Changing weekend effects of air pollutants in Beijing under 2020 COVID-19 lockdown controls | npj Urban Sustainability

Changes in the annual average air pollutants

Figure 1 shows changing percentages of annual average air pollutants in 2020 compared to the previous 2 years. The annual average PM2.5 and SO2 concentrations at 32 stations in Beijing in 2020 were all lower than those in 2018 and 2019. The annual average NO2 and CO concentrations in 2020 were also lower than those in 2018 and 2019 except for station of Donggaocun. For O3 concentration, there were about half stations showing higher values in 2020 than 2018 and 2019. On average of 32 stations, PM2.5, NO2, SO2, CO, and O3 concentrations were 38.13 μg m−3, 29.82 μg m−3, 3.70 μg m−3, 0.64 mg m−3, and 59.80 μg m−3 in 2020 and decreased by 26.75%, 27.31%, 37.87%, 20.20%, and 0.99% compared to 2018 and decreased by 10.86%, 18.20%, 18.76%, 7.94%, and 1.78% to 2019 (Table 2).

Fig. 1: The changes in the annual average air pollutants.

The changing percentages of annual average PM2.5, NO2, SO2, CO and O3 concentrations in 2020 compared to (a) 2018 and (b) 2019.

Full size image

Table 2 The annual average PM2.5, NO2, SO2, CO and O3 concentrations of 32 stations in Beijing in 2018, 2019, and 2020.

Full size table

Changes in weekly cycle of air pollutants

Figure 2 presents the annual average air pollutants of 32 stations for each day of the week in 2018, 2019, and 2020. In 2018 and 2019, PM2.5 time series at most of stations consistently exhibited an evident 7-day cycle: a downward trend from Monday to Thursday/Wednesday, and then rising to Saturday (Fig. 2a and Supplementary Fig. 2). In contrast, no indications of these distinct weekly cycles could be found in 2020. Instead, Saturday was the day of highest PM2.5, followed rather closely by Thursday and Wednesday, while Monday and Sunday were the days of lowest values in 2020. Both NO2 and SO2 concentrations in 2018 exhibited higher values on Friday through Monday, while the larger values occurred from Tuesday through Thursday in 2020 (Fig. 2b, c and Supplementary Figs. 3–4). The seven-day cycle of CO concentrations show very similar patterns to those of PM2.5 in 2018 and 2019, while the weekly cycle in 2020 show almost opposite pattern (Fig. 2d and Supplementary Fig. 5). By contrast, O3 concentrations have different weekly cycle patterns from other air pollutants: they have relatively higher values on Friday through Sunday in 2020, which were similar to those in 2018 but quite different to those in 2019 (Fig. 2e and Supplementary Fig. 6).

Fig. 2: Air pollutants by day of the week.

The annual average (a) PM2.5, (b) NO2, (c) SO2, (d) CO and (e) O3 concentrations by day of the week for the mean of 32 stations in Beijing in 2018, 2019, and 2020.

Full size image

Changes in weekend effects of air pollutants

In previous studies, weekend and weekday have different definitions depending on the signal of weekly cycle. For instance, Simmonds and Keay62 defined Saturday through Sunday as the weekend and Monday through Friday as the weekday. While, some studies63,64,65 took Saturday through Monday as the weekend and Wednesday through Friday as the weekday. In the present study, the larger values of PM2.5, NO2, SO2, and CO in 2018 and 2019 mainly appeared on Friday through Monday. Thereby, we define Friday through Monday as the weekend and Tuesday through Thursday as the weekday to evaluate the changes of weekend effects in 2020 relative to 2018 and 2019. Here, we follow Daniel et al.’s method65 to evaluate the weekend effects and weekly cycle intensities by using weekend effect magnitude (the difference between weekend and weekdays WEM, see Eq. 1 in METHODS) and weekly cycle magnitude (weekly maximum minus weekly minimum, WCM, see Eq. 2 in METHODS).

Supplementary Figs. 7–11 show the WEM of PM2.5, NO2, SO2, CO and O3 concentrations for each year at 32 stations in Beijing. The weekend values of PM2.5 were greater than weekdays at all stations in 2018 and 2019 with the values ranging from 7.05 to 14.38 μg m−3 and 1.98 to 8.72 μg m−3. In 2020, only 3 stations showed larger values on weekend than weekdays, and the WEM fluctuated between −4.10 and 1.14 μgm−3 (Supplementary Fig. 7). The weekend-weekday differences were significant at the 90% confidence level by student’s t-test and Wilcoxon rank-sum test at 21 stations in 2018 and 16 stations in 2019, respectively (Table S1). By contrast, there are no any stations showing significant weekend-weekday differences at the 90% confidence level by student’s t-test and Wilcoxon rank-sum test in 2020. On the average of 32 stations, the WEM of PM2.5 in 2018 and 2019 were 10.30 μg m−3 and 4.98 μg m−3 higher on weekend than weekdays and the differences between weekends and weekdays are both at the 90% significant confidence level by student’s t-test and Wilcoxon rank-sum test (Fig. 3a). Comparatively, weekend was smaller 1.38 μg m−3 than weekdays and the difference was insignificant by student’s t-test and Wilcoxon rank-sum test in 2020. The results show that weekends and weekdays do not differ from each other in Beijing in 2020, implying that the weekend effects of the annual average PM2.5 vanished under lockdown control measures.

