Mapping Foliar C, N, and P Concentrations in An Ecological Restoration Area with Mixed Plant Communities Based on LiDAR and Hyperspectral Data

Round 1

Reviewer 1 Report (Previous Reviewer 3)

Comments and Suggestions for Authors

Dear authors, 

I found your research very interesting about vegetation restoration in degraded lands. I would like to share some insights that I found relevant to me.

First, I found amazing the use of remote sensing's latest technology (Lidar & Hyperspectral), they offer the possibility to obtain new features that could help to understand the complex relationships between different plants communities as is this case (mixed).

Second, I do not understand all the way why do you obtain shallow features and how it works, and you confirm that it does not change much. Also, you have a lot of features to test. It would be great you explain further this approach.

Third, you indicated to use two returns from Lidar, but you do not say which ones, although I suppose they are the first and the last returns. I think is important to be clear with it, because some Lidar sensors has several returns.

Four, the table 6 shows different colors, but they should be the same, because them confuse me a bit, thinking the colors represents some differences. 

Last, I just want to congratulate for your amazing job, I hope you keep improving to help in ecological restoration.

Best regards, 

Comments on the Quality of English Language

I just found one word misspelled in the figure 9. It should be "concentration" instead of "consentration".

Author Response

Dear authors, 

I found your research very interesting about vegetation restoration in degraded lands. I would like to share some insights that I found relevant to me.

First, I found amazing the use of remote sensing's latest technology (Lidar & Hyperspectral), they offer the possibility to obtain new features that could help to understand the complex relationships between different plants communities as is this case (mixed).
Rely 1: Thanks for your time.

Second, I do not understand all the way why do you obtain shallow features and how it works, and you confirm that it does not change much. Also, you have a lot of features to test. It would be great you explain further this approach.

Rely 2: This is an insightful comment. There are two reasons why we use shallow features. Firstly, ecological restoration is usually small-scale and only provides limited samples. Secondly, it is necessary to use physiological features (such as plant height) to explain the variability of C, N, and P, in order to better guide ecological restoration. But the disadvantage of this method is that it requires measuring and verifying the accuracy of shallow features. We have noticed that in recent years, the rise of artificial intelligence has brought new opportunities for monitoring ecological restoration, which can estimate many ecosystem indicators through the deep features of remote sensed images, such as the abstract semantic information extracted by deep network models from remote sensed images and LiDAR point cloud data. We believe that this will be a promising approach to improve estimation accuracy and efficiency. But we haven't tried this method yet.

We have added the above discussion to the 4.3. Limitations and future work.

Third, you indicated to use two returns from Lidar, but you do not say which ones, although I suppose they are the first and the last returns. I think is important to be clear with it, because some Lidar sensors has several returns.

Rely 3: You are right.

The LiDAR (LiAir 220) we use only has 2 echoes. The echo of LiDAR point cloud data is an important feature that can be used to express the structural parameters of vegetation. A laser pulse emitted may return to the LiDAR sensor in the form of one or more echoes. The first laser pulse returned is the most important echo, and it will be associated with the highest elements on the Earth’s surface (such as treetops). When the first echo represents the ground, the LiDAR system will only correspond to the vegetation structure, and the last echo is related to the exposed terrain foot point where the laser pulse is emitted. The first echo usually corresponds to the vegetation structure, and the last echo is related to the exposed terrain model. Therefore, we use the percentage of the number of first echo points to the total number of echo points to estimate the vegetation gap fraction.

We have added the above information into 2.2.1. Remote sensing data. In addition, we add a reference for Table 4 to illustrate how to calculate the LiDAR related feature such as gap fraction.

Four, the table 6 shows different colors, but they should be the same, because them confuse me a bit, thinking the colors represents some differences.

Rely 4: Agree, we have removed the fill background color from the Table 6 and used check marks to indicate which features were employed.

I just found one word misspelled in the figure 9. It should be "concentration" instead of "consentration".

Rely 5: Corrected.

Last, I just want to congratulate for your amazing job, I hope you keep improving to help in ecological restoration.

Best regards,

Rely 6: Thanks again.

Reviewer 2 Report (Previous Reviewer 2)

Comments and Suggestions for Authors

My concerns were properly addressed.

