment of the uncertainties associated with these estimates. Did the authors think about the possibility of developing uncertainty analysis associated with forecast models?

We agree with you about the lack of uncertainty assessment. We have used regression residuals and Q-Q plots to analyse the uncertainty associated with the predictive models to get a clearer picture of the confidence in our estimates.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

1) Please provide study limitations. 

2) Authors may want to draw a figure to represent study steps in detail. 

3) Please discuss the results in detail. 

4) Please compare the model performances with previous studies. 
5) It should be clarified why stepwise regression analysis and the ARIMA model are being considered on the basis of previous literature and justify its election.

6) The authors may want to extend the literature review with recent studies.
7) The authors may want to cite some related papers published in Applied Sciences.

8) The figures can be more clear, the authors may want to make them larger

Comments on the Quality of English Language

Moderate editing of English language is required.

Author Response

1. Please provide study limitations.

We appreciate your suggestion. The limitations of the study will be detailed, including considerations of data constraints and the scope of the methodology applied.

2. Authors may want to draw a figure to represent study steps in detail.

We will create a comprehensive figure illustrating the step-by-step methodology of the study to enhance the visual understanding of the research process.

3. Please discuss the results in detail.

A more in-depth discussion of the results will be provided, highlighting the implications and significance of the findings in relation to the objectives of the study.

4. Please compare the model performances with previous studies.

The performance of our models will be compared with those from previous studies to demonstrate advancements or differences in predictive capabilities.

5. It should be clarified why stepwise regression analysis and the ARIMA model are being considered on the basis of previous literature and justify its election.

We will elaborate on the rationale behind choosing stepwise regression analysis and the ARIMA model, referencing previous literature to justify their selection for our study.

6. The authors may want to extend the literature review with recent studies.

The literature review will be expanded to include recent studies, ensuring a current and comprehensive context is established for our research.

7. The authors may want to cite some related papers published in Applied Sciences.

Relevant papers from the journal 'Applied Sciences' that align with our research will be cited to position our study within the existing body of work.

8. The figures can be more clear, the authors may want to make them larger

We will improve the clarity of the figures by increasing their size and enhancing visual elements for better readability and impact.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

applsci-2728421

Comparative Performance of ARIMA and Stepwise Regression for Forest Fire Prediction in Yunnan Province

Comments to Authors:

Abstract

1.      The title of the study tells us about the comparison of the dual tools used in the study, meanwhile, the abstract tells us about the different story. See Lines 14-18.

2.      Lines 12-13: What is the research problem? Occurrences of Forest Fires or Lack of Prediction Tools? In my opinion, there are a bunch of modern applications available that can easily predict future scenarios. If the problem is forest fire occurrences, then certainly there are natural and manmade phenomena that speed the fire occurrence process.

3.      Why did authors compare or experience ARIMA and Stepwise Regression?

4.      The authors did not talk about the data collection and feeding process to generate results in the abstract. Correction is required.

5.      What is the significance of the study?

Introduction

1.      This section lacks important facts and figures about forest fire occurrences in the world and hence in the study area. Show that your study is need-based. This is a massive correction to upgrade the quality of the introduction section.

2.      Clarify that selected methods are the best option to predict the forest fire. Case studies should be cited to link with the upcoming section of methodology. The link is missing.

3.      The citation numbers 11-15 are outdated. I wonder why the authors included these to clarify fire risk prediction models. The situation is changed as described in these articles from 11-15. The authors are advised to search for the latest citations to validate their arguments.

4.      I also urge authors to relook at their citation portion as given in different sections of the manuscript. Use fresh citations. These should not be older than a decade.

 

Study area and research methodology

1.      Not a single citation was found between lines 76—83. The authors have referred to important parameters in these lines, however, the authors have referenced certain literature. This gives a bad impression.

2.      Published data were taken from 1991-2021 taken from the different agencies. Is it good to analyze published data? As a statistician, published data should not be further analyzed. If so, researchers should take expert opinion to validate the choice of secondary data usage. Therefore, validation and authentication are required.

3.      Subsection 3.2. No citation is given for the formulas (1) to (4). Why do you discuss these here? What was the purpose and limitations?

4.      See lines 98-103. How was the data obtained from the government agencies? In what year, data were collected?

5.      For missing data, why long-term averages were used? This may generate outliers and vanish the aim of the analytical process.

6.      Line 160-162. This is a biased statement. The evidence is also not seen by referring to Figure 3. 1998-2000 the massive outlier is visible. Clarify discussing Figure 3.

