Assessment of Burned Areas during the Pantanal Fire Crisis in 2020 Using Sentinel-2 Images

Round 1

Reviewer 1 Report (Previous Reviewer 1)

This version of the manuscript presents a coherent account of the study and presents a valuable contribution to scholarship in the realms of wetland fire science.

Following are  a few minor comments/suggestions

I) Figure 4 is redundant as the same image appears in Figure 5. Suggest removing Figure 4 or moving to supplementary material.

II) Review the results section to include the material from Figure S2. This is an important aspect of this study (referred to in the discussion) which is not only comparing a Sentinel BA product with Modis and Landsat-derived products but you are also evaluating variables for the Sentinel-derived BA product. There is intrinsic value in this which needs to be captured in the results and discussion sections.

III) The meaning of the first sentence of the discussion is not clear. What is meant by vegetation seasonality? What is meant by wetlands-atmosphere feedbacks? Do you mean wetland-climate feedback? Should this then include fire (i.e. wetland-climate-fire feedback)?  

n/a

Author Response

Reviewer 1

This version of the manuscript presents a coherent account of the study and presents a valuable contribution to scholarship in the realms of wetland fire science.

Author's reply: Thank you very much for the positive comment.

 

Following are a few minor comments/suggestions

5. I) Figure 4 is redundant as the same image appears in Figure 5. Suggest removing Figure 4 or moving to supplementary material.

Author's reply: Thanks very much for your suggestion. Figure 4 was moved to the supplementary material as suggested (please see Figure S2).

 

2. II) Review the results section to include the material from Figure S2. This is an important aspect of this study (referred to in the discussion) which is not only comparing a Sentinel BA product with Modis and Landsat-derived products but you are also evaluating variables for the Sentinel-derived BA product. There is intrinsic value in this which needs to be captured in the results and discussion sections.

Author's reply:  Thanks very much for your suggestion. Figure S2 (now Figure 5) was moved to the results (section “3.3. Burned Area Estimates Validation and Training Variables Importance Assessment”) as suggested, and we have also improved the discussion on the variables’ importance. Please see Lines 280-287 of the revised manuscript.

 

III) The meaning of the first sentence of the discussion is not clear. What is meant by vegetation seasonality? What is meant by wetlands-atmosphere feedbacks? Do you mean wetland-climate feedback? Should this then include fire (i.e. wetland-climate-fire feedback)? 

Author's reply: Thank you very much for your suggestion. We have changed this sentence according to your precise comment. Please see Lines 364-366 of the revised manuscript.

 

Author Response File: Author Response.pdf

Reviewer 2 Report (New Reviewer)

In this study, the authors develop a Google Earth Engine-based approach to map burned area (BA) using Sentinel-2 imagery for the 2020 fires in the Pantanal wetland in Brazil. The Sentinel-2-based BA estimates are compared to estimates from existing MODIS and Landsat-based products. Overall, the approach appears sound and the analysis is well-presented.

However, I do think that one issue should be addressed in greater depth: the fact that the Sentinel-2 method had fairly substantial commission error (~10%) and therefore would tend to overestimate BA. The authors acknowledge this, but this commission error combined with a very low omission error would tend to greatly inflate BA estimates in cases where the minority of the study area burned – for example, in a hypothetical case where 10% of a study area actually burned, this method would estimate that almost 20% burned. Because the primary purpose of mapping BA appears to be estimating total BA in this case, it seems to me that tweaking the methodology to reduce commission error would be greatly desirable, e.g., adjusting the sample sizes for burn and non-burn training samples and/or model parameters for random forest. Otherwise, the authors risk simply trading underestimates of BA (MODIS and Landsat products) for an overestimate.

In the introduction, other recent papers that have used GEE to estimate BA from Sentinel-2 imagery should be acknowledged and the approach used in this manuscript should be distinguished from these prior approaches.

Finally, a table comparing accuracy from each BA classification should be included in the main text so that readers do not have to refer the supplementary material for this important information.

 

Comments by line:

56-57: Confusingly worded. Was the total area covered by water 34% less than average?

58: The first time on record.

59: And therefore greater potential for future fires?

131: This choice to adjust code developed to classify cropland strikes me as unusual. Please justify starting with this code as opposed to code developed for BA mapping specifically.

145: Hudak et al. were focused on burn severity, not burned area mapping.

225-233: This interpretative paragraph should be moved to the discussion.

237-238: Rephrase this sentence to clarify that this is an estimate of BA. Ecosystem effects can be mentioned in the discussion, but should not be mentioned in the results because the analysis does not deal with fire effects.

