Mapping Cork Oak Mortality Using Multitemporal High-Resolution Satellite Imagery
Round 1
Reviewer 1 Report
Forests matter. This manuscript presented a chain-like model to map the cork oak forests. Concretely, the model includes individual tree delineation, vectorization of each tree or tree cluster, tree center identification, and K-mean based clustering. Apparently, the AUTHORs do not do a good job of placing their literature within the context of current literature. As mentioned in ms (In the case of cork oak forests, where the density is generally low and the tree crowns are individualized…), if so, the computer vision-based object detection (e.g., YOLO or mask RCNN) approaches can be more effective to complete such easy task of dead tree mapping. The proposed method is so completed and less effective that I did not see the value of this ms to be published.
Line 94-96. Move to the acknowledgments section.
Line 239-241. Euclidean distance within each polygon? How did you do this? Draw a figure to show how this step was conducted.
Line 438-441, For an imbalanced dataset, I recommend using the macro-average of the F1 score.
Author Response
please see the response in the attached word file
Author Response File: 
Author Response.docx
Reviewer 2 Report
The paper entitled "Mapping cork oak mortality using multitemporal high-resolution satellite imagery" by Catalão et al. addresses the important topic of forest mortality in the Mediterranean region. The paper is well written and interesting to read.
In my opinion, it deserves to be published soon after minor comments have been made:
Figure 4 and Figure 6: The size of the characters in the x/y is very small, please try to increase it. Please consider also to increase the thickness in the plots.
P9,L376: "Dead trees have...". Do you have a possible explanation for the larger dispersion of the dead trees?
P10,L396: I would suggest to eliminate the word "clusters" at the end of the sentence, as it was stated before in the sentence that the ginumbers are referred to "clusters".
P10,L392: typo, remove the space before "ΔNDVI"
Author Response
please see the response in the attached word file
Author Response File: Author Response.docx
Reviewer 3 Report
The manuscript is interesting and okay for those interested in tree mortality research. However, most of the findings seem technical in the manuscript. I would suggest please provide a comprehensive understanding ( why and how? ) of the decrease in tree mortality in Portugal for better knowledge to other researchers.
Line 25–39: I understand the author(s) disused the tree mortality and causes in the Mediterranean region in the first paragraph of the introduction section. However, the author(s) should also explain specifically the cause of tree mortality in the study site since the study site is small in size, and the causes can not be generalized. Proper characterization of the reasons for the increase in tree mortality is needed for better understanding.
Line 40–50: The author(s) was specific about cork oak in this paragraph. Since the author mentioned different forest species in the abstract, what about the other forest species like pine and holm oak? I think, Need to provide a comprehensive overview of the forest species decline and their actual causes in such a small site.
Line 51–57: I understand that the author(s) explained the importance of earth observation data and remote sensing-based indices. Do we need high-resolution data to identify the tree mortality in such small areas in size? Although the author(s) discusses it partially, what was the advantage or disadvantage of not using MODIS or Landsat data to analyze tree mortality by considering the study size, scale, and resolution?
The author(s) did not discuss well why tree mortality is concerned of selected forest species, particularly in the study site or Portugal ( as mentioned in the abstract)?
Sorry, I could not understand the importance of the yellow mark in figure1. It seems non-forest areas.
Line 94–97: I think these lines are stubble in the acknowledgment or grant section. Please remove.
Please add the implication/limitation of such studies using High-resolution satellite data and briefly indicate the advantage or disadvantage of low-medium resolution remote sensing satellite data? Do the author(s) have any solution to analyzing tree mortality using low-medium-high repopulation satellite imagery, especially for large areas?
Please make sure all dash, space, a hyphen, en dash, and capital words would be appropriate throughout the manuscript.
Please make sure the font size is in the table and figure.
Please make sure the abbreviation is clearly defined in the text.
Author Response
please see the response in the attached word file
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
Rephrase the replies and add them to the introduction section to highlight the contributions and novelty of this study.
No descriptions of the training and validation dataset (samples and counts for each).
Author Response
Please see the attachment
Author Response File: Author Response.docx
Reviewer 3 Report
Authors improved the manuscript according to reviewers comments. I think, manuscript can be accepted.
Author Response
Thank you for your comments and suggestions
