Regional Scale Inversion of Chlorophyll Content of Dendrocalamus giganteus by Multi-Source Remote Sensing
, Qingtai Shu 1,*, Zaikun Wu 1, Li Xu 1, Huanfen Yang 1, Zhen Qin 1, Mingxing Wang 1 and Dandan Duan 2Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThis paper proposes an inversion of chlorophyll content of Dendrocalamus giganteus (bamboo) by combining Landsat 8 and LiDAR GEDI satellite data processed by three regression models (Support Vector Machine; BP neural network; and Random Forest). The paper is well-written in all sections with sufficient details. The results are quite convincing so that it deserves publication after the following minor issues are addressed:
1. In the Introduction section, the authors stated that the primary goal of the study is to use RF machine learning technology to estimate chlorophyll content (L118-119). I suggest excluding the term RF in this objective since there are other machine learning classifiers involved in the data analysis.
2. The description of specific objectives can be improved being more straightforward. My suggestion is:
a) to derive a model to retrieve chlorophyll content from a single D. giganteus plant;
b) to derive an optimal model to invert chlorophyll contents of D. giganteus plants; and
c) to produce a distribution mapping of D. giganteus plants in the study area, based on best attributes derived from multispectral Landsat 8 and LiDAR GEDI satellite data.
3. I suggest using the term Dendrocalamus giganteus only at first time it appears in the manuscript. In the rest of the manuscript, it is simpler to use only its abbreviated form (D. giganteus).
4. The title of Figure 1 is too short. My suggestion: Figure 1. Location map of the study area (Xinping County) in China (a) and in Yunan Province (b). The digital elevation model (DEM) and location of field sampling plots in the Xinping County are shown in (c).
5. In Table 4, please, do not use subscriptions for the following terms: DVI, NDVI, SAVI, EVI, Elevation, Slope, and Aspect. The appropriate citations for the vegetation indices must be provided. Authors need to detail how texture features (mean, variance, synergy, contrast, dissimilarity, entropy, second moment, and correlation) were derived in the text.
6. Figure 2 is little information to the scope of the manuscript. I believe it can be excluded.
7. In the Title of Table 5 (Summary of remote sensing faction information extraction from Landsat 8 OLI data), I believe the correct remote sensing data source is GEDI, not Landsat 8.
8. According to the first paragraph of the 2.4 Research Method section, there are four main steps in the flowchart shown in Figure 4. I suggest using different colors to differentiate these steps in this figure.
9. In Figure 5, I suggest plotting the interval of confidence.
10. In Section 3.2 Single Plant D. giganteus model, please identify the meanings of pai, pgap_theta, and pgap_theta_a3. Same for B2_3_SM, B2_3_HO, B2_5_EN in Section 3.4.
11. I missed the comparison of chlorophyll content values obtained by the authors with other values obtained in literature in the Discussion section.
12. The title of Section 5 is Conclusion, not Discussion.
Comments on the Quality of English LanguageEnglish writting is fine.
Author Response
Point 1: In the Introduction section, the authors stated that the primary goal of the study is to use RF machine learning technology to estimate chlorophyll content (L118-119). I suggest excluding the term RF in this objective since there are other machine learning classifiers involved in the data analysis.
Response 1: Thank you for your valuable suggestions. We have corrected this issue.
Point 2: The description of specific objectives can be improved being more straightforward. My suggestion is:
- a) to derive a model to retrieve chlorophyll content from a single giganteus plant;
- b) to derive an optimal model to invert chlorophyll contents of giganteus plants; and
- c) to produce a distribution mapping of giganteus plants in the study area, based on best attributes derived from multispectral Landsat 8 and LiDAR GEDI satellite data.
Response 2: Thank you for your valuable suggestions. We corrected the description of the specific objectives and improved them.
Point 3: I suggest using the term Dendrocalamus giganteus only at first time it appears in the manuscript. In the rest of the manuscript, it is simpler to use only its abbreviated form (D. giganteus).
Response 3: Thank you for your valuable suggestions. We have corrected this issue. In the manuscript, the full name “Dendrocalamus giganteus” was used when it first appeared, and only the abbreviation “D. giganteus” was used in the rest.
Point 4: The title of Figure 1 is too short. My suggestion: Figure 1. Location map of the study area (Xinping County) in China (a) and in Yunan Province (b). The digital elevation model (DEM) and location of field sampling plots in the Xinping County are shown in (c).
