Changes in Air Quality during the COVID-19 Lockdown in Singapore and Associations with Human Mobility Trends

Received: June 11, 2020
Revised: July 2, 2020
Accepted: July 16, 2020

 Copyright The Author(s). This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are cited.

Cite this article:

Li, J. and Tartarini, F. (2020). Changes in Air Quality during the COVID-19 Lockdown in Singapore and Associations with Human Mobility Trends. Aerosol Air Qual. Res. 20: 1748–1758. https://doi.org/10.4209/aaqr.2020.06.0303

HIGHLIGHTS

• NO₂ and SO₂ had the highest two reductions of 54% and 52%, respectively.
• PSI, PM₁₀, PM_2.5, NO₂, and SO₂ significantly decreased nationwide while O₃ increased.
• PM_2.5 and NO₂ were significantly correlated with the mobility trends. 

Keywords: Air pollutant; Anthropogenic pollution; Circuit breaker; SARS-CoV-2.

INTRODUCTION

On the 23^rd January 2020, Ministry of Health of Singapore confirmed the first imported case of coronavirus disease 2019 (COVID-19) caused by the infection of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) (Ministry of Health, 2020a; Wong et al., 2020). On the 7^th of February 2020, the Singaporean government raised the Disease Outbreak Response System Condition (DORSCON) to Orange, slowly implementing measures to limit social gatherings and on the 23^rd of March banned all short-term visitors from entering the country. However, due to the increasing rate of transmission of SARS-CoV-2 in the community, as shown in Fig. S1 in the Appendix, on the 7^th of April 2020 Singapore implemented strict safe distancing measures (Ministry of Health, 2020b; Pung et al., 2020). The Singaporean government named these measures “Circuit-Breaker”, however effectively were analogous to what other Countries referred as “lockdown” or “shutdown” measures. In this paper we, therefore, decided to use the term “lockdown” to describe these measures. During the lockdown period schools were closed, all workers working in non-essential sectors were imposed to work from home, and shops were closed with the exclusion of those providing essential services to the community (Lim et al., 2020; Ng, 2020). People were allowed to exit their home only for purchasing essential products or exercising, no social gatherings were allowed. These measures confined the population at home and reduced the vehicular traffic on the roads and emissions from industrial activities which in turn may had affected the concentration of anthropogenic pollutants in the outdoor air (Dutheil et al., 2020; Filonchyk et al., 2020; Le et al., 2020; Muhammad et al., 2020; Safarian et al., 2020; Xu et al., 2020). Lockdown measures in Singapore were implemented on the 7^th of April 2020 and started to progressively ease on the 12^th of May 2020 (Ministry of Health, 2020c). Hence, we labeled these 35 days from 7^th April to 11^th May 2020 as the lockdown period in this study.

Although the lockdown measures are designed to protect people from the disease, these measures have a severe impact on the economy and people’s daily lives. A silver lining under this scenario may come in the form of health benefits resulting from the reduction of air pollution (Lal et al., 2020). The World Health Organization (WHO) defines outdoor air pollution as a major environmental risk to health, which negatively affects cardiovascular and respiratory health. The WHO classifies particulate matter with diameters less or equals to 10 micrometers (PM₁₀) and 2.5 micrometers (PM_2.5), nitrogen dioxide (NO₂), sulfur dioxide (SO₂) and ozone (O₃) as the major outdoor air pollutants (WHO, 2006). In Singapore the main sources of outdoor pollutants are industries and motor vehicles (National Environment Agency, 2020a). Sporadic elevated haze pollution from neighboring countries during the monsoon seasons (Velasco and Rastan, 2015) or caused by El Niño (Atwood et al., 2013) also have an impact on outdoor air quality. In Europe a report (Myllyvirta, 2020) estimated that the lockdown measures have avoided 11,000 pollution-related deaths in April 2020. The avoided NO₂ and PM_2.5 related deaths in China due to the cleaner air were estimated as 8911 and 3214, respectively (Chen et al., 2020a). The reduction of air pollutant has the potential to decrease the mortality rate of COVID-19 (Brandt et al., 2020; Fattorini and Regoli, 2020; Ogen, 2020) and mitigate the spread of the virus (Jiang et al., 2020; Pani et al., 2020; Sciomer et al., 2020; Yao et al., 2020).

