Air quality during three covid-19 lockdown phases: AQI, PM2.5 and NO2 assessment in cities with more than 1 million inhabitants.

Implemented quarantine due to the ongoing novel coronavirus (agent of COVID-19) has an immense impact on human mobility and economic activities as well as on air quality. Since then, and due to the drastic reduction in pollution levels in cities across the world, a large discussion has been magnetized regarding if the lockdown is an adequate alternative counter-measure for enhancing air quality. This paper aimed at studying the Air Quality Index (AQI), PM2.5, and tropospheric NO2 levels in three lockdown phases (before, during, and after) among 21 cities around the world. Simple before/after comparison approach was carried out to capture the declining trend in air pollution levels caused by the lockdown restrictions. The results showed that the frequency distribution for NO2 is more variable than that for PM2.5, and the distribution is flatter from 2020 to the baseline 2018-2019 period. Besides, AQI, in most of the cities, has varied from high to mild pollution during the lockdown and was moderate before. Although during the lockdown, a reduction of 3 to 58% of daily NO2 concentrations was observed across the cities, an increase was detected in three cities including Abidjan (1%), Conakry (3%), and Chengdu (10%). Despite this mixed trend, the NO2 time series clearly showed the effect of the unlocking phase where the NO2 levels increased in almost all cities. Similarly, PM2.5 concentrations have increased in the post-lockdown period, with 50% of the cities reporting significant positive differences between the lock and the unlock phase. Then, the levels of PM2.5 were higher at the pre-lockdown phase than at any other time exhibiting a "U" shape. In addition, during Ramadan, it was noted that altered patterns of daily activities in some Islamic cities have a significant negative impact on air quality.

Introduction

The COVID-19 pandemic has forced the largest quarantine or self-isolation in human history and led to global lockdown (in different phases). Lockdown has resulted in restricted transportation such as trains, airplanes, and vehicles, closure of industries (temporary), and other primaries as well as the secondary of economic activities. These activities are directly proportional to air pollution (Iverach, Mongan et al., 1976; Dos Anjos Paulino, Oliveira et al., 2014; Alnawaiseh et al., 2015). Several studies have already been done to evaluate the impact of COVID-19 induced lockdown on the concentrations of some pollutants in different countries like China, Canada, India, Brazil, Italy, USA, Kazakhstan, Korea, Morocco, etc. (Adams 2020; Bao and Zhang 2020; Berman and Ebisu 2020; Gautam 2020; (Hoang and Tran, 2021); Kerimray et al. 2020; Li et al., 2020; Nakada and Urban 2020; Sharma and Balyan 2020; Sicard et al., 2020; Wu et al., 2020; Otmani et al., 2020). Although a few studies on India said ambient air pollution level has reduced to 20 years low, and in New Delhi it has reduced to 50% (Gautam 2020; Mahato, Pal et al. 2020), a paper from China highlighted that the lockdown could not avoid the severity of air pollution in some Chinese cities (Wang, Chen et al. 2020).

Recently many studies suggested that air quality improvements were related to partially or fully lockdowns, and the consequent decrease of activities and emissions from sources that cause air pollution including road traffic and industrial activities. According to Sicard et al. (2020), the lockdown in China (Wuhan) and in four European cities (Nice, Rome, Valencia, and Turin) has drastically reduced air pollutant concentrations, particularly NO2 by about 56 % in all cities, and particulate matter by about 42% in Wuhan and 8% in Europe, during the lockdown period compared to previous years. Furthermore, numerous research studies have reported that the lockdown has amplified O3 in the lower atmosphere (Sicard et al. 2020, Zoran et al. 2020). In Southern Europe, the PM2.5 and PM10 levels revealed marginal reductions in Nice (3% and 6%) and Turin (13% and 9%) and increased in Rome (11% and 2%). Unlike PM10 (14%), PM2.5 (6%), large reductions of NO2 levels (27%- 34%), were also observed at most of the Southeast Asian cities during the lockdown phase compared to the 2015-2019 period (Kanniah et al., 2020). In Salé, Morocco, SO2 and NO2 levels reduced by 49% and 96%, respectively, during the lockdown compared to the 2018–2019 averages of the same period, while PM10 reduced by 75% (Otmani et al., 2020). Wang, Yuan et al. (2020) study in 366 cities, China, found that NO2 emission reductions were linked to the transport sector. For instance, the sharp decline in the vehicle and public transport due to the lockdown led to a dramatic reduction in NO2 which mainly originates from combustion processes. However, PM2.5 and PM10 changes during the lockdown period were controversial because their pattern is derived from different emission sources. Sicard et al. (2020) reported that the decrease in PM was counter-balanced by an increase from domestic heating (requiring people to stay at home) and garden activities (e.g. biomass burning). In addition, Otmani et al. (2020) demonstrated that long-range transported aerosol contributions out-balanced the reductions in locally emitted PM10. Li et al (2020) showed in Yangtze River Delta region, China, that even during the lockdown, with primary emissions reduction of 15%-61%, the daily PM2.5 levels, came from background and residual pollutions, are still high. This indicates that despite the extreme reductions in primary emissions, it cannot fully bring down the actual air pollution. Table 1 summarizes some recent studies conducted to assess the impact of countermeasures on air quality during COVID-19 lockdown. It covers the type of dataset used, post-processing techniques, and the most noteworthy findings.

Table 1

Summary of some recent studies on air quality changes due to COVID-19 lockdown in various global cities. It covers the type and the sources of the dataset used, post-processing techniques, and the most noteworthy evaluation measures.