Response 4: Thank you for your valuable suggestions. We have corrected this issue.
Point 5: In Table 4, please, do not use subscriptions for the following terms: DVI, NDVI, SAVI, EVI, Elevation, Slope, and Aspect. The appropriate citations for the vegetation indices must be provided. Authors need to detail how texture features (mean, variance, synergy, contrast, dissimilarity, entropy, second moment, and correlation) were derived in the text.
Response 5: Thank you for your comments and suggestion. We have corrected the problem of abbreviation of vegetation index and index term of DEM extraction in table 4. The vegetation index references related to this study were consulted and cited. At the same time, how texture features are obtained is added in the manuscript.
Point 6: Figure 2 is little information to the scope of the manuscript. I believe it can be excluded.
Response 6: Thank you for your comments. We have deleted the figure 2 that the amount of information in the manuscript is too small.
Point 7: In the Title of Table 5 (Summary of remote sensing faction information extraction from Landsat 8 OLI data), I believe the correct remote sensing data source is GEDI, not Landsat 8.
Response 7: Thank you very much for your feedback. We deeply apologize for the error in the title of Table 5. Based on your feedback, we have corrected it.
Point 8: According to the first paragraph of the 2.4 Research Method section, there are four main steps in the flowchart shown in Figure 4. I suggest using different colors to differentiate these steps in this figure.
Response 8: Thank you for your valuable suggestions. We have used four different colors to distinguish the four main steps of the flowchart according to the first paragraph of the 2.4 research method section.
Point 9: In Figure 5, I suggest plotting the interval of confidence.
Response 9: We took your suggestion. According to your suggestion, we have redrawn the nonlinear fitting 95 % confidence band and prediction band. See Fig.4 for details.
Point 10: In Section 3.2 Single Plant D. giganteus model, please identify the meanings of pai, pgap_theta, and pgap_theta_a3. Same for B2_3_SM, B2_3_HO, B2_5_EN in Section 3.4.
Response 10: Thank you for your reminder. We have specified the meanings of pai, pgap_theta, and pgap_theta_a3 in Section 3.2. Similarly, we have indicated the meanings of B2_3_SM, B2_3_HO, and B2_5_EN in Section 3.4.
Point 11: I missed the comparison of chlorophyll content values obtained by the authors with other values obtained in literature in the Discussion section.
Response 11: Thank you for your comments. We are very sorry here. Under your reminder, we have checked the discussion section. Under the subheading of Section 4.3 of the manuscript, we used the chlorophyll content value of the study area to compare with the results of other literature, and made corresponding supplements. Thank you very much.
Point 12: The title of Section 5 is Conclusion, not Discussion.
Response 12: Thank you for your reminder. We have corrected the title of Section 5 to a conclusion.
Reviewer 2 Report
Comments and Suggestions for AuthorsThe manuscript aims to construct a CCS inversion model using multispectral based vegetation indices in Cunninghamia lanceolata forests. The reviewer appreciates authors’ every effort drawn in the manuscript. Overall comments are as follows.
1. Sampling was conducted from July 12 to 14, 2022. Is the period representative of tree growth?
Is it proper period to collect sample?
2. Based on subsection 2.4.4, this study mentioned that models were evaluated using R2, RMSE, and P. But RSS was used to evaluate the semi-variance function model. Why does this study use RSS?
3. In modeling of SVM, ANN, and RF, the training and validation results were not divided. Is this study trained using all data without test? In machine learning application, checking test or validation performance is essential to evaluate its generalization performance.
Minor points
- Line 25-26, R2, RMSE, and P in test? Or in training?
- Line 17, Full name of GEDI is needed.
- Line 51-52, Rephrase the sentence. The meaning is not clear.
- Line 136, What is the meaning of valley high temperature area? Awkward expression. Mid-mountain warm, alpine cold ..
- Line 145-146, what is the meaning of largest bamboos? Its specific size? Why is it mentioned?
- Line 158, what is DBH? No full name.
- Line 162, what is RTK? No full name.
- Table 1, what is the meaning of “total leaf fresh weight”?
- Line 208, incomplete sentence.
- Line 369-370, incomplete sentence.
- Line 541, Fig. 9 presents importance contribution ratio for RF modeling parameter. In following discussion, Fig. 9 is not associated with model performance and collaborative modeling effect. Needs to be clarify.
- Line 605, Fig. 4 presents research flow chart. Not associated with the discussions. Needs to be clarified.
