Mapping Forest Growing Stock and Its Current Annual Increment Using Random Forest and Remote Sensing Data in Northeast Italy
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThe manuscript entitled “Mapping Forest Growing Stock and Its Current Annual Increment Using Random Forest and Remote Sensing Data in Northeast Italy” uses RF method for mapping forest growing stock and current annual increment using different remote sensing data.
The paper is written well. However, the following comments would improve the quality of the paper.
First of all, citation of references is not presented in the text correctly. (lines 133, 150, etc.)
In Table 1, dates of data collections are not clear. Has ALS derived CHM data been collected for every year (2017-2020)? If yes, why Sentinel 2 images were collected for 2019-2021?
Spatial resolution of CHM and Sentinel 2 are different. Which interpolation method has been used? Does the interpolation method affect the results?
Figure 4 shows that CHM has the most important role in estimation of forest inventory and vegetation indices derived from Sentinel 2 dataset has less impact of the results. Since CHM has high accuracy and spatial resolution, why the use of lower imagery data is considered in this research?
The results obtained from ALS and Sentinel data separately can help to better understanding the role of multisensory data approaches.
Pseudo code for RF method can improve the paper quality.
Also, annual increment value of the forest growing stock in different years would be interesting for the readers.
Author Response
Dear Reviewer, we thank you for the constructive comments. Here the reply. The update paper have been uploaded with the track change.
C1: First of all, citation of references is not presented in the text correctly. (lines 133, 150, etc.)
R1: We thank the reviewer for the comment. We have carefully checked all the reference style both in the text and in the reference list.
C2: In Table 1, dates of data collections are not clear. Has ALS derived CHM data been collected for every year (2017-2020)? If yes, why Sentinel 2 images were collected for 2019-2021?
R2: We have rephrased the text to provide more information about the actual period when lidar data were collected. We have not a lidar coverage for each year, but flights were scheduled in different years to achieve a wall-to-wall coverage of the entire regional area. The table 1 have been changed accordingly. The Sentinel-2 is a median dataset for years 2019-2021. The period 2017-2018 was not included to avoid biases due to the loss of coverage due to the Vaia storm, while 2021 was included to fill some gaps due to cloud coverage. We have added new text to further clarify those issues.
C3: Spatial resolution of CHM and Sentinel 2 are different. Which interpolation method has been used? Does the interpolation method affect the results?
R3: We did not interpolate raster data. We used GIS sampling function (i.e. “zonal statistics” tool in QGIS) to calculate the mean/median data for every independent variable for both the inventory sampling plots (of 13 m of radius) and for the regular grid requested then for the spatialization. We have revised the text to make this clearer in the manuscript.
C4: Figure 4 shows that CHM has the most important role in estimation of forest inventory and vegetation indices derived from Sentinel 2 dataset has less impact of the results. Since CHM has high accuracy and spatial resolution, why the use of lower imagery data is considered in this research?
R4: Thank you for this question. Sentinel-2 MSI (or Landsat TM8) are the most used sensors freely available with a global coverage. As few regions have a full Lidar covarage, we were interested in investigating how the use of Sentinel-2 could help in estimating tree growth and stock. In fact, the S2 resolution of 10x10 meters (20x20 for some IR bands) have been already used in other studies. There are also private platforms with an higher resolution (e.g. Planet Planetscope with 3 m res. for RGB and NIR) but not for free, so covering the entire region would not be feasible. On the other side, since data were averaged at plot (or grid) level the effect of different resolutions is greatly reduced.
C5: The results obtained from ALS and Sentinel data separately can help to better understanding the role of multisensory data approaches.
R5: We thank the reviewer for having raised this point. We agree on that. Indeed, the importance plots were added to highlight the relative importance (in terms of increment of MSE) of each variable if removed from the model. However, we have added the possibility to test different configurations of predictors as a potential advance of this study.
C6: Pseudo code for RF method can improve the paper quality.
R6: We have added a pseudo-code in Appendix A, as requested.
C7: Also, annual increment value of the forest growing stock in different years would be interesting for the readers.
