Mapping Shrub Biomass at 10 m Resolution by Integrating Field Measurements, Unmanned Aerial Vehicles, and Multi-Source Satellite Observations

Round 1
Reviewer 1 Report (New Reviewer)
Comments and Suggestions for AuthorsThe paper “Mapping shrub biomass at 10m resolution by integrating field measurements, unmanned aerial vehicles, and multi-source satellite observations” provided a methodology with cross-scale data to estimate the shrub biomass. The paper’s organization has improved a lot. The study will be helpful for relevant readers. I have some minor suggestions.
1. Line 38-40, the sentence is not clear, suggest rehearse it.
2. Please revise the tables with consistent format.
3. Suggest add scale in Fig. 4 and Fig. 7.
Author Response
The paper “Mapping shrub biomass at 10m resolution by integrating field measurements, unmanned aerial vehicles, and multi-source satellite observations” provided a methodology with cross-scale data to estimate the shrub biomass. The paper’s organization has improved a lot. The study will be helpful for relevant readers. I have some minor suggestions.
Response: Thanks for your positive feedback. We appreciate your comments and suggestions to improve the manuscript. Following them, we have carefully revised the text throughout the manuscript and answered the questions point-by-point. The revisions can be seen in the manuscript.
Other comments:
- Line 38-40, the sentence is not clear, suggest rehearse it.
Response: Thanks for your comments. We revised the sentence as “As an important vegetation type, shrubs play crucial roles in ecosystems [5–7], such as soil conservation, sand fixation, habitat maintaining and so on.” in Lines of 58-62, Page 2.
- Please revise the tables with consistent format.
Response: The format of the tables has been revised following your suggestions.
- Suggest add scale in Fig. 4 and Fig. 7.
Response: We added scale bar in Fig.4 and Fig.7. Here is the revised version.
Figure 4. (a) displays the shrublands and other land-cover types of Helan Mountain, China in 2023. (b-i) represents the zoom-in views of four example regions in the resultant map and the Google Earth images.
Figure 7. (a) represents the estimated distribution map of shrub biomass in the Helan Mountains. (b) displays the corresponding map of standard deviation (SD). (c) illustrates the distribution of EOPC within different ranges of shrub coverage. (d) displays the distribution of EOUB within different ranges of shrub biomass. These analyses were conducted based on the estimated distribution map of shrub biomass and the corresponding map of standard deviation. EOPC denotes the error of one percent coverage of shrub, calculated by SD/mean shrub coverage. EOUB denotes the error of one unit biomass, calculated by SD/mean shrub biomass.
Author Response File: Author Response.docx
Reviewer 2 Report (New Reviewer)
Comments and Suggestions for Authors1. Clarity of Objectives and Methodology:
Ensure the abstract clearly states the objective of the study: "This study aimed to develop a framework for estimating shrub biomass at a 10m spatial resolution."
Emphasize the integration of different data sources: "We integrated field measurements, UAV data, Landsat, and Sentinel-1/2 observations to achieve this goal."
2.Choose keywords that accurately reflect the content and focus of the study: "Consider using keywords such as Remote Sensing, Environmental Monitoring, and Vegetation Mapping alongside existing ones."
3.Results and Findings:
Provide specific results to highlight the study's achievements: "The developed regional-scale prediction model achieved an R2 of 0.62, demonstrating superior performance compared to models using only spectral bands (R2=0.33) or vegetation indices (R2=0.55)."
Mention the practical implications of the findings: "The average uncertainty of biomass estimates was below 4%, with the lowest values (<2%) observed in regions with high shrub coverage (>30%) and biomass production (>300 g/m2)."
4.Significance and Contribution:
Clearly state the significance of the study
Highlight the broader implications
5.Technical Accuracy and Approach:
Ensure technical terms are well-defined or explained: "The Random Forest Regression (RFR) approach was employed for developing the regional-scale prediction model."
Mention the specific methodology used for uncertainty assessment: "Uncertainty in biomass estimates was characterized using standard deviation (SD) mapping via Leave-One-Out Cross-Validation (LOOCV)."
6. In terms of dataset construction, land cover classification, model selection and development, biomass distribution map generation, and spatial distribution analysis, the current study exhibits notable deficiencies in information presentation. Specifically, the dataset construction lacks detailed descriptions of data collection methods, sample selection criteria, and data quantity distribution, which undermines the assessment of the dataset's representativeness and reliability. The land cover classification process lacks clarity in classification methods, algorithms used, and accuracy evaluation, making it difficult to comprehend the extraction process of shrubland distribution. The model selection and development section fails to exhibit the specific comparison process and results, nor does it explicitly state the model selection criteria and evaluation metrics, resulting in insufficient justification for the chosen model. The generation of the biomass distribution map lacks technical details such as geographical scope, resolution, and data processing procedures, impacting the validation of the results. Lastly, the spatial distribution analysis of biomass lacks in-depth analysis content and discussion of its correlation with environmental factors, limiting the understanding of ecosystem functions and biodiversity conservation. These issues necessitate improvement in subsequent research efforts.
7. Conclusion
In the conclusion section, further emphasis can be placed on the connection between this study and the current research background as well as previous works. It should be clearly stated how this study fills the gaps in existing research and the potential contributions it makes to understanding regional ecosystem functions, biodiversity conservation, or climate change responses. Additionally, the limitations of this study should be explicitly pointed out in the conclusion, such as limitations in sample size, insufficiencies in data resolution, and incomplete consideration of environmental factors. Lastly, directions and suggestions for future research should be proposed to further refine and expand upon the current study.
Author Response
Reviewer #2 (Remarks to the Authors):
1.Clarity of Objectives and Methodology:
Ensure the abstract clearly states the objective of the study: "This study aimed to develop a framework for estimating shrub biomass at a 10m spatial resolution."
Emphasize the integration of different data sources: "We integrated field measurements, UAV data, Landsat, and Sentinel-1/2 observations to achieve this goal."
Response: We appreciate your suggestions. We clarified the objectives and methodology throughout the manuscript. In addition, we also revised the text throughout the manuscript to make it clear.