Title	Study area	Type of used datasets	Pre-processing techniques	Evaluation measures	Reference
Temporary reduction in fine particulate matter due to anthropogenic emissions switch-off during COVID-19 lockdown in Indian cities	India (5 cities)	– Hourly PM2.5 data in five Indian cities (Chennai, Delhi, Hyderabad, Kolkata, and Mumbai) were extracted for the period between January 2015 and May 2020.– Data are provided from monitoring stations using beta-attenuation monitors and are available online (https://www.airnow.gov/).	– Divided the lockdown period (25 March 2020 onwards) into four phases regarding the quarantine plan imposed by the Government of India.– Examined impact of lockdown measures on the distribution of PM2.5 concentrations using theoretical probability density function (PDF).	– The lockdown restrictions reduced the concentration of PM2.5, from 19% to 43% (Chennai), 41–53% (Delhi), 26–54% (Hyderabad), 24–36% (Kolkata), and 10–39% (Mumbai).– Cities with larger traffic volumes showed greater reductions.– Aerosol loading decreased during lockdown (April 2020) for Chennai (29–57%), Delhi (11–29%), Kolkata (2–14%), and Mumbai (1–48%). However, Hyderabad showed fluctuations, with an increase of 25% in comparison to 2019, and a decrease of 8% concerning 2018.	Kumar et al. (2020)
Changes in air quality related to the control of coronavirus in China: Implications for traffic and industrial emissions	China (366 urban areas)	– The real-time monitoring data for PM2.5, PM10, SO2, NO2, O3, and CO in the 366 urban areas were provided by China's National Environmental Monitoring Center.– The AQI monitoring values were calculated from the six priority pollutants.	– Before and after the Covid-19 control measures period were compared for changes in air quality index (AQI) and in concentrations of six air pollutants (PM2.5, PM10, CO, SO2, NO2, and O3).	– The transportation sector was linked to the NO2 emission reductions, while lower emissions from secondary industries were the major cause for the reductions of PM2.5 and CO.– The reduction in SO2 concentrations was only linked to the industrial sector.– O3 increase may be related to the observed decreases in fine particles.	Wang, Yuan et al. (2020)
Impact of Covid-19 lockdown on PM10, SO2, and NO2 concentrations inSalé City (Morocco)	Morocco (Salé)	– Continuous ground-based measurements of PM10, SO2, and NO2 after and during the lockdown period.– Low-cost electrochemical sensors were used to collect high-resolution temporal data in real-time of SO2 and NO2.	– Compared the pollutants levels before and after lookdown restriction.– Assessed the inter-period back-trajectory variabilities using the HYSPLIT model.	– The reduction in PM10 (75%), SO2 (49%) and NO2 (96%) concentrations could be attributed to the measures limiting human movement and industrial activities.– Long-range transported aerosol contributions (from the Mediterranean Basin and the near Atlantic Ocean) out-balanced the reduction in PM10 locally emitted.	Otmani et al. (2020)
Spatial and temporal variations of air pollution over 41 cities of India during the COVID‑19 lockdown period	India (41 cities)	– Data of NO2 and Aerosol Optical Depth (AOD) from the TROPOspheric Monitoring Instrument (TROPOMI) and MODIS sensors.	– Using univariate autoregressive moving-average (ARMA) model to quantify the impacts of COVID-19 on the pollution levels.– Adopted paired t-test to compare the mean differences between NO2 pollution levels during different months for the previous (2019) and the current year (2020).	– Significant reduction in pollution, due to the COVID-19 lockdown, in major metropolitan cities.– The 2020 lockdown period mostly occurred during the Summer or pre-monsoon season.– The cities in the northern part of India see elevated pollution during the post-monsoon season due to the combined effect of anthropogenic and atmospheric factors.– Some cities experienced an increase in pollution (less reduction on NO2) due to the fires and biomass burning during the COVID-19 lockdown period.	Vadrevu et al. (2020)
Amplified ozone pollution in cities during the COVID-19 lockdown	Europe (Nice, Rome, Valencia, and Turin) and China (Wuhan)	– Hourly data of NO, NO2, PM2.5, PM10, and O3 concentrations were obtained from the local and regional agencies in charge of air monitoring stations.	– Estimated the lockdown effect on air pollutants levels by the changes within the time series of pollutants.– Computed the changes for each day and station, by calculating the mean bias between the period before and during the lockdown in 2020 and the same period averaged over the 3 previous years (2017–2019).	– The lockdown caused a substantial reduction in NOx in all cities (56%).Reductions in PM were much higher in Wuhan (42%) than in Europe (8%).– The lockdown caused an O3 increase in all cities (17% in Europe, 36% in Wuhan).– The lockdown effect on O3 production was higher than the weekend effect.	Sicard et al. (2020)
The dark cloud with a silver lining: Assessing the impact of the SARS COVID-19 pandemic on the global environment	Global	– Weekly data of CO and NO2 from Sentinel 5-P TROPOMI satellite dataset.– Aerosol Optical Depth data from MODIS satellite dataset.– Temperature and relative humidity hourly datasets from the Copernicus Climate Data Store (CDS), and European Center for Medium-Range Weather Forecast (ECMWF).– Country-wise cumulative cases of COVID-19 affected population and death tolls from WHO.	– Evaluated the impact of the COVID-19 using spatio-temporal vitiation method.– Studied the impact of meteorological variables and air pollution (CO, NO2) on the global infection and spreading rate of COVID-19.	– Observed a substantial reduction in NO2, low reduction in CO, and low to moderate reduction in AOD in the major hotspots of the COVID-19 outbreak (February–March 2020).	Lal et al. (2020)
COVID-19’s Impact on the Atmospheric Environment in the Southeast Asia Region	Malaysia and Southeast Asia (65 air pollution monitoring stations)	– Aerosol optical depth (AOD) observations from Himawari-8 satellite dataset over Southeast Asia SEA.– Tropospheric NO2 column density from Aura-OMI data over Southeast Asia SEA.– Ground-based pollution measurements at several stations across Malaysia.	– Comparing the absolute difference between 2020 and data averaged over 2015-2019 (baseline).– Quantified the accuracy of the Himawari-8 L3 AOD against AERONET using statistical measures such as the root mean square error (RMSE), relative bias (RB), and mean absolute error (MAE).	– In Malaysia, PM10, PM2.5, NO2, SO2, and CO concentrations have been decreased by 26-31%, 23-32%, 63-64%, 9-20%, and 25-31%, respectively.– Decrease (27%-30%) in tropospheric NO2 over areas not affected by seasonal biomass burning.– Major reductions at industrial, suburban, and rural sites.	Kanniah et al. (2020)
Decline in PM2.5 concentrations over major cities around the world associated with COVID-19	Dubai, Delhi, Mumbai, New York, Los Angeles, Zaragoza, Rome, Beijing, and Shanghai	PM2.5 data available through:– The US Embassies in Delhi and Mumbai, India, Beijing;– The AirNow platform for Shanghai, China, and Dubai;– PurpleAir sensors for Rome, Italy, and Zaragoza, Spain.– Rainfall data available through Weather Underground.	– COVID-19 outbreak period (Dec 2019-Mar 2020) compared with 2017−2019 for changes in PM2.5 concentration.	– Variations of PM2.5 depend upon anthropogenic activities, dust events, crop residue burning, and emissions from traffic.– Decline in PM2.5 concentration due to lockdown and to less movement of people to keep social distancing to control the spread of corona virus.	(Chauhan and Singh, 2020)
Effect of restricted emissions during COVID-19 on air quality in India	India (22 cities in different regions)	– Ground-based air quality and meteorological data from a network of air quality monitoring stations across 22 different cities in India for the past four years (2017–2020) from March 16th to April 14th.	– Before/After comparison of the concentrations of PM2.5, PM10, CO, NO2, and AQI for the period of March 16th to April 14th, from 2017 to 2020.– Assessment of the effect of meteorology on the PM2.5 using the WRF-AERMOD model system.	– Compared to previous years (2017−2019), during lockdown periods, PM2.5, PM10, CO, and NO2 concentrations reduced by 43, 31, 10, and 18%, respectively.– O3 shows an increase (17%) and negligible changes observed in SO2.– The mean excessive risks of PM reduced by ~52% nationwide due to restricted activities in the lockdown period.	Sharma et al. (2020a)
Impact of Covid-19 partial lockdown on PM2.5, SO2, NO2, O3, and traceelements in PM2.5 in Hanoi, Vietnam	Vietnam (Hanoi)	– The daily PM2.5, NO2, O3, and SO2 levels were collected from the automatic ambient air quality monitoring station.Sample filter for analysis of he– avy metals in PM2.5 including Al, As, Ba, Ca, Cd, Co, Cr, Cu, Fe, K, Mg, Mn, Mo, Na, Ni, Pb, Ti, Zn, and V.– The hourly meteorological data (precipitation, temperature, boundary layer height, dew point temperature, and surface solar radiation) were obtained from the European Centre for Medium-Range Weather Forecasts (ECMWF).– The relative humidity was computed using the dew-point temperature and temperature data.	– Investigated changes in PM2.5, SO2, O3, and NO2 levels during the lockdown.– Studied the relationship between PM2.5 pollution and the boundary layer height (BLH).– Identified the elemental emission sources using principal component analysis (PCA).	– During the partial lockdown, NO2, PM2.5, O3, and SO2 decreased by 75.8, 55.9, 21.4, and 60.7%, respectively, compared with historical data (2017-2019).– Minor impact of meteorological condition variation before and during the partial lockdown on the changes of pollutants level.– Lower concentrations of Cd, As, Ba, Cu, Mn, Pb, K, Zn, Ca, Al, and Mg were observed during the partial lockdown.	Nguyen et al. (2021)
Severe air pollution events not avoided by reduced anthropogenic activitiesduring COVID-19 outbreak	China (Beijing, Shanghai, Guangzhou, and Wuhan)	– PM2.5 concentrations were estimated from the CMAQ modelling system.– The meteorological inputs were generated using WRF-v3.7.1.	– The WRF-CMAQ modelling system and monitoring data were applied to investigate the impact of lockdown on air quality and emission changes.– Investigated the effect of reduced anthropogenic activities on air pollution following three simulation scenarios.– Studied the changes due to emission changes and meteorology.	– The decreases of PM2.5 in Beijing, Shanghai, Guangzhou, and Wuhan were 9.23, 6.37, 5.35, and30.79 μg/m3, respectively.– The reduction ratios of PM2.5 concentrations were smaller than the reduction ratios of precursor emissions, partially due to the unfavorable meteorological conditions.	Wang, Chen et al. (2020)
The concentration of major air pollutants during the movement control order due to the COVID-19 pandemic in the Klang Valley, Malaysia	– Klang Valley, Malaysia (6 monitoring stations)	– Hourly data of air pollutant (PM10, PM2.5, CO, NO2, SO2, and O3) concentrations and meteorological (temperature, wind speed, wind direction, relative humidity, and solar radiation) were obtained from the Malaysian Department of Environment.– Mobility data provided by using COVID-19 community mobility reports.	– COVID-19 outbreak period (2020) compared with 2018−2019 for percentage reductions in all air pollutant concentrations.– Used bivariate polar plots to show the variation of air pollutant concentrations with wind speed and wind direction for polar coordinates.– Assessed the inter-period back-trajectory variabilities using the HYSPLIT model.	– NO2, CO, PM10, and PM2.5 decrease ranged from 55% to 72%, 13-53%, 10-46%, and 7-45%, respectively, compared with historical data (2018-2019).– O3 concentration increased in the busy areas due to the reduction of NOx levels.	Latif et al. (2021)

All the above-mentioned studies have identified and measured the impact on air quality variations during the lockdown periods in specific cities and regions. However, none of the studies has tried to cover the global ambient air pollution (AAP) issues with available PM2.5 ground-level data and compared the result with the Air Quality Index (AQI). In the current study, the assessment of changes and variations in the AQI, PM2.5, and tropospheric NO2 levels before, during, and after lockdown phases, among 21 cities around the world with different cultures, living levels, industrial activities, societies, and population density shows the role of anthropogenic emission as one of the considerable challenges to achieve the sustainable air quality for all cities. As discussed above the lockdown has drastically reduced air pollution levels whereas the high economic cost of doing so makes it a non-sustainable option for addressing the pollution concern. An earlier study by He et al. (2020) highlighted that in China the environmental regulations implemented to found similar levels of air quality improvement caused by the lockdown can be achieved at a much lower cost.

This attempt aimed at evaluating changes and variations in AQI and PM2.5 and NO2 levels as ambient air pollutants during the lockdown. A comparison was made between the dataset during the lockdown period and data recorded during the same periods in 2019 and 2018. This study appears to be the first to explore air quality during three covid-19 lockdown phases in cities with more than 1 million inhabitants, and to provide necessary responses to the following questions: (i) What are the changes in AQI and PM2.5 and NO2 levels before, during, and after the lockdown periods? (ii) How are the critical pollutants characterizing during the unlock phases in different cities around the world? (iii) Does “before/after comparison” approach enough to investigate temporal variations and trends of air pollutant levels caused by the lockdown restrictions? and (iv) Does the activities of the Holy month of Ramadan in the Islamic cities have effects on PM2.5, NO2, and AQI levels through the lockdown?

