Impact factor (WEB OF SCIENCE - Clarivate)

2 year: 7.664 | 5 year: 7.127

COVID-19Ongoing Research Themes

Changes in air pollution due to COVID-19 lockdowns in 2020: Limited effect on NO2, PM2.5, and PM10 annual means compared to the new WHO Air Quality Guidelines

Cleonilde Maria do Nascimento1,2, Sheilla Andrade de Oliveira2, Otacílio Antunes Santana1, Helotonio Carvalho1,2

1 Department of Biophysics and Radiobiology, Biological Sciences Centre, Federal University of Pernambuco, Recife, Pernambuco, Brazil
2 Department of Immunology, Aggeu Magalhães Institute (IAM), Oswaldo Cruz Foundation (FIOCRUZ), Recife, Pernambuco, Brazil

DOI: 10.7189/jogh.12.05043




Lockdowns have been fundamental to decreasing disease transmission during the COVID-19 pandemic even after vaccines were available. We aimed to evaluate and compare changes in air quality during the first year of the pandemic in different cities around the world, investigate how these changes correlate with changes in mobility, and analyse how lockdowns affected air pollutants’ annual means.


We compared the concentrations of NO2, PM2.5, and PM10 in 42 cities around the world in the first months of the pandemic in 2020 to data from 2016-2019 and correlated them with changes in mobility using Human Development Indexes (HDIs). Cities with the highest decreases in air pollutants during this period were evaluated for the whole year 2020. We calculated the annual means for these cities and compared them to the new World Health Organization (WHO) Air Quality Guidelines. A Student’s ttest (95% confidence interval) was used to evaluate significant changes.


Highest decreases in NO2, PM2.5, and PM10 were between -50 and -70%. Cities evaluated for the whole year 2020 generally showed a recovery in air pollution levels after the initial months of the pandemic, except for London. These changes positively correlated with year-long mobility indexes for NO2 and PM2.5 for some cities. The highest reductions in air pollutants’ annual means were from -20 to -35%. In general, decreases were higher for NO2, compared to PM2.5 and PM10. All analysed cities showed annual means incompliant with the new WHO Air Quality Guidelines for NO2 of 10 μg/m3, with values 1.7 and 4.3 times higher. For PM2.5, all cities showed values 1.3 to 7.6 times higher than the WHO Guidelines of 5 μg/m3, except for New Delhi, with a value 18 times higher. For PM10, only New York complied with the new guidelines of 15 μg/m3 and all the other cities were 1.1 to 4.2 times higher, except for New Delhi, which was 11 times higher.


These data show that even during a pandemic that highly affected mobility and economic activities and decreased air pollution around the world, complying with the new WHO Guidelines will demand a global strategical effort in the way we generate energy, move in and around the cities, and manufacture products.

Print Friendly, PDF & Email

Covid-19 has imposed serious challenges on health authorities and governments [1]. To minimize disease transmission, countries closed borders, imposed quarantines, and used lockdowns, closing schools, stores, restaurants, and offices. With traffic reduction and many industrial activities halted due to the pandemic, air pollution was supposed to be reduced. At the end of February 2020, satellite data from NASA showed significant drops in NO2 levels in China [2]. Later on, NO2 level reduction was observed by the European Union Copernicus Programme, from the European Space Agency, in Europe, and India [3,4]. Lower air pollution levels were also reported in several cities like Madrid, Barcelona [5], São Paulo [6], Milan [7], London, Paris [8], New Delhi [9], Santiago [10], and countries or world regions like China [11,12], India [13,14], Western Europe [15], and the USA [16].

Air pollution is responsible for more than 7 million deaths each year [17]. From the several air pollutants, particulate matter (PM) has been the one most widely associated with adverse health effects. The last air quality report from the European Environmental Agency (EEA) indicates that from more than 427 000 deaths attributed to air pollution in Europe in 2019, 72% were linked to PM2.5 and 9.4% were linked to NO2 [18]. Particulate material, comprising PM2.5 and PM10 as the major components, are mainly derived from combustion processes, including traffic-related and industrial activities. PM is associated with increased risk of death by ischaemic heart disease, cerebrovascular disease, lung cancer, lower respiratory infections, and chronic obstructive pulmonary disease [19]. NO2 is also associated to combustion processes, and calls for more attention about its effects on human health have been issued in countries with large diesel vehicle fleets, like many European countries and India, since these vehicles are the main NO2 source in urban environments [20].

Despite other studies describing the effects of lockdowns on air pollution levels, most of them compared either a few cities or many cities in the same country/region of the world. Only a few studies investigated air quality changes over longer periods in 2020, generally focusing on the February-May period. This study aimed to compare changes in NO2, PM2.5, and PM10 among 42 cities on six continents during the first months of the pandemic, compare them with 2016-2019 data, and check how they would correlate with changes in city mobility indexes. We also aimed to study selected cities with higher changes in air pollution for the whole year 2020, compare their NO2, PM2.5, and PM10 annual means with the 2016-2020 period and analyse them according to the new World Health Organization (WHO) air quality guidelines [21].