For example, the objective was clarified in Lines of 20-22, Page 1 in Abstract “This study aimed to develop a framework for estimating shrub biomass at a 10m spatial resolution by integrating ground measurements, unmanned aerial vehicle (UAV), Landsat, and Sentinel-1/2 observations” and in Lines of 161-163, Page 4 in Introduction “This work aimed to propose a framework to estimate shrub biomass over a mountain region by integrating multi-scale data from field measurements, UAV-based biomass estimation, and satellite observations (Sentinel-1/2, Landsat). ”
The methodology was clarified in the abstract in Lines of 23-25, Page 1 in Abstract “The UAV was used as a novel approach to increase the input samples for simulating shrub biomass based on satellite images.” In Lines of 49-51, Pages 1-2 “This study provides a methodology to accurately monitor the biomass of shrub at the satellite scale by enlarging training samples of shrub biomass with the observations of near-ground UAV as well as the ground measurements.”
2.Choose keywords that accurately reflect the content and focus of the study: "Consider using keywords such as Remote Sensing, Environmental Monitoring, and Vegetation Mapping alongside existing ones."
Response: Thanks for your suggestions. Following your comments, we revised the keywords as “Remote sensing; Vegetation mapping; Environmental Monitoring; UAV; Land cover”
3.Results and Findings:
Provide specific results to highlight the study's achievements: "The developed regional-scale prediction model achieved an R2 of 0.62, demonstrating superior performance compared to models using only spectral bands (R2=0.33) or vegetation indices (R2=0.55)."
Mention the practical implications of the findings: "The average uncertainty of biomass estimates was below 4%, with the lowest values (<2%) observed in regions with high shrub coverage (>30%) and biomass production (>300 g/m2)."
Response: Thanks for your suggestions. The results were revised in the Abstract in Lines of 41-44, page 1 as “The results suggested that the shrub biomass model driven by satellite spectral bands and vegetation indices (R2= 0.62) had superior performance than that only driven by spectral bands (R2=0.33) or vegetation indices (R2=0.55).”
Detail descriptions were added into the results as “The results suggested that the SBVI model achieved the best performance with R2 of 0.62, which was higher than that of the SB model (R2=0.33) and VI model (R2=0.55).” in Lines of 460-464, page 15.
The uncertainty of the method was assessed and the detail description was revised in the results in Lines of 475-485, pages 15-16 as “The accuracy was assessed based on shrub biomass samples across different groups of shrub coverages and biomass (Fig. 6a). The ratio of RMSE to mean shrub coverage was calculated to present the error of one percent coverage of shrub (EOPC) [93]. EOPC decreases with the increase of shrub coverage. Similarly, the ratio of RMSE to mean biomass was calculated to present the error of one unit biomass (EOUB) (Fig. 6b). EOUB decreases along with the increase of shrub biomass. The lowest values of R2 were observed within the range of 20-30% shrub coverage (R2=0.3) and 200-300 g/m2 shrub biomass (R2=0.19). These results indicate the estimation accuracy is relatively lower in the moderate shrub coverage and biomass ranges than that in the high and low levels.”
The implication of the findings was revised in the abstract as “This study provides a methodology to accurately monitor the biomass of shrub at the satellite scale by enlarging training samples of shrub biomass with the observations of near-ground UAV as well as the ground measurements.” in Lines of 47-52, Pages 1-2.
4.Significance and Contribution:
Clearly state the significance of the study.
Highlight the broader implications.
Response: Thanks for your comments. We revised the Introduction to clarify the significance and contribution of this study in Lines of 56-186, Pages 2-4.
We described the significance and contributions from four aspects of (1) the importance of accurately monitoring the shrub biomass at the regional scale by remote sensing technology; (2) the critical roles of selecting optimal remote sensing features on improving the accuracy of shrub biomass estimation by remote sensing; (3) the critical roles of selecting optimal algorithms on improving the estimation of shrub biomass; (4) the reasons of UAV used to improve the estimation of shrub biomass.
We also highlighted that the proposed methodology has the potential to improve the shrub biomass estimation accurately at the regional scale in Lines of 182-186, Page 4 as “This study provided a methodology to accurately estimate the biomass of shrubs at the satellite scale by incorporating near-ground UAV observations. This approach increases the training samples of shrub biomass and addresses the scale mismatch between ground measurements and satellite observations.”
5.Technical Accuracy and Approach:
Ensure technical terms are well-defined or explained: "The Random Forest Regression (RFR) approach was employed for developing the regional-scale prediction model."
Mention the specific methodology used for uncertainty assessment: "Uncertainty in biomass estimates was characterized using standard deviation (SD) mapping via Leave-One-Out Cross-Validation (LOOCV)."
Response: We appreciated your suggestions. The technical terms are defined in the manuscript. In addition, the methods have been revised to make it clear to read. Please see the text in Lines of 384-427, Pages 12-13.
6.In terms of dataset construction, land cover classification, model selection and development, biomass distribution map generation, and spatial distribution analysis, the current study exhibits notable deficiencies in information presentation. Specifically, the dataset construction lacks detailed descriptions of data collection methods, sample selection criteria, and data quantity distribution, which undermines the assessment of the dataset's representativeness and reliability. The land cover classification process lacks clarity in classification methods, algorithms used, and accuracy evaluation, making it difficult to comprehend the extraction process of shrubland distribution. The model selection and development section fails to exhibit the specific comparison process and results, nor does it explicitly state the model selection criteria and evaluation metrics, resulting in insufficient justification for the chosen model. The generation of the biomass distribution map lacks technical details such as geographical scope, resolution, and data processing procedures, impacting the validation of the results. Lastly, the spatial distribution analysis of biomass lacks in-depth analysis content and discussion of its correlation with environmental factors, limiting the understanding of ecosystem functions and biodiversity conservation. These issues necessitate improvement in subsequent research efforts.
Response: We appreciate your comments and suggestions. To make the manuscript clear, we revised the method section totally. In this revision, we added detail information of data collection, sample selection criteria in Lines of 241-264, Pages 7-8. The land cover classification method was clarified in Lines of 302-329, Pages 10-11. The model selection and development were revised in Lines of 383-413, Pages 12-13.
We added more information to describe the generation of the biomass distribution map in Lines of 414-427, Page 13. For the distribution analysis, we clarified that our analyses were conducted along with the gradients of different environment factors in Lien of 435, Page 13. This helps to understand the spatial distribution patterns of the shrub biomass in our study area.