Our aim is to quantify the change in outdoor pollutants concentrations during the lockdown period in Singapore and to evaluate their associations with mobility trends. This would allow us to better understand how anthropogenic activities and various sources in Singapore affect the overall air quality.

METHODS

Air Quality Data

The National Environment Agency (NEA) in Singapore provides Application Programming Interfaces (API)s for real-time weather readings and air quality data. The weather data APIs provide 1-minute intervals data of temperature, relative humidity, precipitations and wind conditions for a total of 64 weather stations across Singapore (National Environment Agency, 2020b).

Data on pollutants were obtained from five monitoring stations distributed strategically throughout Singapore (National Environment Agency, 2020c). The Singaporean government refers to these stations as Central (1°21'N 103°49'E), East (1°21'N 103°56'E), West (1°21'N 103°42'E), South (1°18'N, 103°49'E), and North (1°25'N, 103°49'E) according to their locations in Singapore, and the national level is calculated as the main value among these five stations. We obtained outdoor air quality data from April 2016 and May 2020 using Python. The NEA releases on an hourly basis data regarding pollutants mass concentrations for: 24-h mean PM₁₀ and PM_2.5, 1-h mean NO₂, 8-h mean CO, 24-h mean SO₂, and 8-h mean O₃ and their relative sub-indices for each of the above-mentioned stations. In addition, it provides data on the 24-h Pollutant Standard Index (PSI). The NEA calculates the 24-h PSI value firstly by linearly mapping each individual sub-pollutant (i.e., 24-h PM₁₀, 24-h PM_2.5, 8-h CO, 24-h SO₂, 1-h NO₂, 8-hO₃) to a sub-index ranging from 0 to 500. Then the overall 24-h PSI value is assumed to be equal to the maximum value of any of the pollutants sub-indices (National Environment Agency, 2014). More information regarding the accuracy of the instrumentation used by NEA and the location of the stations can be found on the NEA website (MSS, 2018, p. Annex 3; National Environment Agency, 2020d).

To better analyze the spatial temporal variations of NO₂ and SO₂ concentrations in Singapore during the lockdown period, we obtained data from the European Space Agency (ESA) Sentinel-5P TROPOMI satellite (European Space Agency, 2020). We also obtained the total aerosol optical depth (AOD) at 550 nm data from the Copernicus Atmosphere Monitoring Service (CAMS) (Benedetti et al., 2009; Huijnen et al., 2016; Morcrette et al., 2009). Data were processed and plotted using the Earth Engine Data Catalog (Gorelick et al., 2017). The spatial resolution of the Sentinel-5P TROPOMI satellite is 7 by 7 km² and the satellite scans the earth with a daily revisit frequency. However, for same specific days date were either not available or may have been filtered out due to the presence of clouds. Consequently, since we plotted the average data collected over a relative short period of time, from the 7^th of April to the 11^th of May, the mean concentration calculated for some areas may not be accurate due to missing data (Kramer, 2002).

 
Mobility Data

We use the datasets from both Apple and Google to analyze how changes in mobility affected the air quality. Both datasets show a relative trend of how people movements changed within Singapore each day due to the impact of the COVID-19 pandemic. Google’s dataset is built from data collected from users who allowed Google to access their location information. Google’s data are classified into the following macro categories: workplaces, residential, retail & recreation, grocery & pharmacy, transit stations, and parks (Google LLC, 2020). Google uses as baseline the median value for the corresponding day of the week during the period from the 3^rd of Jan to the 6^th of Feb 2020. The data reflects how the lockdown impacted each category. The data in Apple’s dataset is based on the direction requested by the users in Apple Maps, and classified into three categories: driving, public transit, and walking (Apple Inc., 2020). From both datasets, we selected the mobility parameters which have the potential to impact air quality. The selected parameters are workplaces, residential, and transport station in Google’s dataset and the data of driving and public transit in Apple’s dataset.

In addition to the smartphone data we acquired data regarding the Housing and Development Board (HDB) carpark availability using the API provided by the Singaporean Government (HDB, 2020). Approximately 81% of the whole population in Singapore lives in HDB apartments (HDB, 2018; Li et al., 2020), hence this data provides a meaningful estimation of how many Singaporean used their personal vehicle during the lockdown period. We gathered aggregate data on an hourly basis regarding the total number of residential occupied and free car spots across the whole country. We used this information to estimate at any point in time the percentage of people who commuted from and to their houses using a vehicle. From those data we could not infer how many cars were on the streets, since we only knew when the car left the residential parking and when it returned in the evening.