Comments on the Quality of English Language

There are still some language problems, but overall intelligibility is acceptable.

Author Response

My concerns were properly addressed.

There are still some language problems, but overall intelligibility is acceptable.

Rely 1: Many thanks for your patience and encouragement.

Reviewer 3 Report (Previous Reviewer 1)

Comments and Suggestions for Authors

Thank you for your resubmitted revised manuscript and the effort you have put into this process.

This manuscript is intriguing as it proposes effective technical methods for the rapid monitoring of C, N, and P concentrations in mixed plant communities within ecological restoration areas. In this resubmitted manuscript, the author has addressed the concerns I raised previously.

In this round of review, I have some minor issues that I hope the author can address:

 

1. I noticed that some of the text in Figure 1 is too small, making it difficult to read and discern. Given the small area of the study site, providing the provincial or municipal boundaries where it is located might help readers to locate the study area more easily. Additionally, Figure 1(a) clearly depicts a map of whole China, yet the author labeled it as "in Inner Mongolia, north China." Furthermore, the visualization of Figure 4 is poor, and the x-axis is difficult to read.

   

2. The manuscript mentions that LiDAR features significantly improve the accuracy in estimating C concentration but have limited effects on N and P estimation. More importantly, the author points out spatial distribution differences in C, N, and P concentrations among different plant communities, suggesting further exploration of the ecological processes and influencing factors behind this spatial heterogeneity. For example, analyzing how factors such as soil types, microclimate conditions, and anthropogenic disturbances affect the distribution of C, N, and P in plant communities.

 

3. Considering that ecosystem restoration is a long-term process, would long-term monitoring studies be more valuable?

 

4. Given the limitations in the accuracy of plant community classification using LiDAR and hyperspectral data in this study, how did the author ensure the accuracy of plant community classification to reduce errors in estimating C, N, and P concentrations?

 

5. The manuscript mentions that data collection was conducted in July 2022. Considering that plant growth and nutrient content may be significantly influenced by seasonal variations, did the study consider the effects of seasonal changes on the accuracy of estimating C, N, and P concentrations? In the previous review, the author responded by stating that this limitation had been added in Section 4.3. However, I suggest that the author make an appropriate statement regarding time at the beginning of the data section.

 

In conclusion, I believe this resubmitted manuscript has improved compared to the previous version, and I recommend the author consider the above suggestions to further enhance the research.

Author Response

Thank you for your resubmitted revised manuscript and the effort you have put into this process. This manuscript is intriguing as it proposes effective technical methods for the rapid monitoring of C, N, and P concentrations in mixed plant communities within ecological restoration areas. In this resubmitted manuscript, the author has addressed the concerns I raised previously.

Rely 1: Thanks for your time.

In this round of review, I have some minor issues that I hope the author can address:
1. I noticed that some of the text in Figure 1 is too small, making it difficult to read and discern. Given the small area of the study site, providing the provincial or municipal boundaries where it is located might help readers to locate the study area more easily. Additionally, Figure 1(a) clearly depicts a map of whole China, yet the author labeled it as "in Inner Mongolia, north China." Furthermore, the visualization of Figure 4 is poor, and the x-axis is difficult to read.

Rely 2: We have revised Figure 1, shown as below. As for Figure 4, we have expanded the caption of Figure 4 to give more details of the confirmed features.

Figure 1. Geographical situation of the study area: (a) and (b) in Shaanxi Province, north China; (c) orthophoto view of the study area.

2.The manuscript mentions that LiDAR features significantly improve the accuracy in estimating C concentration but have limited effects on N and P estimation. More importantly, the author points out spatial distribution differences in C, N, and P concentrations among different plant communities, suggesting further exploration of the ecological processes and influencing factors behind this spatial heterogeneity. For example, analyzing how factors such as soil types, microclimate conditions, and anthropogenic disturbances affect the distribution of C, N, and P in plant communities.

Rely 3: Good comments. The driving factors of the spatial heterogeneity of C, N, and P are interesting scientific problem and a key to improving the effectiveness of ecological restoration. We believe the map of C, N, and P could contribute to the further research on such scientific problem.

We have added the above discussion to the 4.3. Limitations and future work.