7.      What is hm2? Looks like a unit of area. Clarify. Also, define all such units to make the manuscript readable.

 

Temporal Characteristics of Forest Fires

1.      This section is overstretched and looks hazy. One cannot figure out why these steps were performed by the authors. Are these performed in the quest to achieve study objectives? It is not clear in this section.

2.      Multiple tests and steps were performed in this section without justification. Their link with the reviewed literature and methodology section is also missing.

3.      Draw a flow chart to summarize the required steps to achieve objectives that need to be utilized in this section.

4.      The authors need to correct this section. Do not go out of scope. Compare this section with others to establish a link.

5.      Many of the outliers were observed in the Figures.

            Conclusion and Recommendations

1.      This section merged two important sections. It is advised to split. Recommendations are made in the light of key findings of the study. Whereas, the conclusion section sums up the study in a professional way.

2.      Highlight the limitations of the study. As some of the data inputs are found missing.

3.      Significance and research contribution should be added as separate subheadings to clarify the importance of the study.

4.      Why concerned agencies, reviewers, or students review this study? Justify.

 

Concluding Review Comments:

An important section, i.e., literature review is missing in the manuscript. This section usually abridges methodology and literature. Some data points are missing. Some sections of the study are underrated like the Introduction and some are overstretched like section 4. Citation rules are not followed and many of the outdated citations were observed. One can also review biased statements in the manuscript. There are so many advanced prediction models available, why authors picked ARIMA and Stepwise regression to predict forest fires in China. What tools did government agencies use to predict forest fires from whom the authors collected data? Therefore, authors are advised to rewrite their manuscript to eliminate all such observed loopholes. Rate every section of the manuscript and make certain modifications to make this manuscript publishable.

Comments on the Quality of English Language

English needs improvement.

Author Response

Comparative Performance of ARIMA and Stepwise Regression for Forest Fire Prediction in Yunnan Province

Comments to Authors:

Abstract

1. The title of the study tells us about the comparison of the dual tools used in the study, meanwhile, the abstract tells us about the different story. See Lines 14-18.

We have revised the abstract to better align with the study title, ensuring consistency in communicating our comparison of ARIMA and Stepwise Regression tools for forest fire prediction.

2. Lines 12-13: What is the research problem? Occurrences of Forest Fires or Lack of Prediction Tools? In my opinion, there are a bunch of modern applications available that can easily predict future scenarios. If the problem is forest fire occurrences, then certainly there are natural and manmade phenomena that speed the fire occurrence process.

Modern applications do exist, but our study aims that by comparing these two models, we can fully understand the dynamics of forest fires from a time series perspective as well as from a multivariate analysis perspective.This comparison is not just a technical comparison, but also an innovative attempt to explore the effectiveness and limitations of the different models in dealing with the same problem, which can promote model optimisation, and by comparing these two models, we can By comparing these two models, a more comprehensive and diversified perspective can be provided for the prediction of forest fires in Yunnan Province.

3. Why did authors compare or experience ARIMA and Stepwise Regression?

ARIMA and Stepwise Regression were compared to evaluate their predictive performances and applicability to forest fire data, capitalizing on their unique statistical strengths.

4. The authors did not talk about the data collection and feeding process to generate results in the abstract. Correction is required.

We have now included a brief on data collection and processing in the abstract to clarify how the results were generated.

5. What is the significance of the study?

The significance of this study lies in the comparison of prediction accuracy and the analysis of meteorological factors that have a strong influence on the number of fires and fire area through stepwise regression. It aims to improve the accuracy and reliability of forest fire prediction.

Introduction

1. This section lacks important facts and figures about forest fire occurrences in the world and hence in the study area. Show that your study is need-based. This is a massive correction to upgrade the quality of the introduction section.

We will incorporate current data on global and regional forest fire occurrences to demonstrate the necessity and timeliness of our study.

2. Clarify that selected methods are the best option to predict the forest fire. Case studies should be cited to link with the upcoming section of methodology. The link is missing.

The superiority of the selected methods will be justified with relevant case studies, establishing a clear connection to the methodology section.

3. The citation numbers 11-15 are outdated. I wonder why the authors included these to clarify fire risk prediction models. The situation is changed as described in these articles from 11-15. The authors are advised to search for the latest citations to validate their arguments.

We will update the citations 11-15 with the latest literature to ensure our discussion on fire risk prediction models is current.

4. I also urge authors to relook at their citation portion as given in different sections of the manuscript. Use fresh citations. These should not be older than a decade.