R7: We agree that annual increment at each year would be interesting. However, in the Italian National Forest Inventory, CAI (m3 ha-1 y-1) is estimated by collecting increment cores by a subsample of trees per plot. It refers to the mean value of tree growth in the last five years (i.e. periodic volume increment). We have clarified this issue in the text. In addition, we don’t own other LiDAR inventories, so calculating annual increment at each year is not feasible.
Reviewer 2 Report
Comments and Suggestions for AuthorsAbstract
Line 12: The introduction to the paper mentions the role of forests in providing multiple goods and services, which is appropriate. However, it would be beneficial to include a couple of specific examples to better contextualize this statement.
Lines 13-14: The need for reliable spatial predictions of forest variables is well noted, but it could be specified which forest variables are the most critical.
Lines 15-16: The description of the study area and the data used (273 plots from the Italian National Forest Inventory) is adequate. However, it would be useful to briefly mention the relevance of this region to the study.
Lines 17-19: The use of a Random Forest model with specific predictors is mentioned, which is clear. However, a brief explanation could be added as to why these particular predictors were chosen.
Lines 20-21: The results of the model for the growth stock and the annual increase are presented with R² and RMSE metrics, which is appropriate. However, it would be useful to give context on whether these values are good or expected.
Lines 22-23: Improvement in validation with a standalone dataset is fine, but it would be beneficial to specify how significant the increase in model performance is.
Lines 24-26: The relative importance of the information derived from LiDAR is mentioned, but it could be expanded with a brief comparison with other techniques to better highlight this conclusion.
Introduction
Line 29: The introduction on the role of forests is general and could benefit from more specific examples.
Lines 30-31: The reference to global, European and national documents is valid, but it would be stronger if specific examples of such documents were mentioned.
Line 32: The multifunctionality of forests is correctly mentioned. It might be useful to mention specific examples of how this multifunctionality manifests itself.
Lines 33-35: The importance of explicit spatial data is well noted, but more context could be given on why it is essential for forest planning and management.
Lines 36-38: The use of sample-based National Forest Inventories is mentioned, which is appropriate. However, a brief description of how these inventories are conducted would be helpful for readers who are unfamiliar with the process.
Lines 39-42: The lack of large-scale detailed spatial layers is correctly mentioned. However, a brief explanation of the implications of this lack of data could be included.
Lines 43-45: The introduction of methods based on Remote Sensing and Artificial Intelligence algorithms is appropriate. However, a brief explanation of what types of data are used in these methods would be beneficial.
Lines 46-50: The mention of Enhanced Forest Inventories (IFS) that combine different data sources is clear. It would be helpful to specify what types of data are typically combined in these inventories.
Lines 51-56: Comparing the technical characteristics and operational constraints of the methods used is fine, but a specific example could be added to better illustrate the differences.
Lines 57-62: The explanation of the Individual Tree Detection (TDI) and the Area-Based Approach (ABA) approaches is adequate. However, the specific challenges associated with each approach could be deepened.
Lines 63-71: The mention of different methods of spatial estimation is appropriate. It would be helpful to include a brief explanation of how Random Forest's algorithm differs from other methods mentioned.
Lines 72-75: The introduction of the Random Forest algorithm and its application in the forestry sector is clear. However, an example from a previous study that has used this algorithm to provide context could be added.
Lines 76-81: The description of previous studies conducted in Italy is adequate. It would be helpful to include more details about the results of these studies to better contextualize the need for the current study.
Lines 82-87: The statement of the purpose of the study is clear. However, it could be strengthened by explaining how this study will contribute to the field of forestry research and what the expectations are in terms of practical impact.
Materials and Methods
Lines 89-91: The description of the study area is clear, but could benefit from a more detailed discussion of the ecological and economic importance of this particular region.
Lines 92-95: The history of forest cover in the region is mentioned, but it would be useful to include specific data on changes in forest cover for context.
Lines 96-102: The description of forest categories and the climate of the region is adequate. However, more details could be added on how these categories and climatic conditions affect forest dynamics and growth.
