During the first period of measurement, prior to the implementation of the COVID-19-restrictive measures, the recorded PM data at the three study sites were in accordance with our classification; the mean PM2.5 and PM10 concentrations were the highest in MGT, which was the most crowded area in terms of both vehicular and pedestrian traffic, while they were the lowest in TKT, the least crowded of the test areas (Fig. 3). However, this ordering of PM concentrations in the three areas was lost during the period of implementation of the restrictive measures, and even reversed in MGT, the site at which the lowest PM concentrations among the three sites were recorded during the second and fourth periods of measurement. Some crowded places or densely polluted metropolitan cities with higher anthropogenic emissions of PM due to high traffic volumes and crowded local human activities showed a greater degree of reduction of PM concentrations when COVID-19-restrictive measures were in force than less urbanized cities or quiet places [30, 31].
The finding of a significant decrease in PM concentrations during the period in which COVID-19-restrictive measures were in effect is consistent with many recent reports worldwide [13,14,15,16,17] and restricted vehicular traffic movement and temporary closure of universities, restaurants, food shops, factories, and industries are common possible explanations for such a finding. On the contrary, only a slight reduction or even an unexpected increase in PM concentrations during lockdown periods has also been reported [18,19,20,21]. These discrepant results could be due to differences in the types of restrictive measures implemented, such as total lockdown, community lockdown, large-scale social distancing and movement control order, differences in the enforcement periods, local meteorological conditions, and the intensity of pre-existing anthropogenic emissions at the respective study sites.
In our study, a noticeable percent reduction (above 70%) was observed at the study sites during the second and third periods of measurement, as compared to the fourth period. Notably, in MGT, a crowded semi-commercial area showed a consistent degree of reduction in PM2.5 in the second period (82.9%) and third period (84.8%) of measurement, which could have been a result of closure of almost all shops and stores at these sites and prohibition of private vehicular movement on the streets, except for those of house owners in the area during these periods. We observed inconsistent changes of the PM concentrations among the three measurement periods during which COVID-19-restrictive measures were in place, and this observation was in line with some previous reports [18, 22].
While some researchers compared the PM data measured during the lockdown periods with data obtained during a period preceding the lockdown in 2020 [32], many investigators compared the PM data obtained while COVID-19-restrictive measures were in effect with those measured at the same time in the previous one or more years [18, 33, 34], and the comparisons were done in terms of percent changes of PM concentrations. It is conceivable that the variations in meteorological parameters during the same period over different years would be relatively less pronounced than the variations between the periods before and during the lockdown within the same year [33].
In Yangon, Stay-At-Home order, Work-From-Home order, and other restrictive measures, including closure of universities, factories, and restaurants allowing take-away service only, were implemented from the second week of April 2020. According to the Road Transportation Administration Department (RTAD), in 2020, there were about 540,000 registered vehicles in Yangon, among which 350,000 automobiles, that is, over two-thirds, were private cars and about 41,000 vehicles were public transport vehicles [27]. Because of the Stay-At-Home and Work-From-Home orders, there could have been a dramatic reduction in the number of both private and public transport vehicles on the road, resulting in a recognizable reduction of the traffic volume.
The percent change in PM concentrations in the second period, that is, May 10 to May 17, were consistently found to be greatly reduced, regardless of the type of comparison (Table 2 and Fig. 4). This period in May is a period of transition from the summer to the rainy season and the percent changes in the PM concentrations were still high even when the comparison was made with the data obtained in the corresponding period of 2019, with similar meteorological conditions. Therefore, such an obvious degree of reduction in this second period of measurement could be directly attributable to the strict obligation of the citizens to conform to the restrictive measures during first wave of COVID-19, resulting in a dramatic reduction in the anthropogenic emissions of PM.
