Comparison of air pollutants and their health effects in two developed regions in China during the COVID-19 pandemic.

Air pollution attributed to substantial anthropogenic emissions and significant secondary formation processes have been reported frequently in China, especially in Beijing-Tianjin-Hebei (BTH) and Yangtze River Delta (YRD). In order to investigate the aerosol evolution processes before, in, and after the novel coronavirus (COVID-19) lockdown period of 2020, ambient monitoring data of six air pollutants were analyzed from Jan 1 to Apr 11 in both 2020 and 2019. Our results showed that the six ambient pollutants concentrations were much lower during the COVID-19 lockdown due to a great reduction of anthropogenic emissions. BTH suffered from air pollution more seriously in comparison of YRD, suggesting the differences in the industrial structures of these two regions. The significant difference between the normalized ratios of CO and NO2 during COVID-19 lockdown, along with the increasing PM2.5, indicated the oxidation of NO2 to form nitrate and the dominant contribution of secondary processes on PM2.5. In addition, the most health risk factor was PM2.5 and health-risked based air quality index (HAQI) values during the COVID-19 pandemic in YRD in 2020 were all lower than those in 2019. Our findings suggest that the reduction of anthropogenic emissions is essential to mitigate PM2.5 pollution, while O3 control may be more complicated.

Introduction

Air pollution in China have attracted attentions due to its frequent regional and local haze events (Li et al., 2017a; Nie et al., 2018; Scott et al., 2018; Shen et al., 2020; Wang et al., 2016, 2019; Xu et al., 2019). The annual averaged PM10, PM2.5, CO and SO2 levels have remarkably decreased due to the efforts of emission reduction and energy optimization (Ma et al., 2016; Maji and Sarkar, 2020; Sun et al., 2018). However, other pollutants including NO2 and O3 are still complicated (Ma et al., 2019; Shen et al., 2020). For example, ~50.7% of the O3 monitoring stations showed a significant positive trend of ≥2 μg m−3 year−1. On the other side, only 20.9% of NO2 monitoring stations were associated with a negative trend of ≥ −2 μg m−3 year−1 (Maji and Sarkar, 2020). In addition, previous studies have illustrated that particulate matters (PM) and gaseous pollutants (SO2, NO2, O3 and CO) can lead to heart disease, lung cancer, respiratory infections, and even premature deaths (Gu et al., 2002; He et al., 2009; Kan et al., 2012). Studies on health effects of air pollution on a large spatial scale in China are relatively limited, especially in some disease pandemic event (Shen et al., 2020). Furthermore, air pollution related health risks with some other social problem, such as population, industry structure, pandemic, is still not yet studied.

At the end of 2019, an epidemic of respiratory disease caused by a novel coronavirus (named as “COVID-19”) started to spread (Huang et al., 2020; Tian et al., 2020; Wang et al., 2020). As is well known, the disease has become a global pandemic with more than 30 million confirmed cases by September 20, 2020, with over 90,000 cases reported in China (https://coronavirus.jhu.edu/map.html). China, the first country facing with this unprecedented challenge, the central and local governments imposed strong restrictions on the movement/activities of people (“lockdown”) from Jan 24 to Feb 10, 2020, to contain the COVID-19 outbreak. Subsequently, the nationwide lockdown provides an experimental opportunity to investigate how the variations of air pollution response to the large reduction of the anthropogenic emissions. Additionally, as the remarkable regional differences of anthropogenic emission sources in the Beijing-Tianjin-Hebei region (BTH) and the Yangtze River Delta region (YRD), it makes the comparison of the atmospheric aerosol evolution processes before, in, and after the lockdown period of COVID-19 in these two regions possible. Previous studies illustrated that coal combustion and traffic-related emissions are the dominant sources in BTH (Li et al., 2017b; Tian et al., 2018; Wang et al., 2019; Xu et al., 2019). Moreover, it has been illustrated that aqueous-phase reactions play an important role in the formation of sulfate, nitrate, and some secondary organic aerosol (SOA) in BTH (Duan et al., 2020; Xu et al., 2017). However, industrial-related coal-combustion PM was reported in some certain areas in YRD (Wang et al., 2018; Wu et al., 2018; Yang et al., 2020). Furthermore, aqueous-phase driven secondary aerosols appear to be very important in enhancing PM1 (particles with an aerodynamic diameter < 1 μm) pollution while organic aerosols (OA) pollution in YRD is due to photochemical processes (Wu et al., 2018). Similar emission-control events have been conducted in China for many times, e.g., the 2008 Beijing Olympic Games (Huang et al., 2010), the 2014 Asia-Pacific Economic Cooperation (APEC) summit (Sun et al., 2016), and the 2016 Hangzhou G20 summit (Li et al., 2018). However, emission-control events mentioned before were all small regional control (Huang et al., 2010; Sun et al., 2016). Thus, the COVID-19 lockdown provides an opportunity to examine how different types of emission sources, how meteorological conditions affect the formation and evolution of air pollution, and what implications have for the effects of nationwide lockdown on the future air quality and human health if China improves the energy structure in the future.

