the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 04 Sep 2023
Response of PM2.5 chemical composition to variations in anthropogenic emissions and meteorological conditions during COVID-19 lockdown
Abstract. PM2.5 is a primary atmospheric pollutant with various sources and complicated chemical compositions that are influenced by various factors, such as anthropogenic emissions (AE) and meteorological conditions (MC). MC have significant impacts on variations of the atmospheric pollutant; therefore, emission reduction policies and ambient air quality are non-linearly correlated, which hinders accurate assessments of the effectiveness of control measures. The online observations of PM2.5 and its chemical composition were conducted in Hohhot, China, from December 1, 2019, to February 29, 2020, to investigate PM2.5 chemical compositions respond to the variation of AE and MC. Moreover, the random forest (RF) model was used to quantify the AE and MC contributions of PM2.5 and its chemical composition during severe hazes and the COVID-19 pandemic lockdown period. During the clean period, MC contributed -124 % to PM2.5 concentrations, indicating that MC promoted PM2.5 dispersion. During severe pollution episodes, MC contributed 49 % to PM2.5 concentrations, indicating that MC promoted PM2.5 accumulation. The inorganic aerosols (SO42-, NO3-, and NH4+) showed the strongest response to MC. MC had significant contributions to the PM2.5 (36 %), SO42- (32 %), NO3- (29 %), NH4+ (28 %), OC (22 %), and SOC (17 %). From the pre-lockdown to lockdown period, AE (MC) contributed 52 % (48 %), 81 % (19 %), 48 % (52 %), 68 % (32 %), 59 % (41 %), and 288 % (-188 %) to the PM2.5, SO42-, NO3-, NH4+, OC, and SOC variations, respectively. The variations of MC (especially the increase in relative humidity) rapidly generated meteorologically sensitive species (SO42-, NO3-, and NH4+), which led to severe winter pollution. This study provides reference for assessing the net benefits of emission reduction measures for PM2.5 and its chemical compositions.
This preprint has been withdrawn.
-
Withdrawal notice
This preprint has been withdrawn.
-
Preprint
(1933 KB)
-
Supplement
(1069 KB)
-
This preprint has been withdrawn.
- Preprint
(1933 KB) - Metadata XML
-
Supplement
(1069 KB) - BibTeX
- EndNote
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-1851', Anonymous Referee #1, 27 Sep 2023
The paper presents a study examining the relationship between PM2.5 chemical composition, anthropogenic emissions, and meteorological conditions during the COVID-19 lockdown. This research offers valuable insights into the effectiveness of emission reduction and its impact on PM2.5 and its chemical composition. The topic is highly relevant to Atmos. Chem. Phys.. However, it is crucial to address more technical details and incorporate more scientific discussion. I recommend the publication of this work after the following concerns are addressed.
- Please justify the use of 1.6 in the conversion of OC to OM.
- Please double-check the correctness of the equations of SOR and NOR.
- Please provide a detailed explanation of the model training process and the hyperparameters.
- Please provide details on how the Cdeweathered is calculated.
- Please provide the scatter plots of the model prediction v.s. observation in the test set.
- Regarding the relative importance of meteorological variables in Figure 5, it would be helpful to clarify whether the values are derived from a single simulation or multiple simulations. Additionally, it would be informative to determine if the results remain consistent when employing a different random state in the train-test split.
- In Figure 5, the importance of meteorological variables is relatively small, whereas in Figure 6, their contribution seems significant. Please clarify this difference. Additionally, could you provide the relative importance of meteorological variables at different pollution levels?
- It would be helpful to have a more in-depth scientific discussion regarding Figure 6 and Figure 7 instead of just reporting the number.
Citation: https://doi.org/10.5194/egusphere-2023-1851-RC1 -
RC2: 'Comment on egusphere-2023-1851', Anonymous Referee #2, 07 Oct 2023
This manuscript by Gong et al. analyzed observations at an individual site and estimated the response of PM2.5 composition to anthropogenic emissions and meteorological conditions using the random forest model method. The effect of COVID lockdown on PM2.5 has been widely examined by many studies such as Huang et al. National Science Review 2021 and Le et al. Science 2020. This work did not provide significant new insights into this topic.
In addition, the observational data sets shown in this manuscript have partially been presented in these authors’ earlier work: Zhou et al. https://acp.copernicus.org/articles/22/12153/2022/, Liu et al. https://www.sciencedirect.com/science/article/pii/S0045653523013231. Surprisingly, these papers are not cited in the present manuscript.
Based on the above major concerns, I could not support the publication of this manuscript in the prestigious journal ACP. I have other comments for this work listed as follows.
- Physical and chemical implications of anthropogenic emissions and meteorological conditions (especially results obtained using the random forest approach).
