Conversion from Forest to Agriculture in the Brazilian Amazon from 1985 to 2021

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Abstract: briefly define land thematic maps... You also present some results not merely mention the method and potential output

More meaningful keywords

The Introduction is short and unstructured. Typical scientific paper structure would be: What is the topic, why is it important, what is known from the topic, what is the research gap, what would you like to do, how can you do it.

Highlight comparison of approach to usual change detection methods, time series approach e.g. markov random field approaches for time series LULC 

Method: how do you consider pixel-level omission/commission errors; and those that are "unlikely/unrealistic" changes such as grassland to forest in short intervals. Clarify this since your example in Figure 1 clearly depicts this e.g. pasture to forest in one year.

How do you deal with edge effects, fragmentation?

Any thoughts and results about benchmarking given the big data analysis and processing nature of your approach?

Figure 6 redundant as conversion length indicated twice

Discussion: Your approach could be best appreciated if there is a github repo or interactive platform that demonstrate 

Can your approach be used in time series continuous variables like biomass?

You presented key results, and these should be discussed in contrast or agreement with existing literature, and should be included in the Abstract and conclusion; but more importantly you produced these results as a research objective (stated in the Intro)

Figure 8 include statistics themselves

Author Response

We would like to thank the reviewer for the time for reviewing the article. We made improvements, and each of suggestions are addressed below:

 

"Abstract: briefly define land thematic maps... You also present some results not merely mention the method and potential output"

We improved the abstract by providing more information about the method and results (lines 1-18).

 

"More meaningful keywords"

We have made small changes to keywords to make them more meaningful (line 19).

 

"The Introduction is short and unstructured. Typical scientific paper structure would be: What is the topic, why is it important, what is known from the topic, what is the research gap, what would you like to do, how can you do it."

We improved the introduction by explaining the framework used in the article, as also highlighting the innovation of the results presented in the article.

 

"Highlight comparison of approach to usual change detection methods, time series approach e.g. markov random field approaches for time series LULC"

We agree that the use of this kind of approach could be interesting to improve land cover classification in the Amazon, and we added a brief discussion about this in the discussion section (lines 414-424). We avoided discussing this matter in the introduction, in order to keep it as objective as possible.

 

"Method: how do you consider pixel-level omission/commission errors; and those that are "unlikely/unrealistic" changes such as grassland to forest in short intervals. Clarify this since your example in Figure 1 clearly depicts this e.g. pasture to forest in one year."

Pixel level errors of omission/commission are discussed in the accuracy assessment, however, since we only do calculations over secondary data, we did not performed any kind of post processing to avoid creating distortions on the original dataset. The original dataset is already obtained by post-processing filters to reduce the amount of errors. The unlikely changes presented in figure 1 are hypothetical values used to illustrate the method (lines 96-97). Of course, that does not mean that the classification data from MapBiomas is completely free from errors that would create unlikely land cover changes, some of these were filtered in our results, as explained in the beginning of the results section (lines 187-192).

 

"How do you deal with edge effects, fragmentation?"

Our article does not perform analysis over edge effects or fragmentation. Both factors are likely to affect deforestation occurrence, however, this should be explored in further studies, and we made the decision to make this article as objective and concise as possible.

 

"Any thoughts and results about benchmarking given the big data analysis and processing nature of your approach?"

It is not clear if benchmarking would  be about how fast the code to process the data was, or if it is about accuracy assessment. By the perspective of code speed, the processing of the data is not optimized, we prioritize the creation of code that would be flexible enough for the user to reproduce the processing with different options, being possible to run it with different hardware specifications. The process takes not more than two days, and it is not meant to be repeated too often, so we believe that code benchmarking and optimization would not benefit the methods or the article in general.

 

"Figure 6 redundant as conversion length indicated twice"

It is not clear if the redundancy comes from repeating conversion length in the title, vertical axis, and the legend. If so, we created the visualization with this repetition in order to make it more accessible and easily understood, with the cost of making redundant elements.

 

"Discussion: Your approach could be best appreciated if there is a github repo or interactive platform that demonstrate "

We provide a github repository with all the code necessary to reproduce the data processing and create the article results. We also provide free access to the data in a public repository. Both links are presented in the data availability statement (line 468). An interactive platform can indeed help the communication of most scientific publications, however, we prioritize easy access to the data and the code necessary to reproduce the results.

 

"Can your approach be used in time series continuous variables like biomass?"

We believe that our data can be used in investigations dealing with time series of continuous variables. We added a brief discussion about this topic in the discussion section (lines 398-406).

 

"You presented key results, and these should be discussed in contrast or agreement with existing literature, and should be included in the Abstract and conclusion; but more importantly you produced these results as a research objective (stated in the Intro)"

We added a section dedicated for discussion of the results (lines 330-426). We also made the findings more explicit in the abstract and conclusion.

 

"Figure 8 include statistics themselves"

The accuracy assessment metrics are presented in figure 7 (lines 292-293).