252: “More detailed”: Meaning that it was able to detect smaller fires?

274-275: This study documents the extent of the fires, not their ecological and hydrological impacts.

277-279: You could be more specific here.

280-284: This is an important point and should be included in the introduction as well, to distinguish this work from other recent studies that have used Sentinel imagery to map BA.

290: Please clarify what the first two percentages represent.

294: Rephrase this clause for clarity.

298: Remove “than underestimation”?

304-308: This is a good opportunity to discuss prior work on detecting BA in wetlands, e.g., those cited in the introduction.

316-319: This should be moved to the results section.

340-341: Or rather, these sensors are the only options for analyzing years prior to 2015.

Author Response

Reviewer 2

 

In this study, the authors develop a Google Earth Engine-based approach to map burned area (BA) using Sentinel-2 imagery for the 2020 fires in the Pantanal wetland in Brazil. The Sentinel-2-based BA estimates are compared to estimates from existing MODIS and Landsat-based products. Overall, the approach appears sound and the analysis is well-presented.

Author's reply: Thank you very much for the positive comment.

 

However, I do think that one issue should be addressed in greater depth: the fact that the Sentinel-2 method had fairly substantial commission error (~10%) and therefore would tend to overestimate BA. The authors acknowledge this, but this commission error combined with a very low omission error would tend to greatly inflate BA estimates in cases where the minority of the study area burned – for example, in a hypothetical case where 10% of a study area actually burned, this method would estimate that almost 20% burned. Because the primary purpose of mapping BA appears to be estimating total BA in this case, it seems to me that tweaking the methodology to reduce commission error would be greatly desirable, e.g., adjusting the sample sizes for burn and non-burn training samples and/or model parameters for random forest. Otherwise, the authors risk simply trading underestimates of BA (MODIS and Landsat products) for an overestimate.

Author's reply: Thank you for your comment. We agree that it is crucial to minimize both types of errors - commission (overestimation) and omission (underestimation). The BA products that we compared showed higher omission and also higher or similar commission errors, leading to a lower overall accuracy compared to our approach. For example, the lowest commission error (4%) was observed in the GABAM product but, on the other hand, it showed an extremely high omission error (~64%). It's important to note that every method has its own set of strengths and limitations, and the choice of approach depends on the specific objectives and requirements of the study. However, in the context of estimating total BA accurately, our approach stands out as an interesting option due to its emphasis on reducing omission errors without greatly overestimating (~10%) while maintaining accurate estimates of BA. We also highlight that the global BA products analyzed in the manuscript are often linked to an underestimate of burned area. These clarifications were added to Lines 323-334 and Lines 388-392 of the revised manuscript.

 

In the introduction, other recent papers that have used GEE to estimate BA from Sentinel-2 imagery should be acknowledged and the approach used in this manuscript should be distinguished from these prior approaches.

Author's reply: Thank you for your comment. Done. Please see Lines 91-103 of the revised manuscript.

 

Finally, a table comparing accuracy from each BA classification should be included in the main text so that readers do not have to refer to the supplementary material for this important information.

Author's reply: Thanks for your suggestion. We have added Table 4 to the revised manuscript, which summarizes the overall accuracy information of each BA classification.

 

Comments by line:

56-57: Confusingly worded. Was the total area covered by water 34% less than average?

Author's reply: Thanks for your suggestion. We have changed this sentence accordingly. Please see Lines 58-59 of the revised manuscript.

 

58: The first time on record.

Author's reply: Thank you for your suggestion. Done. We added “on record” to this sentence.

 

59: And therefore greater potential for future fires?

Author's reply: Thank you for your suggestion. We added your suggestion to the end of this sentence.

 

131: This choice to adjust code developed to classify cropland strikes me as unusual. Please justify starting with this code as opposed to code developed for BA mapping specifically.

Author's reply: Thanks for the comment. This code was solely used to understand the libraries and functions that could be used to apply the Random Forest on a cloud computing platform. This was clarified in the sentence.

 

145: Hudak et al. were focused on burn severity, not burned area mapping.

Author's reply:  Thank you for your correction. We corrected the sentence accordingly.

 

225-233: This interpretative paragraph should be moved to the discussion.

Author's reply: Thank you for the suggestion. As suggested, we have moved this paragraph to the discussion section.

 

237-238: Rephrase this sentence to clarify that this is an estimate of BA. Ecosystem effects can be mentioned in the discussion, but should not be mentioned in the results because the analysis does not deal with fire effects.

Author's reply:  Thanks for the suggestion. Done. We moved the sentence to the discussion section.