7.Conclusion
In the conclusion section, further emphasis can be placed on the connection between this study and the current research background as well as previous works. It should be clearly stated how this study fills the gaps in existing research and the potential contributions it makes to understanding regional ecosystem functions, biodiversity conservation, or climate change responses. Additionally, the limitations of this study should be explicitly pointed out in the conclusion, such as limitations in sample size, insufficiencies in data resolution, and incomplete consideration of environmental factors. Lastly, directions and suggestions for future research should be proposed to further refine and expand upon the current study.
Response: Thanks for your comments. Following your suggestions, we revised the conclusions in Lines of 620-645, Page 22, as “This study proposed a novel approach to improve the estimation of shrub biomass at a regional scale through increasing the shrub biomass samples and reducing the spatial scale mismatch between ground measurements and satellite observations by UAV. The workflow included the following parts: first, a shrubland distribution map was generated for the Helan Mountains, China. Subsequently, the shrub biomass was obtained at the UAV level based on the UAV images and the allometric growth equation for the shrub biomass fitted using the field measurements. Then, the best shrub biomass estimation model at the satellite level was obtained by comparing three RFR models driven by different remote sensing features. This model was used to generate the distribution map of shrub biomass in the Helan Mountains in 2023. Based on the resultant map, we assessed the distribution of shrub biomass along with different environmental factors. The results indicate that using UAV imagery to calculate shrub biomass provides a more convenient and effective method to supplement shrub biomass samples, significantly reducing the workload and costs of fieldwork. Combining remote sensing imagery from different sensors will likely provide more assistance for future research on large-scale shrub biomass estimation. Although this study made some progress in the method, there are still limitations in sample size, data resolution, and incomplete environmental factors considered. In the following works, we will expand this work in other regions to validate and improve the proposed methodology. ”
Author Response File: Author Response.docx
Round 2
Reviewer 2 Report (New Reviewer)
Comments and Suggestions for AuthorsThe article has indeed undergone significant improvements following the major revisions, yet the introduction and discussion sections remain somewhat lacking in depth. Consequently, I would urge the author to revisit and revise these parts thoroughly. Additionally, to enhance the overall quality and competitiveness of the submission, I recommend strengthening the innovative aspects of the research methodology, given that this journal is a renowned international publication. With these further modifications in place, I suggest accepting the article for publication.
Author Response
We appreciate your comments and suggestions. To further improve the introduction and discussion, we revised the introduction and discussion section totally. In this revision, we introduced the study gaps in previous works, clarified the issues still needed addressed, added detail information in the Introduction.
The discussion was revised detailly throughout the manuscript.
The innovative aspects of the research methodology were strengthened in the manuscript in Abstract, Introduction, Method, and Discussion.
Author Response File: Author Response.docx
Round 3
Reviewer 2 Report (New Reviewer)
Comments and Suggestions for AuthorsAfter revisions, this article has improved significantly. I suggest accepting it after enhancing the innovativeness of the methods.
Author Response
We appreciate your comments and suggestions. To further enhance the innovativeness, we revised the abstract, introduction, methods, discussion and conclusions section totally.
We have placed a strong emphasis on revising the Methods section to highlight the innovative aspects of our approach.
Author Response File: Author Response.docx
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
Comments and Suggestions for AuthorsDear authors, while I appreciate the effort put into the study and importance of the topic, I regret that I do not feel that your paper provides the kind of sufficiently novel conceptual insights and conclusions. There are many important issues that are missing in your write-up in terms of the depth of your presentation, poor quality of the English language, rigour of the validation process, novelty of your work, and its potential application to other case studies. Therefore, I feel that the paper, as it stands, would not easily find a suitable publication in another SCI journal without significant revisions.
There are several major issues that need to be addressed and modified:
1. Title. It is risky to use "High-resolution" before you clearly redefine it, as your definition does not align with the understanding of most readers. Furthermore, "arid and semiarid environments" is a broad concept, and your study area may only represent a specific case rather than the entire scope.
2. After reading your abstract and content structure, to be honest, it seems a bit disorganized. I hope the authors can clarify the specific scientific question that the current paper aims to address. Your paper broadly covers "Mapping of shrubland," "Model of biomass," and "Characteristic of distribution," but where is the focus of your research? The depth of your research is also insufficient.
3. There is significant heterogeneity among the multisource remote sensing data, and the authors have not made any targeted efforts to address this, resulting in considerable uncertainty in the results. For example, there are inconsistencies in satellite observation angles, central wavelengths of bands, spectral response functions of sensors, and spatial resolutions between Sentinel-2 and Landsat data. How can these be unified? Additionally, Landsat satellites have different imaging parameters for different sensors (ETM+ and OLI), especially in terms of spectral and radiometric resolutions. How can this be addressed? Furthermore, physical quantities from Sentinel-2 and Sentinel-1 are not consistent. How can they be directly applied?
4. Regarding other data, you chose ESA products and MODIS products for "Land cover data." These two have many differences, such as spatial resolutions, and their release dates are before 2023. How can they be directly applied to the current research? The spatial resolutions of precipitation and temperature data exceed 11km, so how can they be applied to a 10m classification result in a study area of only 2000km2? Additionally, there are too few validation points, which compromises the rigor of the results.
5. What is the innovation of the methodology in this paper? Are there any new theories or insights? In reality, the R2 of the machine learning models is not high and does not demonstrate a clear advantage compared to traditional satellite or drone-based studies.
Comments on the Quality of English LanguageThe paper requires extensive editing to improve its English language usage, as it is currently not written particularly well. While I understand the difficulty of writing in a non-native tongue, I do appreciate the effort put forth. With that said, I recommend enlisting the expertise of a native speaker to review the paper for any grammatical errors or lengthy sentences. This would not only ensure accuracy and clarity but also enhance the overall quality of the written work.
Author Response
Dear reviewers,
Thank you very much for your time and effort in reviewing this manuscript. Your constructive comments on the manuscript are useful. We have carefully addressed all your concerns in this revised version. The revisions are shown using line and page numbers following the tracking version in this response letter.