- There are two discussion sections such as Discussion 4 and 5. Needs to be checked.
- Section 3.1 and 3.2 has same subtitle. Why is it same?
Comments on the Quality of English LanguageModerate editing of English language required.
Author Response
Point 1: Sampling was conducted from July 12 to 14, 2022. Is the period representative of tree growth ?
Is it proper period to collect sample ?
Response 1: In the research area addressed in this paper, the primary focus is on the local economically significant bamboo-Dendrocalamus giganteus. The rainy season in Yunnan is from May to October every year, and the dry season is from November to April of the next year. Among them, the rainfall from mid-June to mid-August is the most, accounting for about 60 % of the annual rainfall. This period of abundant rainfall and favorable climate conditions is critical for the growth of D. giganteus, making it the most representative time to observe changes in its chlorophyll content. In addition, considering the comprehensive factors such as weather, road safety, and representativeness and typicality of field sample plots, it was found that the weather was sunny and cloudless from July 11 to July 16. The road conditions are good. Consequently, field surveys were conducted in the study area from July 12 to July 14 to collect experimental data.
Point 2: Based on subsection 2.4.4, this study mentioned that models were evaluated using R2, RMSE, and P. But RSS was used to evaluate the semi-variance function model. Why does this study use RSS ?
Response 2: In this study, three machine learning models, SVM, BP and RF, were selected to construct the inversion and extrapolation model of chlorophyll content of D. giganteus at regional scale. The leave-one-out cross-validation method was used to evaluate the accuracy of the model, and R2, RMSE and P were selected as the evaluation indexes. These three indexes are the main indexes commonly used to evaluate the prediction accuracy of machine learning models, so that we can quantitatively analyze the performance and prediction ability of the three models to select the best prediction model. When using GEDI footprint points to obtain the continuous distribution of each parameter in the unknown space of the study area through OK interpolation, R2, RSS, range value, nugget value and nugget effect value are mainly selected to comprehensively evaluate the fitting effect of the semi-variance model to find out the optimal semi-variance model of GEDI parameters. Among them, RSS is the sum of the square of the difference between the actual value and the predicted value, indicating the effect of random error. For large sample interpolation data points, the calculation of this value can show the random error of the prediction results of the semi-variance model, and the calculation results are more practical in geostatistics. It is one of the main indicators commonly used to evaluate the OK interpolation to find the optimal semi-variance model.
Point 3: In modeling of SVM, ANN, and RF, the training and validation results were not divided. Is this study trained using all data without test? In machine learning application, checking test or validation performance is essential to evaluate its generalization performance.
Response 3: In this paper, three machine learning models, SVM, BP and RF, are selected, and the estimation accuracy of the three models is evaluated by leave-one-out cross-validation ( LOOCV ). The method is used to participate in training modeling and verification one by one for small sample data. The principle is : based on N overall samples, one sample is extracted as a verification sample each time, and the remaining N-1 samples are used as training samples. Reciprocating until all samples are used as verification samples. Finally, the verification accuracy of the model is calculated according to the predicted results of N cross-validation and the measured values. Compared with K-fold cross-validation and Holdout cross-validation, the verification results are reproducible and not affected by random factors, which makes the model have stronger model generalization ability and stronger robustness.
Minor points
Minor points 1: Line 25-26, R2, RMSE, and P in test ? Or in training ?
Response 1: In lines 25-26, R2, RMSE and P are the results in the test, that is, the model verification accuracy calculated based on the N-fold cross-validation prediction results and the measured values.
Minor points 2: Line 17, Full name of GEDI is needed.
Response 2: Thank you for your reminder. We have added GEDI's full name, Global Ecosystem Dynamics Investigation, to line 17 of the manuscript.
Minor points 3: Line 51-52, Rephrase the sentence. The meaning is not clear.
Response 3: Thank you for your valuable suggestions. We have made a new expression of the sentences in lines 51-52.
Minor points 4: Line 136, What is the meaning of valley high temperature area ? Awkward expression. Mid-mountain warm, alpine cold.
Response 4: Thank you for your reminder. We are very sorry that this sentence is not fluent. We re-stated it under your reminder. Due to the influence of altitude difference, Xinping County has formed three climate types : dry-hot valley high temperature area, semi-mountain warm temperature area and alpine cold temperature area. See lines 140-142 for details.
Minor points 5: Line 145-146, what is the meaning of largest bamboos? Its specific size ? Why is it mentioned ?