 

252: “More detailed”: Meaning that it was able to detect smaller fires?

Author's reply: Thank you for your suggestion. We modified this sentence accordingly.

 

274-275: This study documents the extent of the fires, not their ecological and hydrological impacts.

Author's reply: Thanks for the suggestion. We have changed this sentence to meet the suggestions of both reviewers.

 

277-279: You could be more specific here.

Author's reply: Thanks for the suggestion. Done. Please see the modified sentence.

 

280-284: This is an important point and should be included in the introduction as well, to distinguish this work from other recent studies that have used Sentinel imagery to map BA.

Authors reply: Thank you for your suggestion. We included this point in the introduction as well.

 

290: Please clarify what the first two percentages represent.

Author's reply: Thank you for your suggestion. We modified the sentence.

 

294: Rephrase this clause for clarity.

Author's reply: Thank you for your suggestion. We rephrased the sentence.

 

298: Remove “than underestimation”?

Author's reply: Thank you for your suggestion. We modified the sentence.

 

304-308: This is a good opportunity to discuss prior work on detecting BA in wetlands, e.g., those cited in the introduction.

Author's reply: Thanks for the suggestion. We firmly believe that previous works were already discussed during this section and the previous ones.

 

316-319: This should be moved to the results section.

Author's reply: Thank you for your suggestion. We moved this sentence to the results section and improved the discussion.

 

340-341: Or rather, these sensors are the only options for analyzing years prior to 2015.

Author's reply: Thank you for your suggestion. We modified the sentence.

 

Author Response File: Author Response.pdf

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

An interesting study and one that can potentially contribute to fire science, particularly in wetland/peatland systems which is an emerging problem. The manuscript is generally well written with a few minor grammatical errors (please see attached manuscript). Addressing these will help with the flow of the narrative.

Some other issues that require attention include

Where were the images sourced from? Was it Google Earth or another image repository. Is the code used to process the images available?

One issue that needs addressing is the use of BA products for the temporal analysis. Your results demonstrate that they have relatively poor predictive capacity so your reason for including the temporal analysis needs better justification.

In your conclusion, you state that the Sentinel BA product can be used to calibrate Landsat and MODIS BA products but have not clearly described how this is done. This is an important aspect of the study which needs to be clearly explained.

Comments for author File: Comments.pdf

Author Response

Reviewer 1

Authors: The authors gratefully acknowledge the kind words and constructive comments on the paper entitled “Assessment of Burned Areas During the Pantanal Fire Crisis in 2020 Using Sentinel-2 Images” (fire-2293096), offered by Reviewer 1. We agree with you and have responded to all your comments, as you can find below. All changes performed are highlighted in red in the revised manuscript. We express our sincere gratitude for your review, which helped to improve the quality of the manuscript significantly.

Where were the images sourced from? Was it Google Earth or another image repository? Is the code used to process the images available?

Authors: Thank you very much for your comment. The MSI images were acquired from the GEE platform. This was clarified in section 2.3 of the revised document. Regarding the code availability, we have clarified in the Data Availability Statement that the code is available from the corresponding author by request.

One issue that needs addressing is the use of BA products for the temporal analysis. Your results demonstrate that they have relatively poor predictive capacity so your reason for including the temporal analysis needs better justification.

Authors: Thank you for pointing this out. The reason for including the temporal analysis was better justified in section 2.4 (please see the paragraph before Table 2).

In your conclusion, you state that the Sentinel BA product can be used to calibrate Landsat and MODIS BA products but have not clearly described how this is done. This is an important aspect of the study which needs to be clearly explained.

Authors: Thank you very much for your comment. We have detailed this possible calibration in this sentence. Moreover, following another reviewer comment, we have moved this sentence to the last paragraph of the Discussion section.

Attached manuscript:

It would help the structure of the paper to integrate this paragraph into paragraph two and start the paper with paragraph two about the Pantanal wetlands

Authors: Thank you very much for your comment. These paragraphs were combined according to your comment. Please see the first paragraph of the revised manuscript.

References required for this statement.

Authors: Thank you very much for pointing this out. Keddy et al. (2009) was cited in this statement.

Keddy, P.A.; Fraser, L.H.; Solomeshch, A.I.; Junk, W.J.; Campbell, D.R.; Arroyo, M.T.K.; Alho, C.J.R. Wet and Wonderful: The World's Largest Wetlands Are Conservation Priorities. BioScience 2009, 59, 39-51, doi:10.1525/bio.2009.59.1.8.  