Lines 103-105: Figure 1 is mentioned but not described in detail. It would be helpful to briefly describe what the figure shows and its relevance to the study.
Lines 106-109: The description of the third Italian National Forest Inventory is clear. However, more details on the sampling design could be provided for better understanding.
Lines 110-117: The explanation of the three phases of the inventory is adequate. However, more details could be added on the methods used for the interpretation of aerial photos and the classification of points.
Lines 118-125: The description of dendrometric data collection is detailed, but it would be helpful to explain why plots affected by Storm Vaia were excluded.
Lines 126-129: The description of ALS data and the creation of the digital terrain model (DTM) is clear. However, it would be beneficial to include information on the accuracy of this data.
Lines 130-136: The explanation of how the elevation models and the wooded area mask were derived is adequate. It would be helpful to include a brief discussion of the potential limitations of this data.
Lines 137-145: The description of vegetation indices derived from Sentinel-2 is detailed and clear. However, a brief explanation of how these indices relate to the forest variables being modelled could be added.
Lines 146-159: Table 1 provides a useful summary of the predictors used. However, it would be beneficial to briefly discuss the potential limitations of these predictors and how they were handled in the analysis.
Lines 161-167: The description of the application of the Random Forest (RF) algorithm is clear. However, it would be helpful to explain why 1000 trees were chosen in the model and how this impacts the results.
Lines 168-173: The explanation of how RF generates regression trees and prevents overfitting is adequate. However, a brief discussion of the possible limitations of this approach could be included.
Lines 174-178: The description of the Out-of-bag (OOB) procedure is clear. However, it would be useful to provide more detail on how mean squared errors were calculated and the importance of predictors.
Lines 179-188: The explanation of how the RF model was applied to the covariate data is adequate. It would be helpful to include a brief discussion of how potential biases in the data were handled and any pre-processing performed.
Lines 189-190: Figure 2 is mentioned, but it would be helpful to provide a more detailed description of what the figure shows and its relevance to the analysis.
Lines 191-202: The description of the validation process is clear. However, a brief explanation of why Leave-one-out (LOO) cross-validation was chosen and how independent forest management units were selected for validation could be included.
Results
Lines 203-207: The presentation of GSV and CAI results is clear. However, it would be useful to include more details on the spatial distribution of these results and the implications for forest management.
Lines 208-218: Description of heterogeneity in species composition and environmental conditions is adequate. It would be helpful to include specific examples of how these factors affect GSV and CAI in different areas.
Lines 219-221: Figure 3 is mentioned, but it would be helpful to provide a more detailed description of what the figure shows and how the results are interpreted.
Lines 222-226: The assessment of the relative importance of covariates is clear. However, it would be beneficial to discuss the potential limitations of this analysis and how they might affect the results.
Lines 227-231: The presentation of the accuracy results of the model according to LOO is clear. However, it would be useful to include more details on how these results are interpreted in the context of the study.
Lines 232-243: Validation with forest management units is adequate. It would be useful to include a discussion of the possible limitations of this validation and how the results could be improved in future studies.
Discussion
Lines 246-255: Discussion of the importance of detailed GSV and CAI maps is appropriate. However, it would be useful to include specific examples of how these maps can be used in practice.
Lines 256-272: The comparison of the results with previous studies is clear. It would be beneficial to include more details about methodological differences and how they might affect the results.
Lines 273-299: Discussion of possible improvements to the model and current limitations is appropriate. However, it would be useful to include a more detailed discussion of the implications of these limitations and how they might be addressed in future studies.
Conclusions
Lines 312-321: Conclusions on the importance of reliable estimation of forest biomass and growth are appropriate. However, it would be useful to include more details on the practical implications of these findings.
Lines 322-329: The discussion of possible future improvements is clear. It would be beneficial to include a more detailed discussion of the specific next steps that need to be taken to improve the estimates.
Figures and Tables
The figures and tables are clear and well presented. However, it would be helpful to include more detailed descriptions of what they show and how they should be interpreted in the context of the study.