Herein, to investigate the atmospheric aerosol evolution processes before, in, and after the lockdown of COVID-19 in two typical regions (BTH and YRD), the variations of six criteria air pollutants (PM2.5, O3, NO2, SO2, CO, and PM10) from Jan 1 to Apr 11 in 2020 were compared to the same period in 2019. In addition, the health-risked based air quality index (HAQI) from these six pollutants was also estimated proposed by Hu et al. (2015). The result of this study will provide an experimental opportunity to investigate the air pollution response to the large reduction of the anthropogenic emissions.

Methods

Study areas and data sources

Hourly averaged concentrations of six criteria air pollutants (PM2.5, O3, NO2, SO2, CO, and PM10) from January 1 to April 11 in 2019 and 2020 from 1479 national monitoring sites in 367 cities of China (Figure S1, data from Taiwan, Hongkong, and Macau are not included) were downloaded from website of China Ministry of Ecology and Environment (http://datacenter.mee.gov.cn). The 8-h moving averaged concentrations of O3 were used in this study. Thus, enough data follow the statistical significance and have great confidence that the results of the data might not be caused by chance.

In order to better elucidate the responses of concentrations of six air pollutants to the dramatic changes in human activities during COVID-19 lockdown, we classified the study period into ten stages (Table S1), which covered the period before (Pre-COVID-19, Stages I and II), during (COVID-19 lockdown, Stages III and IV) and after (Recovery period, Stages V–X) the lockdown. This work focuses on the comparison of characteristics of air pollutants in two developed regions, Beijing-Tianjin-Hebei (BTH) and Yangtze River Delta (YRD). There are great differences in the industrial categories between BTH and YRD (Wan et al., 2020). For example, secondary industry (e.g., steel-producing) is leading economy in BTH, while tertiary-industry (e.g., electronic products) is dominant in YRD (Wan et al., 2020).

Meteorological data, including ambient temperature, relative humidity (RH), wind speed (WS), wind direction (WD) and atmospheric pressure, were collected from the European Centre for Medium-Range Weather Forecasts (ECMWF) Reanalysis 5 (ERA5) dataset (Hoffmann et al., 2019). It is the ECMWF's latest reanalysis product and will replace the widely used ERA-Interim. ERA5 provides hourly estimates of a large number of atmospheric, land and oceanic climate variables, with a spatial resolution of 0.25 ° × 0.25 ° and resolve the atmosphere using 137 levels from the surface up to a height of 80 km (https://www.ecmwf.int/en/forecasts/datasets/reanalysis-datasets/era5). In this study, the near-surface (950 hpa) data was used with a constant time point of 06:00 (UTC, 14:00 local time).

Estimation of health effects

In order to evaluate the health effects of varying air quality, we calculated the excess risk (ER) of all six pollutants (ERtotal) as well as the health-risked based air quality index (HAQI) (Hu et al., 2015; Shen et al., 2020). ER of each pollutant is summed up to calculate the ERtotal. After calculation of the ERtotal, the equivalent concentration (C*) of a pollutant by assuming its ER equals to the ERtotal can be estimated, and the HAQI can be calculated in the same manner as AQI, by using C* of the six pollutants (Hu et al., 2015). Technical details can be found in Hu et al. (2015)). The HAQI is proven be a better index to reflect the combined health effects from all six commonly monitored pollutants than AQI (Shen et al., 2017, 2020). More details can be found in Text S1.