Some information about the hydrology of these wetlands and how the drought is impacting the wetlands would be useful. Are the wetlands perched or groundwater fed?

Authors: Thank you very much for your comment. Done. Please see the first sentence of the second paragraph of the Introduction section.

It would be useful to provide some more background information. How much rainfall is typical? How much has it declined during this drought? What about insolation and evapotranspiration rates?

Authors: Thank you very much for your comment. We have edited this sentence to provide background information on precipitation, but no information were found regarding insolation and evapotranspiration rates.

The meaning of this paragraph is not clear.

Authors: Thank you very much for your comment. We have deleted this paragraph from the revised manuscript.

This information should be included when describing the Pantanal.

Authors: Thank you very much for your comment. This information was added to the first paragraph of the introduction section.

What wavelengths or wavelength regions are each of the bands measuring

Authors: Thank you very much for your comment. We have added a table (Table 1) describing these characteristics.

It is better to include wavelengths or wavelength ranges as well.

Authors: Thank you very much for your comment. We have added a table (Table 1) describing these characteristics.

Indicate with dates here

Authors: Thank you very much for your comment. Done. Please see the caption of Figure 2.

Are the BA products using the same algorithm? If so, then yes, the discrepancies may be due to differences in spatial and temporal resolution. If not, then it may be due to spectral resolution and or the differences in the algorithm's used.

Authors: Thank you very much for your comment. Each BA product has a specific input data and algorithm. We have clarified, after Figure 4, that the estimates are based on the original BA products instead of our algorithm.

 

Author Response File: Author Response.pdf

Reviewer 2 Report

A review of 'Assessment of burned areas during the Pantanal fire crisis in 2020 using Sentinel-2 images'.

Summary

The study describes a new method for mapping burned using Sentinel-2 data in a large wetland, a particularly difficult task due to potential confusion of burned areas with wetland. The authors were able to produce maps that were more accurate than those obtained using Landsat and MODIS data. This study makes a good contribution to our existing knowledge on mapping wildfires in a tropical wetland. The introduction section needs to be revised and a focus placed on recent studies on the 2020 Pantanal fires. Some of these studies should include Libonati et al. (2022), Barbosa et al. (2022), Martins et al. (2022), Marques et al. (2021), and Tomas et al. (2021). The authors should also introduce MODIS and Landsat burned products and outline their shortcomings as part of justification for the current study.

Title

Lines 2-3: Modify title to reflect that you are proposing a new burned area mapping approach that uses Sentinel-2 data and results in more accurate estimate of burned area compared to currently available burned area products.

Abstract

Lines 17-18: Rephrase to indicate that a drought that began in the Pantanal wetlands in 2019 continued and worsened through 2020, creating conditions that lead to the disastrous fires in 2020.

Introduction

Lines 37-38: suggested change 'Despite wetlands playing a key role in global climate regulation, they are under increased threat due to their sensitivity to anthropogenic disturbances and climate change. Delete sentence beginning in line 38.

Lines 48-64: Just state how much precipitation was received in the 2019-2020 period and how this compared to longer term records. It would also be helpful to include findings by Barbosa et al., (2022) here. No need to describe data sets such as CHIRPS – It should be enough to cite relevant references. In this section, you also go into great details on how the drought formed but this is unnecessary as it has been described elsewhere, and so the focus should remain on wildfires.

Lines 60-64: No need for a new paragraph as this is a continuation from the previous paragraph.

Lines 71-78: Unclear. Please clarify and rephrase, breaking the Very long sentence (lines 65-68) into two or more sentences.

Lines 78-79: How does unavailability of Landsat imagery leading to an underestimate of burned area? Needs explanation. Sentence is too long and unclear. Split into several sentences and clarify meaning.

Line 82: What is the frequency/availability of MODIS burn products?

Lines 85-91: Suggested change 'The objective of our research was to develop a more accurate map of the area burned in 2020 during the Pantanal fire crisis using Sentinel-2 images and comparing our results with other burned area products'. I would move the rest of the information out to the 'Methods' section.

Lines 90-91: These burn area products need to have been introduced earlier and their deficiencies highlighted before mentioning them here.

 

Materials and Methods

Lines 94-97: suggested change: 'The Pantanal is a seasonally flooded sedimentary basin located within the Upper Paraguay Drainage River Basin (UPDRB) and covers 150,355 km² of the Brazilian territory [10]'.

Line 97: Replace 'The overflooding' with 'Flooding'.

Line 99: Not clear - elaborate on vegetation dynamics of the biome.

Line 102: Delete 'for increasing economic yields'.

