Composite Aerosol Optical Depth Mapping over Northeast Asia from GEO-LEO Satellite Observations

Round 1

Reviewer 1 Report

The paper proposes an interesting research topic and was organized well. But there are some confusions and limitations on the methodology that need be further addressed. Please see the technical comments and questions below.

1. It is not very clear to use the CDFs in Fig.3 as reference to re-scale or adjust the satellite AOD product. In Fig. 2 and Eq.(1), Line 209-210, and Line 222-229, I don’t understand how to use the CDF regression coefficients to adjust the original satellite AOD.

Are you using the AERONET AOD in Eq.(1) to get the regression coefficient a1,a2 and a3 and then apply them to re-scale or adjust the satellite AOD at the different range of AOD?

In  Fig.3, the satellite Himawari-8 CDF results seem very close to the AERONET CDF, is the re-scale factor for Himawari-8 AHI AOD minimum among other satellite product?  Why the RMSE before/after applying CDF method in Table-3 are largest for the Himawari-8 AHI AOD product?

Fig.3, For the small AOD (<1), the CDFs show larger difference, are the rescale factors large for them?

The accuracy of satellite AOD product varies with the surface reflectance over land, such CDF method seems being a statistical way and impossible to improve the composite AOD data physically.

2. The paper uses this IDW/ICW method to estimate the CDF regression coefficients only. Can the IDW and ICW method be used to estimate the AOD in the cloud region where there are no satellite AOD retrievals?

 

3. Line-22, what kind of ‘Coefficients” are referred here?

 

4. Line-41, Strictly the EPA defines the PM2.5 with diameters that are generally 2.5 micrometers and smaller”.

 

5. Section 1. Satellite data. Please give the uncertainty of retrieved AOD for each satellite over land and ocean. I believe that there are some efforts and literatures to evaluate and validate these satellite AOD product.

Please list the Websites for the satellite AOD download in this study (e.g. GOCI, MODIS, and VIIRS).

 

6. Figure 4, Y-label is misled. It might be more accurate with “CDF fitting or re-scaled satellite AOD”. The revision will help separate them from the original satellite retrievals.

 

7. Fig.8(k), please give the altitude and metrological data for those HYSPLIT trajectories. Same question for Fig.9(k).

 

8. Fig.9 (e) and (f), Such blended satellite AOD product may be good for one focused dust event, but it also may make some artificial or false information. For instance, Fig.9(e) did not show the hotspot or high AOD in the Sea of Japan or the east Korean peninsula.

 

9. Table-6, please give the linear regression slope and correlation coefficient for each AERONET site, this will help get the whole picture on the satellite-AERONET data consistency.

 

10. Figure 12. The mean deviations in Fig.12 really depend on the absolute values of AOD. Can you please give the Month-Day average AERONET AOD used for this plot?

 

11. Are the AOD product from the Chinese Polar-orbit Satellites available for this study region? Any chance or comments on their potential fusion or difficulty?

Author Response

Dear reviewer1

Thanks for the review. 

Please see the attachment

Thanks you

Author Response File: Author Response.pdf

Reviewer 2 Report

Review

Composite Aerosol Optical Depth Mapping Over Northeast 2

 Asia from GEO-LEO Satellite Observations

This is a very impressive study both in terms of the scope and science conducted. The authors have done a very major task in blending and evaluating so extensively the numerous AOD products and then evaluating them with AERONET data.

I do not have major issues with the science presented, but I am requesting major revisions to improve the presentation in terms of readability/accessibility of the paper and the clear presentation of the main points/findings of the study:

• I think the paper has so many findings, that it lacks some clarity. Somehow, the key take-away points, thesis, and number of figures and tables need to be reduced some to be easily accessible for Remote Sensing readership to be able to understand and access the study results better. This might mainly consist of targeted moving of sections of text and figures to a supplemental section.
• I recommend some addition to background in the introduction and follow-up in the discussion/conclusions.

Point 1:

I recommend cutting back on the number of tables and figures, and try not listing so many numbers and statistics in the text of the paper, as all of these make reading the paper difficult (again, some of these sections could be moved to the supplementary section, and referred to more generally in the main text).

Overall the writing is of good quality. What I recommend is moving a bunch of the figures (or parts of the multi-panel figures) to supplementary material, so that major take-away points can be made with the remaining tables and figures in the main text.