Data Analysis

To estimate the influence of the lockdown on the outdoor air pollutants concentrations, we firstly had to estimate the concentrations without COVID-19 measures. We call this the counterfactual concentrations. For each parameter at each location, we linearly regressed the data collected over the previous four years (2016, 2017, 2018, and 2019) during the period from the 7^th April to the 11^th of May to test whether there was a consistent change of pollutants concentrations over the years. We used two methods to estimate the counterfactual value for 2020. If the p-value of the linear regression is lower than 0.01 we used the value estimated by the linear regression; alternatively, we used the mean value of the last four years. The reduction was defined as the change from the counterfactual value to the measured one during the lockdown.

We have listed below the reasons why we chose to compare the outdoor concentrations to those measured in the same period of the last four years, instead of using the methodology proposed by (Kerimray et al., 2020; Otmani et al., 2020) who compared the lockdown data to a period of time in 2020 preceding the lockdown.

Firstly, we observed that the difference of meteorological parameters in the same period over different years is much smaller than the variation between two consecutive periods before and after the lockdown due to seasonal trends, as shown in Figs. S2–S5 in Appendix. The total rainfall in April is significantly higher than the rainfall in March, as shown in Fig. S4 in Appendix. Meteorological monthly or seasonal trends affect significantly some air pollutants in Singapore within each year. Hence, the difference between the lockdown period and the period of time preceding it could be inherently different and not related to the COVID-19 pandemic. On the contrary, the difference of the air pollutant between the same period in different years are much smaller.

Secondly, after the first confirmed cases in Singapore on 23 January, Singaporean government started to implement measures to contain the local transmission of the virus, such as raising the DORSCON level to orange (7 February) and an advisory about encouraging people work by telecommuting from home. These early measures may have impacted the air quality in Singapore even before the lockdown. Hence, air quality during a period before the lockdown is not suitable to serve as the baseline.

We used the Wilcoxon signed-rank test to compare the difference of the parameter during between different periods. We performed Spearman’s rank correlation to access the correlation between air quality and mobility. We analyzed the data in the R programming language. We processed and transformed the data with data, table, lubridate, and magrittr packages and visualized them with ggplot2 and patchwork packages. We have published the data sets and source codes we used on a public GitHub repository (https://github.com/
JiayuLIAQ/COVID-19_vs_air_quality_weather/), to allow other researchers to reproduce a similar analysis in other cities or countries.

 
RESULTS

The meteorological parameters in Singapore during the period between the 07^th of April and the 11^th of May from 2016 to 2020 are summarized in Table 1. During the lockdown period in 2020, temperature, relative humidity, rainfall, and wind direction were not significantly different from those recorded during the same period in previous years. Wind speed during the lockdown period was significantly higher than the previous years. However, while the difference was statistically significant, it only increased by 0.8 m s^–1. More detailed comparisons can be found in Figs. S2–S5 in the Appendix.

The Reduction of Air Pollutant Due to COVID-19 Lockdown

The national level of daily PSI and mass concentrations of PM₁₀, PM_2.5, NO₂, CO, SO₂, and O₃ measured from 07 April and 11 May for five consecutive years (2016–2020) are plotted using boxplots in Fig. 1. The filled boxes are the interquartile ranges (IQR) for the air quality values in each year. The horizontal line in the box is the median, and the rhombus dot is the arithmetic mean. Whiskers of the boxes start from the upper and lower limits of the box and end at the length of 1.5 times the IQR or at the maximum and minimum values whichever is reached first.

Fig. 1. National air quality parameters for the period starting on the 7^th of April and ending on the 11^th of May, from 2016 to 2020. ^NS., ^*, ^**, and^*** denote the difference between two samples is at the level of no (Wilcoxon p ≥ 0.05), low (0.01 ≤ p < 0.05), medium (0.001 ≤ p < 0.01), and high (p < 0.001) significances, respectively. The inverted triangle is the counterfactual concentration.