- Problems with the logical relationship between the preceding and following texts. For example, the results of the analysis in the previous paragraph do not correspond to the pattern inputs in the next paragraph. As well as the many descriptions that appear in the overview section that do not have an obvious causal relationship.
- Highlights of this work. It seems to use existing research methods to get the expected conclusions, what is the difference from others’ research? Or what characterizes the choice of Hohhot for AE and MC stripping?
Introduction
The first paragraph
- The relationship between a significant concern and the increasing levels of industrialization and urbanization does not appear to have a direct causal link. It is necessary to re-evaluate the logical connection in the background and clarify the relationship.
- To specifically refer to the pollution process driven by PM2.5, it is recommended to use the term “haze episodes”.
- I do not recommend using “strict” to describe the strategies. I believe these policies are necessary rather than strict.
- The release of the new WHO guidelines in 2021 is indeed a challenge for China in that year or the future, but not for 2019. Please check similar literature citation timeliness issues. Generally use relevant literature from after the policy was enacted to corroborate your points, not older.
- The main focus of this text is to emphasize the impact of anthropogenic emissions (AE) and meteorological conditions (MC) on PM2.5. However, there are many factors influencing PM2.5, and it has been more than ten years since the publication of the reference stating that secondary inorganic aerosols (sulfates, nitrates, and ammonium, SNA) under adverse MC are the driving factor of severe haze in China. Is this still the case in recent years? I suggest the authors re-evaluate the factors affecting PM2.5 in more detail rather than introducing the research topic in a general manner.
In addition, the concepts of quantifying AE and MC impacts are quite different from the perspective of managers, from machine learning, from modeling simulation, and so on. Hopefully, the authors will provide a simple and clear explanation of the concepts studied in this paper.
The second paragraph
- Only NOx reductions why cause the haze episodes to be unexpected?
- Similar to question 5. I don’t understand exactly what the contribution of AE and MC to PM2.5 means in this paper. Especially in adverse meteorological conditions, how to consider the contribution of physical processes such as boundary layer changes to PM2.5?
The third paragraph
- Out of multiple machine learning tools, why choose random forests? And the authors cite many cases from the literature but do not explain why RF can quantify the MC contribution. As with the previous question, what is the meaning of MC contribution here, preferably elaborated in terms of atmospheric chemistry and physics?
The fourth paragraph
- In the introduction, the authors did not write well about the importance of this paper. In addition, what is the rationale for choosing Hohhot as the research site? Is there anything unique about this city?
Data and Methods
- This study observed data for 3 consecutive months, so how were data quality control and quality assurance done?
Results and discussion
3.1 Variations in pollutant concentrations
- There are quite a few problems with Figure 2 and the corresponding paragraph.
- Need to give the whole of WS, RH, T, etc. in the corresponding section. For example, wind speed (WS) should be used for the first occurrence, and WS should be used instead of wind speed only in subsequent sections.
- Many of the subfigures in the figure do not have corresponding textual descriptions, such as P, T, and O3. If these data do not contribute to the analysis in this paper. It is recommended to delete them.
- The color of the wind direction is not easily identifiable, especially from 180-360. 0 and 360 are consistent, so shouldn’t the colors in the chart be consistent? The 360 and 180 used in the chart are very close colors and would give the reader a completely wrong judgment.
- PM2.5 and components have the same units and range of axes, so why not use the same axis for labeling?
3.2 Variations in PM2.5 chemical constituents across different pollution levels
- The previous paragraph roughly divided the sampling period into pre-lockdown and lockdown and preliminarily analyzed the possible causes of the PM2.5 decline and potential source category changes. Why did you mix the haze episodes before and after lockdown again in this section? In addition, the effect of Chinese New Year fireworks sources seems to affect only EP6, which is inappropriate for analyzing all processes Cl, K, and Mg rise. My suggestion is to re-categorize haze and non-haze episodes based on before and after the lockdown to further corroborate the potential source class changes in PM2.5 decline during lockdown through component changes.
3.3 Factors influencing PM2.5 and its compositions
- What are SOR and NOR? How are they calculated? Please double-check all the abbreviations that appear in your text to guarantee that each abbreviation has a corresponding meaning and explanation.
- This section begins with statistics on the relationship of RH, and T to PM2.5. But RH and T are only part of MC, why in Figure 5, parameters such as WD, WS, and P are again used as inputs to RF, and indicators such as Trend, Week, Weekday, Hour, Ox, etc. are considered, which are not part of MC. So, I don’t think there is a logical relationship between the analysis before and after “In conclusion”, and I don’t understand why these indicators are used as inputs to the RF.
3.4 Response of PM2.5 compositions to the variation of AE and MC
- How does RF differentiate between AE and MC?