Choice of the cities

The 42 cities analysed were chosen based on several criteria: 1) air pollution levels of PM2.5 ranging from low to high (Table 1); 2) geographical location to represent different regions of the world and/or different regions within a country; 3) COVID-19 epidemiology in order to represent countries which were highly affected and little affected by COVID-19, and thus would be applying different levels of mobility restrictions during the time frame evaluated; 4) availability of historical air pollution data, since whenever possible, official data were used. For inclusion in the list, the existence of data for NO2, PM2.5, and PM10 for previous years (2016-2019) and for 2020 was checked, as well as the consistency of these data.

Table 1.  Cities analysed on this report, PM2.5 annual mean, HDI, and source of air quality data

We analysed PM2.5 air pollution data for ten cities in China, including Wuhan, Beijing, and Shanghai; three in India, including New Delhi and Mumbai; cities in other Asian countries, such as Seoul, Jakarta, and Hanoi; cities in Europe such as London, Paris, Milan, and Berlin; cities in the Middle East, such as Dubai and Abu Dhabi; cities in Africa, such as Kampala and Addis Ababa; cities in Australia, such as Sydney and Adelaide; cities in North America, such as Los Angeles, New York, and Mexico City; and South America, such as São Paulo and Lima. PM10 and NO2 data were analysed for 28 cities around the world.

Time frame for analysis of air pollution data

Data for NO2, PM2.5 and PM10 were first analysed for the first months of the pandemic, defined as February-April. The choice of this period considered the spread of the pandemic and the consequent lockdowns which happened first in China and then in Europe, USA and all the other regions analysed, according to WHO (21). After these initial months, cities which showed the highest changes in the levels of air pollutants were chosen for a further analysis covering the whole year 2020.

Air quality and mobility data

Air pollution data were obtained from official environmental agencies, websites that reproduce these data, the EEA, US Environmental Protection Agency or US State Department (US Embassies and Consulates). For Berlin, PM2.5 data for 2020 was kindly provided by Dr Paul Herenz (Senate Department for Environment, Transport and Climate Protection). Some of the data for Venice were kindly provided by Patti Salvatore and Luca Zagolin from ARPAV – Dipartimento Regionale Qualità dell’Ambiente-Veneto. NO2, PM10, and PM2.5 data were generally obtained as daily mean values expressed in μg/m3. 2016-2019-monthly means were calculated and compared to 2020 to generate % changes. Table 1 summarizes data source for the analysed cities. PM2.5 annual means were obtained from the 2018 WHO Global Ambient Air Quality Database [22], except for Dubai, Manama, and Addis Ababa, which were calculated. Except for US Embassies and Consulates data, air pollution levels represent the mean for at least three to six stationary air quality monitoring stations, depending on the number of stations available. Whenever possible, monitoring stations were chosen according to their locations in different areas across the cities.

Anonymous Google Community Mobility data were obtained from Google Inc. (USA) [23]. Changes in mobility for transit stations and workplaces were chosen since they better represent changes with potential impacts on air pollution. Google Mobility data were not available for Chinese cities and for Addis Ababa.

Statistical analysis

The Human Development Index (HDI) is composed of indicators of three dimensions of human development: longevity, education, and income. The adjusted Human Development Index (HDIadj) adds other variables such as health, education, income, inequality, poverty, employment, environmental sustainability and was adjusted for the city limits in our study. The HDIadj was then corrected by the City Development Index Method [24,25]. For this work, United Nations Development Programme (UNDP) data for 2019 were used for the calculations.

Discriminant analysis was performed to classify cities within groups by HDIadj. Variables: HDIadj, NO2 (from 2016-2019 to 2020), PM2.5 (from 2016-2019 to 2020), Workplaces Mobility Changes (% change from baseline), and Transit Stations Mobility Changes (% change from baseline) were used in the analysis. The assumptions of discriminant analysis were observed (multivariate normality, homoscedasticity, multicollinearity, and independence) [26]. Regression analyses were used to examine the relationship between environmental variables and mobility variables (R2>0.80, p<0.05, ɛ<0.05), first, between HDIadj (x) and PM2.5 annual mean (y), and second, between PM2.5 (from 2016-2019 to 2020) (x) and Workplaces, and Transit stations (y). In the second situation, PM2.5 data were normalized by (xi + )2−1, according to the DA assumptions [26,27]. Statistical analyses were performed with BioEstat 5.3 package [28]. The data were normalized according to the D´Agostino normality test (p>0.01) [29]. The data equalization from sampling methods to metric of data presentation was carried out by the IPCC Metrics & Methodology [30]. The data comparison between the periods (2016-2019 and 2020) was performed using the t test (95% Confidence, p<0.05). All variables were on a normal distribution.

Regression analysis (R2) was used to indicate trends between NO2 and PM2.5 levels and mobility for the same city in a time series through the year 2020. We prioritized linear regression to standardize the trend between the cities (Y ̂ = β_0+β_1 · X_i ±ɛ i). In this work, the objective of this analysis was not a perfect fit (>R2 possible), but the trend (p<0.05 and ε<0.05) [27].