The citation portfolio will be refreshed with recent literature, adhering to the guideline of using citations from within the last decade.

Study area and research methodology

1. Not a single citation was found between lines 76—83. The authors have referred to important parameters in these lines, however, the authors have referenced certain literature. This gives a bad impression.

We will add citations between lines 76-83 to substantiate references to the key parameters temperature, mean annual rainfall, wind speed and relative humidity.

2. Published data were taken from 1991-2021 taken from the different agencies. Is it good to analyze published data? As a statistician, published data should not be further analyzed. If so, researchers should take expert opinion to validate the choice of secondary data usage. Therefore, validation and authentication are required.

Our rationale for using published data was its relevance and accessibility, which ensured the robustness of the study, and also the consistency of the data for the purpose of carrying out our modelling studies, and we took into account expert opinion when selecting secondary data.

3. Subsection 3.2. No citation is given for the formulas (1) to (4). Why do you discuss these here? What was the purpose and limitations?

Citations for formulas (1) to (4) will be provided, clarifying their relevance and limitations within our methodology.

4. See lines 98-103. How was the data obtained from the government agencies? In what year, data were collected?

Details regarding the acquisition of government agency data, including the year of collection, will be clarified.

5. For missing data, why long-term averages were used? This may generate outliers and vanish the aim of the analytical process.

Long-term averages, calculated on the basis of many years of observational data, are a better reflection of the typical climatic conditions in the region and are considered to be a reliable proxy for climatic characteristics. While long-term averages are valid alternatives, they may not fully reflect recent trends in climate change.

6 Line 160-162. This is a biased statement. The evidence is also not seen by referring to Figure 3. 1998-2000 the massive outlier is visible. Clarify discussing Figure 3.

We will re-examine the statement in lines 160-162 and clarify the discussion around Figure 3, addressing the outlier in 1998-2000.

7. What is hm2? Looks like a unit of area. Clarify. Also, define all such units to make the manuscript readable.

The unit "hm2" refers to a hectare (10,000 square meters), and we will define all units clearly for readability.

Temporal Characteristics of Forest Fires

1. This section is overstretched and looks hazy. One cannot figure out why these steps were performed by the authors. Are these performed in the quest to achieve study objectives? It is not clear in this section.

This section will be streamlined to clearly articulate the steps performed in relation to achieving the study objectives.

2. Multiple tests and steps were performed in this section without justification. Their link with the reviewed literature and methodology section is also missing.

We will add the rationale for performing multiple tests and steps, linking them to the literature and methods reviewed

3. Draw a flow chart to summarize the required steps to achieve objectives that need to be utilized in this section.

A flow chart summarizing the steps toward achieving the study objectives will be included for clarity.

4. The authors need to correct this section. Do not go out of scope. Compare this section with others to establish a link.

This section will be corrected to maintain scope and establish coherence with other sections.

5. Many of the outliers were observed in the Figures.

The presence of outliers in figures will be addressed and discussed.

Conclusion and Recommendations

1. This section merged two important sections. It is advised to split. Recommendations are made in the light of key findings of the study. Whereas, the conclusion section sums up the study in a professional way.

The sections on conclusions and recommendations will be separated for clarity, with recommendations stemming from key findings and conclusions summarizing the study professionally.

2. Highlight the limitations of the study. As some of the data inputs are found missing.

Justifications for the multiple tests and steps performed will be added, linking them to the reviewed literature and methodology.

The limitation of this study is that it is limited to the analysis of historical data and does not take into account the impact of possible future climate change on fire patterns. Also, the predictive accuracy of the model is limited by the quality and availability of the data.

3. Significance and research contribution should be added as separate subheadings to clarify the importance of the study.

Significance and research contributions will be articulated under separate subheadings to emphasize the study's importance.

4. Why concerned agencies, reviewers, or students review this study? Justify.

This study has important applications for relevant government agencies, researchers and students. It provides a scientific basis for decision makers to help them make more informed choices in resource allocation and risk management. In addition, the methodology and findings of this study can be used as teaching cases in courses related to environmental science and forest management.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

In this second version of the article, the authors have improved the content and the way of presenting the results clearly and showing the effectiveness of the proposed method.

 

Author Response

Additions are shown in red, modifications in blue

Reviewer: 1

1. The summary must be reviewed and written again in order to show the relevance of the problem, methodological aspects, and results. For example, you should not include equations in the summary, as the variables are unknown. It is difficult for the reviewer to understand the summary.