Methods and Materials

The essence of our empirical analysis uses comprehensive air pollution data from 42 cities around the world. In total, 3580 atmospheric pollution observations were collected, distributed over 172 days on average for each studied city. There were some basic criteria for the inclusion of the cities for the present study. The primary concern for the inclusion was the availability and consistency of the PM2.5, AQI, and NO2 data in the previous two years (2018-2019) as pre-lockdown; during the lockdown periods (as reported by various media); and in the unlock periods. Finally, twenty-one (n=21) cities were short-listed based on the population concentration, pollutants data availability, and significant lowering of AAP levels during the lockdown periods. Fig. 1 presents the topography of the studied cities and their total population (in thousands) according to the 2018 Revision of World Urbanization (United Nations 2018). The corresponding table summarises each city's annual population and Lockdown history (Table A1). All of these cities (except for Manama city, Bahrain) are urban agglomerations with more than 1 million inhabitants. Cities are classified based on geographical settings, city lockdown history, and ambient pollutants concentration levels. Six of the selected cities are from Africa; four from the Middle East; two from America; and ten from Asia. Some of the cities were affected by other emissions sources such as anthropogenic nature (forest fires), those cities were excluded from the present study.

Fig. 1

Distribution of cities included in the present study and its total population (thousands) in 2018. Annual total population data used in this figure are available in Table A1. The total population of each country is provided from the 2018 Revision of World Urbanization.

Table A1

Annual Population of Urban Agglomerations in 2018 (United Nations 2018) and Lockdown history for the selected cities. Lockdown history information and related policies for the selected cities were collected from global media reports and government announcements.

City, Country	Annual Population in 2018 (thousands)	Lockdown history		References
		Lockdown start	Lockdown end
Abidjan, Côte d'Ivoire	4 921	2020-03-23[1]	2020-05-15[2]	1. "Ivory Coast, Senegal declare emergencies, impose curfews in coronavirus response". Reuters. 23 March 2020. Retrieved 03 June 2020.2. "Côte d'Ivoire: Authorities announce easing of some COVID-19 restrictions in Abidjan May 14".GardaWorld. Retrieved 10 June 2020.
Algiers, Algeria	2 694	2020-03-22[3]	2020-06-13 [4]	3. "Algeria: Government implements lockdown and curfew in Blida and Algiers March 23". GardaWorld. Retrieved 03 June 2020.4. "Covid-19: Partial lockdown maintained until 13 June, total lifting in 4 provinces". www.aps.dz. Algeria Press Service. Retrieved 03 June 2020.
Dhaka, Bangladesh	19 578	2020-03-26[5]	2020-06-14[6]	5. "Coronavirus: Bangladesh declares public holiday from March 26 to April 4". Dhaka Tribune. 23 March 2020. Retrieved 11 June 2020.6. "Bangladesh: Nationwide curfew in place until June 15; Cox's Bazar classified as a red zone"07 Jun 2020. GardaWorld. Retrieved 10 June 2020.
Beijing, China	19 618	2020-02-10[7]	2020-03-24[8]	7. "Beijing locked down as Wuhan virus continues to spread in China".Taiwan News. Retrieved 03 June 2020.8. Guojun He, Yuhang Pan, Takanao Tanaka. COVID-19, City Lockdowns, and Air Pollution: Evidence from China. medRxiv 2020.03.29.20046649; doi: 10.1101/2020.03.29.20046649
Chengdu, China	8 813	2020-02-15[7]	2020-03-24[8]
Bogotá, Colombia	10 574	2020-03-25[9]	2020-06-30[10]	9. "Colombia announces lockdown as coronavirus cases surge". aa.com.tr. Retrieved 7 June 2020.10. "Gobierno amplía aislamiento obligatorio hasta el 1 de julio". El Espectador. 28 May 2020. Retrieved 7 June 2020.
Addis Ababa, Ethiopia	4 400	2020-04-08[11]	2020-06-05[12]	11. "Ethiopia Declares State Of Emergency Over COVID19".Fanabc.com. Retrieved 03 June 2020.12. "En Ethiopie, le Parlement lève l’état d'urgence". Lemonde, in french. 05 June 2020. Retrieved 11 June 2020.
NewDelhi, India	28 514	2020-03-25[13]	2020-06-30[14]	13. "Coronavirus in India LIVE Updates: PM Modi announces 21-day national lockdown, says extremely necessary to take this step". India Today. 24 March 2020. Retrieved 11 June 2020.14. "Lockdown 5.0 Guidelines in India (state-wise): New Lockdown Extension rules announced; night curfew relaxed". The Financial Express. 29 May 2020. Retrieved 11 June 2020.
Kolkata, India	14 681	2020-03-25[13]	2020-06-30[14]
Jakarta, Indonesia	10 517	2020-03-30[15]	2020-06-05[16]	15. "Indonesia: Bali declares state of emergency due to COVID-19 March 30". Garda World. Retrieved 11 June 2020.16. "Indonesia: Jakarta to further ease COVID-19 restrictions from June 5". Garda World. Retrieved 11 June 2020.
Baghdad, Iraq	6 812	2020-03-17[17]	2020-06-14[18]	17. "Iraq: Authorities implement further measures due to COVID-19 from March 17". Garda World. Retrieved 11 June 2020.18. "Iraq: Government extends nationwide curfew until June 13". Garda World. Retrieved 11 June 2020.
Abu Dhabi, United Arab Emirates	1 420	2020-03-26[19]	2020-04-17[20]	19. "UAE: Three-day lockdown scheduled March 26-29". Garda World. Retrieved 15 June 2020.20. "Dubai imposes two-week lockdown as Gulf states step up coronavirus fight". Reuters. 5 April 2020. Retrieved 15 June 2020.
Kuwait City, Kuwait	2 989	2020-03-14[21]	2020-05-31[22]	21. "Kuwait: Government implements nationwide curfew March 22". Garda World. Retrieved 15 June 2020.22. "Coronavirus: Kuwait announces plan to end full COVID-19 lockdown". Gulfnews.com. Retrieved 15 June 2020.
Conakry, Guinea	1 843	2020-04-27[23]	2020-06-14[24]	23. "Guinea: Over 1000 COVID-19 cases confirmed nationwide as of April 27". Garda World. Retrieved 11 June 2020.24. "Guinea: Nationwide state of health emergency extended until June 14 due to COVID-19". Garda World. Retrieved 11 June 2020.
Ulaanbaatar, Mongolia	1 520	2020-03-10[25]	2020-06-30[26]	25. "Mongolia: Government places Ulaanbaatar and other cities on lockdown due to COVID-19". Garda World. Retrieved 15 June 2020.26. "Mongolia: Mongolian government extends COVID-19 emergency measures through June 30". Garda World. Retrieved 15 June 2020.
Kathmandu, Nepal	1 330	2020-03-24[27]	2020-06-14[28]	27. "Nepal locks down for a week to stem coronavirus spread". The Jakarta Post. 24 March 2020. Retrieved 15 June 2020.28. "COVID-19 Information (Updated June 15, 2020)". np.usembassy.gov. Retrieved 15 June 2020.
Islamabad, Pakistan	1 061	2020-03-24[29]	2020-05-09[30]	29. "Coronavirus pandemic: Pakistan to extend lockdown for 2 more weeks as death toll reaches 31". The Statesman. 2 April 2020. Retrieved 15 June 2020.30. "Pakistan: Nationwide lockdown extended through May 9". Garda World. Retrieved 15 June 2020.
Lima, Peru	10 391	2020-03-16[31]	2020-06-30[32]	31. "Coronavirus en Perú: Gobierno anuncia cuarentena obligatoria por 15 días por coronavirus". Gestión (in Spanish). 15 March 2020. Retrieved 15 June 2020.32. "Peru extends nationwide lockdown until end of June". Aljazeera. Retrieved 15 June 2020.
Kampala, Uganda	2 986	2020-04-01[33]	2020-06-05[34]	33. "Uganda: Authorities announce 14-day nationwide lockdown April 1". Garda World. Retrieved 15 June 2020.34. "Uganda: Authorities extend nationwide curfew until June 23". Garda World. Retrieved 15 June 2020.
Khartoum, Sudan	5 534	2020-04-18[35]	2020-06-02[36]	35. "Health Alert: Sudan, Government Announces Lockdown In Khartoum Effective April 18".www.osac.gov. Retrieved 15 June 2020.36. "Sudan: Lockdown in Khartoum state extended through June 2". Garda World. Retrieved 15 June 2020.
Manama, Bahrain	565	2020-03-18[37]	2020-05-05[38]	37. "Bahrain: Entry and domestic restrictions implemented amid COVID-19 outbreak March 17". Garda World. Retrieved 15 June 2020.38. "Bahrain extends closure measures for two weeks starting April 23". Arabnews. Retrieved 15 June 2020.