Lines 103-106: Suggested modification 'Farmers initiate the land transformation process by deliberately setting the natural vegetation on fire to expand pastureland and take advantage of the extremely dry conditions at the end of the dry season in the Pantanal to achieve this. Therefore, both anthropogenic and unusually dry conditions, such as the 2020 drought, can lead to catastrophic wildfires in this unique wetland'.

Page 4, Figure 1

Add country boundaries to provide a better reference location and reflect the Pantanal wetland spans several countries. This will also help reveal to what extent were the non- Brazilian portion of the Pantanal was affected by the wildfires.

Line 130: Elaborate on how monthly composites of Sentinel-2 data were created and pre-processed.

Line 136: What does this mean?

Line 143: How were the mosaics created? What pixel sizes were used considering the bands involved have different spatial resolutions?

Line 145: What pixel size was used?

Line 149: Suggested change: 'We then used the flooded area map created by Pereira et al. [ 10] for 2020 to remove areas that were…….'

Line 157: 'we chose two …… '

Lines 158- 159: 'to compare the burned area obtained using Sentinel-2 data'.

Lines 176-177: What do you mean by 'samples collected visually'. Explain in detail.

Line 185: 'biome over 2020'.

Lines 200-2001: Not clear.

Lines 207-208: 'methods used to generate the products.'

Line 208: Delete 'the' before Landsat, and 'medium spatial resolution' after.

Line 210: 'product uses MODIS images (250m)'.

Lines 214-215: 'resolution (20m) compared to MODIS images (250m), and a better temporal resolution (5 days) compared to Landsat satellites (16 days)'.

Line 220: 'Our results show that 44,998 km²…….'

Line 221: 'fire crisis, resulting in habitat destruction and biodiversity loss'.

 

Discussion

Lines 275-277: Split this very long sentence into several sentences and clarify meaning.

Lines 277: 'The worst recorded fire episode in the history of the Pantanal occurred in 2020'.

Conclusions

Lines 283-284: Suggested change 'The proposed BA mapping approach based on Sentinel-2 images presents advantages over existing BA products available for the Pantanal ……'

Line 285: Suggested change 'refine the BA estimates at the regional scale.'

Lines 288-289: '……Sentinel (2015) and can therefore result in more accurate estimates of BA in the longer term'.

Lines 290-294: Split into several sentences and clarify meaning. What are the implications of your study? What were the limitations of your study and what can you recommend for further study?

Author Response

Reviewer 2

Authors: The authors gratefully acknowledge the kind words and constructive comments on the paper entitled “Assessment of Burned Areas During the Pantanal Fire Crisis in 2020 Using Sentinel-2 Images” (fire-2293096), offered by Reviewer 2. We agree with you and have responded to all your comments, as you can find below. All changes performed are highlighted in red in the revised manuscript. We express our sincere gratitude for your review, which helped to improve the quality of the manuscript significantly.

The study describes a new method for mapping burned using Sentinel-2 data in a large wetland, a particularly difficult task due to potential confusion of burned areas with wetland. The authors were able to produce maps that were more accurate than those obtained using Landsat and MODIS data. This study makes a good contribution to our existing knowledge on mapping wildfires in a tropical wetland.

Authors: Thank you very much for your positive comments. We very much appreciate it.

The introduction section needs to be revised and a focus placed on recent studies on the 2020 Pantanal fires. Some of these studies should include Libonati et al. (2022), Barbosa et al. (2022), Martins et al. (2022), Marques et al. (2021), and Tomas et al. (2021). The authors should also introduce MODIS and Landsat burned products and outline their shortcomings as part of justification for the current study.

Authors: Thank you very much for your comment. Libonati et al. (2022) and Tomas et al. (2021) were already cited in the previous version of the manuscript. We have now cited Barbosa et al. (2022), Martins et al. (2022), and Marques et al. (2021) in the Introduction of the revised manuscript. Moreover, we have also better introduced the MODIS and Landsat shortcomings in this section of the revised manuscript.

Lines 2-3: Modify title to reflect that you are proposing a new burned area mapping approach that uses Sentinel-2 data and results in more accurate estimate of burned area compared to currently available burned area products.

Authors: Thank you very much for your comment. We have changed the title to: “Mapping Burned Areas During the Pantanal Fire Crisis in 2020: a Novel and More Accurate Approach Based on Sentinel-2 Images”

Lines 17-18: Rephrase to indicate that a drought that began in the Pantanal wetlands in 2019 continued and worsened through 2020, creating conditions that lead to the disastrous fires in 2020.

Authors: Thank you very much for your comment. Done. Please see the first sentence of the Abstract section.