References
It would be helpful to review the dates of the cited publications to ensure that the most recent and relevant sources are being used.
The article is clear and well-structured, but could benefit from more specific examples and additional details in various sections to provide clearer context for readers. It would be useful to discuss in more detail the methodological limitations and potential biases in the data, as well as how they could be addressed in future studies.
Author Response
Dear Reviewer. I apologize, please consider only this text. The first its a mistake.
Dear Reviewer, we thank you for the constructive comments. Here the reply. The update paper have been uploaded with the track change. Lines reference are changed.
C1: Line 12: The introduction to the paper mentions the role of forests in providing multiple goods and services, which is appropriate. However, it would be beneficial to include a couple of specific examples to better contextualize this statement.
R1: We have added some examples at the beginning of the introduction, as requested.
C2: Lines 13-14: The need for reliable spatial predictions of forest variables is well noted, but it could be specified which forest variables are the most critical.
R2: We have added some examples, as requested.
C3: Lines 15-16: The description of the study area and the data used (273 plots from the Italian National Forest Inventory) is adequate. However, it would be useful to briefly mention the relevance of this region to the study.
R3: We thank the reviewer for his/her comment. We have added more text to mention the relevance of the region at the beginning of the paragraph about Study Area.
C4: Lines 17-19: The use of a Random Forest model with specific predictors is mentioned, which is clear. However, a brief explanation could be added as to why these particular predictors were chosen.
R4: we have added new text to better explain why we selected the Vegetation indices and LiDAR metrics as covariates in the Material and methods section.
C5: Lines 20-21: The results of the model for the growth stock and the annual increase are presented with R² and RMSE metrics, which is appropriate. However, it would be useful to give context on whether these values are good or expected.
R5: we have slightly changed the text according to the suggestion in order to better clarify this issue. It is worth to say that a wider discussion about this issue is reported in the discussion section.
C6: Lines 22-23: Improvement in validation with a standalone dataset is fine, but it would be beneficial to specify how significant the increase in model performance is.
R6: we have slightly changed the text to better clarify that the performance significantly increased. Anyway, the validation do not change the model itself obviously.
C7: Lines 24-26: The relative importance of the information derived from LiDAR is mentioned, but it could be expanded with a brief comparison with other techniques to better highlight this conclusion.
R7: we have rephrased this section and included also a brief comparison with other techniques as requested.
C8: Line 29: The introduction on the role of forests is general and could benefit from more specific examples.
R8: We have expanded this part providing a wider context and more examples of forest multifunctionality.
C9: Lines 30-31: The reference to global, European and national documents is valid, but it would be stronger if specific examples of such documents were mentioned.
R9: We have expanded this part citing some examples. We have also added more relevant references about the topic.
C10: Line 32: The multifunctionality of forests is correctly mentioned. It might be useful to mention specific examples of how this multifunctionality manifests itself.
R10: We have expanded this part citing some examples. We have also added more relevant references about the topic.
C11: Lines 33-35: The importance of explicit spatial data is well noted, but more context could be given on why it is essential for forest planning and management.
R11: we have better explained the contribution of explicit spatial data in the decision making processes.
C12: Lines 36-38: The use of sample-based National Forest Inventories is mentioned, which is appropriate. However, a brief description of how these inventories are conducted would be helpful for readers who are unfamiliar with the process.
R12: we have slightly changed the text to better explain how NFIs are conducted.
C13: Lines 39-42: The lack of large-scale detailed spatial layers is correctly mentioned. However, a brief explanation of the implications of this lack of data could be included.
R13: We have added a brief explanation of the implications of this lack of data.
C14: Lines 43-45: The introduction of methods based on Remote Sensing and Artificial Intelligence algorithms is appropriate. However, a brief explanation of what types of data are used in these methods would be beneficial.
R14: Data used in the methods ae reported in the cited reference.
C15: Lines 46-50: The mention of Enhanced Forest Inventories (IFS) that combine different data sources is clear. It would be helpful to specify what types of data are typically combined in these inventories.