Data normalization

As air pollutants are values of different orders of magnitude, the correlations between each air pollutant could not be discussed at the same level as equivalent parameters. Only after non-dimensional data process can different physical variables be of significance to make comparisons. Therefore, a Z-score data normalization method has been applied to discuss in relationships between each air pollutant in this study. The method can be described in the following equation:where X is the measured value, Xnormalized is the processed normalized value, while Xmean and SD are the mean values and standard deviation of sample dataset of X during Stage I, respectively.

Results and discussion

Overview of air quality in BTH and YRD during COVID-19 pandemic

In this study, remarkably spatial distributions of the six pollutants levels in most of China were presented in the first four stages of the pandemic (from 1 January to 9 February 2020) (Fig. 1 ). In Stage I, as the whole country was still running normally, high NO2 levels were found over most parts of northern and northeastern, and some parts of southern China, especially in BTH (on average 30 ppb) and YRD (21 ppb). O3 presented the lowest average values of 10 ppb and 12 ppb in BTH and YRD, respectively. However, much higher values for PM2.5, SO2, CO, and PM10 were found in northern and northeastern China. In Stage II, a large number of people were on travel for the Chinese Spring Festival. NO2 had a great decrease in the southeastern part but no obvious change in the northeastern part. However, O3 increased more greatly in the southern part than that in the northern part. Subsequently, the distributions of PM2.5, SO2, CO, and PM10 in northern China consisted with the distribution pattern in Stage I. Stage III was the Chinese Spring Festival week, and most people were forced to stay at home because of the COVID-19 lockdown. As we can see from Fig. 1, NO2 and CO, were reduced about 70% and 30% in most regions of China. However, SO2 showed no obvious variation in most of China. As O3 concentration increased during Stage III owing to significant NO2 reduction, we can infer O3 produced by VOCs through complex photochemical reactions of volatile organic compounds (VOCs) and nitric oxide (NOx), which depends heavily on temperature and solar radiation (Kwok et al., 2015). Moreover, higher SO2 levels in northern part which might be contributed by coal combustion and biomass burning for heat-supply. PM2.5 was still at a very high level in the northern part, especially in BTH, which could reach as high as 120 μg m−3 on average, whereas PM10 was lower than that in Stage II. Some recent studies reveal that this was due to that secondary PM formation offsets the reduction of primary emission, which is further attributed to the enhancement of atmospheric oxidation capacity (Chang et al., 2020; Huang et al., 2020; Tian et al., 2020). Stage IV was still in the COVID-19 strict lockdown period. The mass concentrations of PM2.5 and PM10, together with NO2, SO2, CO, were much lower than those in other three stages, which might be attributed by primary emissions. O3 concentration showed significant increase during Stage IV owing to NO2 reduction, which is consistent with the first three stages. Hence, all the results related above can initially infer that the concentrations of the six pollutants during Stage IV were much lower than those in Stage I–III due to the great reduction of anthropogenic emissions.

Fig. 1

Spatiotemporal variations of PM2.5, O3, NO2, SO2, CO, and PM10 from Stage I to Stage IV (from top to bottom) in 2020. Here O3 value is the 8-h moving average.