Lines 37-38: suggested change 'Despite wetlands playing a key role in global climate regulation, they are under increased threat due to their sensitivity to anthropogenic disturbances and climate change. Delete sentence beginning in line 38.

Authors: Thank you very much for your suggestion. This sentence was changed as suggested. We have also deleted the sentence beginning in line 38.

Lines 48-64: Just state how much precipitation was received in the 2019-2020 period and how this compared to longer term records. It would also be helpful to include findings by Barbosa et al., (2022) here. No need to describe data sets such as CHIRPS – It should be enough to cite relevant references. In this section, you also go into great details on how the drought formed but this is unnecessary as it has been described elsewhere, and so the focus should remain on wildfires.

Authors: Thank you very much for your suggestion. The second paragraph was changed according to you comment and the Barbosa et al. (2022) work was cited. We have kept the drought formation sentence in order to explain that these events are expected to become more frequent. 

Lines 60-64: No need for a new paragraph as this is a continuation from the previous paragraph.

Authors: Thank you very much for your suggestion. Done. These paragraphs were combined.

Lines 71-78: Unclear. Please clarify and rephrase, breaking the Very long sentence (lines 65-68) into two or more sentences.

Authors: Thank you very much for your suggestion. Lines 65-68 were deleted according to Reviewer 1 comment.  Lines 71-78 were completely rephrased for clarity,

Lines 78-79: How does unavailability of Landsat imagery leading to an underestimate of burned area? Needs explanation. Sentence is too long and unclear. Split into several sentences and clarify meaning.

Authors: Thank you very much for your suggestion. This sentence was changed to match your comment and the modifications made on the previous sentence.

Line 82: What is the frequency/availability of MODIS burn products?

Authors: Thanks for your question. The MCD64A1 is a monthly product. This is clarified in section 2.4.

Lines 85-91: Suggested change 'The objective of our research was to develop a more accurate map of the area burned in 2020 during the Pantanal fire crisis using Sentinel-2 images and comparing our results with other burned area products'. I would move the rest of the information out to the 'Methods' section.

Authors: Thank you very much for your suggestion. This change was incorporated to the revised manuscript.

Lines 90-91: These burn area products need to have been introduced earlier and their deficiencies highlighted before mentioning them here.

Authors: Thank you very much for your suggestion. This sentence was moved to section 2.4, as previously suggested, where the BA products are described.

Lines 94-97: suggested change: 'The Pantanal is a seasonally flooded sedimentary basin located within the Upper Paraguay Drainage River Basin (UPDRB) and covers 150,355 km² of the Brazilian territory [10]'.

Authors: Thank you very much for your suggestion. Done. It was changed as suggested.

Line 97: Replace 'The overflooding' with 'Flooding'.

Authors: Thank you very much for your suggestion. We changed to “Flooding” as suggested.

Line 99: Not clear - elaborate on vegetation dynamics of the biome.

Authors: Thank you very much for your suggestion. We revised the sentence to make it clearer for the readers.

Line 102: Delete 'for increasing economic yields'.

Authors: Thank you very much for your suggestion. Done. The words were deleted as suggested.

Lines 103-106: Suggested modification 'Farmers initiate the land transformation process by deliberately setting the natural vegetation on fire to expand pastureland and take advantage of the extremely dry conditions at the end of the dry season in the Pantanal to achieve this. Therefore, both anthropogenic and unusually dry conditions, such as the 2020 drought, can lead to catastrophic wildfires in this unique wetland'.

Authors: Thank you very much for your suggestion. Done. This sentence was modified as suggested.

Page 4, Figure 1. Add country boundaries to provide a better reference location and reflect the Pantanal wetland spans several countries. This will also help reveal to what extent were the non- Brazilian portion of the Pantanal was affected by the wildfires.

Authors: Thank you very much for your suggestion. Done. The figure was changed as suggested.

Line 130: Elaborate on how monthly composites of Sentinel-2 data were created and pre-processed.

Authors: Thanks very much for your suggestion. These changes were incorporated in the second paragraph of section 2.3.

Line 136: What does this mean?

Authors: Thank you very much for your suggestion. We revised the sentence as: “When considering the remote sensing images, pixels are usually composed of several LULCs”.

Line 143: How were the mosaics created? What pixel sizes were used considering the bands involved have different spatial resolutions?

Authors: Thank you very much for your suggestion. We used images with 20 meters of spatial resolution. This was clarified in section 2.3.

Line 145: What pixel size was used?

Authors: Thank you very much for your suggestion. We used images with 20 meters of spatial resolution. This was clarified in section 2.3.

