Identification of Vegetation Areas Affected by Wildfires Using RGB Images Obtained by UAV: A Case Study in the Brazilian Cerrado

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

1. Introduction is way too short. I recommend adding a map of the cerrado and its location in South America (this does not replace Figure 1) some screenshots showing the differences in terms of MapBiomas coverage. Not everybody knows what the cerrado is, so it would be very useful to broaden it a bit more.

2. You mention the UAV has a RGB sensor. Nevertheless, you do not make any substantial comments throughout the article on the caveat of not having infrared in your sensor. Some of the articles you cite have used vegetation indices that use the NIR band, so it is important to mention, from the beginning, that in your case it was not available and that you tried to work with whatever you had access to.

3. Regarding the methodology, nothing to add. It was clearly stated. Perhaps an image showing it as a workflow would be useful

4. Now, regarding results, the article lacks considerable statistical analyses. The box plots /M-index should be a starting point, but should not be the only comparison. 

5. I recommend using spatial analyses, including - but not limited to - map algebra between pre and post burns, and also between different indices, then identifying which zones are more effective than others.

6. Also, you may use classification tools to spatially identify which areas were more affected and if there's considerable spatial coincidence between those across indices.

Therefore, I believe the article should undergo major revision and a variety of analyses, focusing on the spatial distribution of pre and post-fire areas (and the effectiveness of specific indices and bands spatially).

Author Response

Comments 1: Introduction is way too short. I recommend adding a map of the cerrado and its location in South America (this does not replace Figure 1) some screenshots showing the differences in terms of MapBiomas coverage. Not everybody knows what the cerrado is, so it would be very useful to broaden it a bit more.

Response 1: Thank you very much for your comments. Additional information has been inserted into the text of the introduction. To avoid the map becoming repetitive, the map in Figure 1 was updated to contain location information for South America.

 

Comments 2: You mention the UAV has a RGB sensor. Nevertheless, you do not make any substantial comments throughout the article on the caveat of not having infrared in your sensor. Some of the articles you cite have used vegetation indices that use the NIR band, so it is important to mention, from the beginning, that in your case it was not available and that you tried to work with whatever you had access to.

Response 2: Thank you very much for your comments. Comments on the use of RGB sensors have been added in the introduction, and the sensor data used has been provided in the methodology, item 2.2.

 

Comments 3: Regarding the methodology, nothing to add. It was clearly stated. Perhaps an image showing it as a workflow would be useful

Response 3: A diagram of the work methodology is shown in Figure 3.

 

Comments 4: Now, regarding results, the article lacks considerable statistical analyses. The box plots /M-index should be a starting point, but should not be the only comparison. 

Response 4: Thank you very much for your suggestions. The authors would like to inform you that the box-plots show the descriptive statistics of the bands and indices evaluated in the study. With regard to the M separability index, the literature shows that this index is very solid for evaluating images before and after environmental events, as in this study. Examples of studies include:

Pacheco, A.d.P.; da Silva Junior, J.A.; Ruiz-Armenteros, A.M.; Henriques, R.F.F.; de Oliveira Santos, I. Analysis of Spectral Separability for Detecting Burned Areas Using Landsat-8 OLI/TIRS Images under Different Biomes in Brazil and Portugal. Forests 2023, 14, 663. https://doi.org/10.3390/f14040663

Smiraglia, D.; Filipponi, F.; Mandrone, S.; Tornato, A.; Taramelli, A. Agreement Index for Burned Area Mapping: Integration of Multiple Spectral Indices Using Sentinel-2 Satellite Images. Remote Sens. 2020, 12, 1862. https://doi.org/10.3390/rs12111862

Huang, H.; Roy, D.P.; Boschetti, L.; Zhang, H.K.; Yan, L.; Kumar, S.S.; Gomez-Dans, J.; Li, J. Separability Analysis of Sentinel-2A Multi-Spectral Instrument (MSI) Data for Burned Area Discrimination. Remote Sens. 2016, 8, 873. https://doi.org/10.3390/rs8100873

Tran, B.N.; Tanase, M.A.; Bennett, L.T.; Aponte, C. Evaluation of Spectral Indices for Assessing Fire Severity in Australian Temperate Forests. Remote Sens. 2018, 10, 1680. https://doi.org/10.3390/rs10111680

Llorens, R.; Sobrino, J.A.; Fernández, C.; Fernández-Alonso, J.M.; Vega, J.A. Soil Burn Severity Assessment Using Sentinel-2 and Radiometric Measurements. Fire 2024, 7, 487. https://doi.org/10.3390/fire7120487

 

Comments 5: I recommend using spatial analyses, including - but not limited to - map algebra between pre and post burns, and also between different indices, then identifying which zones are more effective than others.

Comments 6: Also, you may use classification tools to spatially identify which areas were more affected and if there's considerable spatial coincidence between those across indices.

Comments 7: Therefore, I believe the article should undergo major revision and a variety of analyses, focusing on the spatial distribution of pre and post-fire areas (and the effectiveness of specific indices and bands spatially).