Thank you for your feedback. The summary will be revised to enhance clarity on the relevance of the research problem, methodological aspects, and results. We will ensure that equations are omitted and that the summary is comprehensible without prior knowledge of the variables.

2. The introduction needs to clarify why to use the ARIMA and Stepwise regression methods for comparison. Why were they chosen?

In the introduction, we will clarify our rationale for choosing ARIMA and Stepwise regression methods. Specifically, we will discuss their respective strengths in modeling and predicting forest fires, and the theoretical and empirical considerations that led to their selection.

3. In general, both in the summary and in the introduction, the problem could be more straightforward.

We will streamline the narrative in both the summary and the introduction to directly address the research problem, making the purpose and objectives of the study more straightforward.

4. In lines 106 to 113, the data set will be used to build the models presented. However, the article did not cover a section on selecting variables associated with forest fire prediction. This part is essential in model building.

We acknowledge the omission and will include a section dedicated to the variable selection process for forest fire prediction, detailing the criteria and methodology used in determining the variables for our models.

5. “Section 3.1 Stepwise Regression Model” was significantly reduced, with insufficient justification to associate with the results.

The section "3.1 Stepwise Regression Model" will be expanded to provide a more comprehensive justification for its use and to better associate the methodology with the results presented in the study.

6. The results are quite consistent, showing fit metrics and comparisons. I just missed an assessment of the uncertainties associated with these estimates. Did the authors think about the possibility of developing uncertainty analysis associated with forecast models?

We agree with you about the lack of uncertainty assessment. We have used regression residuals and Q-Q plots to analyse the uncertainty associated with the predictive models to get a clearer picture of the confidence in our estimates.

 

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

applsci-2728421

Comparative Performance of ARIMA and Stepwise Regression for Forest Fire Prediction in Yunnan Province

Comments to Authors:

 

1.      The authors have not mentioned what exact modifications they have performed. No line numbers are mentioned in the report to further confirm the changes.

2.      Most of the answers of the authors are generalized. For example, the first comment was “The title of the study tells us about the comparison of the dual tools used in the study, meanwhile, the abstract tells us about the different story. See Lines 14-18.” The authors' response is “We have revised the abstract to better align with the study title, ensuring consistency in communicating our comparison of ARIMA and Stepwise Regression tools for forest fire prediction.” My query is “What changes were actually made?” What was the line number?  Should I review the whole abstract or manuscript again? Please rectify.

3.      A very long sentence is found in the Abstract. See Lines 15-18.

4.      Why negative correlation was found? What does it mean? Show negative correlation values.

5.      Lines 108-132. The citations of the regression model are not given. For this purpose, I recommend authors cite and review this meaningful article:

 “DOI: https://doi.org/10.1007/s10668-022-02288-5.

 

Summary of the Comments:

 I just stopped reevaluating the manuscript. It looks like the authors only developed a response report and did not make certain changes to the manuscript.

My recommendations are as follows:

1.      Revisit the comments given in Rounds 1 and 2.

2.      Make certain changes to the manuscript and in the response report mention the exact line number for further verifications.

3.      For Figures or Tables Modifications, show the changes and refer to the authors' response report.

4.      Some of the comments/queries of Report 1 are still unanswered.

 

--BEST OF LUCK--

Comments on the Quality of English Language

English need proofreading.

Author Response

Additions are shown in red, modifications in blue

Reviewer: 3（Round 2）

1.The authors have not mentioned what exact modifications they have performed. No line numbers are mentioned in the report to further confirm the changes.

We thank the reviewers for their suggestions, and we have proposed specific changes as well as mentioning line numbers in our next response.

2.Most of the answers of the authors are generalized. For example, the first comment was “The title of the study tells us about the comparison of the dual tools used in the study, meanwhile, the abstract tells us about the different story. See Lines 14-18.” The authors' response is “We have revised the abstract to better align with the study title, ensuring consistency in communicating our comparison of ARIMA and Stepwise Regression tools for forest fire prediction.” My query is “What changes were actually made?” What was the line number?  Should I review the whole abstract or manuscript again? Please rectify.

Thanks to the suggestions given by the reviewers, we have revisited our response and we will respond in detail to the first round of reviewers' suggestions, including exactly what changes were made and with what line numbers.