Response 5: Respected teacher, in this study we mentioned ' D. giganteus is one of the world 's largest bamboo ', the biggest refers to the average D. giganteus culm height is about 30 m, diameter at breast height is about 15 m, wall thickness is about 14 mm. It is the bamboo species with the largest cultivated area, the widest use and the highest economic value in our province. It is cultivated in tropical and subtropical countries in Asia. Because of its similar bamboo plants such as Dendrocalamus sinicus, Dendrocalamus hamiltonii, Dendrocalamus brandisii, etc., they are also tall and stout, and the planting area is also very wide, so in our study, D. giganteus is one of the largest bamboos in the world.
Minor points 6: Line 158, what is DBH? No full name
Response 6: Dear teacher, DBH refers to the diameter at breast height of D. giganteus. At your reminder, we have added the full name of DBH (diameter at breast height).
Minor points 7: Line 162, what is RTK ? No full name.
Response 7: Dear teacher, RTK is a measurement method used to achieve high-precision positioning in field investigation. It can provide centimeter-level positioning accuracy, and obtain the latitude and longitude coordinates of the center point of the sample plot through data transmission and real-time processing between the reference station and the mobile station. At your reminder, we have added RTK's full name (Real-Time Kinematic) at the time of its first appearance.
Minor points 8: Table 1, what is the meaning of “total leaf fresh weight”?
Response 8: Dear teacher, in our study, ' total leaf fresh weight ' refers to the total weight of each leaf of 137 strains of D. giganteus.
Minor points 9: Line 208, incomplete sentence.
Response 9: Thank you for your reminder. We have completed the sentences in the manuscript.
Minor points 10: Line 369-370, incomplete sentence.
Response 10: Thank you very much for your reminder. We have already completed this sentence.
Minor points 11: Line 541, Fig. 9 presents importance contribution ratio for RF modeling parameter. In following discussion, Fig. 9 is not associated with model performance and collaborative modeling effect. Needs to be clarify.
Response 11: Thank you for your valuable suggestions. We have corrected the errors here in the manuscript and checked the rest of the full text.
Minor points 12: Line 605, Fig. 4 presents research flow chart. Not associated with the discussions. Needs to be clarified.
Response 12: Thank you very much for your reminder. We have corrected the errors here in the manuscript, and checked and corrected all the graphic descriptions of the full text.
Minor points 13: There are two discussion sections such as Discussion 4 and 5. Needs to be checked.
Response 13: Thank you very much for your reminder. We have corrected this error in the manuscript.
Minor points 14: Section 3.1 and 3.2 has same subtitle. Why is it same?
Response 14: Thanks for your reminder. We apologize for this error, and at your reminder we have corrected the subtitle of Section 3.2.
Reviewer 3 Report
Comments and Suggestions for AuthorsDear authors. I am extremely grateful for the opportunity to read the manuscript of the scientific article "Regional scale inversion of chlorophyll content of Dendrocalamus giganteus by multi-source remote sensing". The article is devoted to an interesting and relevant topic, since measuring the chlorophyll content in plants is a key aspect of monitoring their health and productivity. The study fits into the subject of the scientific publication "Forests" (ISSN 1999-4907).
There are several comments.
1. Figure 1. Specify the signatures of China's neighboring countries, geographical features (seas, etc.). The inscription "China" is geographically located somewhere above the territory of Mongolia or Russia - this is incorrect.
2. Figure 1a. The semantic load of the inset map is unclear at the bottom right.
3. Figure 1b. also show the signatures of the neighboring countries. Figure 1c. The inscription of the name of the province should also not be in the air on a white background. The same remark as in paragraph 1 with the name of the country.
4. Lines 213-214. What exactly 76 factors are we talking about? How do they relate to table 1? Have you used all this in your work?
5. The research methodology is not completely clear. How were the remote sensing and field research data correlated? How does the data obtained at the points for individual trees (as the reviewer understood) relate to the average calculated value for a pixel of a landsat satellite image with a resolution of 30*30 m?
6. Describe the field studies in more detail. How the selection was conducted. How many plants were selected from each key site?
7. Why was the GEDI data obtained in one year and the Landsat data obtained in only one day?
8. Figure 4 does not give any idea about the research methodology. How were the blocks of incoming information processed? How is the outgoing information received? It is extremely problematic for another researcher to reproduce the research methodology in full. Section 2 is written quite difficult for analysis and perception.
