Impact of COVID-19 lockdown on the fine particulate matter concentration levels: Results from Bengaluru megacity, India.

Leveraging the COVID-19 India-wide lockdown situation, the present study attempts to quantify the reduction in the ambient fine particulate matter concentrations during the lockdown (compared with that of the pre-lockdown period), owing to the highly reduced specific anthropogenic activities and thereby pollutant emissions. The study was conducted over Bengaluru (India), using PM2.5 (mass concentration of particulate matter having size less than or equal to 2.5 µm) and Black Carbon mass concentration (BC) data. Open-access datasets from pollution control board (PCB) were also utilised to understand the spatial variability and region-specific reduction in PM2.5 across the city. The highest percentage reduction was observed in BCff (black carbon attributable to fossil fuel combustion), followed by total BC and PM2.5. No decrease in BCbb (black carbon attributable to wood/biomass burning) was observed, suggesting unaltered wood-based cooking activities and biomass-burning (local/regional) throughout the study period. Results support the general understanding of multi-source (natural and anthropogenic) nature of PM2.5 in contrast to limited-source (combustion based) nature of BC. The diurnal amplitudes in BC and BCff were reduced, while they remained almost the same for PM2.5 and BCbb. Analysis of PCB data reveal the highest reduction in PM2.5 in an industrial cluster area. The current lockdown situation acted as a natural model to understand the role of a few major anthropogenic activities (viz., traffic, construction, industries related to non-essential goods, etc.) in enhancing the background fine particulate matter levels. Contemporary studies reporting reduction in surface fine particulate matter and satellite retrieved columnar Aerosol Optical Depth (AOD) during COVID-19 lockdown period are discussed.

Introduction

Atmospheric Particulate Matter (PM) is known for its negative impacts on the solar radiation budget, air quality and health. Air pollution is known to affect almost every major human organ system: ranging from cardiovascular, respiratory to neurodegenerative diseases (Kim et al., 2015, Schraufnagel et al., 2019) and is the fourth major cause of mortality world-wide (Murray et al., 2020). In 2016 alone, long-term exposure to PM2.5 caused more than 1.2 million deaths in India (HEI, 2019). Black Carbon aerosols (BC, which is mostly a subset of PM2.5 because of its fine size), are a byproduct of incomplete combustion. Combustion generated aerosols are found to be better indicators of adverse health effects (Janssen et al., 2011), compared to PM2.5. BC is also a strong absorber of solar radiation in the atmosphere and the second most contributor to global warming after CO2 (Ramanathan and Carmichael, 2008). Globally, primary sources of urban PM are traffic, domestic fuel burning, industrial activities and miscellaneous anthropogenic activities (Karagulian et al., 2015). In addition to the above, in the Indian context, emissions from coal-based power plants is also listed as one of the major sources of PM (Guttikunda and Jawahar, 2014). While, construction activity, open garbage burning constitute majority of the miscellaneous activities (Banerjee et al., 2015, Rana et al., 2019).

Cities and governments around the world have implemented policy measures to reduce PM emissions. These measures have varying impacts on air quality and downstream health outcomes. For example, for the 2008 Olympics, Beijing implemented several policy measures that involved a combination of industry closure or relocation, strict traffic control and introduction of new emission standards. These measures were in place from late 2007 to late 2008, which led to improved air quality (Chen et al., 2013), reduced air pollution attributable mortality (He et al., 2016) and increased birth weights (Rich et al., 2015). Recently, a 10-day (21 May to 31 May 2018) strike by truck drivers in Brazil led to about 20% decline in PM10, resulting from a complete pause on truck traffic, and overall reduction of other forms of transport (Leirião et al., 2020). The summer season odd–even car trail (a traffic rationing measure) over Delhi, India resulted in a decrease of traffic related PM2.5 levels by 2% − 74% during traffic peak hours (Kumar et al., 2017).