Line 149: Suggested change: 'We then used the flooded area map created by Pereira et al. [ 10] for 2020 to remove areas that were…….'

Authors: Thank you very much for your suggestion. It was changed as suggested.

Line 157: 'we chose two …… '

Authors: Thank you very much for your suggestion. It was changed as suggested.

Lines 158- 159: 'to compare the burned area obtained using Sentinel-2 data'.

Authors: Thank you very much for your suggestion. It was changed as suggested.

Lines 176-177: What do you mean by 'samples collected visually'. Explain in detail.

Authors: Thank you very much for your comment. We meant identified by visual interpretation. This was explained in section 2.5.

Line 185: 'biome over 2020'.

Authors: Thank you very much for your comment. It was changed as suggested.

Lines 200-2001: Not clear.

Authors: Thanks for your comment. We deleted part of the sentence to make the statement clearer.

Lines 207-208: 'methods used to generate the products.'

Authors: Thank you very much for your comment. It was changed as suggested.

Line 208: Delete 'the' before Landsat, and 'medium spatial resolution' after.

Authors: Thank you very much for your suggestion. Thank you very much for your comment.

Line 210: 'product uses MODIS images (250m)'.

Authors: Thank you very much for your suggestion. It was changed as suggested.

Lines 214-215: 'resolution (20m) compared to MODIS images (250m), and a better temporal resolution (5 days) compared to Landsat satellites (16 days)'.

Authors: Thank you very much for your suggestion. It was changed as suggested.

Line 220: 'Our results show that 44,998 km²…….'

Authors: Thank you very much for your suggestion. It was changed as suggested.

Line 221: 'fire crisis, resulting in habitat destruction and biodiversity loss'.

Authors: Thank you very much for your suggestion. It was changed to “…resulting in a severe ecosystem destruction…”.

Lines 275-277: Split this very long sentence into several sentences and clarify meaning.

Authors: Thank you very much for your comment. This sentence was divided accordingly.

Lines 277: 'The worst recorded fire episode in the history of the Pantanal occurred in 2020'.

Authors: Thank you very much for your comment. This information was revised.

Lines 283-284: Suggested change 'The proposed BA mapping approach based on Sentinel-2 images presents advantages over existing BA products available for the Pantanal ……'

Authors Thank you very much for your suggestion. It was changed as suggested.

Line 285: Suggested change 'refine the BA estimates at the regional scale.'

Authors: Thanks very much for your suggestion. It was changed as suggested.

Lines 288-289: '……Sentinel (2015) and can therefore result in more accurate estimates of BA in the longer term'.

Authors: Thank you very much for your comment. This information was revised as suggested.

Lines 290-294: Split into several sentences and clarify meaning. What are the implications of your study? What were the limitations of your study and what can you recommend for further study?

Authors: Thank you very much for your comment. This sentence was changed according to your comment and moved to the Discussion section according to the other reviewer’s comments.

 

Author Response File: Author Response.pdf

Reviewer 3 Report

This paper describes to detect the severe fire event of 2020 in Amazonian Pantanal using Sentinel images. The purpose is clear and the results are simple and good enough to publish. However, the discussion needs more findings from the result. If the authors cannot improve the discussion section, the paper can not be published.

 

The purpose of this paper is focused and simple & good enough to publish as an original paper. However, the discussion is not well written. The comparison among different satellite products is not good enough, especially which region Sentinel results are superior. Please see the following comments to consider the improvement.

Line 32. where global biomass burning inventories are widely known for having biases on a regional scale.

I am not sure this concluding remark is equivalent to what you did and discussed in this paper. Your discussion needs to clarify what is the advantage on regional improvement from Sentinel images.

Line 36 to 42 & Line 43 to 48.

I think people need to know what is Patanal the first and then understand the importance of wetland. So it can be clear to switch between these paragraphs.

Line 82 to 84.

What is the advantage of the frequency for fire observation in this study? Is there any evidence and advantage to detect the burn area from your study in terms of frequency ? The higher frequency produces better product in this study ? I saw higher resolution shows better accuracy from this paper. How about the frequency ?

Line 151 to 154

The detected area of flooded area is the same in MODIS and Landsat ? The detected area of flooded area can influence your accuracy, because BA can be easily confused by the flood, right ? How about the producer and user accuracy of the flooded area confused with the burn area ? The flooded area is consistent among different sensors (MODIS, Landsat, and Sentinel). If the flooded area is small enough, we don`t need to care much. But you use the mask for this, but other products could not use the same mask. In that case, is it still fair to compare the accuracy ? I mean the flooded area can be the major issue to reduce the accuracy. Then what is still advantage left to use Sentinel images ?

