Effect of COVID-19 on transportation air pollution by moderation and mediation analysis in Queens, New York.

The outbreak of the COVID-19 virus in 2020 has left many changes in the quality of life and environment, including air quality in different parts of the world. As a result of lockdown conditions, the level of air pollution has been changed considerably due to topographic, geographical, and cultural conditions as well as traffic restrictions. Thus, this study aimed to investigate the effect COVID-19 outbreak on improving air quality as a result of changes in traffic volume and traffic patterns in Queens, New York, using the moderation and mediation analysis model structure. In this model, COVID-19 outbreak periods were defined as a moderating variable, traffic volume (number of daily vehicles) as an independent variable and mediator, and air pollution concentration parameters (NOx, PM2.5, and O3) individually as dependent variables. Three-time periods were selected, each representing the duration and severity of traffic restrictions and prohibitions, and these three periods corresponded to 1 February-4 March, 5 March-21 March, and 22 March-15 May. They represented the normal, aware, and lockdown periods, respectively. The result of the study showed that in 2020 compared to the last five consecutive years, PM2.5 and NOx pollutants decreased by 39.2% and 35.8% as a result of the traffic ban due to the COVID-19, but an increase of 15.1% in O3 pollutant was observed in the mentioned period. Although traffic restrictions reduced total traffic volume compared to the same period last year, there has been no significant reduction in the air quality index (AQI). The reduction in NOx concentration leads to more O3 ground levels, and this caused the AQI not to decrease significantly. Finally, the moderation and mediation model results showed that the COVID-19 almost has no significant effect on the correlation between daily traffic and the concentration of NOx, PM2.5, and O3 pollutants as moderator. However, the COVID-19 has a significant correlation with O3 and PM2.5 concentration, and the traffic volume mediation effect is negligible. Therefore, the statistical analysis and models show that the COVID-19 pandemic is an effective traffic volume and air quality parameter.

Introduction

The COVID-19 pandemic has highlighted the importance of ambient air quality and the need for cleaner air, leading to a deeper understanding of environmental threats on a global scale, such as climate change as a result of increased greenhouse gas emissions. According to the World Health Organization (WHO), the number of deaths attributed to air pollution worldwide reached 4.2 million in 2016 (WHO 2018) and estimates that around 7 million people die every year from exposure to fine particles in polluted air (WHO 2018), while the number of casualties from the COVID-19 pandemic to date is about one million people, approximately 14% of deaths from air polluted (WHO 2020). On the other hand, due to the widespread pandemic and the mandatory quarantine in many industrial cities around the world, the emission of air pollutants from human-made resources has changed. The following map shows the air pollution index contributed to ozone and PM2.5 concentrations in the USA over the first positive COVID-19 test until 15 May (the last day of quarantine in some cities) (Fig. 1). It indicates a noticeable change and improvement in the air quality index that is particularly obvious in eastern areas with higher population densities.

Fig. 1

Comparing AQI in 2019 and 2020 (1 Feb–15 may (EPA 2020a))

Rising or falling each of the air pollutants concentrations such as NOx, PM2.5, CO, O3, and greenhouse gases such as CO2 have been in accordance with the culture and lifestyle, the level of industrial activity, geographical conditions, and topography of the areas and also traffic pattern as well as a period of quarantine and forced cessation of industries. For example, NOx pollutants have been declined by an average of 40% in various cities in China and about 20% in different cities in Europe during the COVID-19 lockdown (Bauwens et al. 2020). NOx emissions are directly related to transportation and fossil fuels (Chen et al. 2020). Therefore, declining traffic volume leads to its lower level in many parts of the world. However, as ground-level ozone contamination is often eliminated by NOx, an increased amount of it is reported in some places (Shi and Brasseur 2020). Another study using the WRF-CAM model examined traffic-related parameters and the number of anthropogenic air pollutants within the Yangtze River in China. The results of this study showed a decrease of 29–47% in NOx and 47% − 27% in PM2.5 concentration due to the reduction in industrial activities, the reduction in vehicle miles traveled (VMT) in private cars, and the closure of activities of construction sites (Li et al. 2020). Related studies monitoring PM2.5 concentrations via remote sensing in the north of China have shown a 35% reduction in emissions since the quarantine period and after industry closure (Shi and Brasseur 2020). Unlike other places, the result of studies in London reflects an increase in PM2.5 due to an increase in cooking, burning household gas, and wood during the quarantine period (E.R. Group 2020). In general, air pollutants, PM2.5, CO, and air quality index (AQI) have decreased less than the three pollutants PM10, NO2, and SO2 (Bao and Zhang 2020). A recent study has investigated the relationship between air quality and the COVID-19 pandemic in New York State. The results showed a significant correlation between air quality with the outbreak of the COVID-19 pandemic (Bashir et al. 2020).

