Variability of Major Aerosol Types in China Classified Using AERONET Measurements

Round 1

Reviewer 1 Report

This paper describes the aerosol types and their variability over China, using AERONET data. I have a few general comments and a few minor comments.

 

General comments:

Please identify the data set used in the paper explicitly. I think the data used in the paper is after the screening and time/space spacing? Clustering methods are claimed to be producing the same result. Is there any difference between the two techniques? What authors want to say from Figure 11? Aerosol types and sources remain the same during all the seasons except central china?

Few minor comments:

In abstract “direct sun and inversion measurements of” should be direct Sun measurements and inversion derived parameters

 

Line 90: “Currently there are 47 sites are located in China,” please remove repeated “are”

Line 101: what level AERONET data are used?

 

Line 110-114: Please explain time and space spacing and removal of the data

 

Line 130-152: It would be nice to explain the similarities and differences in the two clustering techniques (if any)

 

Line 163: Please check the number of data points as it is explained earlier in the methods section that the data points were reduced after screening and time and space spacing

English proofreading would be apt for the paper

 

Conclusion main points could be highlighted

Author Response

Please find the attached file for our reply to the reviewer's comments.

Author Response File: Author Response.pdf

Reviewer 2 Report

This paper aims to classify aerosol by type over China using long records of AERONET observations. The authors exploit wisely the dataset and create a convincing classification. The paper is generally well written and the scientific quality is good. The use of this kind of aerosol classification and model are certainly useful for satellite retreivals, as the authors explain (and practiced) at the end of the paper. I have a few questions and remarks however that amount to minor comments:

Unless I am mistaken I didn’t quite see in the paper how the four defined aerosol types relate to observed aerosol species. Obviously dust relates to dust. But for the three other types this is less clear, apart maybe for the absorbing type which probably relates to black carbon. It would be useful to connect this study with ground observations and, if possible, classify/quantify if possible the how/how much sulfate, nitrate, ammonium, primary/secondary organics (or which mixture of these), are part of the other two types: scattering mixed and fine. The authors didn’t mention the impact of relative humidity and the uptake of water, which influence significantly the optical properties of (at least) three of the four aerosol types. How could this impact the classification proposed?

 

Technical comments:

L 345: retrieve, not retrieval Figure 10 : labels/legends are too small L 310 : not covered instead of “uncoverered” L313-314 : this map could become more accurate in the future

 

Author Response

Please find the replies to reviewer 2 in the attached file.

Author Response File: Author Response.docx

Reviewer 3 Report

Review of Zhang and Li 2019 manuscript for Remote Sensing

Overall:

In my view this is an interesting paper that describes work worthy of publication. Its main contribution lies in analyzing variations in aerosol type over China. Potentially, the results can help improve radiative forcing estimates and satellite retrievals of aerosol properties. For the most part, the methodology is sound and the presentation is clear. Nonetheless, there are a few parts where the manuscript needs significant improvements. Please find my specific comments below.

Major:

Line 160: The manuscript says “four nodes essentially best describe the structure of the data“. It would be important to mention, how it was determined that that the SOM method works better with four nodes than with a different number of nodes (3 or 5, etc.).

Lines 162-165: It would be important to clarify why the original K-means and SOM clustering results are discarded, and each data point is reclassified based on the Euclidean distance alone. This could be problematic if, as suggested in Lines 129-130, in some cases data points are assigned to the aerosol type with the closest center, but the closeness is determined not exclusively based on the mean vector. (In principle, it could happen that some data points contribute to determining the final center of a certain class but then are assigned to a different class.) Also, what other parameter is used (in addition to the mean vector) in determining a data point’s closeness to a class center (as hinted at in Lines 129-130)? Finally, in Line 129, “adjust each object” should probably be replaced by “assign each object”.

Figure 7: The station in the westernmost corner of China has data only for the Summer and Fall, but not for Winter and Spring. Some other stations also appear to have data only for some seasons but not for others. It would be important to mention why this occurs.

Lines 278-283: It would be interesting to check whether the changes in aerosol type follow random daily patterns or the frequency of various aerosol types varies systematically through the day (for example, with relatively more absorbing aerosols in the morning or in the afternoon). This topic seems especially interesting in light of Lines 286-289.

Line 306 and Figure 10: The MODIS “continental” and VIIRS “smoke” model SSA values do seem to have a curved spectral dependence in Figure 10. Therefore, this line (mentioning “the lack of curvature”) should be reworded. Also, Figure 10 shows the AOD spectral dependence only for MODIS but not for VIIRS—and so adding a panel on VIIRS AOD dependence would make the figure more complete. Also, references should be added to indicate where the MODIS and VIIRS aerosol model information comes from. Finally, while each MODIS and VIIRS model assumes specific SSA and g values, these models do not assume AOD values—AOD for each location around the Earth is obtained through aerosol retrievals. Moreover, it would be important to clarify how the AOD values displayed in Figure 10 were obtained for each model. Finally, it should be clarified what 660 nm AOD is assumed in obtaining the MODIS g-values, as Table 1 in Remer et al. (2005, J. Atmos. Sci., Vol. 62, pages 947-973) shows that the size distribution of moderately absorbing and strongly absorbing aerosol models vary with AOD.

Minor:

Lines 22 and 306: I suggest changing “curvature shaped” to “curved”.

Line 29: I recommend changing “The aerosol property in China is” to “Aerosol properties in China are”.

Line 48: I suggest changing “made from” to “using”.

Line 49: I suggest chaning “of” to “on”.

Line 51: I suggest changing “practices” to “attempts”.

Lines 69-70: I suggest replacing “it” by “the aerosol population”, and replacing “the aerosol are” by “it is”.

Table 1: It should be clarified that the last data row refers to standard deviation of what physical quantity. (My guess is particle radius.)

Lines 110-111: I suggest inserting “volume” between “data” and “is”, and changing “less” to “smaller”.

Figure 2: It should be clarified what is meant by “DB”, or simply remove “Judging DB index”, as readers at this point don’t know what this refers to.

Lines 142-152: A reference about the SOM method or specific SOM code should be added somewhere in this section.

Line 216: It should be clarified that, as mentioned in Line 331, the mixed types are black carbon mixed with either dust or organic carbon.

Line 219: The word “more” is missing between “much” and “flat”.

Line 220: The letter “s” should be deleted from the end of “confirms”.

Line 228: “Seasonal” should be changed to “season”.

Line 237: The word “the” should be deleted.

Line 272: The word “its” should be change to “their”.

Line 299: The letter “n” should be deleted from the end of “Asian”.

Line 331: Just a typo: The word should be “carbon”, not “carobn”.

Line 345: “Retrieval” should be changed to “retrieve” or to “retrievals of”.

Line 351: The word “even” should be replaced, for example by “and”.

 

Author Response

Please find the replies to reviewer 3's comments in the attached file.

Author Response File: Author Response.docx

Round 2

Reviewer 3 Report

The authors appropriately addressed my comments, and so my recommendation is to accept the manuscript for publication. My only recommendation is to change "are showed" to "are shown" in Line 358.

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.