An Estimation Method for PM_2.5 Based on Aerosol Optical Depth Obtained from Remote Sensing Image Processing and Meteorological Factors

Round 1

Reviewer 1 Report

Thank you for giving me this opportunity to read the manuscript entitled " An estimation method for PM2.5 based on aerosol optical depth obtained from remote sensing image and meteorological factors". The topic of this manuscript is interesting and would be a good contribution to this field. I think it could be considered for publication in Remote Sensing once the following issues are addressed.

1. Please replace the keywords that already appear in the manuscript’s title (e.g., PM2.5) or are too long (e.g., Remote sensing image processing) with close synonyms or other keywords, which will also facilitate your paper to be searched by potential readers.

2. To my knowledge, “back-propagation neural networks (or back propagation neural networks)” are often abbreviated as “BPNN” instead of “BP NN”.

3. Line 37 – 38: Some newly published papers are suggested to be cited as references here. For example the papers titled “Dynamic assessment of PM2. 5 exposure and health risk using remote sensing and geo-spatial big data” and “PM2.5 polluters disproportionately and systemically affect people of color in the United States”.

4. It is good to see the authors comparing their model with other commonly used machine learning models. However, I suggest that the authors try to compare their method with the to highlight the scientific value of the method proposed in the manuscript current more advanced PM2.5 estimation models based on remote sensing data, rather than just comparing a few machine learning models.

5. Please improve the resolution of the Figures used in the manuscript, as the current pictures are not clear and it is hard to read the text information.

6. Some grammatical errors exist. A critical review of the manuscript language will improve readability.

Author Response

Dear reviewer:

We feel great thanks for your professional review work on our manuscript. Based on your comment and request, we have supplemented references here and we have made extensive modification on the original manuscript. A document answering every question from the referees was also summarized and enclosed.

A revised manuscript with the correction sections was attached as the supplemental material and for easy check and editing purpose. If you have any questions, please contact us without hesitation.

Best wishes.

Sincerely

Author Response File: Author Response.pdf

Reviewer 2 Report

The manuscript compared several statistical methods for retrieving PM2.5 concentrations from MODIS AOD and/or weather data and found that the backpropagation neural network (BPNN) offered the best correlation. Does the finding make sense? Yes. But the manuscript is not a quality one in its current form. My concerns and comments are as follow:

Major ones –

• For most BP NN modeling runs, the authors only used two or three hidden layer neurons (HLNs). The small number of HLNs is apparently related to the low dimension of input data sets (AOD and/or meteorology). Is the neural network necessary in this case? To me, the potential of neural network methods cannot be fully utilized. The authors should provide a better justification for the necessity of BP NN.
• (Again, regarding the necessity of BP NN) As per Table 7, the BP NN didn’t show significant improvement as compared to traditional methods such as multi-variable linear regression (MLR). An increase in R by 0.02 does not convincingly justify the use of BP NN for future similar research.
• Why R? Why not adjusted R² (as a measure of correlation)? As per Line 521, many previous studies used R².
• Though the authors stated that a temporal-spatial analysis was performed, I didn’t see any statistical spatial analysis. A descriptive comparison of different regions was given; but for AOD-PM2.5 modeling, no spatial analysis elements (e.g., spatial correlation) were incorporated.  
• Why was Region II included? The region has no ground-based PM2.5 monitoring stations. How did it help with the methodology (BP NN) development (as the title indicated)?

Minor ones –

• The manuscript can be more concise. In particular, section 2.3.2 should be shortened substantially. Section 2.3.4 is uncessary as the definitions of R, RMSE, and SD are common knowledge.
• Table 1: Provide p-values in the top-right half of the table.
• Line 284: What “a” number did you use to determine the number of hidden layer neurons?
• Line 321: Provide a reference for the LIBSVM method.
• Line 399-400: How do you explain the high AOD area in the northwest (Wafangdian)?
• Line 405: Firmly? Provide correlation coefficients here if there are.
• Line 413-427: You don’t have to elaborate on the model tweaking process. They are less relevant.
• Line 446: The increases in R values are too small. BTW, I don’t think it is meaningful to discuss the incorporation of each individual meteorological parameter. That being said, Figure 10 can be removed.
• Line 451: 0.2 should be 0.02; 0.28 should be 0.028.
• Figures 9 and 11: There are hard to read. Increase font size.
• Figures 9 and 11: Why not compare model-predicted PM2.5 versus measured PM2.5? The figures I suggest would be more relevant, IMO.
• Line 468: 0.2 should be 0.02.
• Table 6. A question – why did you decide the best model based on R not RMSE? For nonlinear regression, RMSE is usually preferred.
• Line 523: R=0.81 so a linear relationship between AOD and PM2.5 did exist as per Table 1.
• Lines 543-544: I don’t quite understand. Temperature and humidity were considered, correct?
• Line 563: When you state “a good correlation”, provide numbers.

Author Response

Dear reviewer:

We feel great thanks for your professional review work on our manuscript. We have studied comments carefully and have made correction which we hope meet with approval.The manuscript has been rechecked and the necessary changes have been made in accordance with the reviewers’ suggestions. We consider the meteorological factors as the input and analyze the results.

