Evaluating Visible–Infrared Imaging Radiometer Suite Imagery for Developing Near-Real-Time Nationwide Vegetation Cover Monitoring in Indonesia

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript titled "Evaluating the VIIRS Imagery for Developing Near Real-Time Nationwide Vegetation Cover Monitoring in Indonesia" presents a comprehensive study on the use of VIIRS satellite imagery for monitoring vegetation cover changes in Indonesia. The authors have employed the Normalized Differential Open Area Index (NDOAI) to detect changes in vegetation cover and have compared the performance of VIIRS with Landsat data. The study is well-structured, providing a clear introduction, detailed methodology, and a thorough analysis of the results. The potential of VIIRS in developing an early warning system for vegetation cover change is well-articulated, and the findings are significant for environmental monitoring and policy-making in Indonesia.

 

 

 

Specific Recommendations:

 

1. The manuscript is generally well-written, but some sections could benefit from further clarification. Specifically, the 'Discussion' section could be restructured to more directly connect the findings with the implications for future research and applications.

 

2. The methodology is robust; however, it would be beneficial to include a brief discussion on the limitations of using the NDOAI index, particularly in the context of different vegetation types and environmental conditions in Indonesia.

 

3. The accuracy assessment results are presented clearly, but it would be advantageous to include a sensitivity analysis to show how changes in the threshold value affect the overall accuracy of the system.

 

4. The authors should discuss how their findings compare to other studies that have used different remote sensing technologies for similar purposes. This will situate their work within the broader literature and highlight the unique contributions of their study.

 

5. The figures and tables are informative, but some may need to be revised for better clarity. For instance, Figure 7 could be improved with a more detailed legend to explain the symbols and colors used.

 

6. Ensure that all references are up-to-date and that there is consistency in the citation style throughout the manuscript.

 

7. There are minor grammatical issues and some sentences could be rephrased for better flow. It is recommended that a professional language editing service be used to refine the manuscript.

 

8. The method of data processing in this manuscript is the commonly applied ones, and I would recommend the adding of limitation of the current study in section discussion. Such as adding the new method of machine learning approaches for data processing.

Comments on the Quality of English Language

minor

Author Response

Thank you for your thorough review of our manuscript. We appreciate your constructive feedback and have made the following revisions to enhance the clarity and impact of our work.

Specific Recommendations:

Comments 1: The manuscript is generally well-written, but some sections could benefit from further clarification. Specifically, the 'Discussion' section could be restructured to connect the findings more directly with the implications for future research and applications.

Response 1: Thank you for the comment and recommendation. Our research has resulted in two main findings regarding both the spatial and temporal aspects of satellite image detection. The first key finding relates to the spatial resolution of detection results. It highlights the differences in the minimum detectable patch size between the two datasets, providing insights into their respective capabilities for detecting changes in vegetation cover (Sub-chapter 3.2). The second major finding concerns the temporal resolution of detection results. The Sub-chapter 3.3 examines the differences in the revisit cycles of VIIRS and Landsat related to the change detection results. This temporal comparison underscores the advantages and limitations of each dataset for near real-time monitoring of environmental changes.

Regarding the findings, we have elaborated on potential directions for future research, identifying gaps revealed by our study and proposing methodologies to address these gaps. Then, we revised the title of Sub-chapter 4.3 to “Future Research Improvements and Applications” to better reflect its focus on advancing future research and practical applications. Specific changes include:

            Page 14-15, lines 465-518; lines 528-534

Comments 2: The methodology is robust; however, it would be beneficial to include a brief discussion on the limitations of using the NDOAI index, particularly in the context of different vegetation types and environmental conditions in Indonesia.

Response 2: Thanks for this comment. We have revised the discussion section to provide a clearer explanation of the limitations of our method and future research improvements.

(Sub-chapter 4.3, page 15, paragraph 2).

The following references have been in the list of references.

