COVID-19 Lockdown: Impact on PM 10 and PM 2.5 in Six Megacities in the World Assessed Using NASA’s MERRA-2 Reanalysis

The changes in air quality were investigated in six megacities during the shutdown phases in 2020 and were compared to the same time periods in the previous 10 years (2010 - 2019) using the data of Modern-Era Retrospective Analysis and Research and Application, version 2 (MERRA-2). The concentrations of PM 10 and PM 2.5 were greatly reduced in all megacities during the lockdown in 2020 when compared to the same period in 2019 and in the previous ten years. The highest reduction in PM 10 was recorded in Delhi, and São Paulo (21%, and 15% and by 27%, and 9%), when compared with the concentrations in 2019 and in the period 2010 - 2019, respectively. Similarly, levels of PM 2.5 in Delhi, São Paulo, Beijing, and Mumbai decreased by 20%, 14%, 12%, and 10%, respectively in 2020 when compared to the last ten years. Results indicated that the lockdown is an effective mitigation measure to improve air quality. The MERRA-2 reanalysis dataset could be a vital tool in air quality studies in places with a lack of In-situ observations.


INTRODUCTION
A massive number of viral infection cases were reported at the beginning of December 2019 in Wuhan city of China (Lu et al., 2020), which was identified as a coronavirus on December 31, 2019 (WHO, 2020a).The virus kept on spreading across the whole world covering all countries.On February 2020 WHO declared the disease as Pandemic and named it COVID-19 (Bashir et al., 2020;Bera et al., 2020;Shi and Brasseur, 2020;WHO, 2020a, b, c).WHO announced some measures to contain the COVID-19 pandemic such as isolation, home quarantine, social distancing, restriction of transportation (domestic national, and international), and the lockdown (Rohrer et al., 2020).All countries around the globe enforced lockdown to contain COVID-19 pandemic, which resulted in improving of air quality, as a consequence of reducing local pollutant emissions, particularly in the highly populated cities (Ismail et al., 2021;Luna et al., 2018).
Chronic exposure to atmospheric contamination may represent an encouraging context for the spread of the new virus and there is a significant correlation between air-quality data and infected cases of the COVID-19 (Fattorini and Regoli, 2020).Bauwens et al. (2020) stated that lockdown and reductions in human activities positively affected the environment in China and Western Europe.These findings were supported by similar studies in India (Chhikara and Kumar, 2021;Gupta et al., 2020), the USA (Bashir et al., 2020) and, recently, in Egypt (Abou El-Magd andZanaty, 2021;El-Sheekh and Hassan, 2021).
The current study aims to assess the effects of reduced activity resulting from the spread of the COVID-19 on PM 10 and PM 2.5 concentrations in six major cities around the world, namely: Cairo, Delhi, Mumbai, São Paulo, Beijing, and Tehran, all of which are major epicenters of the outbreak.To achieve this goal, we used time series of aerosol concentrations extracted from Modern-Era Retrospective Analysis, Research and Application, Version 2 (MERRA-2) to see how the particulate matter concentrations in these cities vary during the year using the long time series available from MERRA2.In addition to determining how the reduction of the anthropogenic activities because of the lockdown has been captured by MERRA-2.