Changes in NO2, PM2.5, and PM10 levels in the February-April 2020

We first compared air pollution levels for the first months of the pandemic (February-April) with previous years (2016-2019) to identify the cities most affected by lockdown measures during the first Covid-19 wave. The extent of NO2 reductions varied widely among different cities (Figure 1, panel A). Most Chinese cities showed the highest reductions in February/March, due to earlier restrictive measures. Most other cities imposed lockdowns in March and higher NO2 reductions were observed in April. Among the cities analysed, highest NO2 reductions were observed in some Asian and European cities such as Beijing (-50.1%), London (-53.0%), Paris (-59.7%), Wuhan (-61,2%), Madrid (-61.3%), and New Delhi (-68.2%). The results confirmed the statistical difference in NO2 levels (p<0.05) between the studied periods (2016-2019 and 2020) for 19 out 28 cities with statistically significant changes in NO2 (p<0.05). No statistically significant NO2 changes were found in Moscow, Milan, Rome, Los Angeles, and New York.

Figure 1.  Changes in concentrations of NO2 (panel A), PM10 (panel B), and PM2.5 (panel C) for February-April 2020 compared to 2016-2019. *Cities with P < 0.05. †Compared to 2018-2019. ‡Compared to 2017-2019.

New Delhi (-61.4%), Nanjing (-52.9%), and Wuhan (-52.7%) were the cities with highest decreases in PM10 (Figure 1, panel B). In the case of New Delhi, reductions were progressively higher from February to April. Changes in PM10 in Adelaide and Sydney, were not statistically significant. However, both cities showed much higher PM10 and PM2.5 levels in January compared to 2016-2019 (data not shown) due to extensive wildfires [31]. The results confirmed the statistical difference in PM10 levels (p<0.05) between the studied periods (2016-2019 and 2020) for 16 out of 28 cities. No statistically significant changes in PM10 were observed in Berlin, Rome, Milan, Los Angeles, and New York.

More cities were analysed for PM2.5 (Figure 1, panel C), since data from US Embassies or Consulates were used for cities with no available data. Paris, London, and Berlin showed highest decreases in PM2.5 levels in February (-55.7%, -49.8% and -63.5% respectively). Trends for PM2.5 in these cities were similar to those for PM10, with significantly lower reduction in April. Among other cities with highest reductions on PM2.5 levels were Abu Dhabi (-54%) and New Delhi (-49.6%). Many Chinese cities showed highest decreases on PM2.5 levels in February, including Hangzhou and Wuhan (-51.7% and -46.5, respectively) and March, including Beijing and Shanghai (-52.8% and -42.7%, respectively). The results confirmed the statistical difference in PM2.5 levels (p<0.05) between the studied periods (2016-2019 and 2020) for 18 out of the 41 cities analysed. PM2.5 changes were not statistically significant for Rome, Madrid, Los Angeles, Addis Ababa, Jakarta, Kathmandu, and Ulaanbaatar.

Association between changes in mobility and changes in NO2 and PM2.5 levels in February-April 2020

Anonymous GPS data from mobile phones have been used as a convenient indicator for lockdown measures efficiency [32]. With a few exceptions, most cities were already significantly affected by lockdown/social distancing measures by March 2020, showing decreased mobility indexes, (Figure S1, in the Online Supplementary Document). By April 2020, the effect of lockdowns was more evident, with most cities showing -60% to -80% reductions in mobility. Since the decrease in air pollutants was probably associated with lockdowns, we used mobility changes to evaluate a possible association with changes in air pollution among 16 cities, for NO2 and PM10, and 28 cities, for PM2.5. Discriminant analysis was used to classify cities by HDIadj, according to NO2 and PM2.5 changes, as shown in Figure 2, panel A. In March and April, PM2.5 and NO2 data grouped into cities with similar HDIadj values, forming four different groups. City development is differentially associated to air pollution since pollutants emission varies in developed and emerging countries [33]. A possible association between PM2.5 annual means and HDIadj was also evaluated (Figure 2, panel B) showing a clear negative trend between these variables, with highly developed cities showing significantly lower PM2.5 annual means. A possible association between Google Mobility Indexes and changes in PM2.5 and NO2 levels was then investigated. Figure 3, panel A shows that reductions in PM2.5 levels were clearly associated with changes in Workplace and Transit Stations Mobility according to the city HDIadj, with higher R2 for cities with lower HDIadj. Figure 3, panel B shows the same analysis for NO2. Only HDIadj strata which showed statistical relevance are show. In this case, cities with higher HDIadj showed the best correlation between decreases in NO2 levels and mobility changes.

Figure 2.  (Panel A) Discriminant analysis for cities according to their HDIadj. (Panel B) Association between PM2.5 annual means and HDIadj.

Figure 3.  Association between concentrations of PM2.5 (panel A) or NO2 (panel B) and mobility changes according to cities Human Development Index.

Changes in NO2, PM2.5 and PM10 in 2020 for selected cities.

The 16 cities with highest decreases in air pollution levels over time in February-April and increasing reductions in this period, were further selected to investigate the whole year 2020 compared to 2016-2019. The levels of air pollutants were lower most of the year for London, Paris and Madrid, with highest decreases generally between February-May (-32% to -61%) (Figure, panel A). However, changes in PM2.5 and PM10 for Madrid were not statistically significant (p>0.05). Among Italian cities, NO2 levels were lower during most of 2020. However, changes in PM2.5 and PM10 were not statistically significant (Figure 4, panel B).

