Short-term exposure to ambient air quality of the most polluted Indian cities due to lockdown amid SARS-CoV-2

2.1. Sample cities intended for air pollution study

The present article has focused on 10 most polluted urban centers in India out of the 20 most polluted cities globally (IQAir, 2019) (Fig. 1 ). The list includes the megacity Delhi with 10 million plus population; 2 cities (Ghaziabad and Faridabad) with population more than 1 million; 2 cities (Gurugram and Noida) with population above 6 lakh and 5 cities (Muzaffarnagar; Bulandshahr; Greater Noida; Jind and Bhiwadi) having population 1 lakh and above (Table 1). All of the cities are located in the Indo-Gangetic plain region (northern counterpart) and considered as the most polluted region in India (Fig. 2 ). However, there are four more urban centers listed within in the in the 2019 World Air Quality Report (IQAir, 2019) namely Bandhwari (Rank 10; one of the major landfills in north India located in an ecologically sensitive Aravalli forest area), Lucknow (Rank 11; Capital city of the Indian state of Uttar Pradesh); Baghpat (Rank 15; part of the national capital region located in Uttar Pradesh) and Coraut (Rank 19; located in Varanasi Tehsil of Varanasi district in Uttar Pradesh). These four are excluded from the present analysis because of the fact that, daily continuous air quality monitoring data for all the criteria pollutants, particularly for the study period are not available for those cities.

All the sample cities are located within the National Capital Region (NCR) which is considered as the largest urban agglomeration in India with a decadal growth rate of >20%. The geographical extension of the NCR within which the cities are located is 27.60°N to 29.30°N latitude and 76.20°E to 78.40°E longitude with a total of twenty three districts (Hazarika et al., 2019). Physiographicaly the NCR region belongs to the Indo-Gangetic alluvial plain and sandwiched between two states-Haryana and Uttar Pradesh. The average density of the cities are >6000 persons/sq. km in 2011 and a significant portion of population in the region reside in urban areas (about 60%) which is much higher than the national average. All the sample cities experience semi-arid climate having five major seasons: Summer (Mar–May), Monsoon (Jun–Sept), short Post-monsoon (Oct–Nov), Winter (Dec–Feb) and Pre-monsoon (March–May). Temperatures stuck between 4 °C and 10 °C in winter and 42 °C to 48 °C in summer (Kumar et al., 2017). Average air temperature during the month of March, 2020 and April, 2020 counts 294.35 K (SD: 10.44) and 299.06 K (SD: 10.47) respectively (Fig. 1b and c). Whereas there is no significant variation in the average wind speed during the month of March and April (Max: 7.27 m/s, Min: 1.27 m/2, Avg.: 4.08 m/s; SD: 0.9) this year (Fig. 1d & e). More than 80% of the total annual precipitation occurs during the monsoon months (Perrino et al., 2011). Currently there are a total of 59 ambient air quality monitoring stations across the sample cities (Table 1) all having capacity to monitor and record pollutants.

As far as the selection of the urban areas is concerned, the IQAir regularly publishes air quality reports (IQAir, 2018; IQAir, 2019) every year for regions and cities across the globe based on the concentration of PM_2.5. PM_2.5 is the prevailing pollutant across the cities of India having its source primarily from traffic, power and automated industry and dust (Guo et al. 2017, 2019). Higher concentration of PM_2.5 is also the leading cause of human respiratory problem (Xing et al., 2016; Li et al., 2018; Hopke et al., 2019). COVID-19 is more likely to infect people having respiratory diseases (Chen et al., 2020) and also found to have accelerated spread over the region having higher pollution (Coccia, 2020). Therefore selection of urban areas having more concentration of PM_2.5 pollutants is logical one.

2.2. Data sources

In order to investigate the spatial and temporal scenario of air pollution for 10 most polluted cities across the country during the four week lockdown period (21st March to 20th April) amid COVID-19 daily or hourly concentrations of different criteria pollutants are taken into consideration. Data for a total of eight criteria pollutants namely Particulate Matters (PM_2.5 and PM₁₀), Sulphur Dioxide (SO₂), Nitrogen Dioxide (NO₂), Carbon Monoxide (CO), Ozone (O₃), Ammonia (NH₃) and Nitrogen monoxide (NO) are obtained from the CPCB online web portal of air quality data (https://app.cpcbccr.com/AQI_India/). The data can be used rigorously as the CPCB strictly maintains data quality assurance through its Quality Assurance/Quality Control (QA/QC) programme. Moreover, MERRA-2 (Modern Era Retrospective-Analysis for Research and Applications, Version 2) Aerosol Optical Depth (AOD) data (https://giovanni.gsfc.nasa.gov/giovanni/) has been utilized in this study to show the variation in aerosol concentration in the month of March and April this year from the previous two year.