1 Study Megacities
Cairo, the capital of Egypt, (30°2′N, 31°14′E) is a large metropolitan city with a population of about 21 million (Table 1).It has a high level of air pollution due to the rapid increase in urbanization and industrialization (Abou El-Magd and Zanaty, 2021;Abou El-Magd et al., 2020;Mostafa et al., 2018).On 14 February 2020, the Egyptian government announced the first cases of Corona in the country, and by 19 March 2020, several protective steps were announced, including a partial curfew from 7 PM until 6 AM all over Egypt, along with reducing the number of employees.The suspension of studies in all governorates and the closure of places of worship, public parks, cafes, restaurants, shopping centres, and entertainment places, in addition to the complete closure on the Easter holidays on April 20, and Eid al-Fitr (May 23-25, 2020).
Delhi is the capital of India and is located in the north of India (28°36′36″N, 77°13′48″E), with a population of about 31 million.Mumbai is the financial, commercial, and entertainment capital of India and it lies on the west coast of India (19°07′60″N, 72°87′77″E), with a population exceeded 20 million (Table 1).(https://worldpopulationreview.com/world-cities/).The main sources of air pollution in India are vehicle exhaust, road dust, waste burning, industrial activities, oil combustion, and coal production (Chhikara and Kumar, 2021;Gupta et al., 2020).Indian government enforced the full lockdown from March 25 to April 14, 2020 as the COVID-19 was expanded vigorously (Dhaka et al., 2020).
Beijing is the capital of China and lies in the north region of China (39°90′42″N, 116°40′74″E), with a population of about 22 million.The coronavirus was first broke out in Wuhan in December 2019, therefore, the Chinese government had taken unprecedentedly stringent steps around the country to contain the spread of the pandemic.China has cut off Wuhan from other areas on January 23, 2020 (Pei et al., 2020).Industry and manufacturing remained very limited in Beijing until March, and large-scale production and transportation activities did not resume until late April.Tehran is Iran's capital and most overcrowded city in Iran (35°68′92″N, 51°38′90″E), its population is about 10 Million.Tehran has faced high levels of PM 10 and PM 2.5 due to several industrial activities and the transportation of over 4 million vehicles in the city (Hosseini and Shahbazi, 2016).The first novel case of coronavirus was announced in mid-February 2020 in Iran, which has been accompanied by reinforcing the lockdown in many cities in Iran, including Tehran, to contain the COVID-19 (Faridi et al., 2020).
São Paulo is located 60 kilometers from Brazil's southeast coast (23°55′05″S, 46°63′33″W) at approximately 800 m above the sea level (Molina et al., 2004), and it is the largest metropolitan area in Brazil with 23 million, which constitutes about 20% of the total population of Brazil (Doraiswamy et al., 2017).The first case of COVID-19 was recorded on March 24, 2020 in São Paulo; therefore, the state government ordered a partial lockdown.Air quality in São Paulo metropolitan region and São Paulo city was more influenced by industrial sources (Nakada and Urban, 2020).
Table 1 contains the latitudes and longitudes of the six megacities and the selected months of the lockdown in every city.

2 MERRA-2 Product
Modern-Era Retrospective Analysis and Research and Application, version 2 (MERRA-2) is produced by the NASA Global Modelling and Assimilation Office (GMAO), and it is the first satellite-era global atmospheric reanalysis to assimilate space-based observations of aerosols and represent their interactions with other physical processes in the climate change studies because of its continuous spatial and temporal high-quality resolutions (Delgado-Bona et al., 2020;Bosilovich et al., 2019;Randles and da Silva, 2017), in addition to its usage in many studies related to the particulate matter in different regions in the world (Ma et al., 2021;Raga et al., 2021;Navinya et al., 2020;He et al., 2019;Buchard et al., 2017).
MERRA-2 is a cross-track scanning instrument that provides regular global coverage of measurements.
Where SS2.5 and DU2.5 are the sea salt and dust particles with a size of less than 2.5 µm, and the factor of 1.8 accounts for the conversion of organic carbon into organic matter.The sulphate concentration used for these calculations is assumed to be primarily present in the form of neutralized ammonium sulphate since the GOCART tracer is the mass of the sulphate ion, therefore, it was multiplied by a factor of 1.375 (Ukhov et al., 2020).Regression analysis was used to measure the strength of the relationship between MERRA-2 and surface observation of PM 10 and PM 2.5 .

1 Monthly Means of PM during the
Period 2010-2019 Fig. 1 shows the monthly means of PM 10 and PM 2.5 concentrations over the six megacities during the period 2010-2019, extracted from MERRA-2.This figure can illustrate how MERRA-2 can represent the temporal variations of the particulate matter concentrations, which resulted from various sources and are affected by different meteorological factors.