Reviewer 2 Report

Comments and Suggestions for Authors

This research uses sound methodology to provide spatial and temporal estimates of conversion from forest to agricultural land use in the Brazilian Amazon. Results are extremely informative with respect to the effectiveness of various forest and economic policies aimed at reducing deforestation. Continued exploitation of lands deforested prior to the soy moratorium becomes quite apparent in the results. Continuation of the loophole in the soy moratorium will result in continued land conversion to agricultural uses and reduce future reversion of deforested areas back to forest.

In summary, I think the current manuscript should be published as it provides instructive and relevant analysis of ongoing policies affecting the Brazilian Amazon forests.

Author Response

We would link to thank the reviewer for the time to assess the article, and for the feedback.

 

"This research uses sound methodology to provide spatial and temporal estimates of conversion from forest to agricultural land use in the Brazilian Amazon. Results are extremely informative with respect to the effectiveness of various forest and economic policies aimed at reducing deforestation. Continued exploitation of lands deforested prior to the soy moratorium becomes quite apparent in the results. Continuation of the loophole in the soy moratorium will result in continued land conversion to agricultural uses and reduce future reversion of deforested areas back to forest.

In summary, I think the current manuscript should be published as it provides instructive and relevant analysis of ongoing policies affecting the Brazilian Amazon forests."

We agree with your reflections about the effects of the continuation of the loophole in the soy moratorium. We added a brief discussion about this topic in the discussion section (lines 394-397).

 

Reviewer 3 Report

Comments and Suggestions for Authors

 The paper presents a new dataset designed to identify the transition from forest to agriculture by calculating the time it took for this change between 1985 and 2021. The following improvements are suggested:

Abstract: The abstract offers a comprehensive summary of the study’s objectives, methods, and results, but it lacks specificity in certain areas. For instance, it does not clarify the source of the "high-resolution thematic maps" mentioned. It is recommended that the authors include key information such as the dataset's source, time period, and resolution. Furthermore, although the abstract mentions the significance of the dataset for analyzing land-use change in Brazil, it would benefit from a more detailed explanation of its importance. The abstract should also include the main conclusions of the study.

Introduction: While the paper references reviews on land-use and land-cover (LULC) classification and Brazilian development policies, it could expand on the characteristics of LULC classificatio. The final paragraph should briefly mention the theoretical and practical significance of the research, as well as include an overview of the research framework for the paper.

Methods: In lines 84-85, the description of the conversion length calculation process (Figure 1) could be better represented with a visual diagram to enhance clarity.

Results and Discussion: The Results and Discussion sections should be separated, as they address different aspects. The current section leans more toward presenting the results, so a dedicated Discussion section should be added.

Line 330: The statement about “uncertainty in the boundaries of agricultural areas” (samples 25, 56, 84, 92) being a common issue in satellite image classification should be expanded. How can this uncertainty be reduced? The authors might want to discuss strategies for data processing?

Line 324: Errors like “roads being misidentified as agricultural” and “missing agricultural areas” are mentioned. What image features or data noise led to these errors? Did the authors explore other techniques or methods to minimize these errors? This could be further discussed.

Conclusions: The conclusions section should emphasize the significance of the research. The current content is general. While the dataset can provide valuable insights for studies on land-use change (LULC), its specific applications or impacts are not fully explored. The authors should consider discussing how the dataset could support research in fields like ecology, environmental policy, economics, and others.

Figures: There are quality issues with Figures 4 to 8. The gray background should be replaced with a white background, as the current gray background makes it difficult to clearly view the images and legends.

References: The citation format does not follow the journal's guidelines. The authors should review and adjust it to match the format required by Land.

 

Language Quality: The English language quality is good and easy to follow, but there may be some minor improvements to consider.

 

 

 

Author Response

We would like to thank the reviewer for the time to assess and make good suggestions for improving the article. The review was well organized, clear and constructive.

 

"Abstract: The abstract offers a comprehensive summary of the study’s objectives, methods, and results, but it lacks specificity in certain areas. For instance, it does not clarify the source of the "high-resolution thematic maps" mentioned. It is recommended that the authors include key information such as the dataset's source, time period, and resolution. Furthermore, although the abstract mentions the significance of the dataset for analyzing land-use change in Brazil, it would benefit from a more detailed explanation of its importance. The abstract should also include the main conclusions of the study."

We have made improvements in the abstract. The source of data is better specified now. The results and conclusions are also presented, as a better detail of the importance and innovation of the article (lines 1-18).

 

"Methods: In lines 84-85, the description of the conversion length calculation process (Figure 1) could be better represented with a visual diagram to enhance clarity."

While we agree that diagrams are a good option to present method workflows, we decided that in this case, a visual illustration of the process, followed by a description of the main steps of this process, can be more informative, even tough it can take more time for interpretation. We also added a figure as supplementary material that presents a visual diagram of the whole calculation process.

 

"Results and Discussion: The Results and Discussion sections should be separated, as they address different aspects. The current section leans more toward presenting the results, so a dedicated Discussion section should be added."

We created a dedicated section for discussion, we also organized the results (lines 179-329) and discussion (lines 330-426) in chronological order to improve clarity.

 

"Line 330: The statement about “uncertainty in the boundaries of agricultural areas” (samples 25, 56, 84, 92) being a common issue in satellite image classification should be expanded. How can this uncertainty be reduced? The authors might want to discuss strategies for data processing?