Comments from the Editorial Board Member and Reviewers:
Reviewer #1 (Remarks to the Authors):
Dear authors, while I appreciate the effort put into the study and importance of the topic, I regret that I do not feel that your paper provides the kind of sufficiently novel conceptual insights and conclusions. There are many important issues that are missing in your write-up in terms of the depth of your presentation, poor quality of the English language, rigour of the validation process, novelty of your work, and its potential application to other case studies. Therefore, I feel that the paper, as it stands, would not easily find a suitable publication in another SCI journal without significant revisions. There are several major issues that need to be addressed and modified:
Response: We appreciate your comments and suggestions to improve this manuscript. We make a revision throughout the manuscript including the focus, writing quality, research novelty, and overall organization. The revisions are labelled in the manuscript.
- Title. It is risky to use "High-resolution" before you clearly redefine it, as your definition does not align with the understanding of most readers. Furthermore, "arid and semiarid environments" is a broad concept, and your study area may only represent a specific case rather than the entire scope.
Response: We appreciate your comments. We revised the “High resolution” as “at 10-m resolution” to clearly describe our product. In addition, our research did not cover the whole semi-arid and arid areas, but it is a case study in the arid and semi-arid environments. Therefore, we revised the title as “Mapping shrub biomass at 10m resolution in arid and semi-arid environments by integrating remote sensing observations across multiple spatial scales”.
- After reading your abstract and content structure, to be honest, it seems a bit disorganized. I hope the authors can clarify the specific scientific question that the current paper aims to address. Your paper broadly covers "Mapping of shrubland," "Model of biomass," and "Characteristic of distribution," but where is the focus of your research? The depth of your research is also insufficient.
Response: We appreciate your comments. Our research mainly focused on the accurate estimation of biomass in shrublands. The workflow includes three parts of “Mapping of shrubland”, “Model of biomass”, and “Characteristic of distribution”. The mapping of shrubland was to exclude non-shrubland areas, reduce unnecessary error, which provide a basis for the model of biomass. The spatial distribution analysis was an application of the resultant biomass map, which suggested the application potential of the developed working framework for local carbon budget and ecological management.
- There is significant heterogeneity among the multisource remote sensing data, and the authors have not made any targeted efforts to address this, resulting in considerable uncertainty in the results. For example, there are inconsistencies in satellite observation angles, central wavelengths of bands, spectral response functions of sensors, and spatial resolutions between Sentinel-2 and Landsat data. How can these be unified? Additionally, Landsat satellites have different imaging parameters for different sensors (ETM+ and OLI), especially in terms of spectral and radiometric resolutions. How can this be addressed? Furthermore, physical quantities from Sentinel-2 and Sentinel-1 are not consistent. How can they be directly applied?
Response: Thanks for your comments. According to some differences in bandwidth between MSI, OLI, and ETM+ sensors, it is necessary to match the band reflectance values among the three sensors to construct reliable time series. We used the ordinary least squares regression method proposed by Roy (2016) to match the band reflectance of the three sensors. The spatial resolutions of Sentinel-1 and Sentinel-2 are consistent. Our model was developed at the pixel scale, with the two data types serving as independent training features, without fusing the two datasets. Instead, the fusion was carried out between Landsat and Sentinel-2 bands, with Sentinel-1 data serving as supplementary information. We used the optical and radar bands separately into our biomass estimation model. In this revision, we clarified the text to make the descriptions clear in Lines of 156-159, Page 4.
References:
Roy, D.P., Kovalskyy, V., Zhang, H.K., Vermote, E.F., Yan, L., Kumar, S.S., Egorov, A., 2016. Characterization of Landsat-7 to Landsat-8 reflective wavelength and normalized difference vegetation index continuity. Remote Sensing of Environment, Landsat 8 Science Results 185, 57–70. https://doi.org/10.1016/j.rse.2015.12.024
- Regarding other data, you chose ESA products and MODIS products for "Land cover data." These two have many differences, such as spatial resolutions, and their release dates are before 2023. How can they be directly applied to the current research? The spatial resolutions of precipitation and temperature data exceed 11km, so how can they be applied to a 10m classification result in a study area of only 2000km2? Additionally, there are too few validation points, which compromises the rigor of the results.
Response: We appreciated your comments and suggestions! The land use product of ESA and MODIS was used separately. The ESA product was used as a reference of the land use types at the study area. Through the incorporation of Google Earth imagery and ground-based photographs, we were able to acquire land cover samples accurately. Additionally, we utilized the shrubland definitions within the MODIS dataset as the basis for delineating the shrubland extents within our study area.
Furthermore, despite the low spatial resolutions of the precipitation and temperature datasets, we resampled the 10m results to match their spatial scales to enable the corresponding analyses.
Lastly, we acknowledge the limited number of sample points in our study. To mitigate this constraint, we employed a leave-one-out cross-validation approach, which ensured that all input data could be leveraged as validation datasets.
- What is the innovation of the methodology in this paper? Are there any new theories or insights? In reality, the R2 of the machine learning models is not high and does not demonstrate a clear advantage compared to traditional satellite or drone-based studies.
Response: Thanks for your comments. Our research represents the first large-scale effort to estimate shrubland biomass within arid and semi-arid environments. The focal point of our study was the acquisition of a set of biomass sample points through unmanned aerial vehicle (UAV) imagery, which we utilized to construct our biomass estimation model. Given the significantly coarser spatial resolutions of satellite imagery compared to UAV data, and the relatively small and sparsely distributed nature of shrubs, it has been challenging to reliably capture accurate shrubland biomass information from satellite data alone. Our research aims to provide a novel approach that integrates UAV and satellite imagery, whereby the supplementary sample points can augment the use of satellite data for upscaling shrubland biomass estimation across broader extents.
Comments on the Quality of English Language
The paper requires extensive editing to improve its English language usage, as it is currently not written particularly well. While I understand the difficulty of writing in a non-native tongue, I do appreciate the effort put forth. With that said, I recommend enlisting the expertise of a native speaker to review the paper for any grammatical errors or lengthy sentences. This would not only ensure accuracy and clarity but also enhance the overall quality of the written work.
Response: Thanks for your comments. We refined our manuscript in this revision. Your comments have been highly constructive.
