the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 19 Apr 2024
Observed improvement in air quality in Delhi during 2011–2021: Impact of mitigation measures
Abstract. Assessing long-term air quality trends helps evaluate the effectiveness of adopted air pollution control policies. A decade of SAFAR observations revealed that the trend of particulate matter (PM2.5 and PM10) in Delhi shows a reduction of 2.98 ± 0.53 µg/m3/y (4.91 ± 1.01 µg/m3/y) or overall 29 % (23.7 %) reduction between 2011 and 2021 while vehicles almost doubled but with the implementation of cleaner technologies and stricter industrial regulation. Seasonal negative trends of pre-monsoon (March-April-May; -3.43 ± 1.02 µg/m3/y) and post-monsoon (October–November; -4.51 ± 1.59 µg/m3/y) are relatively higher. The role of trends in dust storms, fire counts and annual rainy days are also discussed. The contribution of meteorology to the trend is estimated using WRF-Chem simulation of PM2.5 for October when maximum stubble burning occurs and gets transported to Delhi. The model is run with the meteorological initial conditions of 2018, 2015, and 2011 while keeping the emissions of 2018 with identical model configuration and found that meteorology contributed 9.8 % in October, while the observed decline in PM2.5 is 35 % (best fit) and 25 % (value). The study identifies the governmental control measures at various levels and green initiatives as the significant contributors to air quality improvement during 2011–2021.
- Preprint
(1793 KB) - Metadata XML
-
Supplement
(550 KB) - BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on egusphere-2024-803', Anonymous Referee #2, 15 May 2024
1. the paper averages out 10 different locations in Delhi and describes the particulate matter scenario and attributes the decrease to the measures taken by the government. It would be better if the authors chose the locations immediately impacted by these government measures, instead of averaging them out. The microenvironments are very different from each other, thus averaging them out wouldn't depict the true picture.
2. 2020 was in full lockdown for few months, followed by partial lockdowns. Similar statewide lockdowns were observed during 2021 as well. So including 2020 and 2021 will skew the results. the anomaly seen in 2020 and 2021 is substantial compared to the other years, thus it doesn't depict the true picture. Please include the following years 2022, and 2023 to see if the trend persists and to rule out the impact of lockdown.
3. why was 2018 chosen as the meteorological base year for comparison?
Citation: https://doi.org/10.5194/egusphere-2024-803-RC1 -
AC1: 'Reply on RC1', Latha Radhadevi, 11 Jun 2024
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2024/egusphere-2024-803/egusphere-2024-803-AC1-supplement.pdf
-
AC1: 'Reply on RC1', Latha Radhadevi, 11 Jun 2024
-
RC2: 'Comment on egusphere-2024-803', Anonymous Referee #1, 15 May 2024
The manuscript evaluates the efficiency of air quality control policies in India. Notable reduction in particulate pollution was seen and the authors were able to estimate trends for the annual pollution levels. The topic of the manuscript is of interest for readers of ACP and overall, the presentation of the results is clear. However, some concerns need to be addressed before I can recommend the MS for publication.
My main concern is that the method for trend estimation is not statistically sound, and it is not capable of answering the questions researchers are trying to solve. The 13-month moving average is claimed to de-seasonalize the data, but it only smooths the variation. The trend probably is not linear and t-test is definitely not a method for calculating a trend. With appropriate trend fitting methods, deseasonalization is not even needed but the seasonal variation can be taken account in the trend calculation.
Specific comments:
Page 7, lines 24-25: averaging does not eradicate inhomogeneity. By averaging, the researchers just assume data “homogenic enough” to get representative city-level value. How justifiable this assumption is proposes another question. I would suggest a sensitivity analysis (perhaps shown in the supplement) where basic statistics would be shown and appropriate trends would be fitted to individual datasets.
Page 9, line 28: Announcing p=0.085 as insignificant is a bit of overkill. Interpretation for p-value should not be based on some artificial threshold value but it should be treated as quantitative measure of significance. See e.g. https://doi.org/10.1080/00031305.2016.1154108 and https://www.nature.com/articles/d41586-019-00857-9
Section 3.4. The argument on the effect of meteorology on PM needs confirmation. The comparison of model results in different time points does not quantify the effect. This could be done with the data by using multivariable statistical models like applied here https://doi.org/10.5194/acp-20-12247-2020 or advanced time series methodology introduced here http://urn.fi/URN:NBN:fi:jyu-201603111829 and here http://dx.doi.org/10.1007/978-3-030-21718-1_4. The same methods can also be applied in Section 3.5. in quantification of the dust storms and in 3.6. to account for stubble burning.