9. What programs/programming languages were used to get the result? The results, for example, mention (line 396) software, but this should be described in detail in section 2.
10. Line 23. "pai, pgap_theta and pgap_theta_a3". This appears in the annotation and then in the results. The research methodology does not say anything about this. The methodology specifies appendix A. However, no additional explanations are provided.
11. Low-quality drawings. It is impossible to read part of the data (especially Figures 7, 8, 9).
12. In section 4, a comparison of the results of the study with other authors is poorly presented.
13. Line 653. Probably the title of the section "Conclusion".
Author Response
Point 1: Figure 1. Specify the signatures of China's neighboring countries, geographical features (seas, etc.). The inscription "China" is geographically located somewhere above the territory of Mongolia or Russia - this is incorrect.
Response 1: Thank you for the expert comments and suggestions. We have redrawn Figure 1, indicating neighboring countries and maritime boundaries. See Figure 1 for details.
Point 2: Figure 1a. The semantic load of the inset map is unclear at the bottom right.
Response 2: Thank you for the expert 's reminder. The lower right corner of Figure 1a is China 's southern cities and the South China Sea region. We have re-marked and clearly expressed in the figure.
Point 3: Figure 1b. also show the signatures of the neighboring countries. Figure 1c. The inscription of the name of the province should also not be in the air on a white background. The same remark as in paragraph 1 with the name of the country.
Response 3: Thank you for the expert 's reminder. According to your suggestion, we marked the location of the neighboring countries in Figure 1b, and deleted the name of the study area marked in Figure 1c. Annotations are made below Figure 1.
Point 4: Lines 213-214. What exactly 76 factors are we talking about? How do they relate to table 1? Have you used all this in your work ?
Response 4: We extracted 73 remote sensing factors from Landsat 8 multi-spectral images and 3 terrain factors from DEM data, a total of 76 feature variable factors, to construct the initial feature variable set at the regional scale. Table 1 counted the basic information of 137 standard long bamboos, such as the total number of samples, total leaf fresh weight, maximum value, minimum value, mean value and other information, to fit the basic model of chlorophyll content of single D. giganteus, and calculate the chlorophyll content at the plot level, which was used as the training sample (dependent variable) of the modeling, and the eigenvalue of 76 characteristic variable factors on the corresponding sample site was extracted as the modeling sample (independent variable), both of which indirectly belong to the relationship between the dependent variable and the independent variable. These characteristic variables are indispensable in the construction of regional scale prediction models and the study of chlorophyll inversion of D. giganteus in the study area. Under your valuable comments, we have supplemented and explained the research methods in detail, as shown in Section 2.4 of the article.
Point 5: The research methodology is not completely clear. How were the remote sensing and field research data correlated ? How does the data obtained at the points for individual trees (as the reviewer understood) relate to the average calculated value for a pixel of a landsat satellite image with a resolution of 30*30 m ?
Response 5: Thank you for the expert comments. In Table 1, we compiled the basic information of 137 standard D. giganteus, such as the total number of samples, total leaf fresh weight, maximum value, minimum value, and mean value. This data was used to fit the basic model of chlorophyll content for a single D. giganteus and to calculate the chlorophyll content at the plot level, which served as the training samples (dependent variable) for modeling. The eigenvalues of the remote sensing characteristic variable factors extracted from the corresponding sample sites were used as the modeling samples (independent variable). Both indirectly relate to the relationship between the dependent variable and the independent variable. The standard plot size set in this study is 30 m*30 m, while the spatial resolution of Landsat8 is 30 m*30 m, which perfectly matches, solving the issue of matching plot size with image pixels, and reducing the uncertainty in estimating the chlorophyll content of D. giganteus.
Point 6: Describe the field studies in more detail. How the selection was conducted. How many plants were selected from each key site ?
Response 6: Thank you for your valuable suggestion. In response to this question we have made further additions in Sections 2.2 and 2.2.1 of the manuscript, thank you.
Point 7: Why was the GEDI data obtained in one year and the Landsat data obtained in only one day ?