Author Response File: Author Response.docx
Reviewer 2 Report
Comments and Suggestions for AuthorsTitle: High-resolution mapping of shrub biomass in arid and semi-arid environments by integrating remote sensing observations across multiple spatial scales
Overview:
In this study, authors used a range of image data and methods to map shrub biomass in the Helan Mountains, China. It is well written and clearly explains the methods, results, etc. I only have minor comments below.
Other comments:
Line 63: If anything I would characterize bands of Landsat and Sentinel as wide, rather than narrow. If detailed spectral characteristics are needed, maybe hyperspectral data would be beneficial?
Table 1: The equations and names don’t quite line up in some places as far as I can tell
Line 188-189 is not a complete sentence.
Line 218 I don’t think I would use “well-liked” here. Maybe something like “widely used”?
Figure 4 I realize the study area has been shown before, but there should still be at least a scale bar here. Check journal guidelines.
Author Response
Dear reviewers,
Thank you very much for your time and effort in reviewing this manuscript. Your constructive comments on the manuscript are useful. We have carefully addressed all your concerns in this revised version. The revisions are shown using line and page numbers following the tracking version in this response letter.
Comments from the Editorial Board Member and Reviewers:
Reviewer #2 (Remarks to the Authors):
In this study, authors used a range of image data and methods to map shrub biomass in the Helan Mountains, China. It is well written and clearly explains the methods, results, etc. I only have minor comments below.
Response: Thanks for your positive feedback. We appreciate your comments and suggestions to improve this manuscript. Following them, we have carefully revised the manuscript point-by-point.
Other comments:
1.Line 63: If anything I would characterize bands of Landsat and Sentinel as wide, rather than narrow. If detailed spectral characteristics are needed, maybe hyperspectral data would be beneficial?
Response: Thanks for your comments. We replaced “narrow” with “wide” in Lines of 64, Page 2. In the future work, we are hopeful to leverage hyperspectral data to improve shrubland biomass estimation models.
2.Table 1: The equations and names don’t quite line up in some places as far as I can tell.
Response: Following your suggestions, we revised the Table 1 in detail in Page 6. The revisions can be seen throughout the manuscript.
3.Line 188-189 is not a complete sentence.
Response: Thanks for your meaningful comments. We revised the sentence in Line 188-189 as “Relevant climate and terrain datasets, including precipitation, temperature, aridity index, and elevation datasets were employed for our subsequent analyses.”
4.Line 218 I don’t think I would use “well-liked” here. Maybe something like “widely used”?
Response: Thanks for your comments. We revised the “well-liked” as “widely used” in Line of 226, Page 9.
5.Figure 4 I realize the study area has been shown before, but there should still be at least a scale bar here. Check journal guidelines.
Response: Thanks for your comments. We checked the journal requirements and added a scale bar to Figure 4. Here is the revised figure:
Figure 4. (a) displays the shrublands and other land-cover types of Helan Mountain, China in 2023. (b-i) represents the zoom-in views of four example regions in the resultant map and the Google Earth images.
Author Response File: Author Response.docx
Reviewer 3 Report
Comments and Suggestions for AuthorsThis paper used various remote sensing images to estimate shrub biomass, and created its biomass distribution map in the Helan Mountains using the most accurate model. I am very interested in the research content of this paper, but there are certain problems with this paper. Therefore, I suggest a major revision of the manuscript before considering whether to publish it in the journal "Remote Sensing". The specific comments are as follows:
Point 1: In this paper, authors used active and passive remote sensing images, and discussed the influence of the optimal feature combination on the estimation accuracy. The introduction section should include a description of the current variable selection methods for different feature combinations. I think authors should add this description, as following: "Combination of multiple feature variables has the problem of data redundancy and complex computation, and the optimization of multidimensional features is important for improving model performance." And cite the paper as follow:
[1] https://doi.org/10.1016/j.jag.2023.103446
Point 2: Lines 103-119, please precisely summarize the information about this study.
Point 3: Lines 166-168, how to obtain the ground in-suit photos in this study?
Point 4: Lines 179-181, this study established 24 sampling plots, but there are 13 samples in the Table 2. Why are the numbers of samples different?
Point 5: Lines 183-186. There are several remote sensing images in this manuscript, but different remote sensing images do not have geographical registration.
Point 6: Lines 187-202. Is this study suitable for using the global climate and topographic data?
Point 7: Lines 318-319, “We have identified four land cover types: forest, barren land, grasslands, and shrublands.” Why is it divided into four categories? How to deal with the misclassification of forests and shrubs?
Point 8: Lines 333-353, the manuscript discussed models constructed from three different variable sets, and selected the best model. The results shown that SB model achieved the worst performance, and the best variables for SB model isn’t contained VV and VH. However, the best variables for SBVI model contained VV and VH. Why not combine the best variables of SB model and VI model as variables for SBVI model?
Point 9: Lines 393-394, “shrub biomass and the aridity index show a positive link up until the index hits 0.2, at which point the two show a negative correlation.” According to Fig. 8c, shrub biomass should be negatively correlated with drought index. Please explain in detail the relationship between biomass and drought index.
Comments on the Quality of English LanguageModerate editing of English language required
Author Response
Dear reviewers,
Thank you very much for your time and effort in reviewing this manuscript. Your constructive comments on the manuscript are useful. We have carefully addressed all your concerns in this revised version. The revisions are shown using line and page numbers following the tracking version in this response letter.
Comments from the Editorial Board Member and Reviewers:
Reviewer #3 (Remarks to the Authors):
This paper used various remote sensing images to estimate shrub biomass, and created its biomass distribution map in the Helan Mountains using the most accurate model. I am very interested in the research content of this paper, but there are certain problems with this paper. Therefore, I suggest a major revision of the manuscript before considering whether to publish it in the journal "Remote Sensing". The specific comments are as follows:
Response: Thank you for your constructive comments to improve the manuscript. We have carefully addressed your comments in this revised version.
Point 1: In this paper, authors used active and passive remote sensing images, and discussed the influence of the optimal feature combination on the estimation accuracy. The introduction section should include a description of the current variable selection methods for different feature combinations. I think authors should add this description, as following: "Combination of multiple feature variables has the problem of data redundancy and complex computation, and the optimization of multidimensional features is important for improving model performance." And cite the paper as follow:
[1] https://doi.org/10.1016/j.jag.2023.103446
Response: Thanks for your comments. The description has been added in Lines of 292-294, Page 10. The suggested paper has been cited in lines 294, Page 10.