Hourly Air Quality Index (AQI) and ambient fine Particulate Matters (PM2.5) concentrations have been downloaded from Air Quality Monitoring stations of embassies of the United States of America located in the selected cities via the AirNow website (https://www.airnow.gov). San Martini et al. (2015) demonstrates the potential utility of continuous PM2.5 data which are currently systematically being collected and made publicly accessible by U.S. diplomatic facilities, while several studies (Jiang et al., 2015; Mukherjee and Toohey, 2016) indicates these data have shown to be of good quality and in good agreement with other observations. The comprehensive evaluation of AQI is based on the pollution index of SO2, NO2, CO, PM2.5, PM10, and O3 (US. EPA 2018). The concentrations of these are converted to a number on a scale of 0–500 using standard formulas developed by the U.S. Environmental Protection Agency (EPA). Yousefi et al., (2019) summarized the EPA's method applied to calculate AQI values as follows: 1) Calculate the moving average concentration of each pollutant based on the averaging duration specified by EPA (8-hours for O3, 8-hours for CO, 1-hour for SO2, 1-hour for NO2, 24-hours for PM2.5 and 24-hours for PM10); 2) Identify the highest concentration of each pollutant among all of the monitors within the reporting area; 3) Find the two breakpoints that contains the observed concentration (); 4) Calculate the index for pollutant p using this equation: , where denote, respectively, to the greater and lower concentration breakpoints of the air pollutant, and and are the AQI values corresponding to those breakpoints concentrations (); 5) Round the index to the nearest integer. Then, the highest AQI value of all air pollutants is reported as the AQI value for the area on that day. Hashim et al. (2021) reported that the AQI is usually divided into six categories: good, satisfactory, moderate, poor, very poor, and severe depending on whether the AQI falls between 0–50, 51–100, 101–200, 201–300, 301–400 and 401–500, respectively.

Datasets collected from January to June 2020 (the latest available data at the time of writing this original manuscript) are analyzed in this study. A total of 21 cities with >75% of validated hourly datasets in a year were selected to calculate a valid aggregated value (24-h average) and subsequent calculations. PM2.5 and AQI data averaged over 2018–2019 (baseline) are also used to detect the Absolute Differences and Relative Changes (in %) between 2020 and the 2-years baseline data. Several choices have been carried out for comparing the 2020 dataset with baseline conditions based on multi-year averaged data. Dantas et al. (2020), Tobías et al. (2020), and Navinya et al. (2020) suggested the comparison with the same values obtained in 2019; Li et al. (2020), and Sathe et al. (2021) with averaged values over the 3 previous years 2017-2019; and Kanniah et al. (2020) with 5-years period 2015-2019. Ambient pollutants concentrations are strongly influenced by meteorological conditions and emission sources, exhibiting thus pronounced seasonal patterns. So, to minimize the meteorological parameters and seasonal variability effects and to explore only the impact of the lockdown, the NO2 and PM2.5 concentrations and AQI values were compared during the same period of 2018, 2019, and 2020 (Gama et al., 2018, Kerimray et al., 2020). These comparisons are based on the assumption that multi-year baseline conditions were long enough to minimize the influence of seasonal variability in results through reducing inter-annual variability in meteorology and atmospheric pollutants concentrations (Gama et al., 2021)

The tropospheric NO2 vertical column density time-series are calculated using the Version 4.0 Aura Ozone Monitoring Instrument (OMI). The Nitrogen Dioxide (NO2) Standard Product (OMNO2) dataset was collected from the available open sources at https://so2.gsfc.nasa.gov/no2/no2_index.html. Nickolay et al., 2019; Krotkov et al., 2017 reported that many studies have commonly used the satellite NO2 data to infer NOx emissions, ground-level NO2, and emissions of co-emitted gases, including particulate matter and greenhouse gases. Similarly, we also used NO2 data averaged over 2018–2019 (baseline) for distinguishing the absolute differences and relative changes between 2020 and the baseline.

Information on lockdown dates and related policies for the selected cities were collected from global media reports and government announcements (Table A1). To face challenges of discrepancies due to any errors in media reporting, retrieving data from a variety of public sources, and differences in time zones of the lockdown starting/ending date, a cross-check analysis has been carried out. The verification process focuses on the accuracy of information by checking and comparing different retrieving sources.

AQI, NO2, and PM2.5 level variations were examined and compared for three periods covering before, during, and after the lockdown periods of the years 2018, 2019, and 2020 for each city. We presume that AAP levels in 2018 and 2019, when social and industrial activities were operating normally, were as usual conditions and any changes in AAP levels in 2020 is mainly the resultant impact of country-specific lockdown. Therefore, a comparison of the PM2.5 and NO2 daily concentrations from January to June of 2018-2019 with the same period of 2020 was carried out. Furthermore, many statistical analyses have been also performed including i) descriptive statistics which summarize the data by simple summary statistics; and ii) time series analysis.

The statistical methods used to analyze our comprehensive data from selected cities have some notable advantages for assessing the relationship between city lockdowns and air pollutants. First, although a simple before/after comparison approach could capture the declining trend in air pollution levels caused by the lockdown restrictions, it can be strongly influenced by meteorological conditions and seasonal changes. This method helps to address this issue by comparing a post-lockdown period with an average of several previous years of data. Second, the primary criterion for the inclusion of the city in the present study was that it has more than 1 million inhabitants. It helps us to examine whether the effects of lockdowns vary across similar types of cities, which runs a common thread through all these cities. For instance, we expected a substantial reduction in air pollution when the lockdown is implemented due to the suspension of industrial activities and reduction of traffic volumes. Lastly, this method uses not only data from monitoring stations but also satellite observations data which has facilitated studies of very large populations, not restricted to urban areas where most of the monitors are. The main steps of the comparison approach for assessing the relationship between city lockdowns and air pollutants and described in more detail in the following subsections. The algorithm and the flowchart indicating the actual processing steps followed in this study are depicted in Fig. 2 a and Fig. 2b, respectively.

Fig. 2

a: Processing algorithm for comparing air quality data collected during the pandemic (2020) with baseline conditions based on 2-year averaged data (2018–2019). The proposed algorithm has three main operational steps and considers parameters such as PM2.5, NO2, and AQI. b: Flowchart of research work activities indicating the actual processing steps followed in this study.

Hypotheses and Limitations

Compiling ambient air quality databases from various cities (countries) presents challenging issues, including the presence (or lack) of monitoring stations, data availability, data coverage, background pollution, meteorological conditions, seasonality of the pollution, and QA/QC (Schwela and Haq, 2020). Therefore, several limitations exist in this study. First, the air quality databases accessible by U.S. diplomatic facilities from different cities have limited comparability due to different locations and percentages of data coverage of the year. Second, air pollution data sets, in some cities, are not available during particular periods. Another sustainable limitation is that the comparison of meteorological parameters observed during the lockdown period against the baseline period was not performed. Finally, the previous two years of data (2018–2019) for the same period is used to define baseline conditions assuming it to represent atmospheric loading of pollutants in the year 2020 in absence of lockdown.

Results

By collecting and processing the daily time series data of AQI values, PM2.5, and NO2 concentrations of 21 cities over the world from Jan. to Jun. 2020, we are dedicated to characterize the pattern of air quality of different cities before (from 1st January 2020 until the start date of the lockdown), during (from the start date of the lockdown until the end date of the lockdown), and after the lockdown periods. The COVID-19 lockdown impact on air quality will be evaluated per type of parameter, namely AQI, NO2, and PM2.5, as already mentioned. Table A2 presents the descriptive statistics of daily PM2.5, NO2, and AQI of the selected cities during the January-June 2020 period for similar periods of the past two years.

Table A2

Overview of summary statistics of daily PM2.5 (μg/m3), NO2 (1E15 molecules/cm2), and AQI variation for the selected cities during the January-June period for each year (2018, 2019, and 2020). Summary of the mean (± standard deviation), number of observations (n), maximum (max) and minimum (min) concentrations of PM2.5 and NO2 and AQI.