Citation: https://doi.org/10.5194/egusphere-2024-803-RC2 -
AC2: 'Reply on RC2', Latha Radhadevi, 11 Jun 2024
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2024/egusphere-2024-803/egusphere-2024-803-AC2-supplement.pdf
-
AC2: 'Reply on RC2', Latha Radhadevi, 11 Jun 2024
Status: closed
-
RC1: 'Comment on egusphere-2024-803', Anonymous Referee #2, 15 May 2024
1. the paper averages out 10 different locations in Delhi and describes the particulate matter scenario and attributes the decrease to the measures taken by the government. It would be better if the authors chose the locations immediately impacted by these government measures, instead of averaging them out. The microenvironments are very different from each other, thus averaging them out wouldn't depict the true picture.
2. 2020 was in full lockdown for few months, followed by partial lockdowns. Similar statewide lockdowns were observed during 2021 as well. So including 2020 and 2021 will skew the results. the anomaly seen in 2020 and 2021 is substantial compared to the other years, thus it doesn't depict the true picture. Please include the following years 2022, and 2023 to see if the trend persists and to rule out the impact of lockdown.
3. why was 2018 chosen as the meteorological base year for comparison?
Citation: https://doi.org/10.5194/egusphere-2024-803-RC1 -
AC1: 'Reply on RC1', Latha Radhadevi, 11 Jun 2024
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2024/egusphere-2024-803/egusphere-2024-803-AC1-supplement.pdf
-
AC1: 'Reply on RC1', Latha Radhadevi, 11 Jun 2024
-
RC2: 'Comment on egusphere-2024-803', Anonymous Referee #1, 15 May 2024
The manuscript evaluates the efficiency of air quality control policies in India. Notable reduction in particulate pollution was seen and the authors were able to estimate trends for the annual pollution levels. The topic of the manuscript is of interest for readers of ACP and overall, the presentation of the results is clear. However, some concerns need to be addressed before I can recommend the MS for publication.
My main concern is that the method for trend estimation is not statistically sound, and it is not capable of answering the questions researchers are trying to solve. The 13-month moving average is claimed to de-seasonalize the data, but it only smooths the variation. The trend probably is not linear and t-test is definitely not a method for calculating a trend. With appropriate trend fitting methods, deseasonalization is not even needed but the seasonal variation can be taken account in the trend calculation.
Specific comments:
Page 7, lines 24-25: averaging does not eradicate inhomogeneity. By averaging, the researchers just assume data “homogenic enough” to get representative city-level value. How justifiable this assumption is proposes another question. I would suggest a sensitivity analysis (perhaps shown in the supplement) where basic statistics would be shown and appropriate trends would be fitted to individual datasets.
Page 9, line 28: Announcing p=0.085 as insignificant is a bit of overkill. Interpretation for p-value should not be based on some artificial threshold value but it should be treated as quantitative measure of significance. See e.g. https://doi.org/10.1080/00031305.2016.1154108 and https://www.nature.com/articles/d41586-019-00857-9
Section 3.4. The argument on the effect of meteorology on PM needs confirmation. The comparison of model results in different time points does not quantify the effect. This could be done with the data by using multivariable statistical models like applied here https://doi.org/10.5194/acp-20-12247-2020 or advanced time series methodology introduced here http://urn.fi/URN:NBN:fi:jyu-201603111829 and here http://dx.doi.org/10.1007/978-3-030-21718-1_4. The same methods can also be applied in Section 3.5. in quantification of the dust storms and in 3.6. to account for stubble burning.
Citation: https://doi.org/10.5194/egusphere-2024-803-RC2 -
AC2: 'Reply on RC2', Latha Radhadevi, 11 Jun 2024
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2024/egusphere-2024-803/egusphere-2024-803-AC2-supplement.pdf
-
AC2: 'Reply on RC2', Latha Radhadevi, 11 Jun 2024
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 390 | 119 | 32 | 541 | 39 | 17 | 16 |
- HTML: 390
- PDF: 119
- XML: 32
- Total: 541
- Supplement: 39
- BibTeX: 17
- EndNote: 16
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1