3. Considering that ecosystem restoration is a long-term process, would long-term monitoring studies be more valuable?

Rely 4: Absolutely. The ecological restoration usually was assessed based on short observation of vegetation coverage. More and more studies have concluded that the restored ecosystem should be proved to be self-sustaining rather than greening. Obviously, long-term data is needed to study the ecological succession after restoration.

4. Given the limitations in the accuracy of plant community classification using LiDAR and hyperspectral data in this study, how did the author ensure the accuracy of plant community classification to reduce errors in estimating C, N, and P concentrations?

Rely 5: From Table 6, it can be seen that the height of plants and the reflectance of leaves in the red-light band are the most important features. The height of different types of plants is usually concentrated in a certain range, for example, poplar trees are usually 10-15m high, and Pinus trees are usually 4-6 meters high. The leaves of different types of plants usually have similar light scattering characteristics, for example, the leaves of poplar are broad-leaved, while the leaves of camphor pine are coniferous, so the reflectivity of camphor pine leaves to light is lower than that of poplar. Therefore, we infer that if it is possible to classify plant communities according to height and leaf shape, reduce inter class differences in data, and then establish estimation models separately, it can improve estimation accuracy.

We have improved the 4.3. Limitations and future work according to explanations above.

5. The manuscript mentions that data collection was conducted in July 2022. Considering that plant growth and nutrient content may be significantly influenced by seasonal variations, did the study consider the effects of seasonal changes on the accuracy of estimating C, N, and P concentrations? In the previous review, the author responded by stating that this limitation had been added in Section 4.3. However, I suggest that the author make an appropriate statement regarding time at the beginning of the data section.

Rely 6: We only collected data once in July 2022. Our research area is located in the northern hemisphere, and July is the growing season for plants. Investigating the growth status of vegetation during this season can better reflect the success of ecological restoration. Therefore, this study did not consider the effects of seasonal changes on the accuracy of estimating C, N, and P concentrations. We have introduced the data collection time and reasons at the beginning of section 2.2.1.

In conclusion, I believe this resubmitted manuscript has improved compared to the previous version, and I recommend the author consider the above suggestions to further enhance the research.

Rely 7: Thanks again.

Round 2

Reviewer 3 Report (Previous Reviewer 1)

Comments and Suggestions for Authors

The author's response addressed my concerns and now I have no further questions.

Comments on the Quality of English Language

Some sentences in the revised manuscript do not conform to common norms for scientific papers, and authors should seek language editors.

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 Authors

This manuscript focuses on using hyperspectral data to discern the distribution of nutrients in ecosystems, aligning well with the scope of this journal and my research interests. After careful review, I look forward to the author's consideration of addressing and responding to my concerns during the revision process:

 

1. The introduction is well-written with excellent logical flow. However, in the penultimate paragraph, the author briefly reviews two current methods for monitoring nutrients using remote sensing technology, followed by the abrupt statement "Overall, studies that discuss the feasibility of mapping the foliar C, N, and P of reported ecosystems with mixed plant communities are lacking" (Line 92). This transition appears disjointed in terms of writing logic. I recommend substantial revisions to the last two paragraphs of this section, delving into the literature gaps through an in-depth review.

 

2. The study area for this research is only 0.6 square kilometers. What is the rationale behind choosing this area for study? How applicable is this monitoring method to larger-scale empirical studies? Are there any limitations?

 

3. Regarding the acquisition and disposal process of field data (Section 2.2.2), were there specific manuals or regulations referenced?

 

4. The text in Figure 3 is too small; please consider improving the layout accordingly.

 

5. I suggest adding a table in the methods section to visually display all data, their types, and sources.

 

6. The author used only R2 and MSE for validation, which are flawed indicators. Are there alternative field validation methods? Can you provide a detailed explanation of "the sample data were randomly split into two subsets with 80% for training and 20% for validation"?

 

7. Given that visualizing the distribution of nutrients is the primary contribution of this study, I recommend a detailed discussion of Section 3.3, explaining the results in terms of spatial distribution.

 

8. Would there be any impact on the effectiveness of this method due to seasonal or monthly variations?

 

9. In Section 4.3, the author briefly outlines the limitations of the method. Could the author provide a detailed list of scenarios where this method is applicable? Similar to prompts in an application manual, in what situations would one prioritize using your method?