The recent studies have only examined the COVID-19 pandemic effect on air quality in the form of trend and mean comparison analysis and have not used statistical models and analyses (such as statistical relation models and two population tests). Also, these researches have not addressed the reason for changes in air pollution concentrations and their relations.

A moderation and mediation analysis is applied to investigate whether the relationship between two variables depends on a third variable’s value. In other words, the presence of a moderator variable can strengthen, weaken, or reverse the nature of a relationship between an independent variable and a dependent variable. Also, a mediator parameter’s presence causes a more indirect effect of two parameters on each other (Hayes 2017).

Therefore, this study aims to investigate the COVID-19 pandemic effect on air quality and its relations by moderation and mediation analysis. The different aspects of COVID-19 pandemic effects on air quality and pollutants and traffic volume (as the critical parameter in air quality and COVID-19 pandemic restrictions) have been performed to provide the conceptual and statistical effect of COVID-19 on air quality condition. Also, meteorological, AQI, and traffic volume daily trend compared with previous years and seasonal change effect on air pollutant concentration have been studied. Queens District of New York has been selected as the study area due to the highest rate of COVID-19 virus infection.

Material and methods

Case study

Queens is one of the five boroughs in the east of New York City. According to available data, the first COVID-19 test in this area was performed on 3 March, and the first positive case was recorded on 5 March (Department of Health and Mental Hygiene (DOHMH) 2020). Queens is one of the most critical boroughs with the highest incidence of the COVID-19 virus among other boroughs (DOHMH 2020). Due to the rapid spread of the virus in New York City, it was quarantined and monitored between 22 March and 15 May.

According to available statistics, 95.9% of vehicles in Queens use gas, 3.7% use diesel fuel, and only 0.2% are electric vehicles (Department of Motor Vehicles (DMV) 2020).

Based on the Environmental Protection Agency (EPA) in 2019 Dataset, Queens was 292 days in good condition, 69 days average, and 4 days unhealthy for sensitive people in terms of AQI (EPA 2020b). Of these, the quality parameters NO2, O3, and PM2.5 were 74, 188, and 103 days, respectively, as the significant parameter of the AQI (EPA 2020b).

Dataset

There are three categories of data used in this study consisting of the concentration of air pollution parameters (including PM2.5, NOx, O3, and CO), daily average wind and temperature, and the volume of traffic passing through the Queens borough. Air pollution concentration, meteorological, and traffic data were gathered daily from the EPA and Department of Transportation (DOT). This study’s air pollution monitoring station was Queens College in Queens, NY (urban, 40.7° N, 74.0° W). Data on the concentration of air pollution parameters were used on a daily basis during the quarantine period (22 March to 15 May) from 2015 to 2019 and on a daily basis during1 Feb to 15 May. Daily traffic volume data was also used during 1 Feb to 15 May. This time period is selected for investigating parameters before the first COVID-19 test in Queens.