R15: we have reported the data combined in these inventories: “data from different sources, like passive (optical), or active (i.e., Synthetic Aperture Radar, or Light Detection and Raging - LiDAR) sensors, and from many platforms (e.g. satellite, plane, drone)”
C16: Lines 51-56: Comparing the technical characteristics and operational constraints of the methods used is fine, but a specific example could be added to better illustrate the differences.
R16: examples illustrating differences in the methods are reported and references are also cited
C17: Lines 57-62: The explanation of the Individual Tree Detection (TDI) and the Area-Based Approach (ABA) approaches is adequate. However, the specific challenges associated with each approach could be deepened.
R17: the specific challenges associated have been deepened
C18: Lines 63-71: The mention of different methods of spatial estimation is appropriate. It would be helpful to include a brief explanation of how Random Forest's algorithm differs from other methods mentioned.
R18: we think that differences between RF and the other methods is already reported. It is a non parametric method (different for the other parametric methods) and is a decision tree method (differently from K-nearest Neighbor.
C19: Lines 72-75: The introduction of the Random Forest algorithm and its application in the forestry sector is clear. However, an example from a previous study that has used this algorithm to provide context could be added.
R19: we have added a reference to provide an example of application
C20: Lines 76-81: The description of previous studies conducted in Italy is adequate. It would be helpful to include more details about the results of these studies to better contextualize the need for the current study.
R20: We have included more details about Italian studies, as requested.
C21: Lines 82-87: The statement of the purpose of the study is clear. However, it could be strengthened by explaining how this study will contribute to the field of forestry research and what the expectations are in terms of practical impact.
R21: we have rephrased the sentence to improve it, as suggested.
C22-24: Lines 89-91: The description of the study area is clear, but could benefit from a more detailed discussion of the ecological and economic importance of this particular region.
Lines 92-95: The history of forest cover in the region is mentioned, but it would be useful to include specific data on changes in forest cover for context.
Lines 96-102: The description of forest categories and the climate of the region is adequate. However, more details could be added on how these categories and climatic conditions affect forest dynamics and growth.
R22-24: We have rephrased the section about study area adding its relevance as a case study and better contextualizing forest trends and dynamics
C25: Lines 103-105: Figure 1 is mentioned but not described in detail. It would be helpful to briefly describe what the figure shows and its relevance to the study.
R25: Figure 1 shows forest areas in the region and the location National forest inventory plots, as described in the caption.
C26-27: Lines 106-109: The description of the third Italian National Forest Inventory is clear. However, more details on the sampling design could be provided for better understanding.
Lines 110-117: The explanation of the three phases of the inventory is adequate. However, more details could be added on the methods used for the interpretation of aerial photos and the classification of points.
R26-27: The sampling design is three-phase, non-aligned, and systematic, as reported. Statistics details (which are out of the scope of this paper) are reported in the cited reference. The method used for visual interpretation is also reported in the document in case the reader needs more information.
C28. Lines 118-125: The description of dendrometric data collection is detailed, but it would be helpful to explain why plots affected by Storm Vaia were excluded.
R28: We have included more details about why plots affected by Storm Vaia were excluded, as requested.
C29: Lines 126-129: The description of ALS data and the creation of the digital terrain model (DTM) is clear. However, it would be beneficial to include information on the accuracy of this data.
R29: added
C30: Lines 130-136: The explanation of how the elevation models and the wooded area mask were derived is adequate. It would be helpful to include a brief discussion of the potential limitations of this data.
R30:. We added as a potential step forward the application of our approach could to trees outside forests not included in the definition of “forest”, such as hedgerows or trees outside the forests
C31: Lines 137-145: The description of vegetation indices derived from Sentinel-2 is detailed and clear. However, a brief explanation of how these indices relate to the forest variables being modelled could be added.
R31: We have slightly revised the text to include expected relationship between indices and modelled variables. More details are however reported in the cited literature.
C32: Lines 146-159: Table 1 provides a useful summary of the predictors used. However, it would be beneficial to briefly discuss the potential limitations of these predictors and how they were handled in the analysis.