 

10. As a validation study, data and its reproducibility are crucial. Would the author consider modifying the Data Availability Statement to include open sourcing of data and code?

 

In conclusion, the manuscript exhibits a cohesive structure and rigorous logic. I look forward to the author's revisions and responses.

 

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript presents a method for mapping foliar C, N, and P concentrations using LiDAR and hyperspectral data. The manuscript is well written and the experiments are thorough. Language needs some polishing, but in general the text can be understood without much effort. I have the following specific suggestions:
- The level of detailing in the “Materials and Methods” section is not enough for proper experimental reproduction. In particular, more details about the preprocessing stage are needed. The bands employed as spectral features should also be specified.
- The caption in Figure 4 needs to be more detailed. Captions should be as self-contained as possible.
 

Comments on the Quality of English Language

Language needs some polishing, but in general the text can be understood without much effort.

Author Response

The manuscript presents a method for mapping foliar C, N, and P concentrations using LiDAR and hyperspectral data. The manuscript is well written and the experiments are thorough. Language needs some polishing, but in general the text can be understood without much effort. I have the following specific suggestions:
- The level of detailing in the “Materials and Methods” section is not enough for proper experimental reproduction. In particular, more details about the preprocessing stage are needed. The bands employed as spectral features should also be specified.

Rely 1: Reasonable suggestions. We have added more detail about the pre-processing of LiDAR data and hyperspectral data, shown as follows.

Two types of remote sensing data, i.e., LiDAR data and hyperspectral images, were acquired from the DJI M600 (Dajiang Baiwang Technology Co., Ltd., Shenzhen, Guangdong, China) on 25 July 2022. For LiDAR data, a total of three flight routes were planned. The flight altitude is 90 m, speed is 5 m/s, horizontal field of view angle is 360°, vertical field of view angle is greater than 20°, and an average point cloud density is 130 points/m². The elevation of the study area was 1200–1296 m, and the number of echoes was 2. The LiDAR 360 data were acquired from GreenValley International (Berkeley, CA, USA) and were used for preprocessing. The preprocessing included creating an aerial strip mosaic, point cloud registration, strip redundancy removal, noise removal, point cloud feature extraction, and point cloud classification [26]. First, mosaic the LiDAR data and then register to minimize the spatial position differences between the points. Then, removal the redundant point cloud data of the overlapping part of the strip and point cloud noise, and then divide the point cloud into ground and non-ground points. Finally extract the point cloud data features. Before extracting LiDAR features, it was necessary to normalize the LiDAR data by subtracting the elevation value Z of each point in the LiDAR point cloud data. Normalization can remove the influence of terrain fluctuations on the elevation value of the point cloud data.

For hyperspectral data, eight flight routes were set up to ensure full coverage of the area. The drone had a flight altitude of 140 m, a spatial resolution of 0.19 m, and a lateral overlap rate of 57%, flying a total of four sorties. The spectral range obtained during each flight was 398–1002 nm, with a total of 112 bands. During the data collection, sensor calibration was carried out using a standard diffuse reflection board, followed by atmospheric correction, geometric correction, band splicing and cropping, image fusion, and other processing. Next, ENVI software (Research Systems, Inc., Boulder, CO, USA) was employed for hyperspectral image preprocessing mainly including radiometric calibration, atmospheric correction, geometric correction, noise and dimensionality reduction, aerial strip mosaicking, clipping, and band fusion [34].

In additions, the bands employed as spectral features also have been be specified.

Given that the spectral reflectance of different plant leaves varies, 112 bands of hyperspectral data were employed as parameters for model construction. In this study, all 112 bands were used, with a wavelength range of 398–1002 nm and a band width of 5.2 nm.