Moderation and mediation analysis

Moderation and mediation analysis has been performed to estimate the effect of the COVID-19 pandemic on the relation between traffic volume and air pollutant parameters. The traffic volume is considered as the independent variable (X), COVID-19 as moderator (M), and air pollutant parameter concentration as dependent variables (Y) in moderation analysis. In the mediation model, the COVID-19 is considered as the independent variable (X), air pollutant parameters concentration as dependent variables (Y), and the traffic volume as mediator.

The conceptual model and statistical diagram are depicted in Figs. 2, 3 and 4.

Fig. 2

Moderation conceptual model

Fig. 3

Mediation conceptual model

Fig. 4

The statistical model of the moderation analysis

The moderation and mediation analyses are done based on the Hayes regression procedure (Hayes 2017; Öztürk and Teksöz 2016). The equation of the analysis is given below:

Moderation analysis:

Air pollutant concentration (,  and ) = C + C1*traffic volume + C2*COVID-19 + C3*(traffic volume * COVID-19)

Mediation analysis:

Traffic Volume = i1 + a * COVID-19

Air pollutant concentration (,  and  ) = i2 + b * traffic volume + c' * COVID-19

The C, C1, C2, and C3 are the moderation regression coefficients and i1, i2, a, b, and c′ are the mediation regression model coefficients. The moderation and mediation models are done by PROCESS macro in SPSS software.

Results

Investigating the general hypothesis

In order to accurately examine the effect of COVID-19 using the moderation and mediation analysis statistical model, three general hypotheses have been investigated to select the best parameters and conceptual structure for this model.

Investigating the meteorological variations

The variations in meteorological parameters are one of the main factors in changing air quality. This issue was examined to determine its effect in 2020 compared to the last 5 years. In this regard, the average daily temperature, wind speed, and relative humidity in the March 22 and May 15 period were investigated by two-sample mean t test analyses between 2020 and each year.

Table 1 demonstrates that in the temperature, considering the significance level of 5%, only the 2017 and 2019 average changes compared to 2020 are slightly significant. In the wind, except for 2015 and 2016, which have considerable difference from 2020, the last 3 years are not significantly different from 2020, and there has been little change in this parameter. Relative humidity data also show that, except for 2015, the last 4 years are not significantly different from 2020. Therefore, it can be concluded that meteorological changes in 2020 are not very noticeable and have little impact. The lack of meteorological conditions’ strong effect on NO2 concentration in New York City has also been confirmed in Goldberg et al. (2020) study.

Table 1

Two-sample mean t test analyses of 2020 with 5-year meteorological parameters on 22 March to 15 May period

Year		2015	2016	2017	2018	2019
Temp (deg F)	t value	1.68	1.43	2.10	0.82	2.12
	p value	0.097	0.155	0.038	0.415	0.036
Wind (knots)	t value	− 4.38	− 3.21	− 1.16	− 1.39	0.02
	p value	~ 0	0.002	0.247	0.168	0.988
Relative humidity (%)	t value	2.04	1.15	0.64	− 0.73	0.80
	p value	0.044	0.258	0.527	0.468	0.425

Effect of seasonal change on air pollutant concentration

Meteorological characteristics of each season can affect the air pollution parameter concentration. For this reason, the average concentration of NOx, PM2.5, O3, and CO has been calculated in winter (1 Feb–20 March) and spring (21 March–15 May) seasons.

Table 2 shows that the average concentration of all air pollutants in winter is almost 2 times of spring. So the seasonal changes have a significant effect on the concentration of air pollutants. Also, the whole lockdown period is in the spring season, and there are no seasonal changes in this period.