City, Country	Parameter	2020			2019			2018
		Mean ± SD	Min-Max	n	Mean ± SD	Min-Max	n	Mean ± SD	Min-Max	n
Algiers, Algeria	AQI	67.25±12.96	47.7- 127.7	143	70.9±13.02	13.17-101.66	102	n.a.	n.a.	n.a.
	PM2.5	19.85±5.99	11.67- 46.82	142	21.74±5.59	2.75- 36.7	102	n.a.	n.a.	n.a.
	NO2	2.13±0.5	1.29-3.69	192	2.33±0.5	1.13-3.71	192	2.12±0.6	0.876- 3.839	192
Addis Ababa, Ethiopia	AQI	76.51±19.74	40.33- 151.33	170	65.29±19.76	32.33- 121.87	171	104.45±26.66	51.07- 183.42	98
	PM2.5	24.54±9.14	9.45- 71.46	171	19.70±8.82	7.75- 51.04	171	39.78±16.39	12.85- 122.5	98
	NO2	0.61±0.12	0.44-0.88	182	0.62±0.92	0.37-0.84	192	0.67±0.11	0.44-0.85	193
Khartoum, Sudan	AQI	133.46±39.9	54.33-321.94	174	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.
	PM2.5	63.32±34.83	14.75-253.55	174	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.
	NO2	0.94±0.13	0.62-1.2	192	1.08±0.24	0.7,1.6	192	1.02±0.18	0.54,1.34	192
Kampala, Uganda	AQI	127.87±29.14	78.5- 200.29	170	131.58±21.42	87.83- 178.83	178	129.20±28.68	38.72- 225.33	172
	PM2.5	54.81±25.43	25.33- 135.25	169	55.99±16.71	29.73- 108	177	56.28±24.49	9.18- 168.16	171
	NO2	0.59±0.11	033-0.98	192	0.62±0.1	.43-0.92	192	0.62±0.14	0.24-0.95	192
Conakry, Guinea	AQI	94.78±44.7	23.18-176.5	175	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.
	PM2.5	37.11±27.48	5.58-103.45	175	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.
	NO2	0.65±0.23	0.39-1.2	192	0.63±0.21	0.29-1.1	192	0.66±0.28	0.27-1.27	192
Abidjan, Côte d'Ivoire	AQI	79.96±32.9	18.6-188.58	146	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.
	PM2.5	28.17±20.61	7.16-125.79	147	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.
	NO2	0.52±0.11	0.26-0.94	192	0.46±0.11	0.2-0.72	192	0.51±0.11	0.24-0.88	192
Abu Dhabi, United Arab Emirates	AQI	71.63±30.52	6.83- 151.04	178	118.83±28.73	68.62-187.37	181	116.32±33.03	48.29-200.25	180
	PM2.5	23.54±13.38	2.81-62.95	178	46.50±18.59	20.91-118.79	181	46.34±22.53	12.58-149.66	180
	NO2	3.01±0.82	1.89-4.73	182	2.76±0.58	3.66-7.56	192	3.22±0.84	1.76-4.99	192
Kuwait City, Kuwait	AQI	104.72±32.37	11.95-205.27	159	106.88±35.05	22.5-262.87	180	120.76±44.91	14.09-257.72	181
	PM2.5	40.01±19.33	3.14-107.45	159	41.48±25.38	5.5-211.62	180	53.63±35.66	3.5-227	181
	NO2	5.34±1.11	3.74-7.84	192	5.28±0.8	3.66-7.56	192	5.10±0.94	3.45-7.52	192
Baghdad, Iraq	AQI	94.64±34.82	30.54-228.2	125	86.58±27.03	37.78-153.47	131	n.a.	n.a.	n.a.
	PM2.5	48.72±31.16	7.87-164.87	125	30.56±14.94	9.86-83.05	132	n.a.	n.a.	n.a.
	NO2	3.87±1.36	2.1-6.73	192	3.64±0.76	2.42-5.39	192	3.87±0.80	2.90-5.83	192
Manama, Bahrain	AQI	112.16±27.36	60.69-177.7	172	119.5±32.32	63.08-292.28	177	n.a.	n.a.	n.a.
	PM2.5	42.07±15.43	16.92-104.45	172	48.30±28.43	17.12-242.71	177	n.a.	n.a.	n.a.
	NO2	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.
Dhaka, Bangladesh	AQI	163.18±62.32	52.08-320.91	171	172.43±64.72	38.25-336.14	180	164.77±65.29	59.82-365.33	181
	PM2.5	98.42±66.32	15-273.83	171	105.14±68.41	9.58-297.14	180	99.19±69.92	16.59-327.16	181
	NO2	4.58±2.37	1.5-9.9	192	4.68±1.64	2.49-7.82	192	5.66±2.10	2.04-10.21	192
Chengdu, China	AQI	117.82±31.77	51.25-176.79	178	126.43±33.46	48-187	173	131.02±36.26	58.78-250.65	171
	PM2.5	57.05±27.17	13-143.86	178	52.75±23.45	11.83-123.5	173	58.32±30.83	15.21-199.13	171
	NO2	5.87±2	2.65-12.54	192	4.65±2.47	0.45-12.36	192	5.7±2.8	0.78-16	192
Beijing, China	AQI	103.36±52.62	15.36-264.5	181	108.39±49.61	15.21-247.58	181	n.a.	n.a.	n.a.
	PM2.5	45.45±39.4	3.4- 211.3	181	49.02±37.49	3.86-204.36	181	n.a.	n.a.	n.a.
	NO2	10.53±4.08	5.5-28.67	192	12.6±4.3	7.2-26.8	192	13.31±3.35	6.07-21.35	192
New Delhi, India	AQI	146.52±51.15	62.29-356.08	178	137.73±56.10	71.58-377.16	145	177.11±58.89	71.12-420.58	178
	PM2.5	77.38±53.97	18.66-317.65	178	102.68±64.35	21.75-345.37	145	106.38±67.66	21.12-378.45	178
	NO2	4.35±1.6	1.7-9.2	192	5.11±1.29	3.13 -8.69	192	5.37±1.07	3.7-9.44	192
Kolkata, India	AQI	124.13±70.98	19.31-335.28	178	137.69±67.44	29.56-343.29	181	161±70.59	56.37-387.95	174
	PM2.5	68.13±62.74	5.05-289.18	178	74.43±65.53	6.33-292.87	181	93.11±76.08	15.5-347.7	174
	NO2	3.18±1.1	1.3-5.8	192	3.35±00.93	1.57-5.54	192	3.79±1.06	1.9-5.7	192
Jakarta, Indonesia	AQI	112.15±21.19	68.56-172	178	110.1±35.78	15.4-171.12	164	90.2±32.68	21.58-155.62	181
	PM2.5	46.90±14.09	23.37-97.5	176	42.86±20.01	5.8-94.37	164	32.5±16.43	5.4-75.58	181
	NO2	1.27±0.37	0.7- 2.5	192	2.3±0.45	0.1-3.3	192	2.1±0.64	0.7-4.3	192
Ulaanbaatar, Mongolia	AQI	111.05±76.96	23.5-321.65	169	125.73±86.69	16.73-430.9	178	n.a.	n.a.	n.a.
	PM2.5	60.96±68.48	5.5-284.04	169	75.07±81.97	3.9-411.09	178	n.a.	n.a.	n.a.
	NO2	0.98±0.24	0.48-1.51	192	1.09±0.28	0.61-1.6	192	0.91±0.21	0.5-1.45	192
Kathmandu, Nepal	AQI	103.69±50.35	10.88-212.95	178	130.92±34.1	41.54-209.79	175	144.76±32.24	44.29-206.12	174
	PM2.5	44.91±33.31	3-150.2	178	58.08±26.36	10.16-137.33	175	72.65±32.8	11.41-145	174
	NO2	1.72±0.49	0.7-3.1	192	2.12±0.5	0.67-3.1	192	2.2±0.5	0.6-3.5	192
Islamabad, Pakistan	AQI	108.24±33.32	58.71-230.54	178	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.
	PM2.5	43.42±23.98	15.79-163.37	178	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.
	NO2	2.74±0.71	1.31-4.52	182	3.01±0.80	1.25-5.02	192	3.24±0.75	1.87-4.97	192
Bogotá, Colombia	AQI	47.84±24.17	1.33-128.5	178	52.46±22.69	5.8-120.45	180	53.44±20.19	12.8-106.33	177
	PM2.5	13.71±8.75	0.66-47.79	178	15.38±8.05	1.8-44.62	180	15.59±7.19	3.28-37.87	177
	NO2	0.94±0.46	0.2-2.5	152	0.82±0.4	0.13-1.9	173	1.09±0.4	0.34-2.5	191
Lima, Peru	AQI	62.61±12.49	37.12- 107.04	179	76.65±23.02	38.3-167.62	174	94.35±23.526	7.8-175.8	134
	PM2.5	17.91±5.56	9.25-36.41	179	24.81±12.07	9.63-87.87	174	33.33±12.67	2.44-105.04	135
	NO2	1.16±0.43	0.47-2.5	192	1.89±0.77	0.8-3.3	192	1.92±0.62	0.9-3.02	192

Air quality vs. population density

For a first visual impression about pollutant patterns and how pollution is recorded, Fig. 3 shows the correlation matrix for AQI and mean concentration levels of NO2 and PM2.5 in 2020 against the population density in each city. The figure reveals the positive correlation of population density with AQI (correlation coefficient, r = 0.50, p < 0.05), PM2.5 (r = 0.59, p < 0.01), and NO2 (r = 0.55, p < 0.05). Borck and Schrauth (2021) argued the effect of population density on urban air pollution in over 400 German districts. They found that increasing population density by one percent increases NO2 by 0.25 percent, PM10 by 0.07 percent, and AQI with an elasticity of 0.11-0.13. Moreover, (Rodríguez-Urrego and Rodríguez-Urrego, 2020) concluded that the most densified population cities reflect high PM2.5 levels.

Fig. 3

Correlation matrix plot between selected variables (PM2.5, NO2, and AQI) and population density. The correlations are visually expressed as Pearson's correlation coefficient, where the p-value is <0.05 for NO2 and AQI, and <0.01 for PM2.5. PM2.5, NO2, and AQI denote the daily average fine aerosols (<2.5 µm in diameter), tropospheric NO2, and Air Quality Index, respectively.

Frequency distributions for PM2.5 and NO2

Frequency distributions for the daily PM2.5 and NO2 concentrations and the peaks during the lockdown, as compared with the baseline (2018-2019) in each city, are shown in Fig. 4, Fig. 5 , respectively. The density plots typically covered varying days from 2020-02-10 to 2020-06-30 considered as a basic lockdown period for all cities. Probability density functions (PDF) were fit to the daily data to determine the shape of the PM2.5 and NO2 concentration distributions (Kumar et al., 2020). Although the extent of right-skewness was apparent in the frequency distribution plots for both PM2.5 and NO2, the density plots provide a more comprehensive visual summary of the estimated shape of the data distributions.

Fig. 4

Density plot of daily PM2.5 concentrations (µg/m3) per city during lockdown period in 2020 (indicates in dashed line) compared to the baseline 2018-2019 (indicates in solid line). Cities are classified based on geographical settings: Africa, Middle East, Asia, and South America. The plots provide a more comprehensive visual summary of the estimated shape of the data distributions.