Response comments 5 to 7: Thank you very much for your comments. The authors would like to thank you for your suggestion. Please note that the work presented here aims to identify areas affected by fires in digital images, and not to classify areas spatially. The M separability index was used to demonstrate that it is possible to carry out this identification in RGB image bands and vegetation indices. In this regard, we would like to point out that no modeling has been carried out to classify images of areas affected by fires, in order to assess the accuracy of this classification. The results obtained here will be used as a basis for future studies aimed at classifying images.

Reviewer 2 Report

Comments and Suggestions for Authors

Overall, the document is well-structured and presents advanced research on wildfire occurrence, providing essential information for proper planning, management, and post-fire rehabilitation, as well as for the continuous monitoring of natural restoration in affected areas.

 

The main advantage of this study is the use of indices obtained with simple RGB devices, avoiding the need for expensive sensors that measure other parts of the spectral region. The effectiveness of these indices is demonstrated in the results and discussed using the M Separability Index, where the best discrimination was obtained for band B and the indices GLI and EXG.

 

To improve the "Methods " section, it is explained that Structure from Motion technique was implemented, but it is necessary to specify which software was used. If QGIS was used, it is important to mention which plugin was utilized.

In addition of the use of M Separability Index, it is recommended to use a value that also represents the accuracy of the detected burn areas.

 

There is a language issue in line 125: "Aquisição de imagens UAV".

Comments on the Quality of English Language

The text needs a review.

There is a language issue in line 125: "Aquisição de imagens UAV".

Author Response

Comments 1: Overall, the document is well-structured and presents advanced research on wildfire occurrence, providing essential information for proper planning, management, and post-fire rehabilitation, as well as for the continuous monitoring of natural restoration in affected areas.

Response 1: Thank you very much for your comments.

 

Comments 2: The main advantage of this study is the use of indices obtained with simple RGB devices, avoiding the need for expensive sensors that measure other parts of the spectral region. The effectiveness of these indices is demonstrated in the results and discussed using the M Separability Index, where the best discrimination was obtained for band B and the indices GLI and EXG.

Response 2: Thank you very much for your comments.

 

Comments 3: To improve the "Methods " section, it is explained that Structure from Motion technique was implemented, but it is necessary to specify which software was used. If QGIS was used, it is important to mention which plugin was utilized.

Response 3: Thank you very much for your comments. The requested information has been inserted into the text.

 

Comments 4: In addition of the use of M Separability Index, it is recommended to use a value that also represents the accuracy of the detected burn areas.

Response 4: Thank you very much for your comments. The authors would like to thank you for your suggestion. Please note that the work presented here aims to identify areas affected by fires in digital images. The M separability index was used to demonstrate that it is possible to carry out this identification in RGB image bands and vegetation indices. In this regard, we would like to point out that no modeling has been carried out to classify images of areas affected by fires, in order to assess the accuracy of this classification. The results obtained here will be used as a basis for future studies aimed at classifying images.

 

Comments 5: There is a language issue in line 125: "Aquisição de imagens UAV".

Response 5: The requested correction has been made to the text.

Reviewer 3 Report

Comments and Suggestions for Authors

The article presents an important contribution to the field of studies on wildfires in savannah areas. The work has an excellent structure, with high-quality figures and current references that adequately support the discussion. To improve the work, I recommend improving the conclusion, which is currently limited to a summary of the results. I suggest that the conclusion highlight the main contributions of the study, explaining how its publication could benefit the field of research, in addition to highlighting the specific contributions to studies in savannahs.

Author Response

Comments 1: The article presents an important contribution to the field of studies on wildfires in savannah areas. The work has an excellent structure, with high-quality figures and current references that adequately support the discussion. To improve the work, I recommend improving the conclusion, which is currently limited to a summary of the results. I suggest that the conclusion highlight the main contributions of the study, explaining how its publication could benefit the field of research, in addition to highlighting the specific contributions to studies in savannahs.

Response 1: Thank you very much for your comments. The requested information has been inserted into the text.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Based on the author's argument that the main point is to prove the M Separability Index is an adequate tool to identify vegetation areas affected by wildfires, my suggestion is that:

1) This is more clearly emphasized throughout the article (especially in its introduction)

2) Discussions and conclusions are more thoroughly detailed (the article is 12 pages long, so there's enough room for that), with citations that support what the authors are proving.

After that, I believe this would be fit for publication.

Author Response

Comments 1: Based on the author's argument that the main point is to prove the M Separability Index is an adequate tool to identify vegetation areas affected by wildfires, my suggestion is that:

Response 1: Dear all, Thank you very much for your comments on the manuscript. The authors inform us that the M index of separability is used in the paper as a tool to evaluate the use of images obtained with UAVs in Cerrado areas affected by wildfires. Therefore, we would like to inform you that the main objective of the work is to “Therefore, this study assesses the feasibility of employing UAVs equipped with RGB sensors to generate high-resolution digital products for identifying and monitoring wildfire-affected areas in the Brazilian Cerrado”.

 

Comments 2: This is more clearly emphasized throughout the article (especially in its introduction)

Response 2: Thank you very much for your comments on the manuscript. The requested information has been inserted into the text.

 

Comments 3: Discussions and conclusions are more thoroughly detailed (the article is 12 pages long, so there's enough room for that), with citations that support what the authors are proving.

Response 3: Thank you very much for your comments on the manuscript. The requested information has been inserted into the text.