Table 2

Mean seasonal comparison of air quality parameters

	NOx	PM2.5	O3	CO
Winter (1 Feb–20 March)	17.1	6.2	0.024	272
Spring (21 March–15 May)	9.1	3.4	0.035	199

Effect of COVID-19 on air pollutant concentration

Initially, the impact of the COVID-19 outbreak on air pollution was proposed as the general hypothesis. The difference in averages for each of the air pollutant concentrations, i.e., NOx, PM2.5, O3, and CO, have been compared annually between 22 March and 15 May over five consecutive years in comparison with the data for similar parameters in 2020. Nitrogen oxides, carbon monoxide, and PM2.5 are three primary air pollutant parameters, and the changes in these parameters can mainly be attributed to the changes in traffic volume and level of services (LOS) and traffic restrictions. Ozone pollution is a secondary pollutant, and its changes are related to changes in NOx emissions (Jhun et al. 2015). According to Table 3, the ratio of NOx, PM2.5, and CO pollutants are negative in the last 5 years compared to 2020, and the reduction in emissions of these pollutants in 2020 is because of decreasing in traffic congestions. Also, the positive ratio of ozone concentration in the last 5 years compared to 2020 is as a result of the reduction in nitrogen oxide pollution as expected.

Table 3

Mean difference with 2020 on 22 March to 15 May period

	Mean difference (%)
Year	2015	2016	2017	2018	2019	5-year-avg
NOx	− 52.1	− 36.1	− 37.8	− 38.5	− 23.0	− 39.2
PM2.5	− 46.1	− 37.1	− 19.4	− 44.1	− 24.1	− 35.8
O3	+ 27.6	+ 15.9	+ 19.7	+ 5.4	+ 9.6	+ 15.11
CO	− 15.8	− 10.0	− 8.6	− 19.6	− 13.2	− 13.5

The effect of COVID-19 on traffic volume and AQI

In the second part of the hypothesis, the effect of COVID-19 on traffic volume and AQI compared to the same time as the previous year was examined. In this study, the changes in traffic volume are considered in three periods. These changes as the normal time periods, aware mode, and complete quarantine for the Queens borough of New York City are illustrated in Fig. 5. For both 2019 and 2020, there are fluctuations in daily traffic due to the weekend and workday traffic patterns. On the other hand, there is a significant decrease in the average volume of traffic during the period of restrictions compared to the normal period in 2020 and the prevalence of COVID-19. Changes in the AQI over the three-time periods of the normal mode, alert mode, and full quarantine as well as traffic restrictions for 2019 to 2020 were examined and shown in Fig. 6. According to this graph, there is a low decrease in the air quality index in 2020 compared to 2019, which was the highest difference between AQI changes in the warning period.

Fig. 5

Daily traffic volume trend comparison in normal, aware, and lockdown periods between 2019 and 2020 years

Fig. 6

AQI 2019 and 2020 trend comparison in normal, aware, and lockdown periods between 2019 and 2020 years

The effect of COVID-19 on the determinant parameter of AQI

Due to the significant reduction in traffic congestion and no significant decrease in the air quality index, the AQI parameter status in this period in 2019 and 2020 was also examined.

The air quality index for Queens Borough is determined by the quality index of PM2.5, NO2, and O3 parameters. In order to compare the contribution of these three pollutants in determining AQI, the number of days that these parameters were the main factors of AQI in each of the periods 1 February–4 March, 5 March–21 March, and 22 March–15 May in 2019 and 2020 is shown in Table 4. According to Table 4, the significant pollutant in most time intervals in 2019 and 2020 was O3 pollution. However, despite the fact that in 29 days of the above period in 2019, NO2 pollution was an indicator in determining AQI. In 2020, this pollutant was not effective due to reduced concentration and emission due to reduced traffic volume and traffic flow stability. Usually, the increase in unhealthy days with ozone pollutants has been linked to a decrease in NO2 pollutants. Also, there are virtually no significant changes in 2020 compared to 2019 in terms of PM2.5 parameter. Due to zero polluted days with NO2 parameter in 2020, almost all of these days have become polluted with O3 because of the inverse relationship between O3 and NO2. Therefore, there are no significant changes in the AQI.