Fig. 5

Density plot of daily NO2 concentrations (1E15 molecules/cm2) per city during lockdown period in 2020 (indicates in dashed line) compared to the baseline 2018-2019 (indicates in solid line). Cities are classified based on geographical settings: Africa, Middle East, Asia, and South America. The plots provide a more comprehensive visual summary of the estimated shape of the data distributions.

The PDF of PM2.5 concentrations was more consistently distributed around the lower values for all cities during the lockdown period (2020), indicating the PM2.5 decline due to lockdown restrictions. It also exhibits that high PM2.5 concentrations (extreme values) were less frequent during the lockdown, typically in Kuwait and New Delhi. Similar to PM2.5, the density plot for NO2 was relatively displaced to lesser concentrations. Besides, the frequency distribution for NO2 is more variable than that for PM2.5 and the distribution is flatter from 2020 to the baseline 2018-2019 and more skewed to the right.

Impact on Air Quality Index (AQI)

Fig. 6 depicts the pattern of variation in AQI during pre-lockdown and post-lockdown days for each studied city. The time series consists of an average of 172-day observation periods and runs from 1st January to 30th June 2020. For each city, the lockdown period caused by the COVID-19 pandemic is indicated by two lines (the first one reflects the start time, and the second line the end of lockdown). As a general trend, AQI values varied with substantial differences among cities and lockdown phases. In general, in almost all the cities, there is a significant improvement in the air quality in comparison to that of the pre-lockdown phase just after few days into the implementation of the lockdown measures. In 88% of the cities with the validated dataset, a decline from 2% to 67% is observed. As shown in Table 2 , the average of AQI at all cities during lockdown is much lower compared to the same period in 2018–2019, while Jakarta city showed an increase (+14%). The reduction of AQI due to lockdown can be recognized in Fig. 7 dealing with the Relative Changes (in %), between 2020 and the 2-years baseline data, for each city. The major negative Changes as observed through AQI data can be seen in Abu Dhabi where the average values compared to the baseline dropped by 42% followed by Islamabad (-38%), Kathmandu (-36%), Lima (-35%), Kolkata (-34%) and New Delhi (-22%). The lower Changes were perceived in Manama (-2%), Baghdad and Dhaka (-5%), Kampala (-6%), Algiers and Bogotá (-7%), and Kuwait (-13%). Additionally, the pre-lockdown period was characterized by lower average values in 2020 (in more than 70% of the cities) compared to the same period in the previous years of 2018-2019. This record needs further investigation for a better understanding of whether the effect is from local and national policies. Actually, nearly all cities are showing mixed trends during the three lockdown phases as some Relative Changes showing positive and negative values. For instance, the time series of daily AQI showed clearly somehow the impact of unlocking phase where AQI values in post-lockdown are higher than in during lockdown period in only 6 cities such as Baghdad, Dhaka, Chengdu, Beijing, Jakarta, and Kathmandu. Otherwise, according to the international AQI standard (Air Quality Index (AQI) basics 2020), when AQI is smaller than or equal to 50, air quality is considered satisfactory and air pollution poses little or no risk; when AQI is larger than 50, air quality becomes worse and the potential of air pollution to affect public health becomes higher (Ma et al., 2020). During the lockdown phase, the average value of AQI ranged between 44.17 in Bogotá and 122.72 in Chengdu. Therefore, the overall situation of air quality in our typical cities is varied from excellent to mild pollution. In contrast, before lockdown, the AQI fluctuated between 52.35 and 213.16, indicated that the air quality was moderately polluted and very unhealthy, respectively.

Fig. 6

Time series of daily AQI, and 15-days moving average concentrations of NO2 (1E15 molecules/cm2) and PM2.5 (µg/m3) from January 1 to June 30, 2020, in 21 cities across the globe. PM2.5 concentrations and AQI are given on the right y-axis and NO2 concentrations on the left y-axis. In green color are represented NO2 concentrations, in blue the PM2.5 concentrations, and in red the AQI values. The green hatched area stands for the lockdown phase and the yellow for the Ramadan period (only for Islamic cities).

Table 2

Average (Ave.), Absolute differences (Diff.), and Relative Changes (Change in %) of AQI during three lockdown periods at typical cities over the world from January to June 2020. Summary data are also subdivided into three phases including Pre, during, and Post lockdown regarding each city lockdown dates. n.a.: Data Not Available.

City, Country	Pre lockdown			During lockdown			Post lockdown
	Ave.	Diff.	Change	Ave.	Diff.	Change	Ave.	Diff.	Change
Algiers, Algeria	76.33	1.89	3%	62.36	-4.80	-7%	66.52	-11.21	-14%
Addis Ababa, Ethiopia	72.34	-0.14	0%	70.96	-13.06	-16%	106.49	-4.10	-4%
Kampala, Uganda	142.46	4.88	4%	103.12	-6.47	-6%	139.12	1.59	1%
Abu Dhabi, United Arab Emirates	61.98	-44.29	-42%	66.89	-48.57	-42%	80.61	-46.27	-36%
Kuwait City, Kuwait	114.98	8.06	8%	96.01	-13.90	-13%	106.16	-21.31	-17%
Baghdad, Iraq	117.87	26.25	29%	79.47	-4.03	-5%	78.07	-18.01	-19%
Manama, Bahrain	111.94	-4.49	-4%	111.76	-1.81	-2%	112.92	-15.54	-12%
Dhaka, Bangladesh	213.16	-7.41	-3%	121.32	-6.48	-5%	86.60	-2.03	-2%
Chengdu, China	140.85	-25.13	-15%	122.72	-20.93	-15%	104.95	-11.85	-10%
Beijing, China	116.71	-21.25	-15%	101.40	-18.09	-15%	98.78	-13.38	-12%
New Delhi, India	183.36	-13.25	-7%	114.35	-31.40	-22%	n.a.	n.a.	n.a.
Kolkata, India	187.42	-15.67	-8%	68.84	-35.19	-34%	n.a.	n.a.	n.a.
Jakarta, Indonesia	109.55	32.56	42%	120.86	15.26	14%	98.70	-13.94	-12%
Ulaanbaatar, Mongolia	192.19	-10.12	-5%	64.04	-9.79	-13%	n.a.	n.a.	n.a.
Kathmandu, Nepal	137.45	-22.63	-14%	85.38	-48.06	-36%	33.21	-57.61	-63%
Islamabad, Pakistan	n.a.	n.a.	n.a.	79.46	-48.97	-38%	89.94	0.87	1%
Bogotá, Colombia	52.35	-9.36	-15%	44.17	-3.19	-7%	n.a.	n.a.	n.a.
Lima, Peru	64.21	-10.90	-15%	60.73	-32.18	-35%	n.a.	n.a.	n.a.

Fig. 7

Relative Changes (in %) of AQI during three lockdown periods at typical cities over the world from January to June 2020 compared to the 2-years baseline data for each city. White space stands for lack of dataset for the city

Impact on NO2 concentrations

Fig. 8 reveals the Relative Changes of the tropospheric NO2 during the lockdown as retrieved from the NO2 concentration underlined in Fig. 6. A detailed analysis of the NO2 levels, during the lockdown, indicates that a 3–58 % reduction in the levels of NO2 across the major selected cities, while positive Changes were observed in three cities ((Abidjan (1%), Conakry (3%), and Chengdu (10%)) as indicated in Table 3 and Fig. 8. High NO2 values > 5E15 molecules/cm2 were observed in two cities including Beijing and Chengdu, while low concentrations < 1E15 molecules/cm2 were recorded in Kampala, Addis Ababa, Conakry, Abidjan, Bogotá, Ulaanbaatar, and Lima. The major reduction values as observed through NO2 data can be seen in Lima where the average values dropped by 58% followed by Jakarta (-33%), New Delhi (-31%), Dhaka (-27%), Islamabad (-24%), and Beijing (-21%). Compared to the baseline 2018-2019, there are mixed trends (increases in some cities while decreases in others) in NO2 after the lockdown period. The daily NO2 mean concentrations clearly increased at 8 cities by +9% in Addis Ababa, +1% in Conakry, +10% in Abidjan, +15% in Kuwait City, +7% in Baghdad, +19% in Chengdu, +9% in Kolkata, and +3% in Ulaanbaatar. The NO2 concentrations declined at 12 cities ranged from -3% in Kampala to -27% in Dhaka, while the NO2 data were not available for Manama city. Despite this mixed trend, the time series of daily NO2 concentrations showed clearly the effect of unlocking phases where the NO2 levels increase in almost all the cities. The NO2 concentrations in post-lockdown are higher than during the lockdown period in 12 cities such as Algiers, Addis Ababa, Kampala, Conakry, Khartoum, Kuwait City, Baghdad, New Delhi, Jakarta, Ulaanbaatar, Islamabad, and Lima.

Fig. 8

Relative changes (%) of NO2 concentrations at before, during, and post lockdown periods (depending on each city lockdown date) in 2020 as compared to 2-years average values indicating that a 3–58 % reduction in the levels of NO2 across the major selected cities. White space stands for lack of dataset for the city.