On 11 March 2020, the World Health Organization (WHO), announced the COVID-19 (Novel Corona Virus Disease) outbreak as a pandemic. Enforced ‘social distancing’, to contain the spread of COVID-19, halted diverse anthropogenic activities across the globe. In response to the pandemic, India imposed a nationwide lockdown on 24 March 2020 (till 3 May 2020 and thereafter activities were relaxed in a phased manner), clamping down on vehicular and human movement for non-essential purposes. This had resulted in a near-curfew scenario in which normal private and public vehicles were banned and only the transport of essential goods and personnel was permitted. People were restricted to their homes and allowed to go out only in case of emergencies or to buy essentials like groceries or medicine. Non-essential commercial services were also on pause during the lockdown, leading to marked reduction in energy/goods consumption and solid waste generation. As a result, major anthropogenic activities potentially contributing to PM emissions, namely, traffic, construction, and possibly waste burning (due to lower waste output) were highly reduced.

Bengaluru (12.97 oN, 77.59 oE), the capital city of the state of Karnataka, located in the southern part of India is known as the ‘Silicon Valley’ of India owing to its IT (Information Technology) infrastructure. Various studies (e.g., Babu et al., 2002, Patil et al., 2013) have reported the air pollution levels of Bengaluru, which is also labeled as a non-attainment city (a polluted city violating the prescribed national air quality standards; MoEFCC, 2019). A recent emissions inventory study by Guttikunda et al. (2019) listed the transport sector (vehicular exhaust and on-road dust resuspension) as one of the primary sources of PM emissions in Bengaluru, contributing 56% and 70% to PM2.5 and PM10 emissions. Leveraging the lockdown situation, the current study attempts to quantify the reduction in fine PM (PM2.5 and BC) over Bengaluru city in relation with the reduced anthropogenic activities, particularly vehicular movement, construction, and other non-essential small-scale industries.

Data, instrumentation and study locations

Data

Real-time PM2.5 and BC measurements made at the Center for Study of Science, Technology and Policy (CSTEP), Bengaluru during the period from 1 March to 22 April 2020 are used in this study. Due to technical issues, BC data is not available from 23 April 2020; for uniformity all the analysis was restricted till 22 April 2020. Hourly open-access PM2.5 data from the PCB’s continuous monitoring sites for the periods 1 March to 22 April 2019 and 1 March to 22 April 2020 is also used to support the inferences drawn based on the CSTEP data and to study the spatial variability (within Bengaluru) in the possible improvements in air quality.

Instrumentation

Beta Attenuation Monitor (BAM-1022)

A BAM (Model: 1022, Met One Instruments, Inc., USA) is used to measure near real-time PM2.5. BAM-1022 uses the beta attenuation technique (beta source: 14C) to measure the mass concentrations of aerosol particles collected onto a glass fiber tape. The accuracy of the BAM-1022 meets the requirements of the USEPA (United States Environmental Protection Agency) class III FEM (Federal Equivalent Method). BAM-1022 (equipped with a manufacturer-supplied 2.5 µm size cut cyclone) operates at 16.7 LPM (liter per minute), with an inbuilt heating arrangement to avoid humidity related errors in the PM2.5 measurements. The detection limit for the analyser is <1 µg m−3 (24 h) and span measurement ranges from −15 to 104 µg m−3. PM2.5 data from the hourly channel of BAM-1022 alone is used for the analysis presented in this paper. More technical details on BAM-1022 can be found at https://metone.com/products/bam-1022/. The BAM-1022 was installed on the CSTEP terrace at a height of ~10 m from the ground and is ~110 m away from the major road.

Aethalometer (AE-33)

A seven-channel (wavelengths: 370, 470, 520, 590, 660, 880, and 950 nm) Aethalometer (Model: AE-33, Magee Scientific, USA) is used to measure real-time BC. AE-33 is a fast-response instrument, which measures the spectral optical attenuation to estimate high temporal resolution BC, with the aerosol sample being collected onto a quartz fiber filter tape. Using wavelength specific mass absorption coefficients, absorbing aerosol mass concentration is estimated at all seven wavelengths. Due to the strong absorption characteristic of BC at 880 nm, the absorbing aerosol mass concentration measured at this wavelength is considered as BC. The dual-spot technique employed in AE-33 compensates for the spot loading effect, which is a typical measurement artifact for any filter-based absorption measurement. AE-33 utilises the spectral light-absorption measurements to apportion BC obtained from fossil-fuel combustion and wood burning sources (Sandradewi et al., 2008a, Sandradewi et al., 2008b). Principally, the total measured BC is the sum of BC from fossil-fuel (BCff) and wood/biomass burning sources (BCbb). A detailed description of the instrument working principle is given in Drinovec et al. (2015). The AE-33 sampling inlet was installed at a height of ~8 m, above the ground level. The flow of the AE-33 is set to 2 LPM and data is logged at 1-minute averaging interval. The instrument is equipped with a 2.5-µm size cut cyclone and a bug filter. A detailed description on the aethalometer measurement uncertainties is given in Backman et al. (2017). For the analysis in the present study, values of BC, BCff, and BCbb measured/estimated at 880 nm are used.