1. 1 Cairo
The highest concentrations of PM 10 over Cairo, based on MERRA-2, were recorded in January and February 2020 with values exceeding 160 µg m -3 followed by levels that were recorded during March and December with approximately 140 µg m -3 (Fig. 1a).On the other hand, the lowest concentrations were observed during the summer season ( June to August) with values ranging from 60 to 95 µg m -3 , which are consistent with the results of Mostafa et al. (2018).The monthly distribution of PM 2.5 followed the same pattern as PM 10 , the highest concentrations of PM 2.5 (40 µg m -3 ) were recorded during winter (December to February), while the lowest concentrations were recorded during the summer of the same period with values lower than 40 µg m -3 (Fig. 1a).
The increase in particulate concentrations during the winter months is attributed to the state of atmospheric stability and the temperature inversions that increase during the winter season (December, January, and February) in Cairo, especially during the night periods (Abou El-Magd and Zanaty, 2021; Mostafa et al., 2018).The other peaks in the Spring months (March, April, and May) are due to the frequent dust storms that occur in this season because of the Khamasin depressions, which are associated with strong hot and dry winds carrying with dust and sand that increase PM 10 and PM 2.5 concentrations over many regions in Egypt (Abou El-Magd et al. 2016;Incecik and Im, 2012).

1. 2 Tehran
Fig. 1b shows the monthly distributions of PM 10 and PM 2.5 over Tehran.The highest concentrations of PM 10 were observed from April to July (80-85 µg m -3 ), while, the lower concentrations were observed in December and January with (30-40 µg m -3 ).The monthly distribution of PM 2.5 has the same behavior as PM 10 , where April-July had high levels (more than 25 µg m -3 ), while, November to January had lower levels (14-20 µg m -3 ).Yousefan et al. (2020) attributed the high concentrations of particulate matter recorded during May to July months to the dust storms that occur during these months which are responsible for the peak concentrations of PM during summer in Tehran.Recently, Ravindra et al. (2022) stated that the natural emission due to dust storms during the summers causes a significant increase the concentrations of particulate matter.

1. 3 Beijing
The highest concentrations of PM 10 and PM 2.5 over Beijing are close to each other during the seasons of the year (Fig. 1c).The highest PM 10 and PM 2.5 concentrations occur during the months of the summer and then fall with values ranging between 60-65 µg m -3 , and 53-58 µg m -3 , respectively.When temperatures and humidity rise, winds contribute to smog during the transport of pollutants from southern industrial areas in a warm temperate (Molina et al., 2004).Whereas, the lower concentrations of PM 10 levels were recorded during the winter months with a value of 50 µg m -3 , and The lower concentrations of PM 2.5 were recorded during January-April with concentrations ranging between 40-44 µg m -3 .These results are in agreement with the results recorded in China (Bauwens et al., 2020).

1. 4 Delhi
Reanalysis of MERRA-2 data showed that the highest concentrations of PM 10 over Delhi were recorded from May to July (150-200 µg m -3 ) followed by levels that recorded during October to December (120-150 µg m -3 ) (Fig. 1d).On the other hand, the lower concentrations were observed during February, March, and September (70-90 µg m -3 ).The highest concentrations of PM 2.5 were recorded in November and December (106 µg m -3 and 90 µg m -3 , respectively).On the other hand, the lower concentrations were observed to occur in the summer months around 50 µg m -3 (Fig. 1d).The high levels of PM 10 and PM 2.5 during the summer months are because of the strong dust storms carrying large amounts of windblown dust from the desert and after this period of the dust storms the rainfall is accompanied by the monsoon season from July to September which reduces levels of PM 2.5 (Molina et al., 2004).However, in the winter season, the strong surface inversions and fog cases over Delhi decrease the ability of the atmosphere to dilute the high polluted emissions (Chhikara and Kumar, 2021).

1. 5 Mumbai
Fig. 1e shows that the highest concentrations of PM 10 over Mumbai occurred during the summer ( June and July) with concentrations of 135 and 145 µg m -3 , respectively.There was a drastic reduction, in August, where levels reached lower than 90 µg m -3 .Moreover, the lower concentrations were observed during winter (December and January) with an average of about 55 µg m -3 .The highest concentrations of PM 2.5 were recorded during November and December to more than 40 µg m -3 , while the lower concentrations were observed during in August (about 20 µg m -3 ).The reduction of PM 10 and PM 2.5 in August could be due to the rainfall associated with the monsoon (Chhikara and Kumar, 2020;Gupta et al., 2020).

