Automated Recognition of Tree Species Composition of Forest Communities Using Sentinel-2 Satellite Data
Round 1
Reviewer 1 Report
The manuscript is devoted to an actual problem and contains new scientific results. However, a number of corrections and clarifications need to be made:
1) L. 167: It is necessary to justify the choice of the NDVI vegetation index for recognizing tree species. The manuscript should be supplemented with appropriate references.
2) L. 186. It is necessary to justify why not all tree types were analyzed in the test sample.
3) The terminology needs to be corrected. In the text of the manuscript, the term "channel" should be corrected to "band" or "range". Such a correction must be made in line 74 and elsewhere in the manuscript.
The term "reflection coefficient" should be corrected to "reflectance" or "spectral reflectance". For example, such a correction must be made in line 154 and other parts of the manuscript.
It is necessary to edit sentences with multiple "of", for example, in lines 16, 57, 102, 103, 288.
4) L. 128. In Figure 1 (left part), it is necessary to remove the buffer zones along the coastline.
5) L. 311-312. Figure 6. Scale bar should be shown.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 2 Report
Automated recognition of tree species composition of forest communities using Sentinel-2 satellite data
This manuscript analyzed the possibilities of tree species discrimination in coniferous-deciduous forests according to Sentinel-2 data using two automated machine learning methods: random forest (RF) and generative topographic mapping (GTM). The RF method showed a higher recognition accuracy. The Authors consider that this approach can be useful for updating forest inventory data and for checking the information on the ground forest inventory.
The topic of the manuscript is of wide interest in remote sensing community and the research design is innovative, I have some minor comments regarding the main Sections of the manuscript.
Review summary
While the topic of this manuscript is in principal suited for Remote Sensing, I am seeing several methodological flaws in this study. My concerns are:
· LN 79: please introduce full name of the abbreviations NDVI, EVI
· Last paragraph should clearly indicate the scientific contributions of the research. Therefore, LN 108 – 112 are not needed in this part of the paper
· Please show spectral bands of Sentinel-2 and dates of remote sensing data in a Table (Section 2.3)
· LN 181: what is a sample?
· LN 301: VSURF is mentioned in this part, but then it also needs to be described in the Methods section. Furthermore, where are the most important variables showed that were obtained with VSURF?
· Adjust the validation results, in some parts of the paper the accuracy is showed within the 0 -1 range, and somewhere 0 – 100. Also, which conclusion can be derived from validation results on a training data. In other words, is it really necessary?
I think that the research design of this manuscript is very good, but still some major changes need to be made, in order to be published in this journal. My only concern is that the reference data is from 2013, and Sentinel-2 imagery are from 2020, which represents a large time gap. However, a very interesting research.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 3 Report
Authors analyzed the tree species discrimination in coniferous-deciduous forests according to Sentinel-2 data using two automated recognition methods: random forest (RF) and generative topographic mapping (GTM). Generally, this manuscript is well-written, well-structured. After minor revision, this manuscript coubld be accepted for publication.
Line 69: in the article, is it appropriate to add the reference [25], suggest to use the appropriate citing format.
Line 129: in figure 1, please add the label of each region in the left sub-figure.
Line 147: in figure 2, please add the longtitute and latitute. E W S or N.
Line 199: in section 2.6, suggest add some formulas of two methods RF and GTM. Existing description is a little vague.
Line 344: in section 4, suggest divide the discussion part into three parts, validation of your results, comparison with other similar studies, limiations and further study.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 2 Report
I am satisfied with the Author's response, and the manuscript has improved from v01 to v02.
From my point of view, it is ready for publishing.