Figure 4.  Concentrations of NO2, PM2.5, and PM10 in 2020 and 2016-2019 for cities in Europe (panel A and panel B), Asia and Middle East (panel C), and North and South America (panel D). For Dubai and Abu Dhabi, 2020 data were compared to 2018-2019 means. P < 0.05 was considered statistically significant.

All the cities in Asia and Middle East showed decreased levels of all air pollutants throughout most 2020, with highest decreases for NO2 and PM10 in February-May (-36% to -68%), in the case of New Delhi, Beijing and Wuhan (Figure, panel C). For Abu Dhabi, highest reductions in PM2.5 were before the pandemic, in January (-62%) and May-October (-41% to -61%). The pattern of changes in PM2.5 levels was similar for Dubai, with highest reductions in January (-60%), and April-October (-26% to -43%).

Among cities in North and South America, overall changes in air pollutants in Los Angeles were not statistically significant (Figure 4, panel D). However, massive increases in PM2.5 (+82% to +144%) and PM10 (+44% to +64%) were observed in September-October due to wildfires [34]. For New York, NO2, PM2.5 and PM10 were lower most of the year, with highest decreases in March-May (-27% to -49%). In the case of Santiago, NO2 changes were higher than other pollutants with highest decreases spread over several months (March-August: -29 to -49%), while changes in PM10 were not statistically significant. The same was also the case for NO2 changes for Buenos Aires, while PM10 was decreased most of the year, similarly to what was observed for PM2.5 in the case of Lima. Overall, the results confirmed the statistical difference (p<0.05) in the data between the studied periods (2016-2019 and 2020) for most of the cities and pollutants analysed.

Association between changes in NO2 or PM2.5 and changes in mobility indexes in selected cities in 2020.

Google mobility data were further analysed for the whole year 2020 (Figure S2 in the Online Supplementary Document). Transit station and workplace changes were still consistently lower up to December for some cities, including Abu Dhabi, London, Milan, and New York. In other cities, such as Dubai, Santiago, Madrid, and New Delhi, transit stations changes were progressively lower until December. A possible association between changes in mobility and changes in air pollution over time was then investigated. Chinese cities were not included due to the lack of mobility data. Many of the cities showed an association between transit stations or workplaces mobility and NO2 changes (Figure 5, panel A), with the strongest associations for Rome (R2 = 0.89 for transit stations, R2 = 0.84 for workplaces), New Delhi (R2 = 0.62 and R2 = 0.61), Madrid (R2 = 0.51 and R2 = 0.63), Paris (R2 = 0.46 and 0.47), and Venice (R2 = 0.42 and R2 = 0.53). London and especially New York showed a weaker association between mobility and NO2 changes. Milan, Los Angeles, Santiago, and Buenos Aires did not show any association between changes in mobility and NO2. An association between mobility indexes and PM2.5 levels was observed only for a few cities (Figure 5, panel B), with strongest associations for New Delhi (R2 = 0.59, transit stations and R2 = 0.65, workplaces) and Lima (R2 = 0.56 and R2 = 0.60), and weaker associations for Milan. There was no apparent association between changes in mobility and PM2.5 for Dubai, Abu Dhabi, Paris, London, Rome, Venice, Los Angeles, Madrid, Santiago, and Buenos Aires.

Figure 5.  Association between concentrations of NO2 (panel A) or PM2.5 (panel B) and mobility changes in March-December 2020. P < 0.01; ɛ<0.05.

Effect of Covid-19 lockdowns in PM2.5, PM10 and NO2 annual means

We also investigated how differences in air pollution levels in 2020 affected annual PM2.5, PM10 and NO2 concentrations, compared to 2016-2019. Figure 6 shows many cities have progressively reduced NO2 levels between 2016 and 2020, which is probably related to efforts to decrease air pollution, especially where NO2 levels have been historically higher. To compare year-to-year variation, percent changes between subsequent years were calculated and the mean change from 2016-2019 was compared to 2020 (Figure 6). Decreases in NO2 annual means were much higher in 2020, especially for London, Paris, and Santiago (-24% to -32%) while most cities showed 2016-2019 decreases in annual means lower than -10%. Los Angeles was the only city with higher NO2 levels in 2020, despite the small difference (+5%). As shown for NO2, most cities showed progressive decreases in PM2.5 annual means in 2016-2019, and higher decreases in 2020. Highest decreases (-18% and -35%) were observed in London, Paris, Wuhan, New Delhi, Abu Dhabi, Dubai, and Lima. However, Rome, Milan, and Venice showed no decrease in PM2.5 annual mean, instead, there was an increase as high as +23% in the case of Milan. An increased PM2.5 annual mean in 2020 was also observed in Los Angeles due to wildfires. In the case of PM10, there was also a general decrease along the years, which was particularly higher in 2020 in the case of London, Wuhan, New Delhi, and Buenos Aires (-19% to -24%). Again, Rome, Milan, and Venice showed increases in PM10 annual mean, but smaller than the increases in PM2.5. PM10 annual mean for Los Angeles was +30% higher in 2020 compared to 2019, in agreement with increases in PM2.5.