1. 6 São Paulo
Fig. 1f shows that the highest concentrations of PM 10  over São Paulo, were observed during the months from August to October with values ranging between 17-22 µg m -3 , which are accompanied by the season of lower precipitation (Nakada and Urban, 2020).On the other hand, the lower concentrations were observed during April and May (about 10 µg m -3 ).The highest PM 2.5 concentrations were happened during September (15 µg m -3 ), while, the lower concentrations were observed during March-May (less than 10 µg m -3 ).

2 Changes in PM 10 and PM 2.5 Levels due to COVID-19
The changes in the monthly mean of PM 10 and PM 2.5 concentrations during COVID-19 lockdown over the megacities during January-June 2020, based on MERRA-2, are illustrated in Fig. 2. In addition, Fig. 3 shows the changes that occurred in levels of PM 10 and PM 2.5 during the lockdown period compared to the same period in the previous 10 years (2010-2019).Tables 2 and 3 summarize the results of the mean concentrations of PM 10 and PM 2.5 , respectively, during the lockdown period and with their percentage changes compared to 2019 and the average of 2010-2019.

2. 1 Cairo
Fig. 2a shows that the monthly average concentrations of PM 10 in January, February, and March 2020 were 109, 96 µg m -3 , and 160 µg m -3 , respectively.The increase in concentrations in March could be ascribed to the frequent dust events that occurred during that month.The monthly average concentration was decreased from April to June, with the lowest concentrations recorded in May 2020 (83.2 µg m -3 ), when the full lockdown applied for some consecutive days.Similarly, the monthly mean of PM 2.5 (the red line) was increased in March 2020 to approximately 45 µg m -3 , and then decreased after that reaching 25.8 µg m -3 in June by the end of the partial lockdown.
Table 2 shows that the monthly average of PM 10 during the period from the beginning of March to the end of June 2020 in Cairo was about 109 µg m -3 , which is lower than the concentrations recorded in 2019 and the past ten years by about 7% and 8%, respectively.The average PM 2.5 concentration was around 32 µg m -3 during the same months, which is lower than the concentrations for 2019 and 2010-2019 by about 0.1% and 7%, respectively (Table 3).The reductions in the concentrations of PM 10 and PM 2.5 over Cairo compared to 2010-2019 are illustrated in Fig. 3a (the top and bottom panels, respectively).

2. 2 Tehran
The average concentrations of PM 10 over Tehran were 30 and 90 µg m -3 , and that of PM 2.5 was 10 µg m -3 and 30 µg m -3 in January April, respectively (Fig. 2b).The lockdown in Tehran was implemented from the beginning of January to the end of February 2020.Surprisingly, due to that time, there was a slight, but insignificant, increase (2.5%) in levels PM 10 compared to 2019, however, there was a slight reduction (4.5%) in the concentrations of PM 2.5 compared with the same period in 2019 (Tables 2  and 3).Moreover, the reductions in the levels PM 10 and PM 2.5 were 5% and 1%, respectively, when compared to the previous ten years (Fig. 3).Surprisingly, changes in PM concentrations in Tehran were substantially smaller than the other cities.These marginal variations in PM 2.5 and PM 10 during lockdown could be attributed to relaxation given to government offices to operate.Moreover, the lockdown was enforced for few weeks, only, in contrast to other cities which enforced the lockdown for months.Therefore, the difference was less remarkable in Tehran, than that observed in cities with the full lockdown (Lam et al., 2022).
The lockdown in Beijing was implemented from the beginning of February to the end of April 2020 (Tables 2  and 3).The average concentrations of PM 10 and PM 2.5 during these months are lower than the levels observed in 2019 by less than 2% (Table 3).Moreover, the concentrations observed in 2020 were lower than those recorded in the previous 10 years (2010-2019) by 15% and 12%, respectively (Fig. 3).
In 2020, the overall economic situation, and social and  economic (e.g. transportation and industrial production) activities were fully suspended in Beijing to control the COVID-19 epidemic, leading to a rapid reduction in the emission of air pollutants from vehicle exhaust and industrial production.During the epidemic control period, the significant decrease in human activity, especially the sharp reduction in traffic emissions (more than a 70% reduction compared with the same period in former years), led to a reduction in primary emissions of particulate matter (Yin et al., 2021).On the other hand, pollutant discharge levels in the Beijing from 2013-2019 were higher than in 2020 (Guo et al., 2021).Air quality is mainly affected by air pollution emissions and meteorological conditions.China began to significantly promote actions by which to control air pollution since 2013 (Guo et al., 2021).With the implementation of the clean air policy in the Beijing, significant declines in the concentrations of PM 2.5 and PM 10 occurred from 2013 to 2019.This could explain why the change between 2020 and 2019 were substantially smaller than that between 2020 and 2010-2019 (Delgado-Bonal et al., 2020).Another explanation could be the increase in frequency of dusty weather where several dusty weather events occurred from between 2010-2019 exceeded the total of the same period in 2020 (Yin et al., 2021).Moreover, the Spring Festival was suspended in 2019 (to the implementation of the clean air policy in Beijing) and 2020 (due to the lockdown), while it was carried out in the previous years (i.e.2020-2018).These results indicate the air quality during 2020 was obviously improved compared with the same period between 2010-2019.