- What is the source of the data for “other key pollution sources have experienced a 22% decrease in SO2 emissions, whereas NOx emissions exhibited a relatively minor variation”?
Conclusion
- Do not use speculation, especially ideas that are not testified to in the text, as the conclusion of the article. For example, “The formation of SO42- in winter was primarily attributed to aqueous-phase oxidation, whereas both aqueous-phase and photochemical oxidation played vital roles in NO3- formation. The formation of SO42- in winter was primarily attributed to aqueous-phase oxidation, whereas both aqueous-phase and photochemical oxidation played vital roles in NO3- formation.” There is no corresponding analysis and evidence in the text.
Citation: https://doi.org/10.5194/egusphere-2023-1851-RC2
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-1851', Anonymous Referee #1, 27 Sep 2023
The paper presents a study examining the relationship between PM2.5 chemical composition, anthropogenic emissions, and meteorological conditions during the COVID-19 lockdown. This research offers valuable insights into the effectiveness of emission reduction and its impact on PM2.5 and its chemical composition. The topic is highly relevant to Atmos. Chem. Phys.. However, it is crucial to address more technical details and incorporate more scientific discussion. I recommend the publication of this work after the following concerns are addressed.
- Please justify the use of 1.6 in the conversion of OC to OM.
- Please double-check the correctness of the equations of SOR and NOR.
- Please provide a detailed explanation of the model training process and the hyperparameters.
- Please provide details on how the Cdeweathered is calculated.
- Please provide the scatter plots of the model prediction v.s. observation in the test set.
- Regarding the relative importance of meteorological variables in Figure 5, it would be helpful to clarify whether the values are derived from a single simulation or multiple simulations. Additionally, it would be informative to determine if the results remain consistent when employing a different random state in the train-test split.
- In Figure 5, the importance of meteorological variables is relatively small, whereas in Figure 6, their contribution seems significant. Please clarify this difference. Additionally, could you provide the relative importance of meteorological variables at different pollution levels?
- It would be helpful to have a more in-depth scientific discussion regarding Figure 6 and Figure 7 instead of just reporting the number.
Citation: https://doi.org/10.5194/egusphere-2023-1851-RC1 -
RC2: 'Comment on egusphere-2023-1851', Anonymous Referee #2, 07 Oct 2023
This manuscript by Gong et al. analyzed observations at an individual site and estimated the response of PM2.5 composition to anthropogenic emissions and meteorological conditions using the random forest model method. The effect of COVID lockdown on PM2.5 has been widely examined by many studies such as Huang et al. National Science Review 2021 and Le et al. Science 2020. This work did not provide significant new insights into this topic.
In addition, the observational data sets shown in this manuscript have partially been presented in these authors’ earlier work: Zhou et al. https://acp.copernicus.org/articles/22/12153/2022/, Liu et al. https://www.sciencedirect.com/science/article/pii/S0045653523013231. Surprisingly, these papers are not cited in the present manuscript.
Based on the above major concerns, I could not support the publication of this manuscript in the prestigious journal ACP. I have other comments for this work listed as follows.
- Physical and chemical implications of anthropogenic emissions and meteorological conditions (especially results obtained using the random forest approach).
- Problems with the logical relationship between the preceding and following texts. For example, the results of the analysis in the previous paragraph do not correspond to the pattern inputs in the next paragraph. As well as the many descriptions that appear in the overview section that do not have an obvious causal relationship.
- Highlights of this work. It seems to use existing research methods to get the expected conclusions, what is the difference from others’ research? Or what characterizes the choice of Hohhot for AE and MC stripping?
Introduction
The first paragraph
- The relationship between a significant concern and the increasing levels of industrialization and urbanization does not appear to have a direct causal link. It is necessary to re-evaluate the logical connection in the background and clarify the relationship.
- To specifically refer to the pollution process driven by PM2.5, it is recommended to use the term “haze episodes”.
- I do not recommend using “strict” to describe the strategies. I believe these policies are necessary rather than strict.
- The release of the new WHO guidelines in 2021 is indeed a challenge for China in that year or the future, but not for 2019. Please check similar literature citation timeliness issues. Generally use relevant literature from after the policy was enacted to corroborate your points, not older.
- The main focus of this text is to emphasize the impact of anthropogenic emissions (AE) and meteorological conditions (MC) on PM2.5. However, there are many factors influencing PM2.5, and it has been more than ten years since the publication of the reference stating that secondary inorganic aerosols (sulfates, nitrates, and ammonium, SNA) under adverse MC are the driving factor of severe haze in China. Is this still the case in recent years? I suggest the authors re-evaluate the factors affecting PM2.5 in more detail rather than introducing the research topic in a general manner.