Note, R₆₈₀ in the table indicates the spectral reflectance at 680 nm; other parameters are analogized. Chlorophyll index—green (CI_green); chlorophyll index—red edge (CI_{red_edge}); double difference index (DD); difference vegetation index (DVI); enhanced vegetation index (EVI); Gitelson and Merzlyak index (GM); green normalized difference vegetation index (GNDVI); land cover index (LCI); modified chlorophyll absorption in reflectance index (MCARI); modified normalized difference (mND₇₀₅); modified simple ratio (mSR₇₀₅); modified soil-adjusted vegetation index (MSAVI); modified triangular vegetation index (MTVI1); normalized vegetation index (NDI); normalized difference vegetation index (NDVI); normalized pigment chlorophyll index (NPCI); plant biochemical index (PBI); photochemical reflectance index (PRI); pigment specific normalized difference of chlorophyll a (PSND_a); pigment specific normalized difference of chlorophyll b (PSND_b); perpendicular vegetation reflectance (PVR); ratio vegetation index (RVI); red edge vegetation stress index (RVSI); the range of leaf reflectance at 680 nm (R₆₈₀); the range of leaf reflectance at 800 nm (R₈₀₀); soil-adjusted vegetation index (SAVI); standardized precipitation index (SPI); simple ratio pigment index (SRPI); triangle vegetation index (TVI); visual atmospheric resistant index (VARI); Vogelmann red edge index 1 (VOG_a); Vogelmann red edge index 2 (VOG₂); water index (WI).

 

The caption in Figure 4 needs to be more detailed. Captions should be as self-contained as possible.

Rely 2: Agree, we have revised the caption of Figure 4, shown as follows:

 

Figure 4. Selection results of the features of: (a) C, (b) N, and (c) P. All features were ranked according to the Z score calculated by the Boruta algorithm. The red, yellow, and green box plots represent the Z scores for the rejected, tentative, and confirmed attributes, respectively. For C, red-edge bands, height variables, and vegetation structure parameters were identified as comparatively important. For N, textural features, height percentiles of 40%–95%, and vegetation structure parameters were deemed significant. For P, spectral features, a height percentile of 80%, and 1 m foliage height diversity were considered crucial.

Reviewer 3 Report

Comments and Suggestions for Authors

Dear authors, 

After reviewed your manuscript, I found some relevant aspects that I would like to share with you, as they are:

- Do you think that selecting a RF feature selection method, could have a negative impact over other two methods?

- Considering that the best model performance (RF 56%) is not high enough, it could affect predictions based on it; so I think it was necessary to include other variables (such as soil properties, water availability, surface and air temperature, etc.)? I am a remote sensing advocated, by the way.

- You did a great work taking so much variables with different types and conditions, so I applaud that valuable effort.

Hope these findings help you.

Best regards, 

Author Response

Dear authors, 

After reviewed your manuscript, I found some relevant aspects that I would like to share with you, as they are:

- Do you think that selecting a RF feature selection method, could have a negative impact over other two methods?

Rely 1: We do not think so. Initially, a total of 112 spectral features, 15 textural features, 33 vegetation indices, 34 height features, and 7 vegetation structure parameters were extracted as parameters for model construction. These features have strong collinearity. Hence, we RF had used feature selection method to select out 32 features. And then, 32 features were used in the three models including Causal band model, Multiple linear regression algorithm, and Random forest algorithm. Therefore, the RF feature selection method will benefit for eliminating collinearity and reducing computational complexity.

- Considering that the best model performance (RF 56%) is not high enough, it could affect predictions based on it; so I think it was necessary to include other variables (such as soil properties, water availability, surface and air temperature, etc.)? I am a remote sensing advocated, by the way.

Rely 2: We agree. We think the plant type is the most important variable that should be included in the context of that the study area is covered by mixed plant communities. Soil properties, water availability, surface and air temperature will also helpful.

We have added this discussion into the section 4.3.

- You did a great work taking so much variables with different types and conditions, so I applaud that valuable effort.

Hope these findings help you.

Best regards, 

Rely 3: Thanks for your precious time.

Reviewer 4 Report

Comments and Suggestions for Authors

A more complete description of the instrumentation – both the LiDAR and the hyperspectral camera is needed.  There is a general lack of sufficient information for a reader to repeat the procedures.  That is a fundamental criterion for publication. 

Specific Comments: 

Line 136: LiDAR 360 data:  LiDAR 360 appears to be a software package, not the LiDAR instrument as suggested in this sentence.  I assume that you acquired the software package from GreenValley, not the data.  Wha was the actual instrument used for data collection, and what are the instrument characteristics?   Was this a full-waveform lidar, multiple pulse lidar, … What was the laser wavelength, the pulse width (or vertical precision), the repetition rate, …. All have a bearing on the character of the results.  It is also important to demonstrate that the authors understand the instrumentation, its capabilities, and its limitations.

