Remote Sensing Classification and Mapping of Forest Dominant Tree Species in the Three Gorges Reservoir Area of China Based on Sample Migration and Machine Learning
, Bin Xu 4, Jiufen Liu 1,2,3, Hongyu Li 1,2,3, Xiaofeng Zhao 1,2,3, Xinping Luo 1,3, Ran Wang 1,2,3, Liyuan Xing 1,3, Chao Wang 1,3 and Honghui Zhao 1,3
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsAccurately obtaining the distribution of dominant tree species is crucial for forest ecosystem assessment. This study proposed a method for transferring single-year sample data to multiple years to achieve multi-year classification and extraction of dominant tree species. The results demonstrated that the CCDC algorithm is feasible for sample migration and the XGB outperformed other machine learning classification techniques, achieving the best classification accuracy. This study might be interesting to the Remote Sensing readership but still required potential improvements before being published. The main issues are as follows:
1) What type of satellite product does this study use? Level 1 or Level 3? If it is not surface reflectance data, does it require data preprocessing?
2) Why does this study use the CCDC algorithm with Landsat images starting from 1986? Since the study focuses on the extraction of dominant tree species starting from 2018, could Sentinel-2 images starting from 2015 be used instead?
3) This study generated seasonal median Sentinel-2 images based on the GEE platform, resulting in four images per year. Considering the phenological characteristics of different dominant tree species, the classification results of the seasonal images vary. The study did not explain how the annual classification accuracy for each tree species was obtained.
4) Although this study utilized the CCDC method to migrate sample data acquired in 2023 to the years 2018-2022, specific quantities of samples for the dominant tree species were not provided. It is unclear whether the migration sample sizes for different years are balanced with the proportion of dominant tree species in the study area, and whether the classification accuracy of dominant tree species is affected by the uneven distribution of samples.
5) In the ‘Discussion’ section, add a mechanistic explanation of why XGB is superior to other classification methods.
The detailed issues:
Line 54: The citation format for the references is incorrect. Change ‘Waster L.T. et al.’ to ‘Waster et al.’, please check the entire manuscript and revise it (Lines 75, 93, 97,114, etc.).
Line 71: Change ‘Sentinle-2’ to ‘Sentinel-2’. Please check the entire manuscript and revise it (Lines 78, 156, 157, 167, etc.).
Line 133: The citation format for the figures is incorrect. Change ‘e.g., Fig. 1’ to ‘Figure. 1’.
Lines 174,198: Change ‘indexes’ to ‘indices’. Ensure consistent wording throughout the entire manuscript.
Line 205: The unit (m2) is missing after 900.
Line 238: Please remove the comma after ‘Name’.
Line 245: The text format is incorrect.
Line 278: Change ’18-22’ to ‘2018-2022’. Similarly, change ’18-19’ to ‘2018-2019’ in line 284.
Line 359: Change the "、" in the table to ",".
Lines 368, 376: "Figure 5" is used at line 368, while "Fig. 5b-d" is used at line 376. There are too many inconsistencies throughout the manuscript. Please review and make corrections.
Line 394: There is a missing space before "Classification models based".
Line 397: Please delete ‘RF algorithms also show strong’.
Line 401: There is a missing ‘%’ after 75.
Line 425: Please provide the classification map of dominant tree species to replace the same background images of (a)-(f) for 2018-2023.
Lines 495-496, 499-500: There are grammatical errors in the usage of "Although...however".
Line 512-513: This sentence is quite difficult to understand. ‘more than 60,000 hm2’ or ‘more than 60,000 images’?
Lines 499-520: "Similarly" is overused. Please use alternative conjunctions.
Lines 523-524: There are grammatical errors in the usage of "Even though...however".
Comments on the Quality of English LanguageThere are some grammatical errors in this manuscript. Please carefully check and revise it.
Author Response
Please see the attachments
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsThe article entitled " Remote sensing classification and mapping of forest dominant tree species in the Three Gorges reservoir area of China based on sample migration and machine learning" presents an excellent method for forest dominant tree classification using inter-annual migration of sample data and classifications such as the Continuous Change Detection and Classification (CCDC) algorithm. This approach achieves excellent R-Squared and RMSE values.
The structure of the research is well-organized and comprehensive, but there are some recommendations provided below:
NBR (Normalized Burn Ratio) is not described in the text.
In line 205, the units for 900 are not specified. Based on explanations in other sections, the readers can assume “900 square meters”.
Line 302 appears to have a typographical error.
Line 397 appears to have a typographical error.
In Figure 8, it is difficult to detect the five dominant tree species. To enhance the understanding of the obtained classification, a close up of the dominant tree species classification is recommended. At a global scale, it is not possible to depict the tree species mapped and classified. Incorporating this detail can also support the excellent values obtained in validation of the research.
Comments on the Quality of English LanguageMinor editing of English language required.
Line 302 appears to have a typographical error.
Line 397 appears to have a typographical error.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for AuthorsThe paper proposed a method combining sample migration machine learning for multi-year tree species classification in China. The article has a certain degree of innovation and is helpful for subsequent papers.
1. Sentinel is spelled incorrectly;
2. 2.2.3 The actual measurement data doesn't seem to be written very specifically; Is the experiment of transferring years in the experimental section feasible?
3. The words and phrases still need to be condensed. I think there are some areas where colloquialism is not scientific enough.
4. Conclusion can supplement quantitative evaluation.
5. The abstract can be further condensed.
6. How to unify the resolution of multi-source data.
7. How did the author consider temporal differences
Comments on the Quality of English LanguageThe words and phrases still need to be condensed. I think there are some areas where colloquialism is not scientific enough.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsThe revised manuscript has been improved following the reviewer's suggestion.