Point 2: Lines 103-119, please precisely summarize the information about this study.
Response: Thanks for your comments. We used more precise sentences to summarize our study. The revisions can be seen in Lines 104-120, Page 3, as “ This work aimed to propose a framework to estimate shrubland biomass over an arid and semi-arid mountain region based on multi-scale data from field measurements, UAV, Sentinel-1, Sentinel-2 and Landsat observations. The Helan Mountains in Ningxia province, China, were selected as the study area. Firstly, we conducted a land cover classification to identify the shrublands and other land cover types in study area. Secondly, the prediction model of shrub biomass was developed in a Random Forest Regression (RFR) approach driven by different predicted variable datasets based on field measurements, UAV, and satellite images. The field measurement data was used to establish the allometric growth equation between shrub biomass and shrub structure parameters. Using the allometric equation, the shrub biomass was determined in UAV data. Using the UAV-based shrub biomass and field measurement data as inputs, the optimal satellite-based biomass estimation model was developed by comparing different predictor variables from Landsat, Sentinel-1, and Sentinel-2 satellites. Thirdly, with the best model, we created a map of the biomass distribution of shrubland in the Helan Mountains. The accuracy of the resultant map was evaluated over various ranges of shrub biomass or shrub coverage. Finally, we assessed the spatial characteristics of shrubland biomass based on the resultant biomass map and the auxiliary datasets of climate and topography.”.
Point 3: Lines 166-168, how to obtain the ground in-suit photos in this study?
Response: Thanks for your comments. As part of our research, we collected extensive ground in-situ photos within the Helan Mountain region during April, July, and August of 2023.We added this point in Lines 174-175, Page 6.
Point 4: Lines 179-181, this study established 24 sampling plots, but there are 13 samples in the Table 2. Why are the numbers of samples different?
Response: Thanks for your comments. We developed an allometric equation relating shrub volume to biomass using the 13 shrub samples detailed in Table 2. We then applied this equation to calculate the total shrubland biomass within the 24 sample plots. We clarified the sample information in the text in lines 181, Page 6.
Point 5: Lines 183-186. There are several remote sensing images in this manuscript, but different remote sensing images do not have geographical registration.
Response: Thanks for your comments. We checked the journal requirements and corrected the Figure 1.
Figure 1. (a, b) Locations of the Helan Mountain in China and Ningxia province, and (c) the distribution of ground truth samples from field measurements, UAV, and visual interpretation.
Point 6: Lines 187-202. Is this study suitable for using the global climate and topographic data?
Response: Thanks for your comments. As different shrub species and their associated biophysical characteristics are often influenced by geographic factors, climatic conditions, and other environmental variables, shrubs may have divergent spectral, vegetation index, and radar signals. Therefore, further research is required to examine the generalizability and scalability of our approach for broad-scale, global applications.
Point 7: Lines 318-319, “We have identified four land cover types: forest, barren land, grasslands, and shrublands.” Why is it divided into four categories? How to deal with the misclassification of forests and shrubs?
Response: Thanks for your comments. Based on our extensive field surveys conducted in 2023, the majority of the study domain comprises four primary land cover types. With respect to the potential for confusion between shrubland and forest classes, our results indicate that shrubs were more often misclassified as bare soil rather than forest cover. However, the overall classification accuracies were deemed acceptable, and this source of error was deemed negligible for the purposes of our analysis. We added a confusion matrix as Fig. S1 to show the accuracy of the map.
Fig. S1 Confusion matrix for land use classification results
Point 8: Lines 333-353, the manuscript discussed models constructed from three different variable sets, and selected the best model. The results shown that SB model achieved the worst performance, and the best variables for SB model isn’t contained VV and VH. However, the best variables for SBVI model contained VV and VH. Why not combine the best variables of SB model and VI model as variables for SBVI model?
Response: Thanks for your comments. In our prior research efforts, we did explore a similar technique, but found that the SBVI model ultimately exhibited superior performance in our context. Furthermore, for each of the candidate models, we employed the RFECV module in Python to rigorously optimize the feature combinations - an approach that has been widely adopted and validated across numerous studies, demonstrating robust feasibility.
Point 9: Lines 393-394, “shrub biomass and the aridity index show a positive link up until the index hits 0.2, at which point the two show a negative correlation.” According to Fig. 8c, shrub biomass should be negatively correlated with drought index. Please explain in detail the relationship between biomass and drought index.
Response: Thanks for your comments. We are revised the statement from "shrub biomass and the aridity index show a positive link up until the index hits 0.2, at which point the two show a negative correlation" with "shrub biomass and the aridity index show a negative correlation."
Comments on the Quality of English Language: Moderate editing of English language required
Response: Thanks for your comments. We refined our manuscript and engaged native-speaker proofreading to further enhance the accuracy and clarity of our work. Your comments have been highly constructive.
Author Response File: Author Response.docx
Reviewer 4 Report
Comments and Suggestions for Authorssee attached file
Comments for author File: Comments.pdf
should be improved
Author Response
Dear reviewers,
Thank you very much for your time and effort in reviewing this manuscript. Your constructive comments on the manuscript are useful. We have carefully addressed all your concerns in this revised version. The revisions are shown using line and page numbers following the tracking version in this response letter.
Comments from the Editorial Board Member and Reviewers:
Reviewer #4 (Remarks to the Authors):
I reviewed the manuscript remotesensing-2870501 entitled « High-resolution mapping of shrub biomass in arid and semiarid environments by integrating remote sensing observations across multiple spatial scales». This paper aims at presenting a framework to estimate shrub biomass with multi-source remote sensing data. The topic of this manuscript has a great significance for accurately estimating shrub biomass. This paper has a complete structure and sufficient results, however, there are several aspects should be considered before potential publication from my perspective:
Response: Thank you for the chance to revise the manuscript. We appreciate your comments and suggestions to improve this manuscript. Following them, we have carefully revised the manuscript point-by-point.
- Line 131, does “height” mean altitude? In that case, I suggest that height should be
modified as altitude.
Response: Thanks for your comments. We agree to replace “height” to “altitude”.
- I think the format and quality of all Tables and all Figures need be improved followed the standard of journal. There should be some space between table and text.
