The manuscript, entitled with "Dust and anthropogenic aerosols vertical distributions over Beijing―dense aerosols gathered at the top of the mixing layer", described the analysis of vertical distributions of anthropogenic, dust and polluted dust at Beijing (capital of China), based on 3 years polarization Raman lidar measurements. The authors presented the annual and seasonal cycle of total AOD and AOD within mixing boundary layer (MBL) and free troposphere (FT), mass concentration of different aerosol types and discussed the possible mechanisms behind. The manuscript is generally well-written and the database of aerosol optical properties at 355 nm are rather valuable for future satellite mission. However, there are issues that need to be fixed or clarified before the consideration of publication.

Major issue

1. The authors speculate too much in the data analysis. For example, in page 18, line 10-12, "The high anthropogenic aerosols mass concentration in the upper air (0.4.0.9 km) over Beijing in summer is mainly caused by the growth of particle hygroscopicity under the influence of southern transport.", this statement cannot be supported by the results presented in the manuscript. As the authors described, all profiles with high relative humidity (> 85%) have been ruled out from the data analysis (see page 6, line 20-21). Then, hygroscopic growth should not be significant. And more importantly, no results of hygroscopic growth factors were shown, how this conclusion can be made? Similar issue is also laid in the analysis of MBL height and aerosol mass concentration (There is a minor issue associated with this). I hope the authors can focus on their own results and start from these results to re-think what they can conclude.

Minor issues

1. Page 5, line 8, the authors should be specific that the gradient of which quantity they used in the MBL height determination. And how do they treat the very shallow nonctual boundary layer height within the incomplete overlap region?

2. Page 6, line 21-22, the authors need to explain why they intended to use lidar-derived MBL height in MBL AOD calculation instead of using ERA-5 MBL height, although they think ERA-5 height is reliable and can be used to evaluate the lidar-derived MBL height.

3. Page 7, line 24, "Bac" -> "BAC"

4. Page 9, line 17, why do the authors think "it suggests a strong sytematic coupliing between ML and FT, ...", instead of they are both modulated by the same mechanism, like regional transport of aerosols.

5. Page 10, line 25, PDR at 532 should be at percentage, namely, 0.082, or adding a percent sign (%) instead. And the authors need to check the manuscript thoroughly, because there are many places with this error.

6. Page 11, line 33, "building warming" -> "building heating".

6. Page 11, line 33, "MEGGA" -> "MERRA". (also in caption of fig. 8)

8. Page 13, line 17, "upper air pollution transport" is more appropriate. 

9. Page 15, line 22, "bottom" should be removed.

10. Page 15, line 29-32, correlation of coefficient cannot be used to determine the goodness of fit for non-linear models. Therefore, it cannot be compared between linear fitting and non-linear fitting, just by looking at correlation of coefficient. The author should either use a different metric to do the comparison or remove such statement.

11. Page 17, line 14-15, the authors should be specific when mentioning "near the ML" or "around ML" (the same page, line 18).

12. Page 18, line 20-21, the authors should clarify why "the bottom dust mass concentration is mainly influenced by transport" instead of by local sources.

13. In caption of fig 2, Check about the conversion factors of dust and anthropogenic aerosols. It's too low for dust and a little high for anthropogenic aerosols (see ref.[1-2]).

References:

1. Mamouri R, Ansmann A. Fine and coarse dust separation with polarization lidar. Atmospheric Measurement Techniques. 2014;7(11):3717-35.

2. Ansmann A, Mamouri R-E, Hofer J, Baars H, Althausen D, Abdullaev SF. Dust mass, cloud condensation nuclei, and ice-nucleating particle profiling with polarization lidar: updated POLIPHON conversion factors from global AERONET analysis. Atmospheric Measurement Techniques. 2019;12(9):4849-65.

The air pollution in Beijing and its surrounding areas is often a hot topic in the field of atmospheric environment. The authors revealed the long‒term vertical distributions of dust (coarse) and anthropogenic aerosols (fine) and their relationships with mixing layer height (MLH). The results found that the anthropogenic aerosols are often accumulate at the top of the mixing layer in summer, while dust mass has a clear seasonal distribution in the vertical direction. These results will enhance our understanding of the sophisticated interactions between dust and anthropogenic aerosols. The long‒term polarization Raman dataset presented here is valuable and surely deserves publication. However, I am afraid the paper cannot be published as it is, as a number of points should be clarified. The detailed comments are as follows:

Major comments:

There are many studies on air pollution in Beijing in the past decade, and the authors need to highlight the innovations of this paper to distinguish it from other studies. In other words, this study revealed the vertical distribution characteristics of anthropogenic aerosols and dust mass concentrations, what kind of research can be conducted in the future based on polarization Raman dataset presented in the manuscript. What is the significance or implications of those research results for air pollution in Beijing.

Other comments:

1. In section 2.2-2.5, the author uses a variety of data. It is necessary to introduce the main purpose of the data and its role in this article before starting.
2. In section 3, the data was from 22 May 2019 to 20 February 2022. But PRL data were missing from 16 November 2020 to 29 May 2021. In the following Fig.3, Fig.5, Fig.7, Fig.9 and Fig.10, the monthly average data was used. It is necessary to clarify which period of time is used for averaging.
3. Page 9, line 8, “Fig1b shows the PRL-derived AOD in the ML and FT”, Fig1b? It should be Fig. 3b.
4. Page 11, line 32, “MERRA” is misspelled.
5. Page 14, line 19, “from the surface to more than 5 km”. Data below 0.25 km and above 5 km are not presented in the manuscript.
6. Page 16, line 31, explain “non-dust”, is it anthropogenic aerosols?
7. Page 17, line 2, “dust aerosol (about 50 µg/m3) is mainly above the ML”. Not only above the mixed layer, the dust concentration is also higher below mixing layer.
8. In section 3.3.2, the concentration of anthropogenic aerosols was low in winter. However, during the observation period, especially at the end of 2019 and the beginning of 2020, epidemic control policy led to the reduction of anthropogenic emissions. Whether these reductions had any effect on the observations of this study? and these effects also remain to be explored.
9. In discussion section, authors found that there is a significant negative correlation between anthropogenic aerosols and MLH in four seasons. Are there any similar observation results or simulation results with the same conclusions as this paper?
10. In section 5, the conclusion should point out the shortcomings in this study and future research perspectives.
11. Suggestion: It would be better to combine Figure 11 and Figure 12 into a single figure, and also for Figure 13 and Figure 14.