Table 4

Number of main pollutants in AQI at three period days

Status	2019			2020
	PM2.5	O3	NO2	PM2.5	O3	NO2
1 February–4 March	16	10	7	23	10	0
5 March–21 March	5	7	5	4	13	0
22 March–15 May	8	30	17	5	50	0
Total	29	49	29	32	73	0

According to the proposed hypotheses in the moderator and mediator analytical model, the effect of the COVID-19 on the traffic volume of Queens Borough and the concentration of NOx, PM2.5, and O3 pollutant parameters were examined. Moderation and mediation analysis was performed using SPSS software and the Hayes regression plugin. The COVID-19 which has three modes as normal, aware, and lockdown (In three time periods: 1 February–4 March, 5 March–21 March, and 22 March–15 May) was defined as 3, 2, and 1 respectively and selected as moderator. Traffic volume parameter (number of vehicles per day) was defined as the independent variable in moderation model and mediator parameter, and air pollution parameter concentrations (NOx, PM2.5, and O3) were defined separately as dependent variables. The results of the moderation analysis were obtained in Tables 5, 6 and  7.

Table 5

Results of the O3 moderation analysis

		Coeff	SE	t	p
Constant	C	0.0409	0.0040	10.12	0.00
Traffic volume (X)	C1	0.0000	0.0000	0.15	0.88
COVID-19 (W)	C2	− 0.0059	0.0034	− 1.73	0.09
Traffic volume* COVID-19 (XW)	C3	0.0	0.0000	− 0.24	0.81

Table 6

Results of the NOx moderation analysis

		Coeff	SE	t	p
Constant	C	2.91	7.53	0.39	0.70
Traffic volume (X)	C1	0.0002	0.0004	0.54	0.59
COVID-19 (W)	C2	5.95	6.3	0.94	0.35
Traffic volume* COVID-19 (XW)	C3	0.0	0.0002	− 0.20	0.84

Table 7

Results of the PM2.5 moderation analysis

		Coeff	SE	t	p
Constant	C	1.51	1.8	0.84	0.40
Traffic volume (X)	C1	− 0.0001	0.0001	− 0.68	0.50
COVID-19 (W)	C2	2.36	1.51	1.57	0.12
Traffic volume* COVID-19 (XW)	C3	0.0	0.0	0.17	0.87

Tables 5, 6 and 7 present the results from the Hayes regression analysis examining the moderation impact of COVID-19 on the relationship between traffic volume and O3, NOx and PM2.5 concentration, respectively.

In the statistical moderation model, the critical parameter is the estimate of C3 coefficient that reveals the moderation impact of the COVID-19 pandemic. The regression coefficient for C3 was zero in all three air pollutant parameters. Thus, the COVID-19 has no meaningful linear effect on the traffic volume and O3, NOx, and PM2.5 concentration relation as a moderator. This means that the relationship between traffic volume and O3, NOx, and PM2.5 concentration remains the same in the normal situation, starting of the pandemic as well as lockdown period. According to t and p values, the COVID-19 pandemic is the essential linear parameter in O3, NOx, and PM2.5 regression models. On the other hand, the COVID-19 pandemic has a significant impact on traffic volume and O3, NOx, and PM2.5 concentrations separately but does not affect their relationship.

The results of mediation analysis are demonstrated in Tables 8, 9 and 10.

Table 8

Results of the O3 mediation analysis

		Traffic volume				O3 concentration
		Regression coefficient	SE	p		Regression coefficient	SE	p
Constant	i1	− 5290	1363	< 0.001	i2	0.042	1.5E − 3	< 0.001
COVID-19	a	16,820	681	< 0.001	c′	− 6.5E − 3	1.9E − 3	0.001
Traffic volume	-	-	-	-	b	− 9.81E − 9	~ 0.0	0.924
	R2	0.86			R2	0.50

Table 9

Results of the NOx mediation analysis

		Traffic volume				NOx concentration
		Regression coefficient	SE	p		Regression coefficient	SE	p
Constant	i1	− 5290	1363	< 0.001	i2	4.320	2.756	0.120
COVID-19	a	16,820	681	< 0.001	c′	4.888	3.458	0.161
Traffic volume	-	-	-	-	b	1.3E − 4	1.9E − 4	0.489
	R2	0.86				0.25