4 Delhi
The average concentrations of PM 10 were 76.42 µg m -3 and 77.53 µg m -3 , while those of PM 2.5 were 52.5 µg m -3 , 56.85 µg m -3 in January and February 2020, respectively (Fig. 2d).These concentrations were significantly reduced after enforcement of the lockdown during March and April; where the average concentrations of PM 10 were 72.09 µg m -3 and 74.76 µg m -3 in March and April 2020.Similarly, average concentrations of PM 2.5 were 37.09 µg m -3 , and 35.58 µg m -3 in these two months, respectively (Fig. 2d).
As illustrated in Tables 2 and 3, the concentrations of PM 10 and PM 2.5 that were observed during the lockdown in 2020 were reduced by about 21% and 14%, respectively, compared to the concentrations recorded in the same period in 2019, and by about 27%, and 20%, respectively, compared to the average duration of the last ten years, respectively, as shown in Fig. 3.

5 Mumbai
Fig. 2e shows the average concentration of PM 10 over Mumbai during January and February were 53 µg m -3 for both months, and during March and April (lockdown months) they were 77 µg m -3 and 96 µg m -3 , respectively.The average concentration of PM 2.5 during January and February was around 38 µg m -3 for both months, and were 32.5 µg m -3 and 35.2 µg m -3 in March and April, respectively.
Tables 2 and 3 illustrate that levels of PM 10 and PM 2.5 over Mumbai, during the lockdown (March-April 2020), were reduced by about 14% and 9%, respectively when compared with the concentrations recorded in the same period in the previous year (2019), and by about 11% and 10% as compared to the average concentrations recorded during the same period in the previous ten years , respectively (Fig. 3).
Table 4 shows the comparison between PM 10 observations at two location: the Egyptian Meteorological Authority in Cairo (Egypt) and Sherif University in Tehran (Iran) with the values extracted from MERRA-2 at the same locations during the period of 2012-2014.
The statistical analysis of the measurements of PM 10 concentration with the values extracted from MERRA-2 during 2012-2014 shows that the root mean square error (RMSE) was 122 and 69 µg m -3 for Cairo and Tehran, respectively, indicative of the reliability of the retrieved results.Also, there was a high correlation between  MERRA-2 and PM 10 measurements at Cairo 0.82 with a mean bias of -3.1 µg m -3 , and the absolute percentage error of -5.6%, showing the applicability of the MERA-2 The results of Tehran were near to those of Cairo since the correlation is about 0.6 and the mean bias is -4.9 µg m -3 .Fig. 4 shows the scatter plots and linear regressions between each reanalysis datasets obtained from MERRA-2 versus ground-based observations of PM 10 and PM 2.5 for Cairo during 2018-2020.The results showed significant spatial agreement between MERRA-2 and groundbased observations (R 2 = 069 and 0.86 for PM 10 and PM 2.5 , respectively, P<0.05).These results confirm the results of Table 3, and could depict that there is an agreement between MERRA-2 and ground-based observations on the global scale with a reasonable degree of accuracy (Lam et al., 2022;Navinya et al., 2020).
Generally, Spatiotemporal analysis of the variations in the levels of criteria air pollutants revealed that there was a significant decline in the concentration of PM 10 and PM 2.5 in the megacities under examination.However, the magnitude of the decline in air pollutant concentrations is different in all the megacities and this is in agreement with the results of Jain and Sharma (2020).However, when compared with the WHO standards (for PM 2.5 = 25 µg m -3 , PM 10 = 50 µg m -3 based on a 24-hours average), Tehran and São Paulo are within the permissible limits for PM 2.5 ; whereas all the megacities violate WHO standards.Similarly, when compared in terms of PM 10 levels, Cairo violates WHO standards, while other megacities are within the permissible limits.
In summary, during and after the COVID-19 pandemic, there should be a concern for environmental protection and preservation.There could be an increased population movement for environmental actions such as clean air and water (Ravindra et al., 2022).Furthermore, Hassan et al. (2022) also linked COVID-19 pandemic with positive impact (clean beaches) and negative aspects (e.g. increase in household waste, reduced recycling).