Response: Thanks for your suggestions. We checked the journal requirements and corrected. The revisions can be seen throughout the manuscript.
- In section 2.2.3, the author set different samples for the research purposes, I suggest that the author should have a detail description to avoid confusing the reader. Line 175, “thirteen shurbs….”, Line 179, “24 sampling plots…..”. Thirteen shrubs should be expressed as Arabic numerals.
Response: We appreciate your comments. We should use more easily comprehensible language to represent the two sampling results, and replace "thirteen" with "13" in Lines of 181, Page 6.
- Table 2, the number of statistical plots should be list.
Response: Thanks for your comments. Table 2 presents the field-sampled individual shrub samples, including the structural parameters and measured biomass data for each individual shrub specimen. Therefore, each data row represents a single shrub sample, and there is no need to introduce the total number. Furthermore, the plots contain multiple shrubs, each with its own structural parameters, as well as the total number of shrubs per plot, as you mentioned. Given the large number of shrubs within each quadrat, the volume of data that would need to be displayed becomes inconvenient to exhibit within the manuscript.
- Line 231 -233, the format of formulas should be modified.
Response: Thanks for your comments. The format of formulas has been corrected. The revisions can be seen throughout the manuscript.
- Figure3, the location of Figure3 d, e, f, g should be correspond to sample points in Figure 3 b, the author should examine carefully. The same problems also occurred in Figure 4, such as “b, c, d, e” “b, d, f, h”
Response: We appreciate your comments. We will check Figure 3. Here is the revised version:
Figure 3. (a) depicts the original unmanned aerial vehicle (UAV) image. (b) represents the classified map of shrublands. (c) illustrates the fishnet constructed based on the UAV imagery. (d-g) represent the zoomed-in views of four sample points in Fig. 3b.
- Figure 5, MAE have not shown in Figure 5. I suggest that the author should modified the histogram to scatter plots for comparing the accuracy of different models.
Response: Thanks for your comments. This was an error on our part, and we will remove the MAE from the manuscript.
.
- the samples number with different groups of shrub coverages and biomass should be added in Figure 6(a)-(b). However, to my knowledge, the scatter plots can descript more details than histograms, I suggest that the author should has some improvement in Figure 6. In Figure 6(c), the author conducted a sensitivity analysis for illustrating the contribution of four variables by the exclusion method, I have some doubts that why the author does not use the importance of variables to analysis the sensitivity of variables, which can be calculated by Random Forest Regression model.
Response: Thanks for your suggestions. Your suggestions are highly constructive. Histograms can more intuitively illustrate the classification accuracy across different ranges within a single figure, allowing for clearer comparisons of group-level differences. While scatter plots can more effectively represent the direct relationship between reference values and predicted values, thereby elucidating the potential relationship between the two variables, the presence of data points from multiple ranges on a single scatter plot can result in a cluttered and less intuitive visualization. This can make it challenging to discern the differences in classification accuracy across the various ranges at a glance. Therefore, in this case, we have opted to utilize histogram to represent the variations in classification accuracy across the different ranges. The histogram format enables a more direct and accessible presentation of the nuances in performance within each range, facilitating a clearer interpretation of the relative strengths and weaknesses of the classification model in different contexts.
However, because the feature importance ranking is unable to be thoroughly captured the individual feature's influence on model accuracy. we would prefer to utilize bar charts to represent the modeling accuracy across different ranges of coverage or biomass. In contrast, sensitivity analysis can more clearly elucidate the nuanced impacts of each feature on modeling precision.
- Figure 8, firstly, the titles of Y-axis should be added. Secondly, the violin plots contain little information for the readers, the mean value and middle value do not show in Figure 8
Response: Thanks for your comments. We included the y-axis label, as well as the median values, in Figure 8. Given the close proximity between the median and mean values, in order to maintain the aesthetic appeal of the figure, we have opted to add only the median line. In this revision, we also added the 25th and 75th percentile lines as additional reference lines. Here is the revised version:
Figure 8. (a) depicts the distribution of shrub biomass under precipitation gradients. (b) illustrates the distribution of shrub biomass under temperature gradients. (c) represents the distribution of shrub biomass within different ranges of aridity index. (d) displays the distribution of shrub biomass along elevation gradients.
- Section 4.1, algorithms. The author seems to discuss the methods in all steps together, like the integration of UAV and satellite data, the land cover classification, and the shrub biomass estimation, and this cause a logical confusion in Section 4.1. therefore, I suggest that the author divides this section into subsections.
Response: We appreciate your comments. We followed your suggestion and divided Section 4.1 into multiple subsections. The revisions can be seen throughout the manuscript.
- Line 455, the author said that the insignificant effects of temperature and elevation on shrubs compared with reference [97]. I think the study area of this manuscript is smaller than reference [97], which means the climatic factors and topography have no significant difference. Therefore, there is meaningless to compare with the results of reference [97]. I suggest the author reconsider the effect factors.
Response: Thanks for your comments. as the differing research scopes have led to our conclusions not aligning with the findings presented in that work. As such, the reference does not adequately support the results of our study.
- Section 4.2, I think the author want to discuss the influence factors of spatial characteristics of shrub biomass and the limitation of the framework for estimating shrub biomass. So, I suggest that the author should divide Section 4.2 into Section 4.2 - “the influence factors for shrub biomass spatial distribution” and Section 4.3 – “limitation”.
Response: We appreciated your comments and suggestions! We divided Section 4.2 into Section 4.2 - “the influence factors for shrub biomass spatial distribution” and Section 4.3 – “limitation”. The revisions can be seen throughout the manuscript.
In the present form, firstly, this research paper is meaningful, which has proposed a detailed framework to integrate UAV and satellite data for estimating shrub biomass, it may be of interest to other researcher with an interesting conclusion. But secondly, this manuscript should be promoted with careful revision, such as more detailed descriptions in Figures and Tables, and discussion. In short, the paper could be structured much better with a major revision is made to address these shortcomings.
Response: Thanks for your suggestions. We appreciate your comments and suggestions to improve this manuscript. We have revised manuscript following your comments. The revisions can be seen throughout the manuscript.
Author Response File: Author Response.docx
Reviewer 5 Report
Comments and Suggestions for AuthorsThe topic of the manuscript is interesting and it is well-organized, with a clear methodological approach and convincing results. Literature references are adequate.