According to your comments, we have made extensive modifications to our manuscript. The reviewer comments are laid out below in italicized font and specific concerns have been numbered. Our response is given in normal font and changes to the manuscript are given in the red text. 

Best wishes.

Sincerely

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Thank you for giving me this opportunity to read the revised version of the manuscript titled "A predicting method for PM2.5 based on aerosol optical depth obtained from Remote sensing image processing and meteorological factors", and for the detailed responses to my earlier comments. I am satisfied with this revised version, and I think it is acceptable now.

Author Response

Dear reviewer:
We are so glad to receive your review soon.We feel great thanks for your professional review work on our manuscript. 
Best wishes.

Sincerely

Reviewer 2 Report

The authors have addressed my comments in their revised manuscript. Yet, there are a few things I'd like the authors to tweak before acceptance for publication.

1. Table 1. The cells along the diagonal line should be filled with "-" not 1. For p values smaller than 0.001, use "<0.001" rather than 0.000.
2. Acc is a new statistic measure that the authors decided to use in the revised manuscript. Please offer a reference. Why Acc instead of others, e.g., AIC (Akaike information criterion)?
3. Still, an increase in R2 with BPNN was fairly minor as relative to traditional regression models. I suggest the authors provide a justification why BPNN should be adopted.
4. I don't understand the authors' response to Comment 11. Both x and y in Figures 9 and 11 should be PM2.5 concentrations in the unit of microgram per m3. The current scale is confusing (PM concentration in a log scale or deviation from average values?).

Author Response

Dear reviewer:

We are so glad to receive your review soon. We feel great thanks for your professional review work on our manuscript. Based on your comment and request, we have supplemented references here and we have made modification on the original manuscript. A document answering every question from the referees was also summarized and enclosed.

A revised manuscript with the correction sections was attached as the supplemental material and for easy check and editing purpose. If you have any questions, please contact us without hesitation.

Thank you again and best wishes. I look forward to hearing from you.

Sincerely

Author Response File: Author Response.pdf

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

Please see attached review. The authors need to become more familiar with the field of aerosol remote sensing before attempting to publish their work.

Comments for author File: Comments.pdf

Author Response

Dear reviewer:

We feel great thanks for your professional review work on our manuscript. This is my team first submission in English, and we are so lucky that you have carefully reviewed our manuscript. These comments are all valuable and helpful for improving our manuscript. We apologize for mistakes in our manuscript. It is really a giant mistake to the whole quality of our manuscript. We feel sorry for our carelessness. We really admire your excellent academic level and stringent academic spirit. Each of your suggestions made us understand a lot, and we have studied comments carefully and have made correction which we hope meet with approval. We sincerely invite you to review our manuscript once again, because your opinion is really important to us. Please give us a opportunity to revise.

According to your comments, we have made extensive modifications to our manuscript. The reviewer comments are laid out below in italicized font and specific concerns have been numbered. Our response is given in normal font and changes to the manuscript are given in the red text in our marked manuscript. The line and page of our reply are based on the version of our manuscript without marked.

Thank you for your consideration. I look forward to hearing from you.

Sincerely,

Jilin Gu

Author Response File: Author Response.pdf

Reviewer 2 Report

The manuscript may be published 

Author Response

Dear reviewer:

We wish to re-submit the manuscript titled “An estimation method for PM2.5 based on aerosol optical depth obtained from remote sensing image processing.” The manuscript ID is 1574059.

The manuscript has been rechecked and the necessary changes have been made in accordance with the reviewers’ suggestions.

We thank you and the reviewers for your thoughtful suggestions and insights. The manuscript has benefited from these insightful suggestions. I look forward to this manuscript closer to publication in the Remote Sensing.

Thank you for your consideration. I look forward to hearing from you.

Sincerely,

Jilin Gu

Author Response File: Author Response.pdf

Reviewer 3 Report

This is a good paper, which is of clear interest for the EO RS community.

The paper is well written and the objectives are clearly stated. However, the methods are not extensively described and the results lack for a thorough discussion. 

Therefore, I recommend to improve description of the methods and discussion of the results, before re-submitting to MDPI. You can find attached some more detailed comments on the annotated PDF.

Comments for author File: Comments.pdf

Author Response

Dear reviewer:

We thank editors and reviewers for the time and effort that they have put into reviewing the previous version of the manuscript. The constructive comments from each reviewer are very valuable for the improving of our researches and the revising of our paper. We have studied comments carefully and have made correction which we hope meet with approval. 

According to your comments, we have made extensive modifications to our manuscript. The reviewer comments are laid out below in italicized font and specific concerns have been numbered. Our response is given in normal font and changes to the manuscript are given in the red text in our marked manuscript. The line and page of our reply are based on the version of our manuscript without marked.

Sincerely,

Jilin Gu

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Please see attached review. The serious flaw with the paper is that only AOD is used as an input.  

Comments for author File: Comments.pdf