Figure 6.  Annual mean concentrations for NO2, PM2.5, and PM10 during 2016-2020 and comparison with the new WHO Air Quality Guidelines. Inner tables show the mean percent change in annual means from year to year for 2016-2019 and for 2020-2019. *Compared to 2018-2019. For New Delhi, data from 2017 were not available for the whole year and were not included in the analysis.


Despite its serious impacts on the world, the COVID-19 pandemic had a positive effect on air pollution. Maximum decreases in air pollution levels described here were, in general, similar to other reports. For NO2, a -58.4% decrease was described for Hangzhou [35], -62.0% for Madrid [5], -53,6% for London, -39.0% for Milan [8], and -60% for Wuhan [36]. For some cities, other reports have found different values for changes in air pollution, what may be related to differences on the choice of monitoring stations, the specific time frame analyzed and/or used for comparison. Thus, a -40% decrease was described for Beijing (-40%), compared to 2018 [36], -51.5% for Paris, compared to 2017-2019 values [8], and -61% for New Delhi, compared to 2017-2019 [14].

Interestingly, even though some cities were studied in other reports, this is the first one to describe changes in air pollution levels due to Covid-19 restrictions in Venice. This is surprising, since NO2 annual mean in Venice is similar to Paris, while PM2.5 and PM10 means are much higher. Besides this, Venice is one of the cities with worst air qualities among more than 300 cities in Europe, based on PM2.5 levels [37], which may be associated to its large dependency on diesel-powered water transport (Vaporetto) and to cruise ships which reach the Venice Lagoon.

For PM10, similar maximum reductions were described for New Delhi (-57%) [14], London (-34%) [38] and Wuhan (-55%) [39]. For PM2.5, highest decreases close to the reported here were observed for Paris (-58%), and London (-56%) [39], New Delhi (-53%) [14], Shanghai (-41%) [40] and Wuhan (-38%) [36]. Despite the absence of statistically significance, the smaller decreases/increases observed for some cities were also described by others for Addis Ababa, Dhaka, Kampala, Jakarta, and Ulaanbaatar [41]. Other authors analysed the effects of COVID-19 restrictive measures in Sydney and Melbourne, but as far as we know, this is the first study to analyse data for Adelaide.

Beijing was the city with highest discrepancies in PM2.5 changes compared to other reports. However, the way data were compared was different among these studies, using comparison against 2015-2019 [39], 2018-2019 [41], 2017-2019 [42] and 2016-2019 for the present report.

Many cities in low- and middle-income countries have poor urban planning, disorganized traffic, lack regular vehicular emission inspection and have older cars compared to high income countries. However, many cities in high-income countries stimulate dispersion of households away from the city centres, emphasize more environmentally friendly and efficient modes of transportation, and restrict the access to city centres creating low and ultralow emission zones [43]. These aspects make the association between air pollutants and mobility vary among cities in low- and middle- compared to high income countries [44]. We used HDIadj to group cities and compensate for such variables, allowing for a better analysis of associations between mobility and air pollution. Sokhi et al. [39] found a positive correlation between mobility data and NO2 concentrations (R2 = 0.513) but did not find a correlation for PM2.5 and argued that atmospheric chemical reactions leading to the formation of secondary aerosols might explain this. Venter et al. [45], only found a correlation between workplaces mobility data and NO2 and attributed this to the fact that, in many countries, PM2.5 is more associated with residential and regional energy generation and agriculture. Despite the reasons addressed by these authors, the approach used here was successful in also confirming an association between changes in mobility and PM2.5 levels. Besides this, the R2 values for both pollutants were higher for most HDIadj strata analysed here. Interestingly, when cities were analysed throughout 2020 and the regression analysis was performed between same city data, more cities showed a relevant association for NO2 than for PM2.5. This is probably linked to differences in the main sources of PM2.5 and NO2 in different cities.

Data analysis for the whole year 2020 shows that, even though many of these cities extended or reapplied lockdowns after this period, as confirmed by the analysis of mobility indexes, the major effect of these measures on air pollution levels was observed on the first months of the pandemic, February-May. An interesting exception is London, where NO2 levels were much lower, during the whole year. Other cities, like Paris and Madrid, also showed lower NO2 levels during the whole year but decreases were much smaller for some of the months. Highest reductions in NO2 were observed for some European cities such as London, Paris, and Madrid, as well as for New Delhi. The general trend for PM2.5 levels at the beginning of the pandemic was quite different compared to NO2. Cities like Paris, London, and Berlin showed decreasing reductions in PM2.5 levels from February to April while only a few cities like Lima, Abu Dhabi, Dubai, and New Delhi showed increasing reductions over these months. The major effects of lockdowns on PM2.5 levels were visible between March and April. London, Beijing, New Delhi, Abu Dhabi, and Dubai, however, showed decreased PM2.5 levels throughout most of the year. In general, decreases in PM2.5 levels were smaller when compared to NO2 which was especially the case for Italian cities. As expected, trends in PM10 were similar to PM2.5, as observed for Paris, London and Berlin, for example, on the first months of the pandemic and for most cities analysed for the whole year 2020.