Here are some suggestions and comments for the Authors:
- The use of the DJI Phantom 4 Pro V2.0 UAV is mentioned only in the caption of Table 2. I recommend including this information in the main text of the manuscript as well.
- In Figure 4, sample points are represented by black dots, but the characters used to indicate the zoom windows (e.g., b, c, d, etc.) do not seem to correspond to the proper map locations. Please, verify and correct this.
- It would be useful to provide information about any planned future developments related to the present study.
- Please review the manuscript for any typographical errors. For example, the correct title of Section 2.2 should be “Data” not “Date.”
Author Response
Dear reviewers,
Thank you very much for your time and effort in reviewing this manuscript. Your constructive comments on the manuscript are useful. We have carefully addressed all your concerns in this revised version. The revisions are shown using line and page numbers following the tracking version in this response letter.
Comments from the Editorial Board Member and Reviewers:
Reviewer #5 (Remarks to the Authors):
The topic of the manuscript is interesting and it is well-organized, with a clear methodological approach and convincing results. Literature references are adequate. Here are some suggestions and comments for the Authors:
Response: We appreciate your positive comments on the manuscript. Here we provide the responses point by point following your suggestions to improve the manuscript.
1.The use of the DJI Phantom 4 Pro V2.0 UAV is mentioned only in the caption of Table 2. I recommend including this information in the main text of the manuscript as well.
Response: Thanks for your comments. This content was originally intended to be included within the main text. However, due to formatting issues, it was erroneously assigned as the title for Table 2. We have now corrected it in manuscript.
2.In Figure 4, sample points are represented by black dots, but the characters used to indicate the zoom windows (e.g., b, c, d, etc.) do not seem to correspond to the proper map locations. Please, verify and correct this.
Response: We appreciate your comments. We have checked and corrected Figure 3. Here is the revised version:
Figure 3. (a) depicts the original unmanned aerial vehicle (UAV) image. (b) represents the classified map of shrublands. (c) illustrates the fishnet constructed based on the UAV imagery. (d-g) represent the zoomed-in views of four sample points in Fig. 3b.
3.It would be useful to provide information about any planned future developments related to the present study.
Response: Thanks for your comments. We included relevant information about the future development of any related plans in the Discussion of Limitation in Lines of 480-485 Page 17-18 as “We hope to utilize a greater diversity of data sources in the future, such as UAV imagery, field sampling data, or high-spectral data with stronger correlations to shrub biomass. By incorporating these additional data inputs, we aim to develop shrub biomass models that can be more broadly applicable across different climatic environments and scalable to larger geographic regions”..
4.Please review the manuscript for any typographical errors. For example, the correct title of Section 2.2 should be “Data” not “Date.”
Response: Thanks for your suggestions. We corrected these errors in our manuscript. The revisions can be seen throughout the manuscript.
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsDear authors, thank you for considering my feedback and making some revisions. However, I regret to inform you that I believe your modifications did not effectively address key aspects of this study, including research objectives, methods, results, and conclusions. I still find several significant issues in your paper that need to be addressed:
1. Firstly, regarding the title, you used "arid and semi-arid environments," but arid and non-arid areas are broad concepts. Is Helan Mountain a typical representative of arid and non-arid environments? Additionally, it seems that you did not highlight the applicability of Helan Mountain as a case study in your paper, especially in the discussion and conclusion sections where you did not mention the model's suitability. Furthermore, if the model in this study is only applicable to regions similar to Helan Mountain, I suggest refraining from using the term "arid and semi-arid environments" in the title, as it could easily lead to misunderstandings.
2. After carefully reading the entire article, I did not see the innovation and theoretical depth of this study reflected in the research methods and results. The paper combines multi-scale data such as drones and remote sensing images and uses machine learning to construct a model for estimating shrub biomass. However, the model's R2 is not high, and compared to traditional satellite or drone studies, it does not demonstrate significant advantages. Thus, it appears that this study lacks practical value.
3. The research methods in the paper also lack scientific rigor. For instance, the study area chosen covers approximately 2,100 square kilometers, but the spatial resolution of precipitation and temperature data is less than 11 km, while the classification result resolution is 10m. The significant disparity in spatial resolution between the two datasets renders direct application through resampling unreasonable. Therefore, I recommend selecting more suitable data for this study to eliminate the impact of spatial resolution differences on the research results.
4. There are also issues with the research data in the paper. Firstly, for the entire study area, it seems that drone image data was only collected at one location, resulting in apparent randomness and contingency in the data. Additionally, comparing the collection locations of drone data with the results of land cover classification, the collection positions are not concentrated in shrubland areas. Using this data as ground truth may lead to errors in the research results.
5. The paper calculates the ratio of RMSE to the average shrub cover to reflect the error of 1% shrub cover (EOPC) and the ratio of RMSE to the average biomass to represent the error of one unit of biomass (EOUB). These are two different types of data. Is this construction scientifically justified as an indicator to assess the model's predictive ability?
6. The depth of analysis in the results section and the completeness of the conclusions are also lacking. While I appreciate the richness of the research content, the main focus of the paper is not sufficiently emphasized, and the hierarchy and key points of the discussion are not clear enough. Furthermore, the conclusion section lacks a thorough summary of the research content. The introduction of the practical value and significance of the results is insufficient and overly broad, lacking conciseness.
Comments on the Quality of English Language
Although the authors have emphasized language revisions, this paper still requires careful editing of the English language expression as some parts lack coherence. Furthermore, there are some formatting errors in the paper, such as symbol labeling errors in Figure 8.d. Therefore, it is recommended that the authors seek assistance from relevant individuals or institutions to thoroughly review the language, images, text, etc., and maintain a rigorous approach.
Author Response
Please see attachment
Author Response File: Author Response.docx
Reviewer 3 Report
Comments and Suggestions for AuthorsAll my concerns have been revised by the authors, I recommend to accept this paper.
Comments on the Quality of English LanguageMinor editing of English language required
Author Response
We tried our best to improve the manuscript and made some changes to the manuscript. These changes will not influence the content and framework of the paper. And here we did not list the changes but marked in red in the revised paper. We appreciate for Editors/Reviewers’ warm work earnestly and hope that the correction will meet with approval.