1. Title Modification: We revised the title to "Forest Fire Prediction in Yunnan Province Based on a Combined ARIMA and Stepwise Regression Analysis Model," to more accurately reflect the research focus on predicting forest fires using a combination of ARIMA and stepwise regression analysis models. See line 2.
2. Global Forest Fire Data Addition: To support our argument, we added data on global forest fires. See lines 35 to 39.
3. Literature Review Addition: We added a literature review section in the article, thoroughly examining existing forest fire prediction models, especially the application and advantages of ARIMA models and stepwise regression analysis in this field. See lines 58 to 101.
4. Research Framework Diagram Addition: To clearly display the research steps and methods, we added a corresponding research framework diagram. See lines 103 to 105.
5. Model Evaluation and Validation: We added a section for model evaluation and validation, including R² values, T-test of the model, and uncertainty analysis. See lines 179 to 214.
6. Study Area Graphics Modification: We updated the elevation map of the study area to provide a more detailed introduction. See lines 242 to 243.
7. Analysis of Meteorological Factors on Forest Fires: We detailed the impact of temperature, annual average humidity, and wind speed on forest fires. See lines 256 to 330.
8. Major Modifications to the Stepwise Regression Model: We significantly modified the stepwise regression model to enhance its predictive accuracy and applicability. See lines 331 to 386.
9. Uncertainty Analysis Addition: We conducted an uncertainty analysis of the stepwise regression model to verify its reliability and accuracy. See lines 382 to 388.
10. Prediction Time Range Modification: We changed the prediction time range to 2024-2033 to better demonstrate the model's predictive accuracy. See lines 456 to 464.
11. Separation of Conclusions and Suggestions: We divided the conclusions and suggestions at the end of the article to more clearly present the research results and application recommendations. See lines 466 to 497.
12. A very long sentence is found in the Abstract. See Lines 15-18.

We have now divided this long sentence into two shorter and clearer sentences, see lines 15-18.

4.Why negative correlation was found? What does it mean? Show negative correlation values.

We appreciate the suggestions from the reviewer. We found that the negative correlation is mainly because as the rainfall and humidity increase, the number of fires and the affected area decrease. This indicates that rainfall and humidity affect the number of fires and the area impacted, and these meteorological factors should be incorporated into the stepwise regression model. In the fire frequency model, the negative correlation values for annual rainfall and average wind speed are -0.481 and -190.484 respectively, while in the model for the affected area, the negative correlation value for annual average rainfall is -7.493. Refer to lines 365 to 371.

5.Lines 108-132. The citations of the regression model are not given. For this purpose, I recommend authors cite and review this meaningful article:

“DOI: https://doi.org/10.1007/s10668-022-02288-5.

Thanks to the reviewers' suggestions, we have revised lines 115 through 120 of the paper to include the literature citations recommended by the reviewers. See lines 115-120

Reviewer: 3（Round 1）

Comparative Performance of ARIMA and Stepwise Regression for Forest Fire Prediction in Yunnan Province

Comments to Authors:

Abstract

1. The title of the study tells us about the comparison of the dual tools used in the study, meanwhile, the abstract tells us about the different story. See Lines 14-18.

We thank the reviewers for their suggestions, and in order to address the mismatch between the abstract and the title, we have revised the title to ensure that the information between the abstract and the title is consistent, thus better communicating to the reader the core content and purpose of this study. See lines 2 through 24.

2. Lines 12-13: What is the research problem? Occurrences of Forest Fires or Lack of Prediction Tools? In my opinion, there are a bunch of modern applications available that can easily predict future scenarios. If the problem is forest fire occurrences, then certainly there are natural and manmade phenomena that speed the fire occurrence process.

Thanks to the reviewers for the suggestion that the main problem of our study is the frequency of forest fires. See lines 12-13. In the introduction we suggest some factors such as natural and anthropogenic phenomena that accelerate the occurrence of fires. See lines 20-22.

3. Why did authors compare or experience ARIMA and Stepwise Regression?

Thanks to the reviewers for this suggestion: in order to better grasp the pulse of the article, we now have the title of the article as a combination of the two models. By utilizing these two models, this study aims to provide a comprehensive understanding of the dynamics of forest fires from the perspective of time series and multivariate analysis. This approach represents an innovative attempt to explore the efficacy and limitations of different models in addressing the same problem. This facilitates model optimization and provides a more comprehensive and multifaceted perspective on forest fire prediction. See lines 85 to 101.

4. The authors did not talk about the data collection and feeding process to generate results in the abstract. Correction is required.

Thanks to the reviewers' suggestions, we have reworked this part of the abstract, see lines 15-22.

5. What is the significance of the study?

Both models allow us to fully understand the dynamics of forest fires from a time-series perspective as well as from a multivariate analysis. See lines 97-107.