Figures 6-9. The equations and R values are too small to see and need to be larger in 6 and 7. This figure could potentially be decreased to show an important subset of plots, with the remaining values move to a supplemental file, but I leave that to the authors. It seems to me that the amount of products you are dealing with is more than most studies have conducted to date, and this should be pointed out.

I can’t really follow what is going on in Fig 8 and 9 they are too small.

 

Point 2:

The introduction overall is excellent. However, it needs 3 additional brief discussions in my opinion. These do not need to be long, just a few sentences and a few references.

1). a discussion on the seasonality of retrievals, and how it is difficult to obtain good AOD retrievals in wintertime basins or over snow cover or desert surfaces is not discussed adequately in the introduction my opinion, as well as the impacts of vertical distribution (easier to look at deep smoky layer for instance than shallow layers near the ground). A few sentences of additional text, both in the introduction and a bit more in the discussion (it is already mentioned in discussion) in terms of implications for the blended products would be good.

Some possible references:

https://www.sciencedirect.com/science/article/pii/S0034425715001108

https://www.mdpi.com/2072-4292/12/9/1524

https://www.sciencedirect.com/science/article/pii/S1352231020305008

 

2. A discussion of how you are using AERONET sites to compare satellite AOD to in situ measurements because it is difficult to quantitatively use surface air pollution stations to compare to satellite AOD, they are completely different measurements. A few references and some discussion would be good about the validity of comparing the satellite measurements with the AERONET sites versus trying to compare satellite with surface PM2.5 sensors.
3. A discussion of any previous attempts at merging multiple AOD products. This is not discussed, but a quick search and I found some papers that have done multisensory AOD retrievals

Review paper that might have some references:

https://www.tandfonline.com/doi/full/10.1080/10643389.2019.1665944?scroll=top&needAccess=true

https://agupubs.onlinelibrary.wiley.com/doi/full/10.1002/2016JD025469

Other minor points

Abstract: Missing commas: Satellite products, however, can be complementary in terms of their accuracy and spatiotemporal comprehensiveness.

Please try to summarize the hypothesized relative effects of the various sources of errors/limitations (clouds, lack of images, shallow layer of pollution, etc) on the validation statistics more succinctly. There are points sprinkled here and there, but I did not find clarity on the findings.

In the discussion, please summarize what is new and different about your AOD blended product versus what has been done in the literature before. This is an impressive study, I am not aware of many extensive evaluations or blending like this, so please point out the novelty of your work if this is indeed the case.

 

Author Response

Dear reviewer2

Thanks for the review. 

Please see the attachment

Thanks you

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Thank you for the detailed responses. The research topic is of interest for the fusions of different satellite AOD retrievals. For the research and academic application,  the accuracy of the new composite AOD is important.

From the correlation between the composite AOD and AEROET-AOD given in Table-4, the composite AOD seem much improved. But, I have a few more questions about the CDF method, and the accuracy of AHI and GOCI AOD as below.

1. Figure 3, the CDF curves between the AERONET and AHI AOD agree very well. Does this mean that the original AHI-AOD agree well with the AERONET-AOD? Does this mean that the original AHI-AOD values are adjusted in very little magnitude in comparison to the MODIS AOD?
2. What is the Eq.(1) used for? According to Line 243-245 in the revised manuscript, the scaling coefficients for the original satellite product are obtained from the polynomial regression between the AERONET-AOD and satellite retrievals for each one segment of CDF-AOD curve.  What is the application of Eq.(1) for each satellite product?

3. In Table-4, the original AOD from AHI and GOCI show poor correlation with the AERONET AOD at Anmyon, Dongsha Island and Beijing in comparison to the MODIS and VIIRS. For instance, the correlation coefficients are mostly less than 0.4 for AHI-AOD, and less than 0.5 for GOCI  while the MODIS and VIIRS AOD show the correlation coefficients of 0.7- 0.9 in Table-4. 

Can you please explain major reasons for the poor AHI and GOCI AOD retrievals except the limited spectral channels?  

Author Response

Dear reviewer

Thanks for the review. 

I think most of Reviewer1's comments are about methodology. I wrote it down in detail in the attached file. 

Thank you

Author Response File: Author Response.pdf

Reviewer 2 Report

The authors have added the needed background information and also have improved the presentation. I believe the paper is ready for publication. 

Author Response

Dear. Reviewer 2

Thank you for accepting my paper.

I think it will be a good paper thanks to your advice (review).

Thank you