During the lockdown period, the PSI was always lower than 50, which is classified by the NEA as “Good” on a 5–point scale with the following categories “Good (PSI: 0–50)”, “Moderate (PSI: 51–100)”, “Unhealthy (PSI: 101–200)”, “Very Unhealthy (PSI: 201–300)” and “Hazardous (PSI: above 300)”. All air pollutant levels during the lockdown period were below the Singapore Targets by 2020 (National Environment Agency, 2011) which are listed in the Table S1 in Appendix. On the other hand, the PSI for the same period of last four years were higher than 50 (“Moderate”) for more than 65% of the time. All the air quality parameters in 2020 during the lockdown period were significantly lower than those in the previous years, with the exception of O₃. Fig. 1 shows that some of the air quality variables (PSI, PM_2.5, CO, SO₂, and O₃) improved over the years, others not (PM₁₀ and NO₂). We used semi-transparent grey lines to depict those values with significantly decreased over the last 4 years (i.e., PSI, PM_2.5, CO, SO₂, and O₃) in Fig. 1. For these parameters, predicted values in the year 2020 are used as the counterfactual concentration and marked as inverted triangles in Fig. 1. We then calculated the reduction by subtracting the average value measured this year from the counterfactual value. We showed these differences as blue arrows in Fig. 1, and summarized the results into Table 2.

During the lockdown period the national PSI decreased by 19% compared to the predicted value for 2020, which suggests that the lockdown due to the COVID-19 pandemic had a positive impact on the reduction of the overall PSI value. The reduction of 29% in the southern area of Singapore was the highest, and 17% in the west and north was the lowest. PM_2.5 dominated PSI in Singapore, which had the third-highest reduction of 29% nationally. The PM_2.5 level decreased by nearly a half in the southern area, which was the highest reduction among the five stations. NO₂ and SO₂had the highest and second-highest reduction regarding percentage change rates, which decreased by 54% and 52% nationally in this year, respectively. The change in PM₁₀, CO, and O₃ concentrations were relatively low when compared to other pollutants. Also, their concentration variations were not consistent among the various locations. Though PM₁₀ and CO were decreasing this year nationally in general, we observed an increase in concentrations for some of the locations. In particular, we observed a nationwide increase of O₃ during the lockdown period, except for the western area. Similarly, increased O₃ concentrations were observed during the lockdown period in several cities around world (Chen et al., 2020b; Collivignarelli et al., 2020; Dantas et al., 2020; Kerimray et al., 2020), which may be explained by a significant reduction of NO₂ concentrations (Sicard et al., 2020).

The different reduction rates among locations could be caused by the spatial distribution of the different pollution sources in Singapore. All the heavy industries, refineries and the harbor in Singapore are located in the south-west part of the island. Singapore is one of biggest harbor in the world. To better understand the spatial variations of outdoor pollutants in the outdoor air we obtained data from the ESA Copernicus Sentinel 5P satellite. Fig. 2 shows the mean value of the NO₂ and SO₂ tropospheric column over Singapore. Satellite images show similar trends to those observed by ground stations and the biggest reductions of NO₂ concentrations were observed in the south and west regions of the island. The biggest reductions of SO₂ concentrations where observed over the island of Pulau Bukom which is an oil and petrochemicals site with manufacturing facilities for fuels, lubricant base oils and specialty chemicals. These results would, therefore, suggest that the COVID-19 and related and concurrent oil crisis may have led to a reduction of petrochemical activities in Singapore. From Fig. 2, it can also be observed that NO₂ concentrations decreased in the east side of Singapore. The reduction could be attributed to the reduced number of flights that landed and departed from Singapore’s airport (located on the east side of the country) and possible due to the decrease in operation of the heavy industry in Malaysia located in Pasir Gudang (north east of Singapore).

Fig. 2. Shows the mean tropospheric column concentrations of NO₂ (left side) and SO₂ (right side) from 7^th of April and 11^th of May 2019 (top side) and 2020 (bottom side). Data were obtained from the ESA Copernicus Sentinel 5P satellite.

Due to the low spatial resolution of AOD data, we could not analyze the spatial variations within Singapore, and we decided to report the data collected in the Southeast Asia region. Comparing to the same period in 2019, the total AOD at 550 nm in Singapore decreased significantly in 2020 during the lockdown. From Fig. 3, we observed that areas around Singapore also experienced a decrease of the AOD during this year. However, capitals of nearby countries such as Kuala Lumpur and Jakarta, showed an increase of the AOD level.

Fig. 3. Shows the mean total AOD at 550 nm from 7^th of April and 11^th of May 2019 (left side) and 2020 (right side). Data were obtained from the Copernicus Atmosphere Monitoring Service (CAMS).