Line 147: whiteboard data:  Was this a calibrated Spectralon panel?  Something else? 

Line 160: fresh leave samples è fresh leaf samples   or   fresh samples. Were the leaf samples taken from the top of the canopy?  That would have been challenging for the trees.  I would think it would be important to specify the location of the collected leaves.

Line 162: Leaf samples were dried …:  I’m curious why a wet weight was not take at this point.  The difference between the wet weight and the dry weight have been a useful measure of leaf water content?  Was that not of interest?

Line 191: The Boruta algorithm …  Need a reference – even if this is a standard procedure.

Line 191: causal bands.  I am not familiar with this process.  Again, references are needed. 

Line 203: The Morphological Point Cloud Filtering method … Need a reference that describes the method and, if there are multiple versions, the specific version used.  Presumably this only applies to the lidar data?  That is not clear.  It reads as if this method was used for the spectral data as well.  Be clear!

Line 209-210:  Spectral band selection.  If there were only 112 bands then all were used.  State that.  If a subset of bands were used, then what were the criteria for choosing the bands?  What was the wavelength range?  What was the sampling interval?  Bands width?  

Line 215-216:  So the first two PCA images were used along with a green, red, and infrared band (unspecified) used to create a false color image.  What wavelengths?  By what method?  How is anyone supposed to repeat this procedure? 

I stopped reading at this point. 

Comments on the Quality of English Language

The quality of the English is quite good.  I found a few minor errors in usage, but overall the text is very readable.  

 

Author Response

A more complete description of the instrumentation – both the LiDAR and the hyperspectral camera is needed.  There is a general lack of sufficient information for a reader to repeat the procedures.  That is a fundamental criterion for publication. 

Specific Comments: 

Line 136: LiDAR 360 data:  LiDAR 360 appears to be a software package, not the LiDAR instrument as suggested in this sentence.  I assume that you acquired the software package from GreenValley, not the data.  Wha was the actual instrument used for data collection, and what are the instrument characteristics?   Was this a full-waveform lidar, multiple pulse lidar, … What was the laser wavelength, the pulse width (or vertical precision), the repetition rate, …. All have a bearing on the character of the results.  It is also important to demonstrate that the authors understand the instrumentation, its capabilities, and its limitations.

Rely 1: Thanks for point out the issue.

1. The LiDAR 360 data is an error. It have been revised into LiDAR 360 software.
2. The detail of the instrument used for data collection have been added into section 2.2.1, shown as follows.

LiDAR data was acquired by a LiAir 220 UAV LiDAR system. A total of three flight routes were planned. The flight altitude is 90 m, speed is 5 m/s, horizontal field of view angle is 360°, vertical field of view angle is greater than 20°, and an average point cloud density is 130 points/m². The elevation of the study area was 1200–1296 m, and the number of echoes was 2. The LiDAR 360 data were acquired from GreenValley International (Berkeley, CA, USA) and were used for preprocessing.

Hyperspectral data were acquired by a S185 hyperspectral sensor (Cubert GmbH, Ulm, Germany). For hyperspectral data, eight flight routes were set up to ensure full coverage of the area. The drone had a flight altitude of 140 m, a spatial resolution of 0.19 m, and a lateral overlap rate of 57%, flying a total of four sorties. The spectral range obtained during each flight was 398–1002 nm, with a total of 112 bands.

Line 147: whiteboard data:  Was this a calibrated Spectralon panel?  Something else? 

Rely 2: Yes. We have revised the mentioned sentence, shown as follows.

During the data collection, sensor calibration was carried out using a standard diffuse reflection board, followed by atmospheric correction, geometric correction, band splicing and cropping, image fusion, and other processing.

Line 160: fresh leave samples è fresh leaf samples   or   fresh samples. Were the leaf samples taken from the top of the canopy?  That would have been challenging for the trees.  I would think it would be important to specify the location of the collected leaves.

Rely 3: Yes. leaf samples taken from the top of the canopy. Most of the trees in the study area are 2-4 meters tall, we had used long rod and hook. For the tree more than 5 meters, we had used ladder, long rod and hook.