In addition, the concepts of quantifying AE and MC impacts are quite different from the perspective of managers, from machine learning, from modeling simulation, and so on. Hopefully, the authors will provide a simple and clear explanation of the concepts studied in this paper.
The second paragraph
- Only NOx reductions why cause the haze episodes to be unexpected?
- Similar to question 5. I don’t understand exactly what the contribution of AE and MC to PM2.5 means in this paper. Especially in adverse meteorological conditions, how to consider the contribution of physical processes such as boundary layer changes to PM2.5?
The third paragraph
- Out of multiple machine learning tools, why choose random forests? And the authors cite many cases from the literature but do not explain why RF can quantify the MC contribution. As with the previous question, what is the meaning of MC contribution here, preferably elaborated in terms of atmospheric chemistry and physics?
The fourth paragraph
- In the introduction, the authors did not write well about the importance of this paper. In addition, what is the rationale for choosing Hohhot as the research site? Is there anything unique about this city?
Data and Methods
- This study observed data for 3 consecutive months, so how were data quality control and quality assurance done?
Results and discussion
3.1 Variations in pollutant concentrations
- There are quite a few problems with Figure 2 and the corresponding paragraph.
- Need to give the whole of WS, RH, T, etc. in the corresponding section. For example, wind speed (WS) should be used for the first occurrence, and WS should be used instead of wind speed only in subsequent sections.
- Many of the subfigures in the figure do not have corresponding textual descriptions, such as P, T, and O3. If these data do not contribute to the analysis in this paper. It is recommended to delete them.
- The color of the wind direction is not easily identifiable, especially from 180-360. 0 and 360 are consistent, so shouldn’t the colors in the chart be consistent? The 360 and 180 used in the chart are very close colors and would give the reader a completely wrong judgment.
- PM2.5 and components have the same units and range of axes, so why not use the same axis for labeling?
3.2 Variations in PM2.5 chemical constituents across different pollution levels
- The previous paragraph roughly divided the sampling period into pre-lockdown and lockdown and preliminarily analyzed the possible causes of the PM2.5 decline and potential source category changes. Why did you mix the haze episodes before and after lockdown again in this section? In addition, the effect of Chinese New Year fireworks sources seems to affect only EP6, which is inappropriate for analyzing all processes Cl, K, and Mg rise. My suggestion is to re-categorize haze and non-haze episodes based on before and after the lockdown to further corroborate the potential source class changes in PM2.5 decline during lockdown through component changes.
3.3 Factors influencing PM2.5 and its compositions
- What are SOR and NOR? How are they calculated? Please double-check all the abbreviations that appear in your text to guarantee that each abbreviation has a corresponding meaning and explanation.
- This section begins with statistics on the relationship of RH, and T to PM2.5. But RH and T are only part of MC, why in Figure 5, parameters such as WD, WS, and P are again used as inputs to RF, and indicators such as Trend, Week, Weekday, Hour, Ox, etc. are considered, which are not part of MC. So, I don’t think there is a logical relationship between the analysis before and after “In conclusion”, and I don’t understand why these indicators are used as inputs to the RF.
3.4 Response of PM2.5 compositions to the variation of AE and MC
- How does RF differentiate between AE and MC?
- What is the source of the data for “other key pollution sources have experienced a 22% decrease in SO2 emissions, whereas NOx emissions exhibited a relatively minor variation”?
Conclusion
- Do not use speculation, especially ideas that are not testified to in the text, as the conclusion of the article. For example, “The formation of SO42- in winter was primarily attributed to aqueous-phase oxidation, whereas both aqueous-phase and photochemical oxidation played vital roles in NO3- formation. The formation of SO42- in winter was primarily attributed to aqueous-phase oxidation, whereas both aqueous-phase and photochemical oxidation played vital roles in NO3- formation.” There is no corresponding analysis and evidence in the text.
Citation: https://doi.org/10.5194/egusphere-2023-1851-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 483 | 139 | 26 | 648 | 71 | 24 | 25 |
- HTML: 483
- PDF: 139
- XML: 26
- Total: 648
- Supplement: 71
- BibTeX: 24
- EndNote: 25
Viewed (geographical distribution)
| Country | # | Views | % |
|---|---|---|---|
| China | 1 | 238 | 36 |
| United States of America | 2 | 195 | 30 |
| Germany | 3 | 41 | 6 |
| undefined | 4 | 28 | 4 |
| France | 5 | 17 | 2 |
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
- 238
Yitian Gong
Haijun Zhou
Xi Chun
Zhiqiang Wan
Jingwen Wang
Chun Liu
This preprint has been withdrawn.
- Preprint
(1933 KB) - Metadata XML
-
Supplement
(1069 KB) - BibTeX
- EndNote