Correlations between Air Quality and Mobility

In order to better understand the relationship between outdoor air pollution and mobility data. We plotted in Fig. 4 the daily average levels of PSI, PM₁₀, PM_2.5, NO₂, CO, SO₂, and O₃ along with the selected mobility trends from HDB, Apple, and Google databases.

Fig. 4. Daily trends of the air quality parameter and the mobility levels since the start of earlier measures on 20 Mar 2020. ^{h, a,} and ^g denote that sources of the dataset are from HDB, Apple, and Google, respectively.

Fig. 4 shows an immediate change in community mobility on the 07^th of April 2020 when the lockdown measures were implemented. Fewer people drove, took public transportation and went to work, while more people stayed at home. Mobility levels declined before and after the day the lockdown measures were implemented but in a much gentler way. This suggests that the measures were effective and widely adopted by the public. The daily averaged HDB car park availability decreased from 54% to 49%. Over the same period, changes in outdoor pollutants concentrations were less pronounced than those observed in the mobility data. We can observe fluctuating downward trends for PM_2.5, and NO₂ levels during the lockdown period, while an increasing trend for SO₂ levels.

To better understand the impact that the reduction in mobility observed in Fig. 3 had on the number of people commuting to and from their home using a motorized vehicle, we plotted the HDB carpark availability in 2020 in Fig. 5. The figure shows that while in January during working days between 9 and 19 up to 70% of the parking spots were available, this percentage reduced to 55% during the lockdown period.

Fig. 5. HDB carpark availability prior and during the lockdown period. We centered the color bar using the percentage of car spaces available on the 7^th of April (first day of the lockdown period) at 11:00 the time of the day when generally the highest percentage of parking spots are available.

The correlations between air quality and mobility parameters were tested via Spearman’s rank test and the results are listed in Table 3. PM_2.5 concentrations were significantly correlated with all the mobility trends and had the highest correlation coefficients with trends of people who visited transit stations comparing to other mobility data. The second-highest coefficient of PM_2.5 was the workplace from Google database, which implies that the shutdown of some industries could contribute to the reduction of PM_2.5. NO₂ only significantly correlated with the mobility trends of car park availability (HDB dataset), driving (Apple dataset), and transit station (Google dataset), but did not with the workplace and residential ones. Both workplaces and traffic are among the main sources of PM_2.5, but NO₂ are more traffic related. For SO₂, we found positive correlations with the human mobility trend in residential places but negative correlations with the other mobility trends. For PM₁₀, CO, and O₃, we did not find any significant correlation with mobility data, while O₃ concentrations showed a negative correlation with car park availability.

It is important to note that in Singapore the great majority of the population uses public transportation to commute and to move around. In 2018, in Singapore there were 957,006 registered vehicles out of which 615,452 were cars (Land Transport Authority, 2018). Consequently, less than 11% of the people living in Singapore owned a car. During the lockdown period the daily bus and train ridership decreased by more than 71% and 75%, respectively. The Singaporean Transport Authority from the 15^th of April announced a that few lines were closed, however, the great majority of the bus lines, and metro trains continued to operate (Land Transport Authority, 2020). Thus, while the number of people going to work and using public transportation decreased significantly the pollution released by public transportation sector did not decrease by the same percentage.

DISCUSSION

Heath Benefits of the Reduction of Air Pollutant

Particulate matter negatively affects the health of people since it can penetrate inside the lungs and those particles with a diameter smaller than 2.5 microns can even enter the blood stream. This contributes to higher risks of developing cardiovascular, respiratory diseases and lung cancer. The World Health Organization (WHO) estimates that a reduction in the annual average concentration of particulate matter (PM_2.5) from 35 to 10 µg m^–3 (WHO, 2018), reductions similar to those observed during the lockdown period, correspond to a reduction in pollution related deaths by around 15%. Consequently, the reduction in outdoor pollutants during the COVID lockdown may have had a positive impact both in the short- and long-term health of Singaporeans. We observed PM₁₀ and PM_2.5 24-h mean concentrations lower than the WHO recommended level which are 50 µg m^–3 and 25 µg m^–3, respectively. Moreover, the measured PM concentrations fluctuated around the recommended annual mean annual mean concentrations which are 20 µg m^–3 for PM₁₀ and 10 µg m^–3 for PM_2.5.