The significance of the study is also to provide a scientific basis for future fire prevention and response strategies, and for decision makers to help them make more informed choices in resource allocation and risk management, see lines 465-473

Introduction

1. This section lacks important facts and figures about forest fire occurrences in the world and hence in the study area. Show that your study is need-based. This is a massive correction to upgrade the quality of the introduction section.

Thank you to the reviewers for their suggestions, and in response to the reviewers' concerns about the lack of important facts and data on the occurrence of forest fires globally and in the study region, in the introductory section of the article we have provided some relevant data. See lines 35-40

2. Clarify that selected methods are the best option to predict the forest fire. Case studies should be cited to link with the upcoming section of methodology. The link is missing.

To address this issue, we detail the benefits of ARIMA and stepwise regression models, see lines 86-96. Case studies supporting these methods: several studies are cited that have successfully used these models for fire prediction. See lines 61-84.

3. The citation numbers 11-15 are outdated. I wonder why the authors included these to clarify fire risk prediction models. The situation is changed as described in these articles from 11-15. The authors are advised to search for the latest citations to validate their arguments.

We thank the reviewers for their suggestions. We are aware of discrepancies between citation numbers 11-15 in the original manuscript and the current, significantly revised manuscript, which may affect the accuracy and relevance of the study. Therefore, to ensure the accuracy and currency of the study, we have re-run the literature search and added new citations (citation numbers [10-16], see lines 47-50)。

4. I also urge authors to relook at their citation portion as given in different sections of the manuscript. Use fresh citations. These should not be older than a decade.

We revisited the citation section of each part of the manuscript based on the reviewers' suggestions: the citations were updated, selected, and integrated. See lines 508 through 676.

Study area and research methodology

1. Not a single citation was found between lines 76—83. The authors have referred to important parameters in these lines, however, the authors have referenced certain literature. This gives a bad impression.

In lines 76-83 we refer to the reviewer's suggestion to cite references to the basic status of Xiuwu County in these lines, see lines 225-232.

38. Deng, X.; Zhang, Z.; Zhao, F.; Zhu, Z.; Wang, Q. Evaluation of the Regional Climate Model for the Forest Area of Yunnan in China. For. Glob. Change 2023, 5, 1073554.

https://doi.org/10.3389/ffgc.2022.1073554.

2. Published data were taken from 1991-2021 taken from the different agencies. Is it good to analyze published data? As a statistician, published data should not be further analyzed. If so, researchers should take expert opinion to validate the choice of secondary data usage. Therefore, validation and authentication are required.

Thanks to the reviewers for their suggestions, the data on the number and area of forest fires in Yunnan Province were obtained from credible sources such as the Yunnan Disaster Reduction Yearbook and the official website of the Yunnan Forestry Bureau. Relevant meteorological data were obtained from the National Meteorological Science Data Center to ensure the reliability of the analyzed information. A paired t-test was also conducted to compare the actual and predicted values, and the results showed no significant difference, thus confirming the accuracy of the model. A residual analysis was also conducted to demonstrate that the distribution of the residuals is generally consistent with the theoretical normal distribution, which supports the validity of the model, see lines 349-385.

3. Subsection 3.2. No citation is given for the formulas (1) to (4). Why do you discuss these here? What was the purpose and limitations?

We add to the underlying formulas what were formulas (1) to (4) in the original manuscript, and formulas (2) to (5) in the current manuscript. The purpose of citing the formulas is to provide the reader with an understanding of the theoretical framework underlying our research methodology, which is used to describe the core concepts of our proposed model or methodology. The main limitation of the model is the requirement for smoothness of the data. See lines 141 to 156

4. See lines 98-103. How was the data obtained from the government agencies? In what year, data were collected?

We obtained data on fires in Yunnan Province, accessed as of November 2023, from the official website of the Yunnan Forestry Bureau. See rows 245-247.

5. For missing data, why long-term averages were used? This may generate outliers and vanish the aim of the analytical process.

Thanks to the reviewers for the suggestion that the use of long-term averages to replace missing meteorological data is standard practice in time series analysis, especially when missing data cannot be otherwise estimated or retrieved. This approach is generally accepted in the fields of environmental science and modeling. The effect of using long-term averages on the results in our study was assessed and found to be small, and an uncertainty analysis of the model was performed to demonstrate the validity of the model, see lines 373-378.

6 Line 160-162. This is a biased statement. The evidence is also not seen by referring to Figure 3. 1998-2000 the massive outlier is visible. Clarify discussing Figure 3.