• Alyasiri, E. A., Wilson, J. L., and James, R. D. Estimating the Population of a Middle Eastern City Based on a Normalized Difference Built-Up Index and Urban Morphology. J. Appl. Geospat. Res. 2023, 14(1), 1-22 http://doi.org/10.4018/IJAGR.313942
• Guha, S., Govil, H., Diwan, P. Analytical study of seasonal variability in land surface temperature with normalized difference vegetation index, normalized difference water index, normalized difference built-up index, and normalized multiband drought index. Appl. Rem. Sens. 2019, 13(2) 024518 https://doi.org/10.1117/1.JRS.13.024518

In the methodology section, we address our rationale for exclusively using the NDOAI algorithm in developing the vegetation monitoring system. In our prior investigation (Setiawan et al. 2016), we employed a comprehensive approach by integrating both the Normalized Differential Open Area Index (NDOAI) and the Normalized Difference Vegetation Index (NDVI). Our study meticulously evaluated the effectiveness of these indices in delineating changes in vegetation cover. The analysis revealed that the NDOAI method successfully detected significant alterations. Therefore, to optimize efficiency and simplicity in the operational procedures for national-scale monitoring endeavors, we decided to utilize the NDOAI vegetation index in the present study exclusively.

(Page 4 lines 156-163).

Comments 3: The accuracy assessment results are presented clearly, but it would be advantageous to include a sensitivity analysis to show how changes in the threshold value affect the system's overall accuracy.

Response 3: We thank R#1 for making this useful comment. We want to share that the threshold of the vegetation cover changes was derived from prior research conducted by Suyamto et al (2021), which involved the examination of high-resolution imagery and field observations. Then, the specific empirical threshold of -100 for ΔNDOAI was implemented to characterize changes in vegetation cover.

As a concern of R#1 about the sensitivity analysis of the threshold, we can explain that we actually have done a sensitivity and sensibility analysis on our approach using probability density function (PDF) and cumulative distribution functions (CDFs) of some geometric features at change-detection thresholds ranging from 50 to 130, as it has been published in the previous paper (Suyamto et al, 2021). The result suggests that a threshold of about 100 was optimal.

The explanation above was added in the method section (Sub-chapter 2.3, page 2, lines 227-234).

Comments 4: The authors should discuss how their findings compare to those of other studies that have used different remote sensing technologies for similar purposes. This will situate their work within the broader literature and highlight their study's unique contributions.

Response 4: Thanks for this remark. The discussion section (passing from page 15 lines 484-503) briefly discusses the intention of this work and the other remote sensing technologies.

We have added the following references in the text.

• Ayhan, B.; Kwan, C.; Budavari, B.; Kwan, L.; Lu, Y.; Perez, D.; Li, J.; Skarlatos, D.; Vlachos, M. Vegetation Detection Using Deep Learning and Conventional Methods. Remote Sens. 2020, 12, 2502.
• Pacheco-Pascagaza, A.M.; Gou, Y.; Louis, V.; Roberts, J.F.; Rodríguez-Veiga, P.; da Conceição Bispo, P.; Espírito-Santo, F.D.B.; Robb, C.; Upton, C.; Galindo, G.; et al. Near Real-Time Change Detection System Using Sentinel-2 and Machine Learning: A Test for Mexican and Colombian Forests. Remote Sens. 2022, 14, 707.

Comments 5: The figures and tables are informative, but some may need to be revised for better clarity. For instance, Figure 7 could be improved with a more detailed legend to explain the symbols and colors used.

Response 5: Thanks for this remark. As suggested by R#1, several figures have been modified to improve clarity.

Comments 6: Ensure that all references are up-to-date and that there is consistency in the citation style throughout the manuscript.

Response 6: Thank you for your comment. We have ensured that all references are up-to-date and reviewed the manuscript to ensure consistency in the citation style.

Comments 7: There are minor grammatical issues, and some sentences could be rephrased for better flow. It is recommended that a professional language editing service be used to refine the manuscript.

Response 7: Thanks for this suggestion. We have conducted a thorough re-check of the language in this manuscript using professional English proofreading.

Comments 8: The method of data processing in this manuscript is the commonly applied ones, and I would recommend the adding of limitation of the current study in section discussion. Such as adding the new method of machine learning approaches for data processing.