Successive decreases in NO2 annual means, and higher decreases in 2020, have driven many of the cities below or close to WHO guidelines (2005) of 40 μg/m3. Milan and New Delhi were the only cities slightly above these guidelines. However, compared to the new stricter WHO guidelines (2021) [46], all the cities analysed showed 2020 NO2 levels 1.7-4.3 times higher than the 10 μg/m3 guideline. Even after successive decreases in PM2.5 annual means and higher decreases observed in 2020, many of the cities still did not comply with the WHO guidelines (2005) of 10 μg/m3. With the new WHO guidelines (2021) of 5 μg/m3, Milan and Venice showed five times higher PM2.5 levels in 2020, Abu Dhabi, Dubai, Lima, and Santiago 4.7-6.8 times higher, Beijing and Wuhan 7.5 times higher, and New Delhi showed levels 18 times higher than the new WHO guidelines. Madrid, Paris, and New York were the only cities with PM2.5 annual means lower than WHO guidelines (2005) but were above the new guidelines (2021), with 2020 pm2.5 levels 1.3-1.9 times higher. In 2020, only London, Madrid, New York, and Buenos Aires complied with the older WHO guidelines (2005) for PM10 of 20 μg/m3. With the new guidelines (2021) of 15 μg/m3, this happened only for New York, while London, Madrid and Buenos Aires were barely above it. Milan, Rome, Venice, and Los Angeles were 1.7-2.2 times higher; Beijing, Wuhan, and Santiago, 3.8-4.2 times higher; and New Delhi, 11 times higher than WHO guidelines. Despite using different data sets, timeframes, and analysing different cities, Carvalho [47] showed that even cities with lower air pollution levels will require additional efforts to satisfy the new WHO Air Quality Guidelines. However, the challenge will be much harder for cities with higher air pollution levels, with many of them in need for at least 90% reduction in PM2.5 levels, compared to 2018 WHO data, including Beijing, Ulaanbaatar, and New Delhi.

It is important to mention that this study has some limitations. Changes in meteorological variables may impact the concentration of air pollutants and a specific compensation for local meteorological factors was not carried out. Instead, we used year to year comparison to account for changes in meteorological variables. Different authors have also reported the impacts of lockdowns on air pollution using a year-to-year comparison as we have done here [36,41], to account for these changes and our results are in line with these as well as other studies which considered specific local meteorological factors [3840]. Another limitation of the study is the fact that Google Mobility data only accounts for Google users, while a fraction of the population was not considered. Besides this, the use of Google Mobility was not possible for Chinese cities, due to lack of data availability, but these data were not available for other mobility indexes as well, such as Apple Mobility.

The results reported here corroborate these conclusions and show that even with the COVID-19 pandemic, the decrease in air pollutant levels was limited if we consider the whole year 2020, since most cities showed a “recovery” in air pollution levels after the most restrictive months of the pandemic.


Lockdowns have been a valuable containment strategy for COVID-19, used even after vaccines were available, to maintain a low numbers of new cases, hospitalizations, and deaths. For many of the cities analysed, 2020 annual means for NO2, PM2.5 and PM10 were much smaller than for 2016-2019, indicating that altogether, even when restrictions were suspended, the increase in the levels of these pollutants was not enough to counteract the decrease previously observed. However, even with the reductions in annual means, many cities were far for complying with the new WHO Air Quality Guidelines, especially for NO2.

Additional material:

Online Supplementary Document


We thank Dr Paul Herenz (Senate Department for Environment, Transport and Climate Protection) for providing PM2.5 data for Berlin and Patti Salvatore and Luca Zagolin (ARPAV – Dipartimento Regionale Qualità dell’Ambiente-Veneto) for providing data for Venice.

Ethics statement: This article is the authors’ own original work, reflecting their own research and analysis and has not been previously published elsewhere. The results are described in the context of existing research. All data sources used are properly disclosed.

Data availability: The sources for data used in this report are mentioned on Table 1.

[1] Funding: Cleonilde Nascimento acknowledges a PhD fellowship from CAPES (Coordenação de Aperfeiçoamento de Pessoal de Nível Superior), Brazil. Helotonio Carvalho acknowledges a Research Grant from Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq, Brazil, Grant No 423733/2016-8). The authors acknowledge CAPES for the support for the publishing fees.

[2] Authorship contributions: HC: conceived the study, coordinated the work, and wrote the first version of the manuscript. CN and HC: conducted literature research, collected data, and conducted data analysis. OS, CN, and HC: Conducted statistical analysis and produced the figures. SO: revised data and revised the first version of the manuscript. All authors revised the final manuscript and have read and agreed to the published version of the manuscript.

[3] Disclosure of interest: The authors completed the ICMJE Disclosure of Interest Form (available upon request from the corresponding author) and disclose no relevant interest.


[1] World Health Organization. Listings of WHO’s response to COVID-19. Accessed: 29 January 2022.

[2] Patel K. Airborne nitrogen dioxide plummets over China. Accessed: 13 January 2022.

[3] European Space Agency. Coronavirus lockdown leading to drop in pollution across Europe. Accessed: 13 January 2022.

[4] European Space Agency. Air pollution drops in India following lockdown. Accessed: 13 January 2022.