Pollution control board (PCB) PM2.5

To study the spatial variability and region-specific PM2.5 levels, data from Central and State Pollution Control Boards’ (CPCB/SPCB’s) Continuous Ambient Air Quality Monitoring Stations (CAAQMS) was downloaded (https://app.cpcbccr.com/ccr/#/caaqm-dashboard-all/caaqm-landing/data) and analysed from four stations (of the 10 monitoring stations installed in Bengaluru), based on the data availability. The selected CAAQMS stations are (i) Peenya (PNY) (ii) BTM Layout (BTM), (iii) Hebbal (HBB) and (iv) Hombegowda Nagar (HMB). CAAQMS use reference-grade instrumentation for all criteria pollutant ambient concentration measurements. A quality check algorithm is applied on the CAAQMS data before its use. The algorithm includes removal of negative, fill and spurious values. This algorithm flagged ~2% data (hourly PM2.5). Daily aggregates are made only if ~75% of the data in a day is available after cleaning (Kumar et al., 2018). The geographical spread of the study stations is shown in Fig. 1 . Meteorological fields (temperature, relative humidity, wind speed and wind direction, rainfall) from these stations are also used.

Fig. 1

Geographical spread of the study sites (black dots) superimposed on Open Street Map (OSM-black and white). The gray lines indicate roads and the gray patches indicate water and green bodies. The red line indicates the Bengaluru Urban boundary. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Global precipitation measurements (GPM)

Daily accumulated rainfall data used in the present study is acquired from gridded (0.1° × 0.1° spatial resolution) GPM precipitation products. GPM is a next-generation satellite mission for observing the global precipitation characteristics with enhanced spatiotemporal resolution. GPM carries the first spaceborne dual frequency precipitation radar (Ku band at 13.6 GHz and Ka band at 35.5 GHz) and a multichannel GPM microwave imager. Several studies (e.g., Tang et al., 2016) compared GPM rainfall data with that of ground-based measurements and found good agreement. More technical details of GPM can be found elsewhere (Skofronick-Jackson et al., 2017) and are not repeated here.

Meteorological parameters

Data on meteorological variables (ambient temperature, Wind Direction and Wind Speed) are acquired from the meteorological monitoring system installed at PCB CAAQMS sites. All the meteorological sensors (National Institute for Standards and Technology-USA traceable) are mounted on a telescopic 10 m (above ground level) tower, sensor data directly interfaced with the Data Acquisition System along with pollutants data. Analysis of meteorological data from BTM is shown (based on data availability) in the current study. As the reference grade monitors measure dry PM2.5, data on relative humidity (RH) is not shown. Daily mean values are generated by averaging hourly data.

Study sites

All the study sites (Fig. 1) are located within Bengaluru’s administrative boundary. The population of Bengaluru was estimated to be ~10 million in 2015 (Guttikunda et al., 2019). CSTEP is located close to the outer ring road in North Bengaluru. CSTEP’s proximity to the major road and a railway track makes it susceptible to traffic related pollution. Situated in the north-west part of the city, PNY is an industrial region. It houses small, medium and large-scale industries. Because of their contribution to high pollution levels in the city, most of the industries in this area are being shifted to outer regions of Bengaluru. BTM is one of the popular residential and commercial neighborhoods in South Bengaluru due to its proximity to the outer ring road and other important locations of the city. HBB monitoring station is located inside the green campus of Veterinary College, which is adjacent to one of the busiest major roads in North Bengaluru. HMB is also a residential layout near central Bengaluru.