During the third period of measurement, that is, August 31 to September 6, as described earlier, the extent of reduction of the absolute PM concentrations was even more pronounced (Fig. 3). Moreover, the percent reduction was also high when the comparison was made using the data obtained during the first period, before the onset of the COVID-19 pandemic, as reference (Fig. 4). However, the percent changes in this period relative to the data obtained during the corresponding period of 2019 showed only minimal reduction or even a percent increase (Table 2). Another interesting finding was that when the percent reductions among the five data sources, namely PurpleAir (2019), PurpleAir (2020), MGT, SOK, and TKT, were compared, the values were almost equal for both PM10 and PM2.5 concentrations (Fig. 4). This finding suggests that in 2019, even in the absence of COVID-19-restrictive measures, an obvious percent reduction occurred during the period from August 31 to September 6. In fact, the third period fell during the rainy season in Yangon, when precipitation of PM by rain could occur. Consequently, in the third period of measurement, when COVID-19-restrictive measures were in force, seasonal influence could have been the predominant factor contributing to the reduction in ambient PM pollution, although reduced anthropogenic emissions due to the restrictive measures could also have contributed to the reduction.
The fourth period of measurement, namely December 7 to December 14, fell within the winter season. In contrast to the case during the rainy season, the temperature inversion phenomenon, a favorable condition for ambient PM concentrations, commonly occurs during the winter season. Moreover, burning of dry leaves during the winter season could also be a possible additional source of PM. Such seasonal factors could have been responsible for the lower percent reductions in the fourth period as compared to the two preceding periods of measurement (Fig. 4). However, the percent reductions could not be greatly determined by seasonal pollutant dispersion, because unlike in the third period, there was an inconsistency in the percent reductions in the fourth period; the percent reductions of PM2.5 and PM10 in TKT were lower than those in PurpleAir (2019), whereas those in the remaining data sources were greater (Table 2 and Fig. 4). A renovation project at the Thanlyin Bridge, about 7–8 km away from the TKT site, resumed in November and this emission source could be a possible reason for the highest PM concentration and lowest percent reduction in the quiet residential quiet area. During the second wave of COVID-19, although the COVID-19-restrictive measures were enforced again, many factories and construction sites, restaurants, and food shops resumed their business after being cleared to do so according to the guidelines set by the MOHS. Although prohibition of mass gatherings was still in effect, there was a resurgence of human activities. Therefore, the decline in the percent reductions in the fourth period could be attributable, at least in part, to some relaxation of the restrictive measures, with a lower degree of compliance with orders by the citizens and in part, by the presence of weather conditions that favor PM dispersion.
Some studies have also taken into consideration seasonal variations while describing the changes in PM concentrations during the COVID-19 pandemic. In the study reported by Hashim et al. (2020), the investigators compared the average PM concentrations in Baghdad, Iraq, during five periods; the first period before the enforcement of a lockdown, and the remaining four periods during partial or total lockdown. They observed that the PM2.5 and PM10 concentrations were the lowest during the first partial and total lockdowns among the five periods. They speculated the following possible reasons for this finding; the citizens’ compliance with the lockdown measures during that first lockdown period contributed to the large decline of PM concentrations during that period, and the dry hot climate during the summer resulted in the relative increase of PM concentrations in the subsequent lockdown periods [18]. In a report from Thailand [32], the ambient PM concentrations were compared among three measurement periods; pre COVID-19, early COVID-19, and while a work-from-home order was in place. An unexpected increase in the ambient PM concentrations was noted in the early COVID-19 period, during which only personal hygiene measures were encouraged, without other strict restrictive measures, and there was also the seasonal transition from winter to summer.
Meteorological factors, such as the ambient temperature, relative humidity, wind speeds, precipitation, radiation, and ambient pressure could also exert an influence on the ambient PM concentrations [35]. In our study, the ambient temperature and relative humidity were measured and their correlations with the PM concentrations were evaluated. In spite of revealing significant level of P value, only weak correlations (r < 0.25) were noted during all the four periods of three study sites (r = 0.11, P = 0.003 between PM concentration and temperature; r = − 0.06, P = 0.001 between PM concentration and relative humidity). A previous study also showed that during the COVID-19 pandemic, these two meteorological parameters were only weakly correlated with the PM concentrations (r < 0.25), indicating that they contributed little to the ambient PM concentrations [36]. Although we found an obvious difference in the temperature and relative humidity between the COVID-19 pandemic year (2020) and the previous year (2019) (Table 1), the aforementioned correlations were also weak for 2019 data (r = 0.19, P = 0.004 between PM concentration and temperature; r = − 0.1, P = 0.001 between PM concentration and relative humidity). These findings could be attributable to the contributions of other meteorological factors rather than the two aforementioned variables to the ambient PM concentrations.