Changes in Air Quality in Southeast Asia

In southeast Asia, NO₂ levels decreased by 27, 30, 22, 34, 34% in Kuala Lumpur, Singapore, Bangkok, Jakarta, Manila based on satellite images during 10 to 24 April in 2020 comparing to the average of the same period from 2015 to 2019 (Kasturi Devi Kanniah et al., 2020). The neighboring nation of Malaysia reported a national-wide drop of PM_2.5 between 11.3 and 23.7% during 18 March to 14 April 2020 compared to 14 to 17 March 2020 (Abdullah et al., 2020). Suhaimi et al. (2020) showed a reduction of PM_2.5, NO₂, CO, and SO₂ by 3 to 36%, 43 to 68%, 1 to 48%, and 2 to 48% according to different regions between 18 March and 21 April 2020 in Malaysia (Suhaimi et al., 2020). Nadzir et al. (2020) reported that CO concentrations decreases ranging from 40.5 to 47.5%, but PM_2.5 levels showed a both increase and decrease according to different regions in Malaysia during the period from 18 March to 08 April 2020 (Nadzir et al., 2020). In general, the reported reductions of air pollutants during the lockdown were comparable with the results we obtained. Compared to those studies in Southeast Asia, the reductions of PM_2.5, NO₂, and SO₂ concentrations in Singapore were higher, while the reductions in CO were lower than in the neighboring regions, respectively.

Limitation

In Singapore outdoor pollutants concentrations vary as a function of those pollutants generated within the country and those from neighboring countries. In this study, we did not account the contribution of air pollutants from neighboring countries. The lockdown periods of Singapore and other neighboring countries were overlapped with each other, which may have had a mutual impact on the air quality of this region. Another limitation of this study was that while we gathered information on mobility from various sources, we did not have sufficient data to estimate the exact reduction of vehicular traffic on the roads. We also did not have data on how much some industrial activities (e.g., port, refineries, airport, power generation) decreased during the lockdown period. Consequently, while we observed that they play a significant role in generating harmful outdoor pollutants, we were not able to correlate a decrease in emission from industrial activities with a reduction in outdoor concentrations. Finally, while we observed a significant decrease in outdoor pollutants, it would be difficult to estimate the positive health effect on the population since the great majority of the population stayed indoor during the lockdown period and was exposed to the cleaner air only for a short period of time.

CONCLUSIONS

In this study, we used nationwide data of PSI, PM₁₀, PM_2.5, NO₂, CO, SO₂, and O₃ from 2016 to 2020 to determine how outdoor pollutants concentrations varies during the lockdown period (from the 7^th April to the 11^th May) in Singapore. Detailed spatial distribution characteristics of NO₂ and SO₂ were analyzed using data from the European Space Agency Copernicus Sentinel 5P satellite. Mobility datasets from the Singaporean Housing Development Board, Apple, and Google were used to identify the influence factors of the improvement of air quality. The main findings are listed below:

1. Nationwide, the reductions in PSI, PM₁₀, PM₅, NO₂, CO, and SO₂ were 19, 23, 29, 54, 6, and 52%, respectively, while O₃ increased by 18%.
   
   
2. The levels of PSI, PM₅, NO₂, and SO₂ decreased during the lockdown in all the five areas of Singapore, while PM₁₀ in the north and CO in the east showed opposite trends. O₃ decreased only in the west but increased in the other four areas. The southern and western areas had the most substantial reduction of air pollution in Singapore, which is due to the local spatial distribution of heavy industries and the airport.
   
   
3. The levels of PM₅ and NO₂ were significantly correlated with the reduction trend of human mobility data. The reduced visits to workplaces and transit stations had the highest two correlation coefficient with PM_2.5, and the reduction of traffic-related mobility data, such as, car park availability and driving, correlated with NO₂.
   
   
4. The satellite data showed that the slowdown in the aviation, refining, and port activities contributed to the reduction of NO₂, SO₂ and AOD concentrations.

ACKNOWLEDGMENTS

This research was funded by the Republic of Singapore's National Research Foundation through a grant to the Berkeley Education Alliance for Research in Singapore (BEARS) for the Singapore-Berkeley Building Efficiency and Sustainability in the Tropics (SinBerBEST) Program. BEARS has been established by the University of California, Berkeley as a center for intellectual excellence in research and education in Singapore.

DISCLAIMER

The authors declare that they have no conflict of interest. 

Aerosol Air Qual. Res. 20 :1748 -1758 . https://doi.org/10.4209/aaqr.2020.06.0303