2.3. Methods employed and analysis procedure

In order to understand the overall improvement in air quality during the lockdown period in comparison to the pre-lockdown period and proceeding years the National Air Quality Index (NAQI) introduced by CPCB (2015) has been utilized. The CPCB also prescribed the Indian National (Ambient) Air Quality Standards (INAQS) for each of the pollutants and are highlighted in Appendix 1.

The newly framed, NAQI was launched by CPCB in October 2014 in order to broadcast air quality to the common public in a lucid way. The detailed calculation procedure of NAQI is available at the National Air Quality Index - India Environment Portal (www.indiaenvironmentportal.org.inCPCB, 2015). The CPCB has also developed an excel template for calculation of the same and is available at http://www.arthapedia.in/docs/AQI-Calculator-aug15.xls. In the present article we will highlight the method very briefly.

The calculation of NAQI is based on data of eight criteria pollutants (namely PM₁₀, PM_2.5, SO₂, NO₂, CO, O₃, Pb and NH₃) of which data for at least three pollutants is desirable including one being either PM₁₀ or/and PM_2.5. In the present article, we have excluded Pb because of the fact that the data for Pb is rarely available for all the sample stations used in the study. Based on different ambient concentrations range of each of the criteria pollutants corresponding health break points are specified (Appendix 2). Predominantly two steps are involved in NAQI calculation- Formulation of sub-indices for each criteria pollutant and aggregation of the sub-indices to obtain the NAQI.

For each pollutant (_i) the sub index (NAQI_i) is calculated as (Equation (1))-

where, Ci implies concentration of pollutant ‘i’; B_HI and B_LO are breakpoint concentrations greater and smaller to Ci; IN_HI and IN_LO are corresponding NAQI values.

The overall NAQI is the maximum NAQIi, and the corresponding pollutant is the predominant pollutant. The NAQI has six classes of air quality where each category is associated with certain health impacts specified in Appendix 2.

The MERRA-2 is a NASA atmospheric reanalysis data released by the NASA Global Modeling and Assimilation Office (GMAO) that was launched in 2017 (Randles et al., 2017) replacing the original MERRA of 1980 (Rienecker et al., 2011). The MEERA-2 uses the data assimilation system of the upgraded version of Goddard Earth-observing System Model, Version-5 (GEOS-5) (Randles et al., 2017). In the present analysis the Aerosol optical depth (AOD) products from MERRA-2 are evaluated using available independent ground-based in situ and remote sensing products. MERRA-2 AOD has 8 instantaneous values a day that the 8 instantaneous values are taken every 3 h (from 00:00 to 21:00 UTC). In the present case MERRA-2 instantaneous AOD at 06:00 (UTC- Coordinated Universal Time) were chosen as it corresponds to 11:30am (Indian Standard Time (IST). The MERRA-2 AOD for the month of March and April from 2018 to 2020 is used to analyze the variation during the 2 months as well as comparison with the preceding two years.

3.1. National Air Quality Index (NAQI) at the 10 most polluted cities prior to and during the lockdown

In the present slot we have considered 6 week window period-two weeks before (10th of March to 23rd of March) the commencement of lockdown (i.e. 24th of March); during the three week 1st phase full-fledge lock down (24th of March to 13th of April) as well as 1st week of the 2nd phase partial lockdown (14th of April to 20th of April). Fig. 2 highlights the air quality standard as revealed by the NAQI in different weeks belonging to the pre-lockdown, first phase lockdown and partial lockdown stage. In compassion to the other cities across India (Fig. 2a and b), the 10 selected cities show worse condition of air quality particularly in comparison to the south Indian and coastal cities. In south India strong wind and in coastal cities the land-sea breeze disperses a large amount of the emissions coming out of the industries, vehicles and ports. The sampled cities of the NCR region are landlocked with higher degree of convective activity during pre-monsoon months (Tiwari et al., 2015) and hence higher concentration of ambient air pollutants leading to added NAQI.