Results

Days from 24 March (nationwide lockdown start date) through 22 April 2020 (30 days) constitute the period of reduced anthropogenic activity and are termed as event days (ED). To compare and contrast the pollution levels during ED against regular days, the study used data before the lockdown period (1 to 23 March 2020, termed as control days (CD).

Meteorological parameters

The study period (CD and ED) spanned the months of March and April, both of which belong to the Indian summer season. Based on this, any observed differences in the pollution levels during CD and ED cannot be attributed to the synoptic scale variations in air-pollution levels. To rule out any possibility of modulation of air pollution levels by local meteorological conditions during CD and ED, we investigated the daily mean temperature (oC), wind speed (WS, m s−1), and wind direction (WD, o) for the study period, using the data collected by PCB weather station. Daily accumulated rainfall (mm) is acquired from Global Precipitation Measurements (GPM) satellite observations. The temporal variation in the daily mean meteorological parameters is shown in Fig. 2 for the study period. There is no gross difference in the meteorological conditions during CD and ED, except in rainfall. There are three days during ED for which, the daily accumulated rainfall is greater than 10 mm, while it is only one day during CD. On 6 April, ~50 mm of daily accumulated rainfall was recorded. Mean temperature and WS values during CD and ED are ~22.6 ± 0.9 °C, 1.0 ± 0.26 m s−1 and 22.8 ± 0.2 °C, 0.8 ± 0.12 m s−1 respectively. During the whole study period, the WD is maintained at ~160°.

Fig. 2

Temporal variations of the meteorological parameters during pre-lockdown period (control days, CD) and lockdown period (event days, ED).

PM2.5 and BC measurements at CSTEP

The hourly PM2.5 and BC measured at CSTEP are shown as a daily box and whisker plot in Fig. 3 . ED (lockdown days) is shaded in gray. In the box plot, the solid red dot represents the mean, the range of the box indicates the 25th and 75th percentile values and the central line indicates the 50th percentile (median) value. The range of the whiskers indicates the 9th and 91st percentile values and the ‘+’ symbol indicates the outliers in the distribution. Qualitatively, it can be inferred from the plot that PM2.5 levels during CD and ED have not varied greatly. In contrast, the BC levels were observed to be significantly low during ED compared to that in CD. Within the study period, PM2.5 exhibited lowest values on the 7th (median and mean values of ~18 µg m−3) and 10th of April (daily median value ~17 µg m−3 and mean value ~18 µg m−3), possibly due to the combined effect of lockdown and wet-removal by the rain on the previous days. Meanwhile, such a dip is not seen in BC. The mean ± standard deviation (median ± IQR) values of PM2.5 for CD is 38 ± 15 µg m−3 (36 ± 17 µg m−3), while it is 30 ± 16 µg m−3 (28 ± 13 µg m−3) for ED (Table 1 ). By removing data for the rain affected days (7th and 10th of April), the ED PM2.5 aggregates (mean and median) just increased by 1 µg m−3. These numbers translate to a reduction of ~20% (8 µg m−3) in mean and ~23% (8 µg m−3) in median PM2.5 values with respect to CD. Welch’s t-test was performed to understand the statistical significance of the reduction, which revealed that the daily mean and daily median PM2.5 reductions were statistically significant (p < 0.05). The mean ± standard deviation (median ± IQR) values of BC for CD was 7.0 ± 4.7 µg m−3 (6.1 ± 4.6 µg m−3), while for ED, it was 3.1 ± 3.1 µg m−3 (2.7 ± 1.7 µg m−3), which turns out to be a 55% reduction in BC. The t-test results show that the reduction in BC was highly significant (p < 0.01).

Fig. 3

Daily variation of PM2.5 and BC measured at CSTEP for the study period. Shaded area corresponds to the lockdown period (event days, ED). The solid red dot represents the mean, the range of the box indicates the 25th and 75th percentile values, and the center line indicates the 50th percentile (median) value. The range of the whiskers indicates the 9th and 91st percentile values and the ‘+’ symbol indicates the outliers in the distribution. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Table 1

Mean PM2.5 for pre-lockdown period (control days, CD) and lockdown period (event days, ED) observed at various study stations. Numbers in brackets indicate the number of data availability days. Percentage reduction is calculated with respect pre-lockdown period.