Response 8: Thanks for this recommendation. We have improved the text to better explain the limitation of the approach as well as the future research improvements in the discussion section (page 15 lines 484-503; page 15 lines 506-540)

We want to highlight that the simplicity of the image differencing approach in the NDOAI algorithm makes it nearly ideal for national-scale near real-time monitoring, as a straightforward and efficient method is essential for processing large datasets across extensive regions. However, addressing the limitations of seasonal variability is crucial, especially in arid areas like the southeastern part of Indonesia, where long-term monitoring over at least one year is required for accurate results. In comparison, remote sensing technologies like machine learning offer advanced capabilities for change detection but come with increased complexity and computational demands. Machine learning algorithms can handle various data types and be trained to recognize subtle patterns and changes in vegetation cover, potentially providing more accurate and detailed results. However, these methods require substantial computational resources, expertise in algorithm training and validation, and extensive labeled datasets for training.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

2.1 Overview and general recommendation:

The study is well written and a topic I am very interested in. The writing is very clear and the experiment is nicely described. The topic and study area are special. The specific comments are as follows.

2.2 Comments:

1. In section 2.4.1 Sampling Design for Accuracy Test, the author used various disturbance properties, why is the >70%, >40%, >30%, >20%, and <5% of a VIIRS pixel size, it is not the equal interval.

2. the units of spatial resolution are not consistent, especially the singular and plural.

3. the last paragraph of the section 1 Introduction is not clear.

4. Figures of the article are not consistent and standard.

5. the equations don’t have the number, additionally, the terms of equations are also not standard.

Comments on the Quality of English Language

No

Author Response

Thank you for your thorough review of our manuscript. We appreciate your constructive feedback and have made the following revisions to enhance the clarity and impact of our work.

 

Overview and general recommendation:

The study is well written and a topic I am very interested in. The writing is very clear, and the experiment is nicely described. The topic and study area are special. The specific comments are as follows.

Thanks for this meaningful remark.

Comments:

Comments 1: In section 2.4.1 Sampling Design for Accuracy Test, the author used various disturbance properties, why is the >70%, >40%, >30%, >20%, and <5% of a VIIRS pixel size, it is not the equal interval.

Response 1: Thank you for the remark. The interval is referred to in the previous reference. It shows a significant change in vegetation cover.

Evaluating the pixel-level accuracy of change detection is challenging because we are comparing a change map at a coarser resolution to a Landsat-based reference map. Our objective is to identify areas of vegetation disturbance accurately, and the smaller the detectable percentage of disturbance, the more effective the method. To achieve this, we defined the level of vegetation disturbance using various disturbance proportions, such as >70%, >40%, >30%, >20%, and <5% of a VIIRS pixel size.

These proportions were chosen to capture a wide range of disturbance intensities, rather than using equal intervals, to understand better the detection capabilities across different levels of vegetation cover disturbance.

By selecting these particular thresholds, we can categorize the disturbance level into: (1) significant disturbances, the >70% threshold ensures that areas with substantial disturbance are identified, providing a benchmark for high-intensity changes, (2) intermediate disturbances: the >40%, >30%, and >20% thresholds help in evaluating the algorithm's performance across varying moderate levels of disturbance, which are crucial for early intervention and management, (3) minimal disturbances: the <5% threshold is included to test the algorithm's sensitivity to very low levels of disturbance, then it is categorized as no-change.

Comments 2: The units of spatial resolution are not consistent, especially the singular and plural.

Response 2: Text has been revised accordingly (lines 84, 97, 98, 99, 138, 139).

 

Comments 3: the last paragraph of the section 1 Introduction is not clear.

Response 3: Thanks for this comment. We have modified the last paragraph to explain more about the objective of this study. (Page 3, lines 103-109)

The important remark of the last paragraph (Introduction section) is the necessity to assess the effectiveness of VIIRS data for monitoring vegetation changes nationally, which is required to establish a robust early warning system for preventing forest disturbances in Indonesia.

Comments 4: Figures of the article are not consistent and standard.

Response 4: Thanks for this remark. All figures were modified accordingly.

 

Comments 5: The equations don’t have the number; additionally, the terms of equations are also not standard.

Response 5: Thanks for this comment. We added the number to the equations in the text (equations 1 and 2).