CONCLUSION
In conclusion, the impact of the short-term controlled measures empowered by different governments due to the COVID-19 pandemic helped in improving air quality over the megacities, which was proved by high-resolution reanalysis data.However, the reduction in emissions of particulate matter was not significant in all cases as the lockdown was partial.Moreover, the spatial dispersion of particulate matter varied with the geographical location and the time of the year as well as meteorological parameters, and/or long-range transport of pollutants.Furthermore, these reductions are evident not only from the comparison of PM concentrations before and during the lockdown but also when comparing levels of PM observed in 2020 with those during the same period in 2019 and in the previous 10 years.Through this study, the MERRA-2 reanalysis dataset has captured the changes in PM 10 and PM 2.5 during the short period of lockdown in different cities around the world, and thus it could be a vital tool in air quality studies in places with a lack of Insitu observations.Therefore, it is concluded that satelliteretrieved aerosol maps provide the possibility and capability of monitoring and characterizing spatiotemporal distribution of the surface PM with a reasonable degree of accuracy.
This decline in the concentration levels of PM and con- sequently improvement in air quality in the examined megacities could be attributed to the complete travel lockdown and suspension of all the non-essential travel activities in all the cities.Moreover, meteorology also played an essential role in the dispersion of air pollutants in all the megacities.A potential challenge is the degradation in air quality, probably to higher levels than the business-as-usual scenario attributable to increased usage of private modes of transportation in comparison to public transport due to the perception about the public transport in terms of lack of safety due to COVID-19 infection and less management by the agencies in terms of proper sanitization.This problem could be easily solved by implementation of 'odd-even' strategy for work offices and private vehicles, a quick transition from fossil fuelsbased energy system to renewable and cleaner energy alternatives, plantation of trees with high surface area to volume ratio in the pollution hotspot areas, and regulating construction and demolition activities in urban areas.
This study is beneficial to the policymakers, epidemiologists, and environmentalists to access and analyze the effect of various factors on improving air quality so that future infrastructure and strategy can the planned accordingly.Integrated planning that caters to the environment and economy together came out as an essential suggestion from the experts, and they also highlighted to align all the policy decisions in line with the Sustainable Development Goals (SDGs).Modeling is urgently needed to identify the health measures and other contributing factors.In summary, the present study highlighted the importance of reductions in emissions of PM in improving air quality and this was clear during the lockdown, throughout the world.Moreover, as a blessing in disguise, it indicates that the environment is self-healing during the lockdown.

Fig. 3 .
Fig. 3. Relative changes of PM 10 (top panel) and PM 2.5 (bottom panel) occurred during the lockdown of each city in unit (%), with respect to the last 10 years (2010-2019) extracted from MERRA-2.

Fig. 4 .
Fig. 4. Regression analysis between MERRA-2 dataset and surface observation of PM 10 (Top panel) and PM 2.5 (lower panel).The small squares indicate the monthly mean of PM (2018-2020).The solid line is the line of best fit.

Table 1 .
The studied megacities, their countries and their latitudes and longitudes that used to extract the MERRA-2 data.

Table 4 .
Statistical comparison for PM 10 observations between ground stations and MERRA-2 for Cairo (Egypt) and Tehran (Iran) during the period of 2012-2014.RMSE means root mean square error.