Line 324: Errors like “roads being misidentified as agricultural” and “missing agricultural areas” are mentioned. What image features or data noise led to these errors? Did the authors explore other techniques or methods to minimize these errors? This could be further discussed."

We expanded the discussion of uncertainties in the discussion section (lines 407-426). However we tried to be brief and concise in relation to technical details of satellite image classification errors, in order to make the article as objective as possible. 

 

"Conclusions: The conclusions section should emphasize the significance of the research. The current content is general. While the dataset can provide valuable insights for studies on land-use change (LULC), its specific applications or impacts are not fully explored. The authors should consider discussing how the dataset could support research in fields like ecology, environmental policy, economics, and others."

We improved the conclusions with more information about the importance and new opportunities that may arise from our results (lines 427-446).

 

Figures: There are quality issues with Figures 4 to 8. The gray background should be replaced with a white background, as the current gray background makes it difficult to clearly view the images and legends.

We modified the figures and made the improvements as suggested, making the figures more clear and easy to visualize.

 

References: The citation format does not follow the journal's guidelines. The authors should review and adjust it to match the format required by Land.

We reviewed the references, now they should be in the journal guidelines.

Reviewer 4 Report

Comments and Suggestions for Authors

 (1). The introduction

§  More explicitly define the research gap. While the context is well-detailed, the novelty of the methodology and dataset should be highlighted earlier.

§  Consider briefly discussing the global relevance of Amazonian deforestation to climate change, biodiversity, and carbon dynamics to enhance the paper's scope.

(2). Methodology:

§  Add a summary flowchart outlining the seven steps of conversion length calculation for easier comprehension.

§  To improve the clarity and accessibility of the methodology section, please include a flowchart summarizing the key steps.

(3). Results and Discussion:

·         To provide better geographical context for the study, please include a map of Brazil showing the location of the Amazon biome and the specific study area within it.

 ·         In Figure 4, please include the names of key locations (e.g., states or major regions) on the map to provide better spatial context.

·         The current combination of Results and Discussion makes it challenging to distinguish between the presentation of findings and their interpretation. please divide this section into two sections

(4). Conclusions:

·         Add actionable insights, such as how the data can guide sustainable land management or inform global climate mitigation strategies.

(5). References:

 Include more citations from studies outside Brazil to underscore the global applicability of the findings.

·         Ensure all cited datasets and tools, such as MapBiomas and Google Earth Engine, have proper acknowledgment.

Author Response

We would like to thank the reviewer for the time for reviewing the article. We made improvements, and each of suggestions are addressed below:

 

"More explicitly define the research gap. While the context is well-detailed, the novelty of the methodology and dataset should be highlighted earlier."

We improved the introduction with a better explanation of the methodology and its novelty (lines 20-60).

 

"Consider briefly discussing the global relevance of Amazonian deforestation to climate change, biodiversity, and carbon dynamics to enhance the paper's scope."

We mentioned the possibility of using the new data in investigations about carbon stocks in the introduction (lines 53-60) and in the discussion (lines 398-406). However, to keep the introduction as objective as possible, we avoided expanding the scope of the article.

 

"Add a summary flowchart outlining the seven steps of conversion length calculation for easier comprehension."

"To improve the clarity and accessibility of the methodology section, please include a flowchart summarizing the key steps."

We decided to keep the visual illustration and the following description of the main steps, since it contains more text and more details than we could represent in flowcharts. However, we added a detailed flowchart of the calculation process as supplementary material.

 

"To provide better geographical context for the study, please include a map of Brazil showing the location of the Amazon biome and the specific study area within it."

We agree that this is an improvement,, and added a map of Brazil and the Amazon biome in figure 4.

 

"In Figure 4, please include the names of key locations (e.g., states or major regions) on the map to provide better spatial context."

We agree that this is an improvement, and added the name of the states inside the Amazon biome for improving the spatial context of figure 4.

 

"Add actionable insights, such as how the data can guide sustainable land management or inform global climate mitigation strategies."

We agree that this is an improvement, and added (in the discussion) some examples of how the data can be used in land management, specially in relation to carbon stocks and forest restoration (lines 394-406).

 

"Include more citations from studies outside Brazil to underscore the global applicability of the findings."

Since the temporal pattern of forest->pasture->agriculture seems very specific to the Brazilian Amazon, we could not find similar studies in other countries, although we agree that it would be an improvement to the article. However we decided to avoid extrapolating our results to other countries.

 

"Ensure all cited datasets and tools, such as MapBiomas and Google Earth Engine, have proper acknowledgment."

We made sure to cite Mapbiomas (line 66) and Google Earth Engine (line 74).

Round 2

Reviewer 4 Report

Comments and Suggestions for Authors After carefully reviewing the revised manuscript, I am pleased to confirm that the authors have addressed all the necessary adjustments and incorporated the feedback provided during the first round of the review process. The revisions have significantly improved the quality and clarity of the manuscript, and I believe it now meets the standards required for publication.    I therefore recommend that the manuscript be accepted for publication.     Best regards,