Station/Location	Mean PM2.5, μg m−3		% reduction
	CD	ED
CSTEP	38 (23)	30 (30)	20
Peenya (PNY)	45 (23)	35 (28)	22
BTM Layout (BTM)	30 (20)	26 (30)	14
Hebbal (HBB)	32 (23)	26 (22)	19
Hombegowda Nagar (HMB)	32 (19)	26 (19)	19

Fig. 4 shows the daily variation of the BC components (BCff and BCbb). During CD, BCff contributed ~90% to the total (BC). Clearly, there is a large reduction in the BCff during ED (compared to that in CD), which translated into BC reduction. The mean ± standard deviation (median ± IQR) values of BCff for CD was 6.2 ± 4.4 µg m−3 (5.4 ± 4.2 µg m−3), while for ED, it was 2.3 ± 2.2 µg m−3 (2.0 ± 1.3 µg m−3) (Table 2 ). This indicates a reduction of ~63% (3.9 µg m−3, highly significant) in the mean and median BCff during ED. Quantitatively, there was no reduction in BCbb values (CD mean: ~0.79 µg m−3; ED mean: ~0.81 µg m−3) due to the lockdown situation; in fact, a marginal increase was observed.

Fig. 4

Daily variation of BCff and BCbb measured at CSTEP for the study period.

Table 2

Mean and standard deviation (SD) of Black Carbon (BC) and its constituents (measured at CSTEP) for pre-lockdown period (control days, CD) and lockdown period (event days, ED). Numbers in brackets indicate the number of data availability days. Percentage reduction is calculated with respect pre-lockdown period.

Parameter	Mean ± SD, μg m−3		% reduction
	CD	ED
BC	7.0 ± 4.7 (23)	3.1 ± 3.1 (30)	56
BCff	6.2 ± 4.4 (23)	2.3 ± 2.2 (30)	63
BCbb	0.79 ± 0.49 (23)	0.81 ± 1.1 (30)	−3

Fig. 5 shows the diurnal variations (computed from hourly median values) in PM2.5, BC, BCff and BCbb for CD and ED periods. All parameters exhibited the classical bi-modal diurnal variation as reported by various earlier studies (e.g. Sreekanth et al., 2018). For PM2.5, there was a consistent decrease during ED, throughout the day (except for few early hours), maintaining a uniform diurnal amplitude (maximum-minimum, ~ 16 µg m−3) during the study period. In the case of BC (and BCff), in addition to the reduction in the absolute values, diurnal amplitudes also shortened (~7.1 µg m−3 during CD vs 2.9 µg m−3 during ED). The morning peak (rush hour) in BC was reduced by ~6.2 µg m−3, while the evening peak reduced by ~4.3 µg m−3. There was no difference in the diurnal structure in BCbb for CD and ED.

Fig. 5

Diurnal variation in median PM2.5, BC, BCff, BCbb. The red line represents pre-lockdown period (control days, CD), while the green line represents the lockdown period (event days, ED). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Pollution control board (PCB) PM2.5

Fig. 6 shows the box plots obtained from PCB measured PM2.5 at four different locations in the city. The box plots are derived from the daily median PM2.5 (median is chosen rather than the mean, to avoid any possible influence of outliers on central tendencies) values. Making use of the multi-year measurements, data for the year 2019 is also shown in Fig. 6 to better understand whether the observed differences in CD and ED PM2.5 are a result of the lockdown situation or a regular annual phenomenon (even though CD and ED periods are part of the same Indian summer monsoon season). Among the study sites, PNY has the highest PM2.5, with a mean PM2.5 during CD (ED) of ~45 (35) µg m−3), owing to its industrial nature. The box plots indicate that the PM2.5 levels during the period from 24 March to 22 April 2019 are always higher than or similar to that in the period from 1 March to 23 March 2019. This observation confirms that the reduction in the observed PM2.5 during ED can be attributed to the lockdown situation. PNY (an industrial location) observed the highest reduction in PM2.5 levels (~10 µg m−3) during ED, followed by HMB, HBB (~6 µg m−3) and BTM (~4 µg m−3) (see Table 1). When translated to percentage reduction with respect to the CD conditions, PNY recorded the highest (~22%), followed by HMB, HBB (~19%) and BTM (~14%). t-test results revealed that the reductions observed only over PNY and HMB are statistically significant (p < 0.05).