In the week-1 (17th - 23rd March) of the pre-lockdown phase (Fig. 2b inset) air quality in the eight cities out of the ten remained in the poor category but in the subsequent week+1 (24th – 30th March) in the lockdown phase (Fig. 2c inset) air quality all of these cities restored to good or satisfactory level. Over 50% improvement in air quality has noticed for Bhiwadi, Noida, Delhi, Faridabad, Muzaffarnagar and Ghaziabad in the 3 weeks of first phase lockdown in comparison to the two weeks pre-lockdown phase (Table 2 , column 11) and at all cities in the first phase of lockdown the improvement is >38% than its previous. In the partial lockdown phase (Week+4) air quality standard (Fig. 2f inset) became little inferior in comparison to its preceding weeks (Week+1 to Week+3) of the lockdown phase (Fig. 2c, d, 2e inset). As stated earlier, after three weeks of lockdown due to letting off of certain controlled industrial activity and necessary transportation in some cities is the possible cause for the slight increase in NAQI on the partial lockdown week. Nevertheless, in this partial lockdown phase as much as 40% reduction in NAQI has been noticed in Bhiwadi, Faridabad, Noida, Greater Noida, Ghaziabad and Delhi (Table 2, column 13). However, improvement in air quality during the entire 4 week lockdown period is relatively less in Jind and Bulandshahr. There may be some localized cause for such lesser reduction in NAQI in the two cities. The attenuation in NAQI during the subsequent weeks of the lockdown periods is for the most part coupled with the variation of prevailing pollutants, mainly PM₁₀ and PM_2.5 discussed afterward.

3.2. Alterations in the concentrations of PM₁₀ and PM_2.5 in pre and post weeks of lockdown announcement

In comparison to the other pollutants, particulate matter (PM) is the foremost one in most of the Indian cities (Guo et al. 2017, 2019). It has also been observed that a nearly uniform meteorological condition for pollutants dispersion prevails during the month of March to April in Indian subcontinent (Tiwari et al., 2015; Yadav et al., 2014). In order to portray the changes in the concentration of PM₁₀ and PM_2.5 in the most polluted cities 24 h readings for each day of a week are taken and averaged for respective weeks and subsequently plotted as Box-and-Whisker Plots (Fig. 3 and Fig. 4 ) and the results of the same are shown in the Supplementary Material section (Supplementary 1 and 2).

As a general trend, during the first phase three week lockdown period all of the cities have witnessed substantial reduction of PM₁₀ as PM_2.5 concentration below the permissible limit (Figs. 3 and 4). In particular, the two primary pollutant concentrations have shown substantial decrease in the first and second week (+1 Week and +2 Week) of the lockdown when the medium concentration of the two pollutants are reduced up to about −49% to −57% respectively. However, during the 3rd and 4th week of lockdown (+3 Week and +4 Week) there was partial relaxation on necessary transportation and controlled industrial activity particularly power plants outside the COVID-19 infected area (Red and Orange Zone) which has caused the concentration of pollutants appreciably. Increase in the concentration of both PM₁₀ and PM_2.5 during the 3rd and 4th week of lock down in the entire 10 sample cities may also be associated with the increase of convective activity. This is because in the month of April and May particularly the north India due to convective activities PM₁₀, MP_2.5 and O₃ are generally remains at its highest level (Yadav et al., 2019). However, for all the cities during the month of April concentrations of the two pollutants however much lower than the previous years and outlined in detail in the subsequent section. This may be due to the fact that in this year, there is no significant variation in the average wind speed (SD 0.9) signifying lesser convective activity during the month of March and April as evident from Fig. 1d and e. When comparing between the cities pertaining to the concentration of primary pollutant PM₁₀ the reduction was substantial for the city Gurugram, Jind, Bhiwadi, Ghaziabad and Noida during 24th March to 06th April (+1 week and +2 Week) (Fig. 3). Nearly similar patterns can also be noticed in these cities in case of PM_2.5 concentration also for the first two weeks of the lockdown period (Fig. 4). Noticeably, the outskirt cities located within the National Capital Region namely Gurugram, Gaziabad, Noida, Faridabad and Greater Noida the concentration of the two primary pollutants have reduced to an average of 39% during the first phase of lockdown period (+1 week to +3 Week).