Table 10

Results of the PM2.5 mediation analysis

		Traffic volume				PM2.5 concentration
		Regression coefficient	SE	p		Regression coefficient	SE	p
Constant	i1	− 5290	1363	< 0.001	i2	1.238	0.651	0.060
COVID-19	a	16,820	681	< 0.001	c′	2.569	0.827	0.002
Traffic volume	-	-	-	-	b	− 4.7E − 5	4.6E − 5	0.306
	R2	0.86				0.26

The results of these tables show that according to the coefficient “a,” there is a significant correlation between the COVID-19 pandemic and traffic volume. The coefficient “c′” results illustrate a significant correlation between the COVID-19 pandemic and O3 and PM2.5 concentration, but this relationship is not significant in the NOx parameter. Coefficient “b” indicates the direct effect of traffic volume on the air pollutant concentration, showing that traffic volume almost has no significant relationship with O3, NOx, and PM2.5 parameters. The direct and indirect effect of traffic volume on air pollutant parameter concentration has been shown in Table 11.

Table 11

The direct and indirect effect of traffic volume on O3, NOx, and PM2.5 parameters

	O3		NOx		PM2.5
	Coefficient	p	Coefficient	p	Coefficient	p
Direct effect (c′)	− 6.5E − 3	0.001	4.888	0.161	2.569	0.002
Indirect effect (a*b)	1.65E − 4	0.159	2.190	0.247	− 7.91E − 1	0.154

The results of Table 11 demonstrate that the direct effect between COVID-19 and O3 and PM2.5 concentration is significant, and the indirect impact of COVID-19 is much lower than direct effect in O3 and PM2.5 concentration. In other words, the COVID-19 pandemic has a substantial relationship with the concentration of O3 and PM2.5 parameters, and the traffic volume has a low effect as the mediator (Hayes 2017; Öztürk and Teksöz, 2016). Thus, the COVID-19 pandemic has a direct and significant effect on O3 and PM2.5 concentration without traffic volume effect because their p value is lower than 0.05.

Conclusion

Numerous researches have studied the environmental effect of COVID-19 pandemic lockdown in different countries. One of the most critical environmental issues that have been affected by COVID-19 is air pollution. The impact of COVID-19 lockdown on air quality in some parts of the world has been investigated using different methods, such as simple statistical analysis of data, satellite images, and historical maps of contaminated areas. In this study, conceptual moderation analysis models have been made by investigating the dataset and three general hypotheses. These hypotheses have generally investigated the concentration of pollutant parameters and meteorological dataset since 5 years ago, the trend of changes in the number of vehicles per day, and air quality index in 2020 compared with 2019, as well as the number of days that each pollutant parameter is the main contributor in AQI. The results showed the direct impact of COVID-19 pandemic lockdown on traffic congestion and pollutant concentrations, while due to the inverse relationship between NOx and O3 parameters, no significant change was observed in the 2020 AQI compared to 2019 and no significant change in wind and temperature in 2020 comparing to 5 years ago. As a result, in the proposed conceptual model of moderation and mediation analysis, the impact of COVID-19 as the moderator and traffic volume as a mediator on the number of vehicles and its relation to the concentration of pollutants (NOx, PM2.5, and O3) in three time periods (1 February–4 March, 5 March–21 March, and 22 March–15 May) based on statistical analysis was examined. Hayes regression results for the three pollutants NOx, PM2.5, and O3 were almost identical, showing almost no effect of the COVID-19 pandemic on traffic volume and O3, NOx, and PM2.5 concentration relations as moderator. Also, the COVID-19 pandemic has a significant correlation with traffic volume and PM2.5 and O3 parameters, and there is no significant mediator effect of traffic volume on COVID-19 and O3, NOx, and PM2.5 concentration relation. Therefore, the statistical analysis and models show that the COVID-19 pandemic is an effective parameter in traffic volume and air quality.