In order to investigate the atmospheric aerosol evolution processes in the two developed regions (BTH and YRD) during the COVID-19 pandemic, the differences in the six major pollutants during the same period in 2019 and 2020, have been compared in this study (Fig. 2 ). Briefly, in BTH, the entire period-average mass concentration of PM2.5 in 2020 was 88% of that in 2019, whereas O3 increased 7%. NO2, SO2, CO, and PM10 were 81%, 67%, 89%, and 77% in 2020 of those in 2019, respectively. The implication is that anthropogenic emissions were largely reduced in BTH in the first season in 2020 when compared to those in 2019. For the cases, the PM2.5 concentrations increased gradually in the first three stages in 2020 and then dramatically decreased to 50% of Stage III from Stage IV to VII. As the heating supply ended on March 15 in the northern part of China, PM2.5 dropped to 20% of that in Stage III from Stage VIII to X. In the meantime, CO also showed a decrease from Stage VIII to X. PM2.5 in Stage III (Spring Festival week) in 2020 was even higher than that in Stage III of 2019, while the PM2.5 levels were similar in 2019 and 2020 for both Stage I and II. In Stage III, the differences of CO in 2019 and 2020, suggests there might be a different PM2.5 formation pathway during the Spring Festival of 2020. In comparison to the gradually increasing O3 in 2019, O3 in Stage III dramatically increased to two-fold of that in Stage II of 2020, indicating photochemical processes may act as a contributor to PM2.5 formation in Stage III. As expected, O3 was keeping at a high level in the other stages in BTH, and NO2, SO2, CO, and PM10 were at lower concentrations in 2020 when compared to those in 2019.

Fig. 2

Variations of PM2.5, O3, NO2, SO2, CO, and PM10 during the same period (from Stage I to X) for 2019 and 2020 in both BTH and YRD.

However, in YRD, the entire period-average mass concentration of PM2.5 was much lower than that in BTH in both 2019 (54 μg m−3 in YRD and 71 μg m−3 in BTH, respectively) and 2020 (39 μg m−3 in YRD and 63 μg m−3 in BTH, respectively). Besides, the concentrations of SO2, CO and PM10 were also significantly lower than those in BTH in both 2019 (SO2, CO, and PM10 were 47%, 74%, and 67% of that in BTH, respectively) and 2020 (SO2, CO, and PM10 were 56%, 70%, and 60% of that in BTH, respectively), whereas NO2 levels were similar between BTH and YRD (88% and 80% of that in BTH in 2019 and 2020, respectively). Regarding the differences in the industry-structure of these two regions, BTH was more polluted from winter to early spring, and more SO2 and CO were released to the atmosphere.

Implications of the variations of PM2.5, O3, NO2, and CO for the economic activities during the COVID-19 pandemic

Fig. 3 displays the temporal variations of normalized ratios of PM2.5, O3, NO2, and CO by their mean values from pre-COVID-19, in BTH and YRD, respectively. The normalized ratios of primary concentrations, e.g., NO2 and CO, present better correlation (r2 of 0.47, Figure S2) in BTH than that in YRD (r2 of 0.14, Figure S2), which suggests that there are great differences in emission sources between BTH and YRD. The significant difference between the normalized ratios of CO and NO2 during COVID-19 lockdown (Stage III–IV) in both BTH and YRD, along with the increasing PM2.5, which can be initially inferred indicated the dominant contribution of secondary processes on PM2.5 during those stages. The normalized NO2 and CO were kept less than 1 during the whole recovery period from Stage V to X. Since NO2 and CO are two primary emissions directly related to anthropogenic activities, illustrating that the economic activities/anthropogenic emissions in BTH were still not fully recovered during this period. In comparison, the normalized CO and NO2 in YRD were close to those in Stage I, suggesting that the recovery of economic activities in YRD was better than that in BTH.

Fig. 3

Temporal variations of normalized ratios of PM2.5, O3, NO2, and CO by their mean values of pre-, during- and after-lockdown in (a) BTH and (b) YRD.