R32: We have added new text to discuss potential limitation of used VIs in the discussion section, citing appropriate literature.
C33: Lines 161-167: The description of the application of the Random Forest (RF) algorithm is clear. However, it would be helpful to explain why 1000 trees were chosen in the model and how this impacts the results.
R33: 1000 trees were chosen to balance the calculation time with the correctness of the results, as reported in the cited reference.
C34-35: Lines 168-173: The explanation of how RF generates regression trees and prevents overfitting is adequate. However, a brief discussion of the possible limitations of this approach could be included.
Lines 174-178: The description of the Out-of-bag (OOB) procedure is clear. However, it would be useful to provide more detail on how mean squared errors were calculated and the importance of predictors.
R34-35: We thank the reviewer for the comment. We decided to reduce the technical details about functioning of the model since the computational details are well-consolidated. However, we included relevant references where the reader can find more details about specific procedures of the model.
C36: Lines 179-188: The explanation of how the RF model was applied to the covariate data is adequate. It would be helpful to include a brief discussion of how potential biases in the data were handled and any pre-processing performed.
R36: pre-processing of data was described and discussed extensively in the previous paragraphs, adding new text (e.g. filling gaps due to data coverage).
C37: Lines 189-190: Figure 2 is mentioned, but it would be helpful to provide a more detailed description of what the figure shows and its relevance to the analysis.
R37: Figure 2 is a flowchart as reported in the caption and it describes the phases of data processing with the details of used software.
C38: Lines 191-202: The description of the validation process is clear. However, a brief explanation of why Leave-one-out (LOO) cross-validation was chosen and how independent forest management units were selected for validation could be included.
R38. We thank the reviewer for his/her comment. Discussion about validation methods is reported in the discussion: “LOO has been often used for accuracy assessment of model outputs [11], but, as it is sensitive to outliers, it may not be representative of the model’s performance and the estimation accuracy may be over optimistic. For this reason, we validated our model results also using data from independent an independent dataset (i.e., forest management units).”
C39: Lines 203-207: The presentation of GSV and CAI results is clear. However, it would be useful to include more details on the spatial distribution of these results and the implications for forest management.
R39: We agree that details on the spatial distribution of these results are relevant. For this reason, we have included and revised Figure 3. New text about implications for forest management is include in the discussion section.
C40: Description of heterogeneity in species composition and environmental conditions is adequate. It would be helpful to include specific examples of how these factors affect GSV and CAI in different areas.
R40: Examples are reported discussing implications of species and environmental conditions on tree growth: e.g., “CAI is higher in the endalpic area like GSV but also in the esalpic area where most of the forests have been established on fertile ex-agricultural land following secondary successions and are characterized by fast growing species such as Robinia pseudoacaica, Acer spp, Fraxinux spp.”
C41: Lines 219-221: Figure 3 is mentioned, but it would be helpful to provide a more detailed description of what the figure shows and how the results are interpreted.
R41: See R39.
C42: Lines 222-226: The assessment of the relative importance of covariates is clear. However, it would be beneficial to discuss the potential limitations of this analysis and how they might affect the results.
R42: Discussion about limitations of our approach has been expanded in the discussion section, reporting also potential improvements of the study.
C42-43: Lines 227-231: The presentation of the accuracy results of the model according to LOO is clear. However, it would be useful to include more details on how these results are interpreted in the context of the study.
Lines 232-243: Validation with forest management units is adequate. It would be useful to include a discussion of the possible limitations of this validation and how the results could be improved in future studies.
R42-43: see R38
C44: Lines 246-255: Discussion of the importance of detailed GSV and CAI maps is appropriate. However, it would be useful to include specific examples of how these maps can be used in practice.
R44: we have added more text to clarify this issue
C45: Lines 256-272: The comparison of the results with previous studies is clear. It would be beneficial to include more details about methodological differences and how they might affect the results.