Fig. 3: Weekend effect magnitudes of air pollutants.

The annual average (a) PM2.5, (b) NO2, (c) SO2, (d) CO and (e) O3 concentrations weekend effect magnitudes for the mean of 32 stations in Beijing in 2018, 2019, and 2020. The weekend refers to Friday through Monday, and weekday is Tuesday through Thursday. The weekend effect magnitude is computed by subtracting the average weekday value from the average weekend value.

Full size image

The values of NO2 and SO2 concentrations on weekend were larger than weekdays at 32 and 31 stations and the weekend-weekday difference were significant at the 90% confidence level by Student’s t-test and Wilcoxon rank-sum test at 14 stations in 2018 (Supplementary Figs. 8–9, Tables S2–3). In 2020, NO2 and SO2 concentrations on weekend were smaller than weekdays at 32 and 28 stations and the differences were significant at the 90% significant confidence level by Student’s t-test and Wilcoxon rank-sum test at 19 and 25 stations. On average of 32 stations, NO2 and SO2 concentrations on weekend were larger 2.66 μg m−3 and 0.84 μg m−3 than weekday in 2018 at the 90% significant confidence level by Student’s t-test and Wilcoxon rank-sum test (Fig. 3b, c). In 2020, NO2 and SO2 concentrations on weekends were lower 1.91 μg m−3 and 0.43 μg m−3 than weekdays and the differences were at the 90% significant confidence level by student’s t-test and Wilcoxon rank-sum test.

The weekend values of CO were greater than weekdays at all stations in 2018 and 2019 while there were only 2 stations showing larger values on weekends than weekdays in 2020 (Supplementary Fig. 10). Student’s t-test and Wilcoxon rank-sum test reveal that the weekend-weekday differences were significant at the 90% confidence level at 28 stations in 2018 and 18 stations in 2019, respectively (Table S4). By contrast, there are only 1 station showing the 90% significance of difference by student’s t-test and Wilcoxon rank-sum test in 2020. On the average of 32 stations, CO were 0.10 mg m−3 and 0.04 mg m−3 higher on weekend than weekdays and the differences between weekends and weekdays are both at the 90% significant confidence level by student’s t-test and Wilcoxon rank-sum test in 2018 and 2019 (Fig. 3d). Comparatively, weekend was smaller 0.03 mg m−3 than weekdays and the difference was insignificant by student’s t-test and Wilcoxon rank-sum test in 2020. Similar to PM2.5, the weekend effects of the annual average CO in 2018 and 2019 vanished in 2020.

The weekend values of O3 were greater than weekdays at 30 stations in 2018 and 32 stations in 2020 (Supplementary Fig. 11). The weekend-weekday differences were significant at the 90% confidence level by student’s t-test and Wilcoxon rank-sum test at 5 and 8 stations in 2018 and 2020 (Table S5). On the average of 32 stations, the WEM in 2018 and 2020 were 2.43 μg m−3 and 2.09 μg m−3 higher on weekend than weekdays and the differences were both insignificant by student’s t-test and Wilcoxon rank-sum test (Fig. 3e).

Conclusively, the annual average PM2.5, NO2, SO2, and CO concentrations on weekends were higher than weekdays in 2018, but they showed opposite pattern in 2020. Further analysis reveals that the reductions of PM2.5, NO2, SO2, and CO on the weekends were larger than weekdays in 2020 relative to 2018, and 2019 (Table 3). In other words, the changing weekend effects of PM2.5, NO2, SO2, and CO in 2020 are mainly attributed to more reductions on the weekends than weekdays.

Table 3 The averaged reduction percentages of PM2.5, NO2, SO2, CO, and O3 concentrations for 32 stations on weekends and weekdays in Beijing in 2020 relative to 2018 and 2019.

Full size table

Diminishing weekly cycle magnitude of air pollutants

Moreover, we detect the changes in WCMs of PM2.5, NO2, SO2, CO and O3 concentrations in 2020 compared to the previous 2 years (Fig. 4 and Supplementary Figs. 12–16). The WCMs of PM2.5 concentrations in 2020 were smaller than those in 2018 and 2019 at nearly all stations (Supplementary Fig. 12). The average WCMs of PM2.5 concentrations for 32 stations in 2018 and 2019 were 19.31 and 11.20 μg m−3, while the average value in 2020 was only 5.94 μg m−3 and decreased by 69.24% and 47.21%. The average WCM of NO2, SO2, and O3 for 32 stations in 2020 decreased by 29.69%, 54.82%, and 32.05% compared to 2018, and 4.49%, 5.27%, and 21.05% to 2019. These results demonstrate that WCMs of air pollutants largely diminished in 2020 compared to the previous 2 years. We notice that the WCM of CO concentrations in 2020 decreased 33.97% compared to 2018, but increased by 15.96% compared to 2019.

Fig. 4: Weekly cycle magnitudes of air pollutants.

The annual average (a) PM2.5, (b) NO2, (c) SO2, (d) CO and (e) O3 concentrations weekly cycle magnitudes for the mean of 32 stations in Beijing in 2018, 2019, and 2020. The weekly cycle magnitude is defined as the difference between the weekly maximum and minimum.

Full size image