Line 176 to 189.

Your validation data was prepared on Sentinel images. Is it fair to validate with other remote sensing images ? It seems not fair to compare using the training data prepared from Sentinel to say that Sentinel is the superior to any other images... It is the best to use ground samples from any other sources (not Sentinel).

Line 235 to 237. 

Visually, you compared the results on Figure 5. Why don`t you compare the result in smaller regions grouped by smaller segments of ecological boundaries? or climate boundaries? or watershed boundaries ? or topographic effect ? This visual overall comparison is too rough to compare. It is better to compare the results in smaller regions to find what is the actual reason to show the difference among sensors. The current result is too rough.

Line 273 These differences incorporate limitations in each product, leading to underestimates or overestimates, particularly in regional analysis

It has NO meaning from this discussion sentence. It is too general to say any value of your research. For sure, different sensors use different algorithms to detect BA. But what do you want to say and compare from that here?  If you want to discuss from your results, you should say the following points.

1) what is the advantage of higher frequency of Sentinel image based BA ?

 2) what is the advantage of higher resolution of Sentinel image based BA ?

 3) which region has more difference among sensors (do not report from your visual inspection or your feeling. please report by area, actual numbers). Then which region is actually improved by Sentinel product and you need to discuss why it happened.

Author Response

Reviewer 3

Authors: The authors gratefully acknowledge the kind words and constructive comments on the paper entitled “Assessment of Burned Areas During the Pantanal Fire Crisis in 2020 Using Sentinel-2 Images” (fire-2293096), offered by Reviewer 3. We have properly responded to all your comments, as you can find below. All changes performed are highlighted in red in the revised manuscript. We express our sincere gratitude for your review, which helped to improve the quality of the manuscript significantly.

This paper describes to detect the severe fire event of 2020 in Amazonian Pantanal using Sentinel images. The purpose is clear, and the results are simple and good enough to publish. However, the discussion needs more findings from the result. If the authors cannot improve the discussion section, the paper cannot be published.

Authors:  Thank you very much for your comment. The Discussion section was improved according to your comment and the other reviewer’s comments. Please see section 4 of the revised manuscript.

 

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The manuscript presents well and communicates a coherent narrative regarding a comparison of burn area extent in the Pantanal. Many of the original issues have been addressed in this revised manuscript.

My main concern is, that the calibration section has been removed and this was the most novel aspect of the study. To possibly compensate for this, I would suggest a more detailed explanation of how the RF was adjusted from the original study and the contribution of each of the indices to the BA calculation. 

Otherwise, there are some minor grammatical errors.

Comments for author File: Comments.pdf

Author Response

Reviewer 1

Authors: The authors gratefully acknowledge the kind words and constructive comments on the paper entitled “Assessment of Burned Areas During the Pantanal Fire Crisis in 2020 Using Sentinel-2 Images” (fire-2293096), offered by Reviewer 1. We have responded to all your comments, as you can find below. All changes performed are highlighted in red in the revised manuscript. We express our sincere gratitude for your review, which helped to improve the quality of the manuscript significantly.

Here are some specifics comments:

“My main concern is, that the calibration section has been removed and this was the most novel aspect of the study. To possibly compensate for this, I would suggest a more detailed explanation of how the RF was adjusted from the original study and the contribution of each of the indices to the BA calculation.”

• Thank you for your suggestions. The calibration section has not been removed from the revised version. We moved the paragraph previously located in the conclusion section to the discussion section.
• Adjustments in the RF from the original study include cloud filtering specific to Sentinel-2A/MSI images, spectral bands selection, and new endmembers for the Linear Spectral Mixing Model (LSMM) used in the classification model. This information was added on lines 132-134.
• We have added a boxplot (Figure S2, Supplementary Materials) showing the importance of the variables in the RF model. Based on this information we added a discussion on lines 310-313.
• Some minor grammatical errors (line 26) were rectified, and the abbreviation “GEE” was previously introduced on lines 122-123.

 

Reviewer 2 Report

Thank you for the hard work you put into revising your manuscript! I believe is much improved and look forward to seeing it published.

Author Response

Reviewer 2

Authors: The authors gratefully acknowledge the kind words on the paper entitled “Assessment of Burned Areas During the Pantanal Fire Crisis in 2020 Using Sentinel-2 Images” (fire-2293096), offered by Reviewer 2. We express our sincere gratitude for your review, which helped to improve the quality of the manuscript significantly.