Table 3

Reduction in PM2.5 (μg/m3) and NO2 (1E15 molecules/cm2) levels in 2020 as compared to 2-years average values. Absolute differences (μg/m3 for PM2.5 and 1E15 molecules/cm2 for NO2) and relative changes (%) of PM2.5 and NO2 concentrations at before, during, and post lockdown periods (depending on each city lockdown dates). n.a.: Not Available.

City, Country	Pre lockdown						During lockdown						Post lockdown
	Ave.		Diff.		Change		Ave.		Diff.		Change		Ave.		Diff.		Change
	PM2.5	NO2	PM2.5	NO2	PM2.5	NO2	PM2.5	NO2	PM2.5	NO2	PM2.5	NO2	PM2.5	NO2	PM2.5	NO2	PM2.5	NO2
Algiers, Algeria	24.13	2.67	0.23	0.11	1%	4%	17.60	1.74	-2.17	-0.38	-11%	-18%	19.53	1.76	-5.25	-0.31	-21%	-15%
Addis Ababa, Ethiopia	22.60	n.a.	-0.87	n.a.	-4%	n.a.	22.73	0.61	-6.24	0.00	-22%	0%	38.05	0.79	-6.74	0.07	-15%	9%
Kampala, Uganda	67.42	0.58	5.63	-0.06	9%	-9%	36.94	0.07	-3.82	-0.02	-9%	-3%	56.05	0.70	-5.01	-0.03	-8%	-3%
Conakry, Guinea	n.a.	0.74	n.a.	0.02	n.a.	3%	15.18	0.56	n.a.	0.01	n.a.	3%	11.03	0.91	n.a.	0.00	n.a.	1%
Khartoum, Sudan	n.a.	0.87	n.a.	-0.10	n.a.	-10%	53.68	1.02	n.a.	-0.21	n.a.	-17%	57.90	1.08	n.a.	-0.24	n.a.	-18%
Abidjan, Côte d'Ivoire	n.a.	0.59	n.a.	0.04	n.a.	7%	18.35	0.53	n.a.	0.01	n.a.	1%	22.21	0.43	n.a.	0.07	n.a.	19%
Abu Dhabi, United Arab Emirates	19.92	3.70	-21.08	0.51	-51%	16%	21.77	2.66	-23.19	-0.49	-52%	-16%	26.94	2.39	-24.03	-0.37	-47%	-14%
Kuwait City, Kuwait	45.54	6.25	3.04	0.31	7%	5%	35.05	4.45	-10.89	-0.21	-24%	-4%	41.45	5.49	-14.76	0.71	-26%	15%
Baghdad, Iraq	49.83	5.16	16.32	0.51	49%	11%	39.83	2.83	11.03	-0.38	38%	-12%	78.07	3.77	42.31	0.24	54%	7%
Manama, Bahrain	42.41	n.a.	-5.15	n.a.	-11%	n.a.	42.00	n.a.	-4.29	n.a.	-9%	n.a.	41.60	n.a.	-9.34	n.a.	-18%	n.a.
Dhaka, Bangladesh	151.43	7.08	-5.78	-0.02	-4%	0%	51.25	2.85	-3.94	-1.08	-7%	-27%	29.45	1.98	-1.16	-0.73	-4%	-27%
Chengdu, China	61.69	6.15	-30.17	-2.22	-33%	-27%	48.75	6.18	-17.32	0.55	-26%	10%	58.19	5.65	12.92	0.91	29%	19%
Beijing, China	56.71	13.36	-27.03	-3.46	-32%	-21%	49.02	10.61	-16.42	-2.85	-25%	-21%	39.31	9.49	-7.92	-2.24	-17%	-19%
New Delhi, India	115.29	5.49	-15.95	-0.09	-12%	-2%	44.26	3.48	-27.09	-1.58	-38%	-31%	n.a.	3.49	n.a.	-0.36	n.a.	-9%
Kolkata, India	121.30	4.30	-15.25	-0.09	-11%	-2%	21.69	2.36	-15.32	-0.49	-41%	-17%	n.a.	2.12	n.a.	0.18	n.a.	9%
Jakarta, Indonesia	41.09	1.62	15.65	-0.22	62%	-12%	48.78	1.61	8.62	-0.79	21%	-33%	63.84	2.20	19.00	-0.46	42%	-17%
Ulaanbaatar, Mongolia	131.11	0.99	-14.12	0.07	-10%	8%	20.32	0.94	-6.90	-0.05	-25%	-5%	n.a.	1.33		0.04	n.a.	3%
Kathmandu, Nepal	66.15	1.59	-19.67	-0.39	-23%	-20%	31.56	1.98	-27.43	-0.51	-46%	-20%	9.32	1.35	-21.76	-0.27	-70%	-16%
Islamabad, Pakistan	n.a.	3.13	n.a.	-0.36	n.a.	-10%	25.39	2.12	n.a.	-0.66	-60%	-24%	30.32	2.67	-0.17	-0.60	-1%	-18%
Bogotá, Colombia	15.55	1.22	-2.91	0.19	-16%	19%	12.22	0.74	-1.36	-0.19	-10%	-20%	n.a.	0.71	n.a.	-0.24	n.a.	-25%
Lima, Peru	18.74	1.33	-5.21	0.17	-22%	14%	17.34	0.96	-15.06	-1.29	-46%	-58%	n.a.	2.02	n.a.	-0.47	n.a.	-19%

Impact on PM2.5 concentrations

For the present study dedicated to the impact of COVID-19 lockdown on air quality, a significant reduction in PM2.5 concentrations was observed for almost all the studied cities around the world (Table 3) when comparing the averages values recorded during this period in 2020, with those recorded during the same period in 2018-2019. As reported in Fig. 9 maximum decline in PM2.5 concentrations is shown in Islamabad (-60%), Abu-Dhabi (-52%), Kathmandu, and Lima (-46%); while the lower decrease was obtained for Dhaka (-7%), Manama, and Kampala (-9%), Bogota (-10%) and Algiers (-11%). As per PM2.5 pollution is significantly related to morbidity and mortality in the world ( Emmanouila et al., 2016; Attademo et al., 2017; (Samuel Mwaniki Gaita et al., 2016)), these reductions in their levels may confirm the temporary environmental benefit as a positive impact of COVID-19 lockdown (S.M. Ali et al., 2021; Rodríguez-Urrego and Rodríguez-Urrego, 2020). Fig. 6 shows the time series of daily PM2.5 concentrations from January 1 to June 30, 2020, in 21 cities across the world. There was an obvious valley point for most cities that the impact of COVID-19 lockdown is well demonstrated and observed. The maximum decline, during the lockdown, is shown in Kolkata, Ulaanbaatar, Dhaka, Conakry, New Delhi, Beijing, and Baghdad. While an increase in PM2.5 concentrations was observed in Addis Ababa and Jakarta. In addition, for the analyzed dataset, a very good correlation was obtained between PM2.5 concentrations and AQI values. The Pearson correlation coefficients (r) ranged from 0.88 in Beijing to 0.98 in Bogotá (p<0.01).

Fig. 9

Relative changes (%) of PM2.5 concentrations before, during, and post lockdown periods (depending on each city lockdown date). Reduction and increase in PM2.5 (μg/m3) levels in 2020 as compared to 2-years average values. Maximum decline in PM2.5 concentrations was showed in Islamabad, Abu-Dhabi, Kathmandu, and Lima, while the lower decrease was obtained for Dhaka, Manama, Kampala, Bogota, and Algiers. White space stands for lack of dataset for the city.

Impact of the lockdown during Ramadan month in Islamic Cities

On the other hand, a special focus on the variability of air quality during the Ramadan period has been carried out. Ramadan is one of the Islamic holy months in a year and lasts for 29 to 30 days (one lunar cycle). This year, Ramadan started on 24 (±1 day) April 2020 and it was very different from observing lockdown restrictions for population safety. To determine the impact of Ramadan observance on air quality, PM2.5 and NO2 concentration differences between the month of Ramadan (24 April to 23 May) and the shoulder months on either side were calculated. The two shoulder months were defined as 24 March to 23 April and 24 May to 25 June, respectively. For almost all investigated Islamic cities, both of these shoulder periods coincided largely within the lockdown period. Table 4 shows the effect of Ramadan on AQI and concentrations of PM2.5 and NO2 in some Islamic cities during the lockdown phases. The available data show that Ramadan has positive and negative impacts on air quality parameters arising from changes in people's activities and reduce unnecessary actions.

Table 4

Averages of AQI, PM2.5 (μg/m3), and NO2 (1E15 molecules/cm2) before, during, and after the month of Ramadan 2020 (24 April to 23 May).