Thanks to the reviewer's suggestion, we have discussed Figure 4 (Figure 3 in the original manuscript) in detail . See lines 259 through 278. 1998-2000 experienced extreme drought conditions that greatly increased the potential for forest fires. Drought dries out forest vegetation, making it more flammable and susceptible to fire. Such extreme weather conditions could have resulted in atypical spikes in fire incidence that would show up as outliers in the data. See lines 262-264.

7. What is hm2? Looks like a unit of area. Clarify. Also, define all such units to make the manuscript readable.

Thanks to the reviewer's suggestion that the unit "hm2" refers to hectares (10,000 square meters), we will clearly define all units for ease of reading. See lines 35-39.

Temporal Characteristics of Forest Fires

1. This section is overstretched and looks hazy. One cannot figure out why these steps were performed by the authors. Are these performed in the quest to achieve study objectives? It is not clear in this section.

We responded to this suggestion of the reviewers and we extensively revised the manuscript.

Correlation analysis of meteorological factors (Section 4.2.1), see lines 331 to 348

Model fitting and prediction (Section 4.2.2), see lines 349 to 363

Analysis of the results of the model (Section 4.2.2), see lines 363 to 372

Accuracy assessment of the model (Section 4.2.2), see lines 373-379

Perform an uncertainty analysis of the model (Section 4.2.3), see lines 380 to 386.

2. Multiple tests and steps were performed in this section without justification. Their link with the reviewed literature and methodology section is also missing.

In response to this suggestion from the reviewers, we have revised this section extensively and detailed the rationale and methodology for undertaking this step in section 4.2, see lines 380-386.

3. Draw a flow chart to summarize the required steps to achieve objectives that need to be utilized in this section.

We mapped the flow as suggested by the reviewers, see lines 103 through 104.

4. The authors need to correct this section. Do not go out of scope. Compare this section with others to establish a link.

Thanking the reviewers for this suggestion, we have recorrected this section and created links to other sections in an effort to be more logical, see lines 257 through 326.

5. Many of the outliers were observed in the Figures.

Thanks to the reviewer's suggestion, the outliers are mainly due to disasters of natural conditions and the impact of government policies, see lines 262 to 264

Conclusion and Recommendations

1. This section merged two important sections. It is advised to split. Recommendations are made in the light of key findings of the study. Whereas, the conclusion section sums up the study in a professional way.

At the suggestion of the reviewers, we split the conclusions and recommendations. See lines 467-497.

2. Highlight the limitations of the study. As some of the data inputs are found missing.

Justifications for the multiple tests and steps performed will be added, linking them to the reviewed literature and methodology.

Thanks to the suggestions made by the reviewers, we have emphasized the limitations of the study in the conclusion section of the paper, see lines 481 to 483.

In response to some missing data, we did a t-test of the study, see lines 373 to 379

Uncertainty analysis see lines 380 to 386

Residual tests see lines 427 to 431

3. Significance and research contribution should be added as separate subheadings to clarify the importance of the study.

We appreciate the reviewer's suggestion to emphasize the significance of the study and research contributions under separate subheadings. However, we believe that integrating these discussions into the conclusion section could simplify the manuscript and directly link the findings to their implications, which may be possible by providing a coherent narrative. The conclusion section of our manuscript succinctly synthesizes our main findings.

4. Why concerned agencies, reviewers, or students review this study? Justify.

We thank the reviewers for their suggestions. This study is of great significance to related organizations, reviewers, and students by developing a prediction model for forest fires in Yunnan Province. ARIMA and stepwise regression analysis models were used to comprehensively analyze the meteorological factors affecting forest fires, and innovative prediction methods were proposed. It helps the government and related organizations to develop more effective fire prevention strategies and countermeasures to reduce the incidence and impact of forest fires, and it provides a valuable educational resource that enhances the understanding of forest fire prediction and its importance among students in the field.

Author Response File: Author Response.pdf

Round 3

Reviewer 3 Report

Comments and Suggestions for Authors

Predictive Modeling of Forest Fires in Yunnan Province: An Integration of ARIMA and Stepwise Regression Analysis

 

Comments to Authors

I must congratulate the authors for their compliance and corrections made to upgrade the quality of their manuscript.

Some of the minor corrections are suggested as follows:

1.      The last paragraph of the introduction should be added to the remainder of the manuscript with a contribution and novelty approach.