[5] JM Baldasano. COVID-19 lockdown effects on air quality by NO2 in the cities of Barcelona and Madrid (Spain). Sci Total Environ. 2020;741:140353. DOI: 10.1016/j.scitotenv.2020.140353. [PMID:32593894]

[6] LYK Nakada and RC Urban. COVID-19 pandemic: Impacts on the air quality during the partial lockdown in São Paulo state, Brazil. Sci Total Environ. 2020;730:139087. DOI: 10.1016/j.scitotenv.2020.139087. [PMID:32380370]

[7] MC Collivignarelli, A Abbà, G Bertanza, R Pedrazzani, P Ricciardi, and MC Miino. Lockdown for COVID-2019 in Milan: What are the effects on air quality? Sci Total Environ. 2020;732:139280. DOI: 10.1016/j.scitotenv.2020.139280. [PMID:32402928]

[8] MC Collivignarelli, C De Rose, A Abbà, M Baldi, G Bertanza, and R Pedrazzani. Analysis of lockdown for CoViD-19 impact on NO2 in London, Milan and Paris: what lesson can be learnt? Process Saf Environ Prot. 2021;146:952-60. DOI: 10.1016/j.psep.2020.12.029. [PMID:33390670]

[9] S Mahato, S Pal, and KG Ghosh. Effect of lockdown amid COVID-19 pandemic on air quality of the megacity Delhi, India. Sci Total Environ. 2020;730:139086. DOI: 10.1016/j.scitotenv.2020.139086. [PMID:32375105]

[10] AR Toro, F Catalán, FR Urdanivia, JP Rojas, CA Manzano, and R Seguel. Air pollution and COVID-19 lockdown in a large South American city: Santiago Metropolitan Area, Chile. Urban Clim. 2021;36:100803. DOI: 10.1016/j.uclim.2021.100803. [PMID:33614404]

[11] K Chen, M Wang, C Huang, PL Kinney, and PT Anastas. Air pollution reduction and mortality benefit during the COVID-19 outbreak in China. Lancet Planet Health. 2020;4:e210-2. DOI: 10.1016/S2542-5196(20)30107-8. [PMID:32411944]

[12] H Fan, Y Wang, C Zhao, Y Yang, X Yang, Y Sun, and S. Jiang. The Role of Primary Emission and Transboundary Transport in the Air Quality Changes During and After the COVID-19 Lockdown in China. Geophys Res Lett. 2021;48:GL091065. DOI: 10.1029/2020GL091065. [PMID:34230714]

[13] S Sharma, M Zhang, J Anshika Gao, H Zhang, and SH Kota. Effect of restricted emissions during COVID-19 on air quality in India. Sci Total Environ. 2020;728:138878. DOI: 10.1016/j.scitotenv.2020.138878. [PMID:32335409]

[14] Y Sathe, P Gupta, M Bawase, L Lamsal, F Patadia, and S Thipse. Surface and satellite observations of air pollution in India during COVID-19 lockdown: implication to air quality. Sustain Cities Soc. 2021;66:102688. DOI: 10.1016/j.scs.2020.102688. [PMID:33391979]

[15] L Menut, B Bessagnet, G Siour, S Mailler, R Pennel, and A Cholakian. Impact of lockdown measures to combat Covid-19 on air quality over western Europe. Sci Total Environ. 2020;741:140426. DOI: 10.1016/j.scitotenv.2020.140426. [PMID:32593893]

[16] JD Berman and K Ebisu. Changes in U.S. air pollution during the COVID-19 pandemic. Sci Total Environ. 2020;739:139864. DOI: 10.1016/j.scitotenv.2020.139864. [PMID:32512381]

[17] World Health OrganizationAir Pollution. Accessed: 17 January 2022.

[18] European Environmental Agency. 2021. Air quality in Europe – 2021 report. Accessed: 17 January 2022.

[19] AJ Cohen, M Brauer, R Burnett, HR Anderson, J Frostad, and K Estep. Estimates and 25-year trends of the global burden of disease attributable to ambient air pollution: an analysis of data from the Global Burden of Diseases Study 2015. Lancet. 2017;389:1907-18. DOI: 10.1016/S0140-6736(17)30505-6. [PMID:28408086]

[20] H Moshammer, M Poteser, M Kundi, K Lemmerer, L Weitensfelder, and P Wallner. Nitrogen-Dioxide Remains a Valid Air Quality Indicator. Int J Environ Res Public Health. 2020;17:3733 DOI: 10.3390/ijerph17103733. [PMID:32466201]

[21] World Health Organization. Listings of WHO’s response to COVID-19. Accessed: 29 January 2022.

[22] World Health Organization. WHO Global Ambient Air Quality Database (update 2018): Accessed: 5 January 2022.

[23] Google Community Mobility Reports. Accessed date: 5 January 2022.

[24] United Nations Development Programme. Global Human Development Indicators. Accessed date: 5 January 2022.

[25] Holden M. World City Development Index: Global Urban Observatory. In: Michalos A.C. editor Encyclopedia of Quality of Life and Well-Being Research. Dordrecht: Springer, 2014. Pp. 7254-7257.

[26] Härdle W, Simar L. Applied Multivariate Statistical Analysis. Berlin: Springer International Publishing, 2019.

[27] Zar J. Biostatistical analysis. New Jersey: Prentice Hall, 1999.

[28] Instituto Mamiraua. BioEstat 5.3. 2007. Available: Accessed date: 5 January 2022.

