R-MFNet: Analysis of Urban Carbon Stock Change against the Background of Land-Use Change Based on a Residual Multi-Module Fusion Network
Round 1
Reviewer 1 Report
This paper improved the deep-learning classification by using the attention mechanism and ASPP and mapped Landsat images at 3 typical years while analyzing the relationship between land cover change and carbon stock change.
Overall, the method is ok. Ablation experiments should be added.
The proposed method compared two deep learning methods. Other methods such as the UNet can be compared.
Since GEE provides a time series of Landsat images, why did you select these three years (2001, 2009, 2020)? If mapping more annual land cover maps, will the carbon stock model estimation results be improved?
The validation data are not explained.
Languages should be polished.
Author Response
Thank you. Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
Thank you so much for inviting me to review this manuscript. The topic and method applied seems interesting and under the scope of the journal. In this manuscript, the major concerns are:
1. There are no spaces between the main text and the citation, and the text does not use proper spaces, punctuation and initial format.
2. Incorrectly formatted formulas and variables.
3. Check the formatting of secondary headings, tertiary headings, figure and table headings in articles for compliance with journal requirements.
4. Words following ";" are generally lowercase.
5. Language and grammar errs are sever.
6. Land use types in articles are not named consistently.
7. Discussion and conclusion has major problem.
Materials and Methods
1. Line 158: Soil type richness in which?
2. Figure 1: Please refer carefully to journal requirements for figure caption formatting.
3. Line 173-174: Inaccurate citation format and website information unrelated to the content of the article.
4. Table 1: "Water body" and "Construction land" appear in an awkward position in the table. This punctuation after “distribution” should be removed.
5. Line 216: Incorrect citation author format.
6. Line 227: This place uses the "from. . to..." has an ambiguous meaning.
7. Table 2: What does "Ci-above..." mean, this needs to be clearly explained.
8. Line 359: Which ones?
9. The four variables on the right-hand side of the equation (12) are in a different format to those in Table 2.
Results
1. Table 3: There are six land use types, and only three land use types are listed in Table 3.
2. Line 440: The description of the distribution in this section could be more detailed, it is too rough with just the direction.
3. Line 444-446: Is this phrase shown in Figure 8, I can't find a basis for this phrase. If you need topographic support in your text, why not create an elevation map in the study area?
4. Line 454: Figure serial number error.
5. Line 457: Read and read.
6. Line 458-462: This section repeats the analysis above, or perhaps the changes from 2001-2009 should have been analysed above.
7. Line 466: I suggest that each value is followed by the unit.
8. Line 470-471: The phrase means the area of change in grassland, whereas your value is written as the area of grassland in 2009.
9. Line 472-474: Have a proper break in the sentence.
10. Figure 8: Please align the borders in the legend.
11. Line 522-526: Which year?
12. Line 545: What does "them" mean here?
Discussion
Poor discussion because there is no adequate deep interpretation and argue with literature.
1. Line 604-612: The discussion section is needed to support your research with the findings of others, rather than presenting their results.
2. Line 616-632: Specific policies, the passage has no supporting material.
3. Line 632-627: What is the relevance of the conclusions in this literatures to your discussion?
Extensive editing of English language required
Author Response
Thank you. Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Table 2: Since you mapped land covers in three years, the number of samples in the years (2001 and 2009) should also be included.
Table 4: Show the meanings of the per-class accuracy, user’s or producer’s accuracy.
Table 7: Do not use kappa*100 because kappa ranges from -1 to 1.
Languages can be improved by native speakers.
Author Response
Dear editor:
Thanks for your letter and for reviewer’s comments concern our manuscript. Those comments are valuable and helpful for revising and improving our paper, as well as the important guiding significance to our researches. We have studied all comments carefully and have made correction. Revised portion are marked in red in the paper. The main corrections in the paper and the responds to the reviewers’ comments are as flowing:
Reviewer 1
- Table 2: Since you mapped land covers in three years, the number of samples in the years (2001 and 2009) should also be included.
Response: Thank you very much for your suggestion. We have improved the information in the table by including sample information from 2001 and 2009, and changed the name of the table to make it logical, as shown below:
Table 2. Total samples, training samples and verified samples in Zhengzhou city.
|
|
Total number of samples |
Training samples |
Validation samples |
||||||
|
2001 |
2009 |
2020 |
2001 |
2009 |
2020 |
2001 |
2009 |
2020 |
|
|
Farmland |
22544 |
26505 |
39935 |
6786 |
7934 |
11920 |
15758 |
18571 |
28015 |
|
Woodland |
4856 |
15550 |
11577 |
1468 |
4742 |
3562 |
3388 |
10808 |
8015 |
|
Grassland |
1160 |
1391 |
2036 |
352 |
427 |
723 |
808 |
964 |
1313 |
|
Water body |
12544 |
5998 |
19026 |
3789 |
1785 |
5603 |
8755 |
4213 |
13423 |
|
Built land |
54357 |
39380 |
72366 |
16371 |
11846 |
22398 |
37986 |
27534 |
49968 |
|
Other land |
689 |
543 |
2620 |
228 |
154 |
760 |
461 |
389 |
1860 |
- Table 4: Show the meanings of the per-class accuracy, user’s or producer’s accuracy.
Response: The accuracy of each of the categories we mention in the text is the precision which is analogous to the user’s accuracy.
|
Actual Predict |
0 |
1 |
|
0 |
TN |
FN |
|
1 |
FP |
TP |
This is the formula we use to calculate the accuracy of each land type in the code.
The formula for user’s accuracy and producer’s accuracy in remote sensing image classification is:
We have reviewed the accuracy calculation formula for each type of object in the code and confirmed that it belongs to user’s accuracy, and we have marked it in the text.
- Table 7: Do not use kappa*100 because kappa ranges from -1 to 1.
Response: We multiply the kappa coefficient by 100 to gain a clearer understanding of the performance improvement of each module, we have modified it.
Author Response File: Author Response.pdf
Reviewer 2 Report
Thank you very much for inviting me to review this manuscript. After times revision, the manuscript has improved significantly. Thus, I have no appending comment.
Author Response
Dear editor:
Thanks for your letter and for reviewer’s comments concern our manuscript. Those comments are valuable and helpful for revising and improving our paper, as well as the important guiding significance to our researches. We have studied all comments carefully and have made correction. The main corrections in the paper and the responds to the reviewers’ comments are as flowing:
Reviewer 2
Thank you very much for inviting me to review this manuscript. After times revision, the manuscript has improved significantly. Thus, I have no appending comment.
Response: Thank you for your valuable suggestions on the manuscript, which greatly improved its quality. Thank you again for your appreciation of our work.
Author Response File: Author Response.docx