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

See word content for details

Comments for author File: Comments.pdf

Author Response

Thank you for your thorough review of our manuscript. We appreciate your constructive feedback and have made the following revisions to enhance the clarity and impact of our work.

 

Comments and Suggestions for Authors

See word content for details.

This paper assesses the potential of VIIRS satellite imagery in developing an early warning system for monitoring vegetation change in Indonesia and compares it with the effectiveness of applying Landsat in identifying vegetation change.

In light of these findings, it is recommended that the authors reflect on the following questions before publishing their article:

Comments 1:  Is the introductory section of the article excessively lengthy and poorly structured? Does it effectively convey the significance of the NFMS or does it merely highlight the significance of assessing the significance of VIIRS satellite imagery in the study of vegetation change in Indonesia?

Response 1: Thanks for this insightful comment. Our study focused on developing a near-real-time vegetation cover change monitoring system to support the National Forest Monitoring System (NFMS).

The first two paragraphs explain the NFMS's importance in providing information on the status and trends of forest resources. This information is essential for REDD+ implementation by enabling measurement, reporting, and verification (MRV). Additionally, the NFMS underscores the need for an early warning system to monitor various forest disturbances effectively.

In the third paragraph, we discuss the necessity of developing an early warning system for vegetation cover change to support the NFMS. This is crucial due to the various forest disturbances in Indonesia, including complete deforestation, partial degradation, and broader loss of vegetation cover. Given that deforestation and forest degradation can be defined differently, this study aims to develop a near-real-time monitoring system for vegetation cover change.

To clarify, the objective stated in the latter paragraph is: "This study aimed to rigorously evaluate a vegetation cover change algorithm based on VIIRS data to develop a robust early warning system for preventing forest disturbances. The goal was to contribute to more effective forest management and conservation efforts in Indonesia."

We have modified the last paragraph to explain more about the objective of this study. (Page 3, lines 103-109)

Comments 2: What are the advantages of NDOAI over traditional vegetation indices such as NDVI, FVC, etc?

Response 2: In the methodology section, we address our rationale for exclusively using the NDOAI algorithm in developing the vegetation monitoring system. In our prior investigation (Setiawan et al. 2016), we employed a comprehensive approach by integrating both the Normalized Differential Open Area Index (NDOAI) and the Normalized Difference Vegetation Index (NDVI). Our study meticulously evaluated the effectiveness of these indices in delineating changes in vegetation cover. The analysis revealed that the NDOAI method was particularly successful in detecting significant alterations. Therefore, to optimize efficiency and simplicity in the operational procedures for national-scale monitoring endeavors, we decided to exclusively utilize the NDOAI vegetation index in the present study (page 4 lines 156-163).

 

Comments 3: Is the information depicted in some of the diagrams in the article clear and accurate? For example, in Figure 2, relevant details from the figure could be added, or additional text could be added to enhance the accuracy of the depicted information.

Response 3: Thank you for the comment. Some figures were modified accordingly.

 

Comments 4: In the article Δ NDOAI implements a specific empirical threshold of -100 to describe changes in vegetation cover, is this empirical threshold reasonable to apply in Indonesia?

Response 4: We thank R#3 for making this useful comment. That’s a good point. We want to share that the threshold of the vegetation cover changes was derived from prior research conducted by Suyamto et al (2021). We can explain that we have done a sensitivity and sensibility analysis on our approach using probability density function (PDF) and cumulative distribution functions (CDFs) of some geometric features at change-detection thresholds ranging from 50 to 130. The result suggests that a threshold of about 100 was optimal, as it has been published in the previous paper (Suyamto et al, 2021).

 

The explanation above was added in the method section (Sub-chapter 2.3, page 2, lines 227-234).

 

Comments 5: Is the information depicted in Figure 4 clear and accurate, and is Δ NDOAI greater than the threshold represented visually and accurately in the figure?

Response 5: Yes, that's correct. The figure serves as a visual aid to demonstrate how the change in the NDOAI value pattern is determined by comparing it with the ΔNDOAI. Specifically, a change occurs when the ΔNDOAI exceeds the specified threshold.

 

Comments 6: Is the writing of formulas or abbreviations in the article standardised? Will the writer please check if the font, slant, etc. are standardised?