3.3. Changes in the 24 h concentrations of other pollutants at the selected cities before and during the lockdown

During the pre-lockdown weeks (10–23rd March) all cities (except Jind for PM₁₀) exceed the annual PM₁₀ and PM_2.5 standard of 100 μg/m³ and 60 μg/m³ respectively but during the lockdown weeks (24th March to 13th April) pollution level reduced below the INAQS (except Bulandshahr for PM₁₀) (Fig. 5 a and b). Combining all cities, the average PM₁₀ concentration during the lockdown and partial lockdown period is as much as −49% and −12% respectively (Appendix 3). Whereas the PM_2.5 averaged concentrations is reduced by about −46.9% and −31.5% during the consecutive lockdown periods. Exceptionally in Bulandshahr, Jind and Muzaffarnagar PM₁₀ concentration during the partial lockdown period has increased in comparison to the pre-lockdown period (See Appendix 3, column 9, 10 & 11). However, in the case of PM_2.5 in these cities the considerable decrease (−9%–25.7%) during the partial lockdown period has increased in comparison to the pre-lockdown period. An increase of only PM₁₀ in these cities may be due to the increase in the typical crustal elements (like-Fe, Al, Ti) and dust transport particularly in the urban-traffic sites as evident from several other studies (Chenery et al., 2020; Contini et al., 2014). In Bulandshahr a similar pattern of increase can also be noticed in case of NO and NH₃. NH₃ also found to have increased in Noida, Greater Noida and Gurugram during the partial lockdown period. These cities, located in the NCR Delhi region are counted among the fastest growing Indian cities, usually recognized to be dusty due to lots of construction activities. The regional meteorological factor in association with operation of few coal based industries and biomass burning may be some other factor. Apart from the particulate matter pollution, CO and NO₂ are found to have higher concentration in these cities (Fig. 5g and d) particularly during the pre-lockdown period which has reduced to about −30% and −57% respectively during the three week lockdown period. In the partial lockdown period the trend continues and their concentration has reduced up to −21% and −51% respectively (Appendix 3). In these cities the two pollutants namely CO and NO₂ are largely emanate from traffic, manufacturing industry and power plants and hence their concentrations remained higher. During the partial lockdown period certain relaxation has been given for restricted public transportation, controlled industrial activity and hence their concentration during the partial lockdown period has increased slightly. The 24 h mean concentration of SO₂ and NH₃ for all cities has remained below the permissible limit even during the pre-lockdown period (Fig. 5e and f). This is because all of these cities are landlocked located in northern India, but majority of the SO₂ emission source are from shipping activities and NH₃ emanating from non-agricultural sources is insignificant (Sutton et al., 1995). However, slight reduction (>-18%) in mean concentration of these two pollutants can be noticed during the lockdown periods. Concentration of O₃ has increased during the consecutive lockdown period (Fig. 5g). Generally, the month of April to August is considered as the period of high concentration of O₃ particularly in south-east Asia due to an increase in Insolation (Gorai et al. 2017). Increase in the average air temperature during April, 2020 as evident from Fig. 1b and c supports the fact. The concentration of O₃ has also augmented in these industrial and transport dominated cities possibly due to the decrease in NO (Fig. 5c) which resulted in lesser O₃ consumption (titration, NO + O₃NO₂ +O₂) and consequent increase of O₃ concentrations.

3.4. Diurnal variation of PM₁₀ and PM_2.5 concentrations in three selected sample day belonging to pre-lockdown, lockdown and partial lockdown period

Real-time hourly concentrations of PM₁₀ and PM_2.5 for all the 10 cities for three sample dates from pre-lockdown (10th March), lockdown (27th March) and partial lockdown (20th April) are considered for detecting the pattern of variation (Fig. 6 ; Supplementary 3). 24 h mean concentration and standard deviation (SD) considering all stations are calculated for each sample date to show the pattern of variation between sample dates. The daily mean PM₁₀ in 10th March (pre-lockdown) was as high as 165.8 (SD 81.8) (Fig. 6a), this has dropped to nearly 70% (avg. 43.2; SD 27.8) only on 4th day of the lockdown (27th of March) (Fig. 6c) and remained below the permissible limit. This pattern of variation remains almost the same for PM_2.5 also, however the drop in the mean concentration between 10th of March (pre-lockdown) and 27th of March (lockdown) is almost −72% (Net variation 60.7) (Fig. 6b and d). On 20th of April (partial lockdown) both the PM₁₀ (Fig. 6e) and PM_2.5 (Fig. 6f) again attained its pre-disposition when the average concentration has risen to 122.7 (SD 66.4) and 48.4 (SD 34.5) respectively. Interestingly, very good consistency with lesser standard deviation can be observed in both the pollutants on 27th of March (lockdown), which was less consistent during 10th of March (pre-lockdown). The hourly concentration of PM₁₀ and PM_2.5 also shows significant differences among the cities. Particularly, cities like Ghaziabad, Muzaffanagar, Gurugram, Faridabad and Greater Noida are closer to intensive emission sources from manufacturing and power plant industries leading to more concentration of pollutants. However, heavy PM pollution typically associated with sluggish weather conditions with gentle wind and hence site specific climatic conditions is one of the key influencing factors for such variation.