Fig. 4 showed different relationships between each air pollutant. CO, originated from primary sources (Khalil and Rasmussen, 1988), can help to distinguish the primary and secondary sources of PM2.5. Therefore, the variation of normalized PM2.5 with normalized CO is displayed in Fig. 4a. In BTH, the normalized PM2.5 in heavy-polluted periods (Stage I–III) increased from 1 to 1.6, however, there is no significant change in the normalized CO, which indicates that partial PM2.5 was likely attributed to the secondary processes in Stage II and Stage III (Spring Festival week). During the less-polluted periods (Stage IV to Stage X), the normalized PM2.5 and normalized CO was basically on the 1:1 line in BTH, suggesting the PM2.5 pollution was mainly from primary emissions. In comparison, in YRD, the normalized PM2.5 shows a poor correlation with normalized CO, indicating different sources, including primary emissions and secondary formation. However, the normalized PM2.5, as a function of O3, shows a remarkable difference between the heavy-polluted periods and less-polluted periods for both BTH and YRD. O3 in the heavy-polluted periods varied from 1 to 1.8, which were 2–3 times lower that in the less-polluted periods. Meanwhile, the normalized PM2.5 increased with the increasing of the normalized O3 in the heavy-polluted periods. While in the less-polluted periods, there is no significant relationship between the normalized PM2.5 and O3. Together with the relationship between PM2.5 and CO discussed above, we can infer that PM2.5 in the heavy-polluted periods were mainly associated with both primary and secondary formation but primary emission contributed most to PM2.5 in the less-polluted periods. Actually, O3 pollution depends on PM2.5 concentrations and its chemical composition (Li et al., 2019). The increase of O3 could be driven by decreasing PM2.5 concentrations due to HO2 and NOx radicals scavenged by fine particles (Li et al., 2019; Wang et al., 2017). However, when compared with PM2.5, the origins of and factors influencing O3 are more complex.

Fig. 4

The relationships between (a) normalized PM2.5 and normalized CO, (b) normalized PM2.5 and normalized O3, (c) normalized PM2.5 and normalized NO2, and (d) normalized O3 and normalized NO2 in BTH and YRD.

The relationship of the normalized PM2.5 and the normalized NO2 was shown an opposite result (Fig. 4c). Negative relationships between NO2 and O3 can be found in Fig. 4d. Furthermore, the normalized O3 displayed a negative correlation with the normalized NO2, especially in BTH (Fig. 4d). The implication is that more primary emissions may produce more O3 through photochemical reactions with NO2 during the heavy-polluted periods, while high O3 concentrations in the less-polluted periods may attribute to higher intensive of more VOCs emissions.

Figure S3 illustrates the maps of mean values of temperature (T), relative humidity (RH), wind speed (WS), and wind direction (WD) at 950 Pa in China from the EPA5 reanalysis data. Stage III in BTH was under a cold (average T of −0.8 °C), high moisture (average RH of 63%), stable (average WS of 1.5 m s–1 with regional cycling air masses) meteorological condition. Subsequently, a huge amount of primary emissions was accumulated in BTH in Stage III. In addition, the Stage III (Spring Festival week) in BTH was of particular interest. In Stage III, most factories were shut down or set at a lowest operating efficiency. In addition, the vehicle emissions have also been reduced to the lowest because of lockdown. However, the averaged mass concentration of PM2.5 (120 μg m–3) was even higher than those in Stage I and II, which was the highest during the entire period in 2020. Hence, it aroused a nationwide concern of people.

Moreover, photochemical processing may play a significant role in the increase of PM2.5 pollution as displayed in Figure S4. CO and SO2 in Stage III were comparable to those in Stage I and Stage II, however, NO2 varied greatly and showed obvious negative correlations with PM2.5 and O3, respectively. The implication is that NO2 was oxidized to nitrate in PM2.5, especially organic nitrate, such as RO2NO2 (PAN) and RONO (methyl nitrate). Actually, the reduction of NOx emission could lead to an increase of organic nitrate due to the increase of O3 and consequently a major availability of OH radical that can oxidize VOCs (Huang et al., 2021). However, there is no evidence to support the observation of organic nitrate in gaseous and particulate phase in this study.