2.      What is the purpose of Figures 1 and 2? What are their benefits and key features? These must be explained as text. In this way, the authors are advised to explain all Tables and Figures of the manuscript as text.

3.      It is not advisable to commence a new subheading right after Tables or Figure Captions. Correct your manuscript.

4.      Include Summaries of each and every section of the manuscript in the last paragraph. For Example, the authors can write about the research gap in the last paragraph of the review section. Correction is required.

5.      The numbering of the Figures is also not correct. See Figures 2 and 4. Lines 243 and 276. Correct all such minor errors for Figures and Tables, headings, and subheadings.

6.      See lines 347-348. Note should come after the Figure Caption.

7.      Recommendations should be added right after the key findings of the research. Therefore, put key findings of the study as per the study scope, objectives, and results. After that, the authors are suggested to include recommendations as separate subheadings.

8.      After recommendations, the authors should include Research contribution and research impact as separate subheadings.

9.      The conclusion should be the distinct and last section of the manuscript. Clarify that objectives are achieved and what new have been found. This section should sum up the whole study and conclude all the sections of the manuscript.

10.  Proofread your manuscript well and submit a similarity report to the Handling Editors.

 

 

--- GOOD LUCK ---

Comments on the Quality of English Language

Minor corrections are required.

Author Response

Comments to Authors

I must congratulate the authors for their compliance and corrections made to upgrade the quality of their manuscript.

Some of the minor corrections are suggested as follows:

1. The last paragraph of the introduction should be added to the remainder of the manuscript with a contribution and novelty approach.

Thanks to the reviewers' suggestions, we have added contributions and novel approaches to the rest of the last paragraph of the introduction. See lines 65 through 73; 113 through 119. lines 515 through 525.

2. What is the purpose of Figures 1 and 2? What are their benefits and key features? These must be explained as text. In this way, the authors are advised to explain all Tables and Figures of the manuscript as text.

Thanks to the reviewer's suggestion, Figure 1 describes the influence of anthropogenic and natural factors on fire, clarifying their roles in fire events. Figure 2 summarizes the framework of our study and details the steps of the study. The strength of Figure 1 is that it clearly depicts the influence of natural and human factors on fires and helps to understand the causes. Meanwhile, Figure 2 clarifies the hierarchical structure and flow of our study. We have explained all the tables and graphs in the manuscript thoroughly in the main text. The description of Figure 1 is found in lines 45 through 61; the description of Figure 2 is found in lines 123 through 134.

3. It is not advisable to commence a new subheading right after Tables or Figure Captions. Correct your manuscript.

Thank the reviewers for their suggestions, and we have recorrected the manuscript accordingly, see lines 62 to 74, 121 to 135, 188 to 195, 257 to 268; 298 to 308; 372 to 380; and 491 to 501.

4. Include Summaries of each and every section of the manuscript in the last paragraph. For Example, the authors can write about the research gap in the last paragraph of the review section. Correction is required.

Thanks to the reviewers' suggestions, we have summarized each section of the manuscript in the last paragraph, see lines 65 to 73, 113 to 119. 253 to 259; 293 to 298.

5. The numbering of the Figures is also not correct. See Figures 2 and 4. Lines 243 and 276. Correct all such minor errors for Figures and Tables, headings, and subheadings.

Thanks to the reviewer's suggestions, we have renumbered the figures to correct them and checked for errors in the figures and tables, titles and subheadings. See lines 271 to 272.

6. See lines 347-348. Note should come after the Figure Caption.

Thanks to the reviewer's suggestion, we have placed the notes after the figure title, see lines 385 through 386.

7. Recommendations should be added right after the key findings of the research. Therefore, put key findings of the study as per the study scope, objectives, and results. After that, the authors are suggested to include recommendations as separate subheadings.

We thank the reviewers for their suggestions, which we have included as separate subheadings after the study results, see lines 514 through 527.

8. After recommendations, the authors should include Research contribution and research impact as separate subheadings.

We thank the reviewers for their suggestions, which we followed with separate subheadings for research contribution and research impact. See lines 528 through 549.

9. The conclusion should be the distinct and last section of the manuscript. Clarify that objectives are achieved and what new have been found. This section should sum up the whole study and conclude all the sections of the manuscript.

Thanks to the reviewers' suggestions, we have revised the conclusion section, see lines 551 through 566.

10. Proofread your manuscript well and submit a similarity report to the Handling Editors.

We thank the reviewers for their suggestions and will carefully check the manuscript and send a similarity report to the handling editor.

Author Response File: Author Response.pdf