3.5. Changes in the aerosol optical depth (AOD) over the last three years

In aim to supplement the particulate matter concentration during the lockdown and non-lockdown period as outlined in the earlier subsections, we have also tried to detect the real-time concentration of particulate air pollution with the aid of aerosol remote sensing for the preceding three years (from 2018 to 2020) for the identical window period (i.e. March and April) (Fig. 7 ). From the time series AOD maps for month March (Appendix 4) it is quite evident that the AOD for the year 2020 (Fig. 7e) is more than its previous years 2018 (Figs. 7a) and 2019 (Fig. 7c). As lockdown imposed at the last week of March this year, therefore while considering the entire month for AOD, the effect of the lockdown week (24th march to 31st March) suppressed while deriving in the average concentration for the entire month as the first three weeks of March this year have counted quite more AOD. However, during the month of April in the current year the AOD amount reduced to about 35.5% and 17.5% and in comparison to April 2018 and April 2019 respectively (Table 7). The reduction of AOD during the month of April this year in comparison to April 2018 for the cities ranged from −32.81% (Greater Noida) to-40.68% (Bhiwadi). This is a clear indication that the lockdown has lead to up-gradation of air quality as a result of the sizable reduction aerosol optical thickness concentration and also been documented in Gautam (2020); Dutheil et al. (2020) etc.

3.6. Discussion and conclusion

The lockdown has had significant impacts on the air quality of the 10 most polluted urban centers located within NCR region with enhanced NAQI ratings. The NAQI outcome also conveys that about 70% of the cities across India air quality have restored to either good or satisfactory categories during the lockdown episode. Combining all the selected cities, among the selected pollutants PM₁₀ and PM_2.5 have witnessed maximum reduction during the three week lockdown period counting as much as −49% and −46.9% respectively. Pollutants like CO and NO₂ have reduced to about −30% and −57% respectively during the full-fledged lockdown period. On the contrary there is insignificant decrease in SO₂ and slight increase in O₃ concentration during the period. Drastic drop in the diurnal concentration of PM₁₀ and PM_2.5 to almost −70% can be noticed in 27th of March (Lockdown period) in contrast to that of the 10th of March (Pre-lockdown phase). Real-time concentration of aerosols as revealed from the profile of the monthly averaged AOD for the month of April in the year 2018 and 2019 show much higher value throughout India than this year. The observation again reveals lessening of the pollutant concentration. The framework adopted in this study can be easily transferred to not only examine similar lockdown effects on the air quality of other urban centers as well as industrial clusters, but also for continuous monitoring and image-extracted datasets, which can extend and improve spatio-temporal air quality assessments.

The policy planning and its implementation in this country in most of the cases are limited to the Tier-1 and Tier-2 cities. Yet, several medium and small urban centers of this country are enlisted in the world's most polluted cities. Therefore, there is a need for a closer look at the state of air pollution, particularly the most polluted Tier-3 cities in order to quantify the extent of pollution. Moreover, with the boosting of population and ever-rising urbanization trend particularly in and around the metropolitan cities, especially over the industrial clusters of the Indo-Gangetic Plain there is a need for revisiting the policy planning to combat the air pollution threat. The lockdown amid the COVID-19 pandemic has given us a rare window when the atmospheric pollution and particles are far less than at any point in the last 2–5 years especially for the industrial sites. Although the air quality have bettered yet, this improvement does not necessarily guarantee sustained purer air quality and there is a high likelihood that the environment shall return to its former degraded status once all lockdown measures are lifted and industrial production is increased to make up the economic losses engendered by such a period of closure. Therefore, integrative plans to recover the air quality are sorely required, together with better regulatory and sound technological interventions. The spate of newspaper articles, blogs and tweets about the cleanlier air quality during this lockdown must be used to press on for coordinated and augmented efforts to truly improve and sustain the environmental health, critically reexamining current management practices, which are quite unsustainable and exclusionary.

Of course lockdown has brought threat to the national economy but certainly the environmental refurbishment process also goes on hand by hand. Consequently if a city lockdown undeniably improved the air quality, it could be taken as an effective unconventional measure to restrain air pollution. The baseline report outlined in the present manuscript will definitely add on the researcher, planners and people at large to think for alternative policy interventions so as to manage the ever rising air pollution. However, an increase in the ground level monitoring network for these most polluted cities may support reporting air quality in finer terms. In these selected cities and all other resembling cities across India urbanization along with population boost is an increasing trend accompanied by ever rising automation and industrialization. Hence, these cities need to start pollution control planning by anticipating the possible environmental threat. There is a dearth of study devoted to emission inventory for most of these polluted Tier-3 cities. Therefore, detailed inventory for these medium and small cities having environmental threats need to be focused in order to assess the pollution trend and identify localized pollution hotspots (source contribution).