Health effect

In this study, the health risks from these six pollutants were also estimated. Here we used the threshold value of ambient air quality standards of China (CAAQS) Garde II. Figure S5 presents averaged excess risks (ERs) of individual pollutants and the total ERs in the entire study duration in 2019 and 2020. Overall, the total ER decreased slightly, from 3.04% in 2019 to 2.92% during the COVID-19 epidemic in 2020, which is due to the national lockdown strategy and emission reduction in each industrial section. The main health risk factors were PM2.5, PM10, NO2, of which PM2.5 contributed the most, accounting for 53% in 2019 and 54% in 2020, followed by PM10 with the proportion of 45% in 2019 and 44% in 2020, and finally NO2, accounting for 2% in 2019 and 1% in 2020. In addition, the ERs of other pollutants (SO2, O3, and CO) can be ignored. The averaged HAQI values from January 1 to April 11 in both 2019 and 2020 in most of China are shown in Fig. 5 . Overall, it was found that the averaged HAQI values in 2020 was 75.9, a decrease of 14.4% over the same period in 2019 (88.7). Relative higher averaged HAQI values (>100) in both 2019 and 2020 were observed in northern China, especially in BTH, indicating higher health risks in these regions. Apparently higher HAQI values (>100) can be found in BTH during the same period in both 2019 and 2020. However, almost all people in YRD lived in healthy air with the HAQI values less than 100. Obvious declines of the HAQI values in 2020 in BTH and YRD have been observed with decreases of 24.5% and 18.1% when compared to those in 2019, respectively. Furthermore, these results related above show that there are still great challenges existing in air pollution of China. We can infer that obvious declines of the health risks exposed to air pollution from 2019 to 2020 indicated the response to the large reduction of the anthropogenic emissions.

Fig. 5

Distribution of annual mean HAQI values in most of China in (a) 2019 and (b) 2020.

Fig. 6 illustrates the HAQI values in different stages for both BTH and YRD in 2019 and 2020. The HAQI values in 2020 for all stages were lower than those in 2019, except Stage III in BTH, which showed the value in 2020 was 1.3 times higher than that in 2019. The differing results might be explained by the dramatic increase of PM in Stage III as discussed above. Besides, Shen et al. (2020) also found that HAQI values were responsible for other socioeconomic factors, such as total population, population density and built-up area, secondary industry. Interestingly, in YRD, the HAQI values decreased with the decreasing of PM concentrations but increasing of O3, indicating the relatively negligible O3 contributions to HAQI. Actually, the health effect of O3 could not be negligible in reality due to their contribution to respiratory tract disease and cardiovascular disease. This is because the lower β values of O3 in the HAQI estimation method. Therefore, the parameters of HAQI estimation should be adjusted in the future work.

Fig. 6

HAQI values in different stages in BTH and YRD for both 2019 and 2020.

Conclusion

In this study, we analyzed ambient monitoring data of six air pollutants including PM2.5, PM10, NO2, SO2, CO from January 1 to April 11 in both 2020 and 2019 to investigate the aerosol evolution processes before, in, and after the lockdown period of COVID-19 pandemic. The six ambient pollutants concentrations were found to be much lower during the COVID-19 lockdown because of the declines of anthropogenic emissions. Regarding the differences in industrial structures of BTH and YRD, BTH was more polluted with more SO2 and CO release. The significant difference between the normalized ratios of CO and NO2 during COVID-19 lockdown, along with the increasing PM2.5, indicated the dominant contribution of secondary processes on PM2.5. NO2 showed obvious negative correlations with PM2.5 and O3, suggesting a large number of NO2 was oxidized to nitrate in PM2.5. In addition, the national total ER decreased slightly during the COVID-19 pandemic, which is due to the national lockdown strategy and emission reduction in each industrial section. The most health risk factor was PM2.5 followed by PM10 and NO2, while other pollutants (SO2, O3, and CO) can be ignored. Obvious declines of the averaged HAQI values in BTH and YRD have been observed with decreases of 24.5% and 18.1%, respectively. Our results will provide experimental opportunity to investigate the air pollution response to the large reduction of the anthropogenic emissions.

Credit author contribution statement

Junfeng Wang, planed and designed the research, conducted the statistical analysis, contributed to writing the manuscript. Yali Lei, planed and designed the research, conducted the statistical analysis, were responsible for reviewing and editing the manuscript. Yi Chen, were responsible for reviewing and editing the manuscript, supervised the study. Yangzhou Wu, conducted the statistical analysis. Xinlei Ge, supervised the study. Fuzhen Shen, conducted the statistical analysis. Jie Zhang, conducted the statistical analysis. Jianhuai Ye, were responsible for reviewing and editing the manuscript. Dongyang Nie, were responsible for reviewing and editing the manuscript. Xiuyong Zhao, were responsible for reviewing and editing the manuscript. Mindong Chen, supervised the study. All authors read and approved the manuscript.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.