R45: details about methodological differences are reported: e.g. “... Chirici et al. [26] achieved an R2 of 0.47 applying RF algorithm to RS covariates (without ALS) and a RMSE% equal to 68.70. Puliti et al. [47], including ALS among covariates, reached an accuracy similar to ours when considering management units as ground truth (R2=0.83), but these authors used ALS data with a lower density than ours (7.45 points m-2)...”. We are available to further improve this part in case the reviewer thinks it is needed.
C46: Lines 273-299: Discussion of possible improvements to the model and current limitations is appropriate. However, it would be useful to include a more detailed discussion of the implications of these limitations and how they might be addressed in future studies.
R46: We have improved the discussion about limitations of our approach and how they might be addressed in future studies
C47: Lines 312-321: Conclusions on the importance of reliable estimation of forest biomass and growth are appropriate. However, it would be useful to include more details on the practical implications of these findings.
R47: we have added new text to clarify potential use of our maps at the beginning of the discussion section
C48: Lines 322-329: The discussion of possible future improvements is clear. It would be beneficial to include a more detailed discussion of the specific next steps that need to be taken to improve the estimates.
R48: we have added new text to clarify next steps that need to be taken to improve the estimates in the discussion section
C49: The figures and tables are clear and well presented. However, it would be helpful to include more detailed descriptions of what they show and how they should be interpreted in the context of the study.
R49: we have improved quality of figures and tables
C50: References
It would be helpful to review the dates of the cited publications to ensure that the most recent and relevant sources are being used.
R50: We have revised the reference list adding new relevant and recent papers.
C51: The article is clear and well-structured, but could benefit from more specific examples and additional details in various sections to provide clearer context for readers. It would be useful to discuss in more detail the methodological limitations and potential biases in the data, as well as how they could be addressed in future studies.
R51: We thank the reviewer for the positive and very detailed review. We have widely revised the paper trying to solve to all the issues raised. Not sure if the word “less..” at the end of the comment is a typo or a residual of a copy-paste process.
Reviewer 3 Report
Comments and Suggestions for AuthorsI reviewed the manuscript ‘Mapping Forest Growing Stock and Its Current Annual 2 Increment Using Random Forest and Remote Sensing Data in 3 Northeast Italy’. It contains many positive aspects but some issues should be address before the manuscript could be consider for publication in any journal. Please, find bellow main and specific comments. Some spelling mistakes were found.
Main comments
The o objective is not clear. What is the gap the paper is looking to address? The introduction needs to be restructured. Some information could be separated and developed in distinct paragraphs. The article has citation errors throughout that need to be carefully revised. The discussion needs to be improved. The authors do not make clear the importance of the findings themselves. Instead, they focus part of the discussion on the importance of the general approach. I think a more specific discussion highlighting the main findings is missing. Citations need to be carefully reviewed throughout the article. I would also suggest a brief approach to the ecological importance of aboveground biomass in the context of decision-making, which could be dealt with in both the introduction and the discussion (see for example Maciel et al 2021).
Specific comments
L41 ‘In the last decades …’ seems to be a new paragraph.
L42 maybe the authors could be more specific about forest attribute means
L48 what do you mean by forest variable?
L51 which ones?
L56 please be specific about the limited in space and time
L57 I guess it could be a new paragraph?
L63 could the author cite some variable?
L69 consider to move it to sentence above
L76-80 this paragraph needs further working. Maybe the authors could discussion some examples of this in Italy
L85-87 should be move to introduction
L93 which trend is common in there?
L100 what the 9% means?
L113 ‘first phase’s points’ what do you mean?
L126 fix this error
L129 ‘resolution is 16 points’ why?
L129-130 needs to be clear
L138 see general rule for writing abbreviations
L142 another reference error
L163-165 another reference error
L217 check spelling
the map in figure 3 needs improvement
information is needed to interpret the graph in figure 4
Maciel, E. A., Martins, V. F., de Paula, M. D., Huth, A., Guilherme, F. A., Fischer, R., ... & Martins, F. R. (2021). Defaunation and changes in climate and fire frequency have synergistic effects on aboveground biomass loss in the brazilian savanna. Ecological Modelling, 454, 109628.