City, Country	One month before Ramadan			During Ramadan			One month after Ramadan
	PM2.5	NO2	AQI	PM2.5	NO2	AQI	PM2.5	NO2	AQI
Algiers, Algeria	18.57	1.72	64.35	16.64	1.66	60.30	17.99	1.87	63.34
Abidjan, Côte d'Ivoire	18.77	0.52	61.86	17.82	0.54	62.03	23.47	0.45	74.71
Abu Dhabi, United Arab Emirates	22.24	2.53	68.27	23.03	2.15	72.13	34.07	2.43	97.25
Kuwait City, Kuwait	26.95	4.43	82.00	41.70	4.53	106.55	41.99	4.86	108.65
Baghdad, Iraq	34.26	2.30	95.29	23.73	3.16	74.03	80.71	3.28	80.71
Manama, Bahrain	41.56	n.a.	111.13	44.08	n.a.	117.59	39.30	n.a.	107.47
Khartoum, Sudan	78.31	0.89	149.09	53.31	1.01	121.79	60.90	1.12	130.54
Islamabad, Pakistan	28.08	1.99	84.98	27.15	2.36	83.01	31.58	2.77	92.73
Jakarta, Indonesia	48.05	1.74	122.85	51.10	1.53	128.95	54.90	1.91	97.14

Discussion

The present study has attempted to investigate the impact of forced lockdown on environmental components like PM2.5, NO2, and AQI before, during, and following lockdown periods (three lockdown phases). From the obtained results, it could be noted that the governments’ decisions in response to COVID-19 have varied impacts on the air quality with substantial differences among pollutants and also among cities. Then, a sudden decrease in economic and industrial activities due to the COVID-19 lockdown has resulted in a decrease in emissions of pollutants worldwide. Therefore, the differences observed between cities can be explained by considering their emission pollutant sources and the national lockdown policies mainly in case of non-integrated responses to the pandemic including partial/total lockdown and time lag in the full implementation of the countermeasures (Faridi et al., 2020; Zalakeviciute et al., 2020).

Hashim et al. (2021) argued that Baghdad (Iraq) recorded the highest NO2 emission before the lockdown due to high population density, traffic pollution, and industrial activities and that the transportation restriction and industry emission slowdown, during the lockdown, led to a clear decrease in NO2 emission. In Jakarta (Indonesia), despite the lockdown countermeasure, a remarkable increase in PM2.5 and NO2 concentrations have been noted. Wibowo et al. (2020) associated PM2.5 increase to the mobility of vehicles that remained in operation during the pandemic because the Jakarta area has still been active outdoors with strict health protocols and to the use of private vehicles instead of public transportation. Das et al. (2021) set out that within 10 days after lockdown, air quality in Kolkata (India) started to improve dramatically and became much healthier after 30 days of lockdown. They noted that the reduction of anthropogenic activities (namely tanneries, chemical, and cast-iron industries), of traffic, as one of the biggest concerns in Kolkata, and a complete ban on construction activities have accelerated the reduction of air pollution. Recent studies also reported that NO2 (Sathe et al., 2021) and PM2.5 (Kumar et al.,2020) reductions in Indian cities are comparable to the reductions observed in Asian and European cities. Paradoxically, while Biswal et al. (2020) and Vadrevu et al. (2020) argued that the highest reduction of tropospheric NO2 in New Delhi may be due to a reduction in vehicular emissions, which is the major pollutant source in the region, they estimated that in some Indian regions, there was an increase in NO2 concentrations observed during the lockdown. It could be related to the contribution of biomass burning sources identified as a major source of NO2 in the absence of significant anthropogenic activities.

The relative changes of NO2 concentrations in Fig. 8 call attention to the fact that NO2 levels in Chengdu (China) have increased during the defined lockdown period (from Feb. 25th to Mar. 24th). As such, the beginning and end of the lockdown were defined as represented by the control actions required by the Level-1 public health emergency response (i.e., Level-1 control actions regarding the municipal-level lockdown policies of different durations). Wang, Wen et al. (2020) explained that the lockdowns (namely period with Level-1 control actions) led to decreases in ambient NO2 concentrations, but the duration of the lockdown significantly affected the rebound of NO2 concentrations. For PM2.5, the relative reductions were less significant compared with NO2 during the period with Level-1 control actions. Accordingly, Chengdu had lifted the restrictions in late February and an increasing number of people began commuting normally. Not surprisingly, that led to substantial increases in NO2 concentrations. In contrast, Beijing still enforced the most stringent lockdown controls, and the daily NO2 emissions were reduced by 21% (Fig. 8 and Table 3). As a general trend, primary pollutant concentrations showed a decrease within few days after the lockdown and Kumar et al. (2020) stated that cities with larger traffic volumes showed mostly greater reductions. Since PM2.5 and NO2 are, commonly, the pollutants of main concern in urban cities with more than 1 million inhabitants, these results suggest that their concentrations during the lockdown period have been drastically reduced compared to previous years (baseline), recording a mean reduction of 7-60% for PM2.5 and 3–58% for NO2. The NO2 levels have been gradually regained in some cities in the unlock period and return-to-work days as argued by (Kanniah et al., (2020)) and Zhang et al. (2020), while PM2.5 has shown mixed trends as it increases in some cities and decreases in others. These are largely driven by changes in emission sources which are likely to have been not directly or less affected by the lockdown. PM2.5 levels are related to the strength of several sources like vehicle exhaust, industrial activities, thermal power generation, and biomass burning emissions, while NO2 levels were considerably linked with the cities' efforts to the cessation of industrial and transportation activities.

The air quality index (AQI) classification calculation mainly uses SO2, NO2, CO, O3, PM10, PM2.5, and other pollutant concentration values to convert into corresponding indexes. AQI can be used for environmental status assessment, trend evaluation, and providing timely and accurate air quality (Guo et al., 2019). Based on daily mean concentrations during the investigated period, the Pearson's coefficient (r2) between AQI and PM2.5 showed a significant positive correlation suggesting that PM2.5 was the main factor influencing air quality. This is in agreement with Yan et al. (2016) who pointed out that PM2.5 was the major pollutant affecting air quality in China and Beijing. Bao and Zhang (2020) showed by estimating the effects of COVID-19 related travel restrictions on air pollution that a reduction in both AQI and PM2.5 was partially mediated by human mobility. Bringing together different AQI observed trends, it can be seen that the overall situation of air quality during lockdown at our typical cities is varied from excellent to mild pollution while before lockdown it was categorized as moderate polluted and very unhealthy. In contrast, the air quality began to worsen during unlock phase in 50% of the cities where AQI ranged from 33.21 to 139.12.

Furthermore, this study also discussed the pollutant concentration changes during the Ramadan period. The integrated effect of the Holy month of Ramadan and the lockdown in 2020 resulted in a decrease and an increase in air quality parameters levels. Abou El-Magd et al. (2020) showed that a NO2 decrease was observed in late May in Egypt as this period coincided with the end of the Holy month of Ramadan and Breaking the Fast “Eid Al-Fitr” holiday together with a weekend. Contrarily as Mostafa et al (2021) discussed, from April 1 to May 1, 2020, most of the plants resumed their operation to provide the essential products needed during the Holy month of Ramadan and that led to a slight increase in NO2 emissions over Cairo and Alexandria during that period. These positive and negative impacts on air quality were probably due to differing preventative measures and the degree of the population adhered to the countermeasures immediately.

Finally, the vision, or worldview, that an analyst uses to judge the impact of the COVID-19 outbreak points out that the air quality in the investigated cities has greatly improved during the lockdown period and the concentration of various pollutants has been significantly reduced. However, it is noteworthy to think that this situation is always temporary. In this spot, this study advocates that major policy for future pollution abatement would be related to the transport and industrial sectors that may reduce the ambient particulate matter concentrations although it may not have a significant effect on some secondary ambient air pollutants such as O3.

Conclusion

This study aimed to estimate the impact of COVID-19 lockdown on NO2, PM2.5, and AQI in 21 cities around the world. A before/after comparison approach was used to investigate temporal variations and trends of air pollutant levels caused by the lockdown restrictions. In this vein, our attempt has focused on the evolution of AQI, NO2, and PM2.5 levels during three periods covering before, during, and after the lockdown periods. The main conclusions of this study could be summarized as follows:

The frequency distribution of PM2.5 and NO2, expressed by the probability density functions, is more variable for NO2 than that for PM2.5 and is flatter from 2020 to 2018-2019 periods.

AQI in most of the studied cities, during the lockdown, has varied from high to mild pollution while before lockdown, the level of air quality could be classified as moderate to very unhealthy;

PM2.5 displayed higher levels at the pre-lockdown phase, in 50% of the investigated cities, than at any other time indicating significant positive differences between the lock and unlock phase;

The integrated effect of the Holy month of Ramadan during the lockdown period led to positive and negative impacts on air quality parameters arising from the changes in people's activities and the degree of their adhesion to the countermeasures.

In short, the worldwide lockdown due to the COVID-19 outbreak has drastically reduced air pollution and thus improved the air quality over the investigated cities. Analyzing Relative Changes in AQI, PM2.5, and NO2 levels during different lockdown periods is necessary to improve the understanding of the impact of emission reductions on air quality. Moreover, the role of air mass trajectories and meteorological factors has neither been evaluated nor quantified in this study. Authors believe that further investigations are needed in the future for revealing the degree of the pollution reemergence over urban cities particularly after the reopening of their economies.

Ethical approval

Authors declare that this current research paper is original and has neither been published earlier nor it has been sent to any other journal for publication.

Consent to participate

Not applicable.

Consent for publication

Not applicable.

Availability of data and materials

The data used in this work are available in U.S. embassies and consulates repository, https://www.airnow.gov/international/us-embassies-and-consulates/.

Funding

This research received no external funding.

Authors’ Contributions

Conceptualization, A.B., and B.B.; methodology, A.B., A.W., R.M.S., M.T., and B.B.; data acquisition: R.M.S. and B.B.; formal analysis, A.B., A.W., and M.T.; investigation, A.B., A.W., R.M.S., M.T., and B.B.; writing—original draft preparation, A.B., A.W., R.M.S., M.T., and B.B.; visualization, A.B. and M.T. All authors have read and agreed to the published version of the manuscript.

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.