Fig. 6

PCB-measured PM2.5 at various stations in Bengaluru. The red box represents data for 1–23 March, and the green box represents data for 24 March to 22 April. Data for 2019 is also shown to substantiate that the observed difference in PM2.5 between pre-lockdown period (control days, CD) and lockdown period (event days, ED) is due to the lockdown situation. Data from only four PCB CAAQMS stations (out of 10 in Bengaluru) is shown due to the unavailability of PM2.5 data from rest of the stations. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 7 shows the PM2.5 diurnal variations across the PCB sites. The diurnal variations clearly exhibit both morning and evening peaks (BTM has multiple peaks). Clear spatial variability is seen in the levels and shape of the diurnal variations. PM2.5 during ED is consistently low across all times of the day and across all sites compared with that of CD (except for a few early morning hours over HMB and HBB). The highest diurnal amplitude (~32 µg m−3) is observed over PNY, followed by HBB (~24 µg m−3).

Fig. 7

Diurnal variations in median PM2.5 measured by PCB at various stations in Bengaluru. The red line represents the pre-lockdown period (control days, CD), while green line represents the lockdown period (event days, ED). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Discussion

PM2.5 and BC levels reduced during the COVID-19 nation-wide lockdown period (ED) over the city of Bengaluru. Few recent articles have also reported improvement in air quality and CO2 levels in similar COVID-19 lockdown situations across the globe (e.g. Mahato et al., 2020, Calma, 2020, Myllyvirta, 2020). In the present study, the observed PM2.5 reductions are not statistically significant over all the five study sites located within the Bengaluru city boundary (Fig. 1). During the lockdown period, a near-curfew scenario was observed with all non-essential travel, activities (e.g., construction), and industries being strictly prohibited; only emergency or essential services (e.g., hospitals, sanitation), travel (e.g. ambulance, food, dairy and grocery delivery trucks, goods trains), industries (e.g. food processing units) were exempt from the lockdown. The absolute values of PM2.5 reduction was attributed to the lockdown situation, assuming that, the PM contribution from natural sources was almost constant across the study period (both control and lockdown periods are part of the 2020 summer monsoon season).

Guttikunda et al., (2019) have listed travel exhaust, domestic emissions, industries, dust, waste burning, generator sets, and brick kilns as the major contributors to PM2.5 emissions over Bengaluru (in addition to any natural sources such as open fires, sea-salt, biogenic, dust storms etc.). During March, dust is identified as a leading contributor to PM2.5 over Bengaluru. Because of the multi-source nature of PM2.5 and the fact that the source strengths of some of the major sources (e.g. dust, domestic emissions, generator sets) are independent of the lockdown situation, the reductions observed in their levels are relatively small (15–22%) and non-significant (over 2 study sites). In contrast, the observed PM2.5 reductions during lockdown (ED) over CSTEP, PNY and HMB are significant. In the case of PNY, it is understood that most of the industries were shut during the lockdown and a considerable decrease was expected and observed. Over HMB, which is a residential site, a statistically significant 19% reduction was observed. BTM, which is again a residential site, experienced a non-significant reduction. Over residential areas, in general, domestic emissions dominate over other emission sources. HBB, which is located next to one of the busiest major roads, observed a non-significant decrease (~19%) in PM2.5 levels. The non-significant reduction can be partly attributed to the atmospheric residence times of the fine particulate matter. In general, particles have lifetimes of ~1 week to 10 days (based on their size). Changes in pollutant levels during the nationwide lockdown have been reported from several other cities across India (e.g., Kumar et al., 2020). On the basis of real-time PM2.5 data, Navinya et al., 2020a, Navinya et al., 2020b studied PM2.5 distribution over 17 cities in India to assess the reduction in PM2.5 levels during the lockdown period compared to pre-lockdown levels. Ahmedabad (a city in western India) showed the highest reduction of 67.7%, while Mumbai (a city along the west coast of India) experienced a reduction of only 0.9% (Navinya et al., 2020a, Navinya et al., 2020b). Mor et al. (2020) reported nearly 30% (6 µg m−3) reduction in PM2.5 over Chandigarh (a city in northern India) during the lockdown period. Jain and Sharma (2020) examined PM2.5 levels from 38 monitoring stations located across New Delhi (Indian national capital, located in northern India) and reported an overall reduction of around 41% during the lockdown period. Sharma et al. (2010) evaluated changes in PM2.5 levels over seven different cities across Rajasthan (largest Indian state, located in northwestern India). Ajmer showed the highest (47.62%), whereas Udaipur showed the least reduction (22.52%) in lockdown PM2.5 levels (Sharma et al., 2010). Few researchers also reported reduction in columnar particulate matter during the lockdown period. Based on satellite aerosol optical depth (AOD) measurements over north India, Mahato and Ghosh, (2020) reported a reduction of around 32–40% in columnar aerosol burden during April 2020 (lockdown period) compared to the columnar aerosol burden in April 2018. The impact of COVID-19 lockdown on AOD has also been reported by Lokhandwala and Gautam, (2020).