Response 7: Thank you very much for the expert 's comments. The acquisition time of GEDI L2B product data is from January 1, 2022 to December 31, 2022, and the imaging time of optical image Landsat 8 OLI is September 24, 2021, because the distribution form of L2B spot data is a profile distribution along the GEDI running track. In the experimental design, in order to ensure that the GEDI L2B product data evenly covers the entire study area, the study selected the L2B product data from January 1, 2022 to December 31, 2022 as the experimental remote sensing data. In our study, the research object is the local main economic D. giganteus. The rainy season of Yunnan is from May to October every year, and the dry season is from November to April of the following year. Among them, the rainfall from mid-June to mid-August is the most, accounting for about 60 % of the annual rainfall. During this period, the rainfall is abundant and the climate is suitable. It is the main growth period of D. giganteus, and it is also the most representative period to reflect the change of chlorophyll content of D. giganteus. In addition, considering the comprehensive factors such as weather, road safety, and the representativeness and typicality of field sample setting, after screening, it is found that during the period from July 11 to July 16, the weather is sunny and cloudless. The road conditions are good. Therefore, from July 12 to July 14, the field survey was carried out in the study area to collect experimental data. In order to match the sampling time of the field survey, the Landsat 8 OLI image with the best image quality in the near time was selected ( September 24, 2021 ).
Point 8: Figure 4 does not give any idea about the research methodology. How were the blocks of incoming information processed? How is the outgoing information received? It is extremely problematic for another researcher to reproduce the research methodology in full. Section 2 is written quite difficult for analysis and perception.
Response 8: Thank you very much for the expert 's comments. We are sorry that the research method is not clear enough. In your reminder, we use four different colors to distinguish the four main steps of the flow chart. And the materials and methods of Section 2 were modified, as detailed in Section 2.4 of the article.
Point 9: What programs/programming languages were used to get the result? The results, for example, mention (line 396) software, but this should be described in detail in section 2.
Response 9: Thank you for your valuable suggestion. In this study, SVM and BP models were realized by MATLAB R2023a software, and RF model was realized by Python3.10 version programming, and then the remote sensing inversion of chlorophyll content at regional scale was successfully realized. In the three models, we use the fitrsvm function in the statistics and machine learning toolbox of MATLAB R2023a software to train the SVM model. The kernel function selects the Gaussian kernel, and optimizes the hyperparameters through the random search optimizer. Randomly sample within the predefined range of the hyperparameters and evaluate the performance of each sampling point. For the BP model, we use the Neural Network Toolbox built in MATLAB R2023a software to realize the regression modeling of BP neural network. The RF model mainly uses Python 's sklearn, pandas, and numpy libraries to realize the construction of the estimation model and the inversion of the chlorophyll content of the D. giganteus in the study area. Among them, the RF model mainly calls the RandomForestRegressor module, the SVM model mainly calls the fitrsvm module, and the BP model mainly calls the MLPRegressor module. Under your valuable suggestions, we have done a detailed description in Section 2.4.3, thank you.
Point 10: Line 23. "pai, pgap_theta and pgap_theta_a3". This appears in the annotation and then in the results. The research methodology does not say anything about this. The methodology specifies appendix A. However, no additional explanations are provided.
Response 10: Thanks for the expert 's reminder. According to your reminder, we have added notes to the manuscript, as detailed in Table 6, Figure 6, and Appendix A.
Point 11: Low-quality drawings. It is impossible to read part of the data (especially Figures 7, 8, 9).
Response 11: Thanks for the expert 's comments. According to the initial requirements of the journal, we provided images with a resolution of 300 dpi, which led to a lack of clarity. According to the advice of experts, we have redrawn the map to make its data clear and will increase the resolution to 600 dpi.
Point 12: In section 4, a comparison of the results of the study with other authors is poorly presented.
Response 12: Thank you for your comments. We are very sorry here. Under your reminder, we have checked the discussion section. Under the subheading of Section 4.3 of the manuscript, we used the chlorophyll content value of the study area to compare with the results of other literature, and made corresponding supplements. Thank you very much.
Point 13: Line 653. Probably the title of the section "Conclusion".
Response 13: Thank you very much for the expert 's reminder. We have modified the title to ' Conclusion '.
Round 2
Reviewer 2 Report
Comments and Suggestions for AuthorsAll comments have been incorporated into the revised manuscript.
Comments on the Quality of English LanguageMinor editing of English language required
Author Response
We are very grateful to the experts for their hard work on our manuscripts. We have checked and corrected the full text of the manuscripts again. Thank you for your valuable suggestions and patient guidance to make our manuscripts better.
Reviewer 3 Report
Comments and Suggestions for AuthorsAccept in present form
Author Response
Thank you very much for your valuable suggestions and patient guidance for our manuscript. We take this opportunity to check and correct the full text of the manuscript again. Thank you for your hard work and make our manuscript better. Thank you.