Response 6: Thank you for this meaningful suggestion. We thoroughly reviewed to confirm that all formulas and abbreviations adhere to the standard formatting guidelines. Yes, the writing of formulas and abbreviations in the article has been standardized. We have ensured consistency in font, slant, and other formatting aspects throughout the guideline.

 

Comments 7: In the article, different percentages of disturbance are used to define the level of vegetation disturbance; is there a theoretical basis for the thresholds chosen for accuracy monitoring? Why not 60 or 80 percent?

Response 7: We thank the reviewer for making this suggestion. We share concerns about the level of vegetation disturbance using various disturbance proportions, such as >70%, >40%, >30%, >20%, and <5% of a VIIRS pixel size.

Evaluating the pixel-level accuracy of change detection is challenging because we are comparing a change map at a coarser resolution to a Landsat-based reference map. Our objective is to identify areas of vegetation disturbance accurately, and the smaller the detectable percentage of disturbance, the more effective the method. To achieve this, we defined the level of vegetation disturbance using various disturbance proportions, such as >70%, >40%, >30%, >20%, and <5% of a VIIRS pixel size. These proportions were chosen to capture a wide range of disturbance intensities, rather than using equal intervals, to better understand the detection capabilities across different levels of vegetation cover disturbance.

By selecting these particular thresholds, we can categorize the disturbance level to: (1) significant disturbances: the >70% threshold ensures that areas with substantial disturbance are identified, providing a benchmark for high-intensity changes, (2) intermediate disturbances, the >40%, >30%, and >20% thresholds help in evaluating the algorithm's performance across varying moderate levels of disturbance, which are crucial for early intervention and management, (3) minimal disturbances: the <5% threshold is included to test the algorithm's sensitivity to very low levels of disturbance, and then it is categorized as “no-change”.

 

Comments 8: Are there some differences between VIIRS and Landsat satellites in identifying the time of change at specific thresholds, and do these errors affect the applicability of VIIRS satellites for monitoring vegetation change in Indonesia?

Response 8: Thank you for this valuable remark. The findings regarding the temporal differences between VIIRS and Landsat satellites in detecting changes are indeed very interesting. Our results indicated that when a threshold value greater than 20% is used, VIIRS can identify devegetation 4.5 days earlier than Landsat. Additionally, for areas with 40% openness (10 hectares) and 70% openness (18.75 hectares), VIIRS detects changes 25.4 days and 54.8 days faster than Landsat, respectively.

This result has been included on page 12, lines 384-396.

Author Response File: Author Response.pdf

Reviewer 4 Report

Comments and Suggestions for Authors

The document titled 'Evaluating the VIIRS Imagery for Developing Near Real-Time Nationwide Vegetation Cover Monitoring in Indonesia' makes a significant contribution to the monitoring system for forest cover and its changes.

Overall, the document is well-structured, but there are some sections that require amendments.

In line 3, it appears that the authors intend to protect or mitigate forest cover rather than reduce natural vegetation cover.

Additionally, Table 1 and line 329 provide redundant information, as they both repeat the values presented in the aforementioned table.

Moreover, line 463 contains redundancy with 'observed observations'

While the referenced literature is acceptable, it is advisable to incorporate new and up-to-date sources to enhance the content of the document.

Comments on the Quality of English Language

In line 3, it appears that the authors intend to protect or mitigate forest cover rather than reduce natural vegetation cover.

Moreover, line 463 contains redundancy with 'observed observations'

Author Response

Reviewer 4 (R#4)

Thank you for your thorough review of our manuscript. We appreciate your constructive feedback and have made the following revisions to enhance the clarity and impact of our work.

 

Comments and Suggestions for Authors

The document titled 'Evaluating the VIIRS Imagery for Developing Near Real-Time Nationwide Vegetation Cover Monitoring in Indonesia' makes a significant contribution to the monitoring system for forest cover and its changes.

Overall, the document is well-structured, but there are some sections that require amendments.

Thanks for this meaningful remark.

Comments 1: In line 3, it appears that the authors intend to protect or mitigate forest cover rather than reduce natural vegetation cover.

Response 1: Thank you for your comments.