Comments on the Quality of English Language
The article could benefit from an English revision
Author Response
Dear Reviewer, we thank you for the constructive comments. Here the reply. The update paper have been uploaded with the track change.
C1: The o objective is not clear. What is the gap the paper is looking to address? The introduction needs to be restructured. Some information could be separated and developed in distinct paragraphs. The article has citation errors throughout that need to be carefully revised. The discussion needs to be improved. The authors do not make clear the importance of the findings themselves. Instead, they focus part of the discussion on the importance of the general approach. I think a more specific discussion highlighting the main findings is missing. Citations need to be carefully reviewed throughout the article. I would also suggest a brief approach to the ecological importance of aboveground biomass in the context of decision-making, which could be dealt with in both the introduction and the discussion (see for example Maciel et al 2021).
R1: We thank the reviewer for his/her comment. We have revised the introduction to make the added value of our work clearer and to better focus our aims. In addition, we have revised the discussion reporting the potential use of our results for decision making. We have also reported the implications of biomass estimation at large scale for fire modeling. We have also cited the suggested reference.
We have carefully checked all the reference style both in the text and in the reference list.
C2: L41 ‘In the last decades …’ seems to be a new paragraph.
R2: We have revised the text and divided the new paragraph with the previous one.
C3-C4: L42 maybe the authors could be more specific about forest attribute means. L48 what do you mean by forest variable?
R3-R4: We have provided some examples of forest attributes in brackets. We have also removed the term “forest variable” and substituted it with “forest attribute” throughout the manuscript for consistency.
C5: L51 which ones?
R5: we have revised the text to make it clearer.
C6: L56 please be specific about the limited in space and time
R6: we have revised the text to make it clearer.
C7: L57 I guess it could be a new paragraph?
R7: We agree. We have separated the two paragraphs.
C8: L63 could the author cite some variable?
R8: Agree, added as examples
C9: L69 consider to move it to sentence above
R9: Agree. Moved.
C10: L76-80 this paragraph needs further working. Maybe the authors could discussion some examples of this in Italy
R10: we have revised this section adding a wider literature review of existing studies in Italy
C11: L85-87 should be move to introduction
R11: We put this paragraph here for concluding the theorethical and technical introduction and introducing our work.
C12: L93 which trend is common in there?
R12: It is refered to the "forest expansion". We have rephrased this sentence to make it clearer.
C13: L100 what the 9% means?
R13: A mistake. Removed.
C14: L113 ‘first phase’s points’ what do you mean?
R14: A mistake. Corrected.
C15: L126 fix this error
R15: we thank the reviewer for the comment. However we do not see any error at line 126. We are available to make further revisions in case the reviewer suggests how to improve the text.
C16: L129 ‘resolution is 16 points’ why?
R16: Corrected.
C17: L129-130 needs to be clear
R17: we have revised the text to make it clearer.
C19: L138 see general rule for writing abbreviations
R19: Removed the 4 abbreviations, because not necessary
C20: L142 another reference error
R20: The reference seems correct, the text point to the flowchart where the VIs extraction is depicted in the left (Google Earth Engine rectangle). If the reviewer refers to citation formatting, we have carefully checked the reference style throughout the manuscript
C21: L163-165 another reference error
R21: If the reviewer refers to citation formatting, we have carefully checked the reference style throughout the manuscript.
C22: L217 check spelling
R22: Corrected.
C23: the map in figure 3 needs improvement
R23: We have improved the thematization and legend as requested and made the legend more clear.
C24: information is needed to interpret the graph in figure4
R24: Agree, we added a description.
C25: Maciel, E. A., Martins, V. F., de Paula, M. D., Huth, A., Guilherme, F. A., Fischer, R., ... & Martins, F. R. (2021). Defaunation and changes in climate and fire frequency have synergistic effects on aboveground biomass loss in the brazilian savanna. Ecological Modelling, 454, 109628.
R25: We have added the suggested reference.