We had recorded the latitude and longitude of each sample. We have made a statement about the data sharing at the end of this manuscript.

The data presented in this study are available on request from the corresponding author.

 

Line 162: Leaf samples were dried …:  I’m curious why a wet weight was not take at this point.  The difference between the wet weight and the dry weight have been a useful measure of leaf water content?  Was that not of interest?

Rely 4: We are very sorry. We did not determine the leaf water content. We focus on the C, N and P concentration. The leave should be dried before determine the C, N and P concentration.

Line 191: The Boruta algorithm …  Need a reference – even if this is a standard procedure.

Rely 5: The reference has been added.

Kursa, M.B.; Rudnicki, W.R. Feature selection with the Boruta package. Journal of Statistical Software 2010, 36, 11, doi:10.18637/jss.v036.i11.

 

Line 191: causal bands.  I am not familiar with this process.  Again, references are needed. 

Rely 6: Causal bands, namely, causal absorption features (nm) related to foliar biochemicals including C, N, and P [68].

Huber, S.; Kneubühler, M.; Psomas, A.; Itten, K.; Zimmermann, N.E. Estimating foliar biochemistry from hyperspectral data in mixed forest canopy. Forest Ecology and Management 2008, 256, 3, 491-501, doi:https://doi.org/10.1016/j.foreco.2008.05.011.

 

Line 203: The Morphological Point Cloud Filtering method … Need a reference that describes the method and, if there are multiple versions, the specific version used.  Presumably this only applies to the lidar data?  That is not clear.  It reads as if this method was used for the spectral data as well.  Be clear!

Rely 7: Yes. The Morphological Point Cloud Filtering method only was used for LiDAR data. We have revised the mentioned sentence, shown as follows.

The morphological point cloud filtering method [35] was used to distinguish the ground information in the LiDAR data and extract the variables highly correlated with the plant community.

Lohmann, P.; Koch, A.; Schaeffer, M. Approaches to the filtering of laser scanner data. International Archives of Photogrammetry and Remote Sensing 2000, 33, 540-547.

 

Line 209-210:  Spectral band selection.  If there were only 112 bands then all were used.  State that.  If a subset of bands were used, then what were the criteria for choosing the bands?  What was the wavelength range?  What was the sampling interval?  Bands width?  

Rely 8: In this study, all 112 bands, which are collected by a S185 hyperspectral sensor (Cubert GmbH, Ulm, Germany), were used, with a wavelength range of 398–1002 nm and a band width of 5.2 nm. We have added this explanation into section 2.3.2.

Line 215-216:  So the first two PCA images were used along with a green, red, and infrared band (unspecified) used to create a false color image.  What wavelengths?  By what method?  How is anyone supposed to repeat this procedure? 

Rely 9: We have added a formula and refence for the creation of the false color image.

And the formula for calculating the pseudo-color image is as follows [26]:

                 B_pseudo-color=(B_Green+B_Red+B_NIR)/3

Tang, J.; Liang, J.; Yang, Y.; et al. Revealing the structure and composition of the restored vegetation cover in semi-arid mine dumps based on LiDAR and Hyperspectral Images. Remote Sensing 2022, 14, 4, 978-978, doi:https://doi.org/10.3390/rs14040978.

 

I stopped reading at this point. 

Rely 10: Thanks for your precious time.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Thank you for the author's response, which has addressed some of my concerns. However, I found that the author's reply did not directly address certain issues. Since these overlooked concerns are crucial in scientific matters:

Regarding the Point 6 I raised during the first review, concerns about validation and dataset usage. The opacity of the method and data processing may impede the reproducibility of the study.

I have always been concerned about the applicability of this method. In the Point 9 I mentioned during the first review, I suggested that the author add a statement similar to a user manual, but it does not seem to be presented in the revised manuscript.

Instead, I found some new confusing sentences, such as "Finally, our approach is more suitable for semi-arid regions with less diverse vegetation; when applied to other study areas, careful parameter adjustments are necessary" (521-522). I expect the author to declare the applicability conditions of this method, but the author's simple statement "careful parameter adjustments are necessary" seems to be ineffective.