We want to confirm that our study focused on developing a near-real-time vegetation cover change monitoring system to support the National Forest Monitoring System (NFMS). We have structured the content in the paragraph section to emphasize the importance of this study.

The first two paragraphs explain the NFMS's importance in providing information on the status and trends of forest resources. This information is essential for REDD+ implementation by enabling measurement, reporting, and verification (MRV). Additionally, the NFMS underscores the need for an early warning system to monitor various forest disturbances effectively.

In the third paragraph, we discuss the necessity of developing an early warning system for vegetation cover change to support the NFMS. This is crucial due to the various forest disturbances in Indonesia, including complete deforestation, partial degradation, and broader loss of vegetation cover. Given that deforestation and forest degradation can be defined differently, this study aims to develop a near-real-time monitoring system for vegetation cover change.

To clarify, the objective stated in the latter paragraph is: "This study aimed to rigorously evaluate a vegetation cover change algorithm based on VIIRS data to develop a robust early warning system for preventing forest disturbances. The goal was to contribute to more effective forest management and conservation efforts in Indonesia."

We have modified the last paragraph to explain more about the objective of this study. (Page 3 lines 103-109)

Comments 2: Additionally, Table 1 and line 329 provide redundant information, as they both repeat the values presented in the aforementioned table.

Response 2: Thank you for your comment. I've revised the text to enhance clarity and highlight the trend observed in the table.

(Page 11 line 413)

 

Comments 3: Moreover, line 463 contains redundancy with 'observed observations'

Response 3: Thank you for your comment. We have revised the text. The paragraph has been modified to better explain about the future research improvements in discussion section

(Page 15 lines 484-503; page 15 lines 506-540)

 

Comments 4: While the referenced literature is acceptable, it is advisable to incorporate new and up-to-date sources to enhance the content of the document.

Response 4: We've incorporated additional references into the text.

 

Comments on the Quality of English Language

I line 3, it appears that the authors intend to protect or mitigate forest cover rather than reduce natural vegetation cover.

Moreover, line 463 contains redundancy with ‘observed observations’

A: Our response is as stated above.

 

 

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

I could see the manuscript was much improved. It can be accepted for publication. Congratulations to all co-authors.

Comments on the Quality of English Language

minor

Author Response

Dear Reviewer,

Thank you very much for your positive feedback and for recognizing the improvements in our manuscript. Your insightful comments and suggestions were invaluable in enhancing the quality of our work.

We appreciate your time and effort in reviewing our manuscript and are grateful for the opportunity to contribute to the field.

Sincerely,
Yudi Setiawan
On behalf of all co-authors

Reviewer 2 Report

Comments and Suggestions for Authors

1. in page 4 of 19, the expression and description of equation (1) are not consistent.

2. the font and size of the equation (2) and figure 3 are also not consistent, variables should be the italics font. And also check other figures.

Comments on the Quality of English Language

N/A

Author Response

Dear Reviewer,

We appreciate your thorough review and attention to detail, which have greatly improved our manuscript.

 

Comments and Suggestions for Authors:

 

Comment 1. In page 4 of 19, the expression and description of equation (1) are not consistent.

Response 1. Thank you for your valuable feedback. We have reviewed the expression and description of equation (1) on page 4 of 19 and acknowledge the inconsistency. We have revised the text for clarity and accuracy. Here is the updated version:

 

..where and represent the reflectance values in the shortwave infrared (SWIR) and near-infrared (NIR) regions of the electromagnetic spectrum, respectively. The SWIR range typically covers wavelengths from approximately 1.0 to 2.5 µm, while the NIR range covers wavelengths from approximately 0.7 to 1.3 µm.

(page 4, lines 146-149)

 

Comment 2. The font and size of the equation (2) and figure 3 are also not consistent, variables should be the italics font. And also check other figures.

Response 2. Thank you for pointing out the inconsistencies in the font and size of equation (2) and Figure 3. We have reviewed these elements and made the necessary adjustments to ensure consistency throughout the manuscript.

(page 5, lines 198-200; Fig. 2, Fig. 3, Fig. 4, Fig. 12, Fig. 13, Fig. 14)

Author Response File: Author Response.pdf