[29] RB D’Agostino, A Belanger, and RB D’Agostino. A suggestion for using powerful and informative tests of normality. Am Stat. 1990;44:316-21.

[30] Krey V, Masera O, Blanford G, Bruckner T, Cooke R, Fisher-Vanden K, et al. 2014: Annex II: Metrics & Methodology. In: Climate Change 2014: Mitigation of Climate Change. Contribution of Working Group III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change [Edenhofer, O, Pichs-Madruga, R, Sokona, Y, Farahani, E, Kadner, S, Seyboth, K, et al. Eds.]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA. 2014.

[31] AM Graham, KJ Pringle, RJ Pope, SR Arnold, L. Conibear, and H Burns. Impact of the 2019/2020 Australian megafires on air quality and health. GeoHealth. 2021;5:GH000454. DOI: 10.1029/2021GH000454. [PMID:34723045]

[32] N Oliver, B Lepri, H Sterly, S Deletaille, R Lambiotte, and S Deletaille. Mobile phone data for informing public health actions across the COVID-19 pandemic life cycle. Sci Adv. 2020;6:eabc0764. DOI: 10.1126/sciadv.abc0764. [PMID:32548274]

[33] A Wang, A Zhang, EHW Chan, W Shi, X Zhou, and Z Liu. A Review of Human Mobility Research Based on Big Data and Its Implication for Smart City Development. ISPRS Int J Geoinf. 2021;10:13 DOI: 10.3390/ijgi10010013

[34] The Guardian. California’s wildfire hell: how 2020 became the state’s worst ever fire season. Accessed: 12 January 2022.

[35] L Wang, M Li, S Yu, X Chen, Z Li, and Y Zhang. Unexpected rise of ozone in urban and rural areas, and sulfur dioxide in rural areas during the coronavirus city lockdown in Hangzhou, China: implications for air quality. Environ Chem Lett. 2020;18:1713-23. DOI: 10.1007/s10311-020-01028-3. [PMID:32837481]

[36] X Shi and GP Brasseur. The response in air quality to the reduction of Chinese economic activities during the COVID-19 outbreak. Geophys Res Lett. 2020;47:GL088070. DOI: 10.1029/2020GL088070. [PMID:32836516]

[37] European Environmental Agency. European city air quality viewer. Accessed: 12 January 2022.

[38] E Vega, A Namdeo, L Bramwell, Y Miquelajauregui, CG Resendiz-Martinez, and M Jaimes-Palomera. Changes in air quality in Mexico City, London and Delhi in response to various stages and levels of lockdowns and easing of restrictions during COVID-19 pandemic. Environ Pollut. 2021;285:117664. DOI: 10.1016/j.envpol.2021.117664. [PMID:34380230]

[39] RS Sokhi, V Singh, X Querol, S Finardi, AC Targino, and MF Andrade. A global observational analysis to understand changes in air quality during exceptionally low anthropogenic emission conditions. Environ Int. 2021;157:106818. DOI: 10.1016/j.envint.2021.106818. [PMID:34425482]

[40] C Gao, S Li, M Liu, F Zhang, V Achal, and Y Tu. Impact of the COVID-19 pandemic on air pollution in Chinese megacities from the perspective of traffic volume and meteorological factors. Sci Total Environ. 2021;773:145545. DOI: 10.1016/j.scitotenv.2021.145545. [PMID:33940731]

[41] A Benchrif, A Wheida, M Tahri, RM Shubbar, and B Biswas. Air quality during three covid-19 lockdown phases: AQI, PM2.5 and NO2 assessment in cities with more than 1 million inhabitants. Sustain Cities Soc. 2021;74:103170. DOI: 10.1016/j.scs.2021.103170. [PMID:34290956]

[42] F Fu, KL Purvis-Roberts, and B Williams. Impact of the COVID-19 Pandemic Lockdown on Air Pollution in 20 Major Cities around the World. Atmosphere. 2020;11:1189 DOI: 10.3390/atmos11111189

[43] FJ Kelly and T Zhu. Transport solutions for cleaner air. Science. 2016;352:934-6. DOI: 10.1126/science.aaf3420. [PMID:27199415]

[44] L Rosa, F Silva, and C Analide. Mobile Networks and Internet of Things Infrastructures to Characterize Smart Human Mobility. Smart Cities. 2021;4:894-918. DOI: 10.3390/smartcities4020046

[45] ZS Venter, K Aunan, S Chowdhury, and J Lelieveld. COVID-19 lockdowns cause global air pollution declines. Proc Natl Acad Sci U S A. 2020;117:18984-90. DOI: 10.1073/pnas.2006853117. [PMID:32723816]

[46] World Health Organization. Ambient (outdoor) air pollution. Accessed: 12 January 2022.

[47] H Carvalho. New WHO global air quality guidelines: more pressure on nations to reduce air pollution levels. Lancet Planet Health. 2021;5:e760-1. DOI: 10.1016/S2542-5196(21)00287-4. [PMID:34774116]

Correspondence to:
Prof. Helotonio Carvalho, PhD
Department of Biophysics and Radiobiology
Biological Sciences Centre
Federal University of Pernambuco
[email protected]