BC and PM2.5 (over CSTEP) recorded significant reduction. CSTEP, located close to the outer ring road experiences significant vehicular emissions. During control days, the average BC contribution to PM2.5 is ~18%; moreover, the BC reduction during the lockdown period is primarily dictated by BCff. All of these arguments and inferences coalesce into the conclusion that the observed reduction in PM2.5 levels are mostly due to suppressed vehicular activity and coal-based industrial activities during ED. BCbb, which corresponds to wood/biomass burning activities, is unaltered during the lockdown period, suggesting near constant biomass cooking emissions (mostly happening in the slum areas of Bengaluru) and background. As the lockdown situation is being implemented nation-wide, the transported anthropogenic component could be less than that of the pre-lockdown period. A similar study over Bhubaneswar (a city along the east coast of India) reported a 47% reduction in BC during the lockdown period when compared to the pre-lockdown level (Panda et al., 2020). Moreover, the generally present prominent bimodal peaks in the diurnal variation of BC were not observed during the lockdown period. Another similar study over European cities revealed a 37%-72% reduction in BC over a five-week period of restricted traffic in certain areas of the cities (Titos et al., 2015). The reduction in concentrations however were very local and did not improve the air quality in other areas of the city where the controls were not implemented. A week-long nationwide strike of truck operators that occurred in India in January 2009 provided another such model for a natural experiment. During the strike, data from another south Indian city, Hyderabad, revealed consistent reduction in BC and PM2.5 due to a complete stop of truck traffic (Sharma et al., 2010).

Even though the lockdown was implemented as a mitigation strategy for a pandemic and not for improving the air quality, the unintended consequences here are a welcome side-effect. Our analysis could be used in deciding which aspects of the lockdown could continue in the post-pandemic situation, as a deliberate strategy for emissions reduction. Across the world and also in India, universities and schools are shut down with classroom instruction moving online, corporate offices are implementing work-from-home using online tools, and conferences and professional meetings are being held via webinars. Governments may consider which of these aspects can continue and contribute to lowered emissions. Implementing work-from-home policies alone may lead to co-benefits of air quality improvement and climate-change mitigation by cutting down transport-related emissions (Irwin, 2004).

Summary

This study examines the impact of a nation-wide COVID-19 lockdown on the air quality of Bengaluru, India. Concentrations of ambient PM2.5 and Black Carbon (BC) at several locations in the city were used to compare the levels in the pre-lockdown (control days, CD) and during-lockdown (event days, ED) periods. Data for a total of 53 days (23 CD and 30 ED) from five different locations in Bengaluru were analysed. During the lockdown (ED), the daily PM2.5 levels reduced by ~15–22% (with respect to CD). A spatial variability was observed in the reductions of PM2.5 in terms of magnitude and statistical significance. Concentration reductions were more pronounced in BC, with ED levels ~55% lower than CD levels. This reduction was completely driven by reduction in the fossil-fuel burning component (BCff) which constitutes ~90% of the total BC during CD. Specifically, there was a 63% reduction in BCff levels during ED.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.