Yangon city indeed suffers from PM-related air pollution. During the period from January 25 to January 29, 2018, we assessed the ambient PM2.5 and PM10 concentrations in seven townships of the city, and found that the mean PM concentrations were over the WHO guideline limits [23, 24]. The average annual PM2.5 concentration (weighted by the population) in the city in 2019 was 31 μg m−3, exceeding the annual mean PM2.5 exposure threshold of 10 μg m−3 set by the WHO. In regard to the ranking of regional capital cities of the world according to the PM2.5 exposure level, Yangon is placed 19th out of 85 capital cities across the world, and ranks third among cities in the Southeast Asia region [4]. In our study, the daily mean PM2.5 and PM10 concentrations measured in the period prior to the onset of the COVID-19 pandemic were consistently over the WHO-recommended limits (Fig. 5). Conversely, the mean daily PM2.5 and PM10 concentrations were below the set limits on almost all seven days of both the second and third periods of measurement, when COVID-19-restrictive measures were in effect. Therefore, in contrast to the previous years, Yangon city experienced a profound improvement of PM-related air quality in 2020, and this appears to be, in all probability, due to the restrictive measures proposed for COVID-19 containment. Moreover, as compared to previous reports from studies conducted worldwide, the percent reductions in our study were relatively higher. This could be due to the fact that the city, in which no prior proactive air pollution control measures were in place, experienced a rapid and effective restriction of anthropogenic emissions for a very first time during the period of enforcement of COVID-19-restrictive measures (Fig. 6).
The United States Air Quality Index (USAQI) is the most widely used for assessment of the ambient air quality. The average 24-h PM2.5 concentrations are converted into six categories of AQI, where higher values indicate a higher health risk [4]. In our study, the air quality during the first period of measurement prior to the onset of COVID-19 fell into either the unhealthy category (55.5–150.4 μg m−3) or the category of “unhealthy for sensitive group (USG),” such as children, elderly persons, and patients with cardiovascular disease (35.5–55.4 μg m−3). However, improvement in air quality became apparent during the second and third periods of measurement, when COVID-19 lockdown was in place, and the AQI category became moderate (12.2–35.4 μg m−3) or even good (0–12.1 μg m−3), i.e., the air quality became satisfactory and posed little or no risk. Unfortunately, on a few days during the fourth period, the AQI was categorized as USG.
In addition, as fine particles are more harmful than coarse particles, higher PM2.5/PM10 ratios may result in serious air pollution, whereas the lesser the ratio, the lesser the possibility of poor air quality [30]. A higher ratio implies predominant contribution of PM2.5, which is generally ascribed to primary pollution by anthropogenic emissions, while a lower ratio suggests a greater contribution of coarse particles, which mainly arises from natural sources [37]. In a study from South Korea, after the implementation of social distancing, the PM2.5/PM10 ratio decreased from 0.66 to 0.4 in Seoul and 0.68 to 0.54 in Daegu city, and this finding was explained by a decrease in anthropogenic emissions [17]. In our study also, the PM2.5/PM10 ratios declined during the COVID-19 measurement periods as compared to the pre-COVID-19 values; from 0.82 to 0.79 in MGT, from 0.86 to 0.82 in SOK, and from 0.88 to 0.82 in TKT. This finding indicates that when the COVID-19-restrictive measures were in place, the PM-related air quality improved, and more specifically, that reducing human activities favors reduction of the anthropogenic sources of PM2.5 rather than PM10. A study in Wuhan city showed that not only did the mass concentration of PM2.5 reduce, but its chemical composition also altered during the period of enforcement of COVID-19-restrictive measures [38]. Although we only assessed the mass concentrations of PM in this study, we propose to analyze the PM compositions in Yangon cut using a high-volume sampler in the future [39].