Introduction

In early December 2019, the novel coronavirus pneumonia named COVID-19 by the World Health Organization (WHO) was initially identified in Wuhan, the capital city of Hubei province (Huang et al. 2020; Guan et al. 2020; WHO, 2019). As of May 28, 2020, COVID-19 had spread rapidly on a global scale with 5,491,678 confirmed cases, 349,190 confirmed deaths in 217 countries, areas or territories, as reported by the WHO (https://www.who.int/emergencies/diseases/novel-coronavirus-2019). To curb the spread of COVID-19, Wuhan and other cities of Hubei province announced lockdowns on January 23 and 24, 2020, respectively, followed by other provinces in China, by declaring the highest level of a public health emergency on January 26, 2020 (Tian et al. 2020). As part of the national emergency response, the Chinese government established nationally strict curfews and isolation policies, named ‘lockdown’, for instance:

  • extending the Spring Festival holiday,

  • suspending all public transportation including city buses, subways, ferries and long-distance coaches,

  • closing outbound channels at airports and railway stations,

  • closing schools and entertainment venues,

  • banning public gatherings,

  • enhancing social distancing, and

  • issuing stay-at-home orders and closing non-essential businesses.

Future analyses will evaluate the extent to which these approaches have been effective, and whether potential benefits have outweighed the costs, e.g., economic losses. Non-essential businesses in most areas resumed work after February 10, although non-essential businesses in Wuhan remained closed until March 20.

The all-out massive efforts of the Chinese government by sealing off all cities in a bid to curb the spread of the COVID-19 provide an unprecedented opportunity to study the possible effects of anthropogenic emission on air quality. NASA and European Space Agency (ESA) pollution monitoring satellites, i.e., the Tropospheric Monitoring Instrument (TROPOMI) on ESA’s Sentinel-5 satellite, detected significant decreases in NO2 over China based on the data from January 1–20, 2020, before the quarantine, to February 10–25, during the quarantine (NASA 2020). There is evidence that the changes were at least partly related to the economic slowdown following the coronavirus outbreak. By comparison of NO2 values in 2020 with the average values detected at this time of year from 2005 to 2019, NASA determined that NO2 concentrations in eastern and central China in 2020 were 10–30% lower than those normally observed for this time period (NASA 2020).

On the other hand, despite reduction in activities at the outbreak of COVID-19, the Beijing-Tianjin-Hebei cluster remained shrouded in smog during the Spring Festival holiday in 2020 despite the expectation that air pollution would remain low (MEE 2020; China Daily 2020). On Wednesday, February 5, 2020, hourly concentrations of PM2.5—fine particles with an aerodynamic diameter lower than 2.5 μm—in most districts in Beijing exceeded 200 μg m−3 with worsening haze over the Lantern Festival on Saturday, February 9 (MEE 2020). It was estimated that from January 24 to February 2, 2020, pollutant emissions, with 77% fewer freight vehicles and 39% fewer cars on the roads, decreased by at least 40% in the 28 major cities in the Beijing-Tianjin-Hebei (BTH) cluster (MEE 2020). However, emissions from industries such as steel, nonferrous metals, glass, cooking and heating in the BTH cluster remained generally unchanged, despite a suspension of activities by many restaurants and labor-intensive factories (MEE 2020). There was a decrease of only 10% in emissions by steel plants and coal-burning power generation sectors with no decrease in other smokestack industries. About 10 million rural households in the BTH region relied on coal for heating, with some shift to cleaner energy like natural gas. Coal consumption for heating increased as reflected by more than 10% enhancement of CO concentrations, relative to last year in the BTH rural areas. Another major factor for poor air quality in the BTH region during the cold periods was the unfavorable meteorological conditions for pollutant dispersal, such as mixing layer heights of only 300 to 500 m, half to one-third of their normal heights in the same season in MEE (2020).

Elevated PM2.5 is a major factor controlling regional climate and human health (van Donkelaar et al. 2010; Yu et al. 2014; Wang et al. 2014a; Rosenfeld et al. 2019; Pope 2000; Seinfeld and Pandis 2016; Chen et al. 2013). The main driver of decreases in PM2.5 in China between 2013 and 2017 was strict emission control policies (Zhang et al. 2019; Ding et al. 2019). Efforts to improve air quality during large international events, such as the 2008 Beijing Olympics, the 2014 Beijing Asia–Pacific Economic Cooperation (APEC) summit, the 2015 Grand Military Parade, and the 2016 Hangzhou G20 summit, have included short-term closure of power plants and factories, restriction of traffic, and reduction in construction activities (Xing et al. 2011; Liu et al. 2016; Li et al. 2017; Wang et al. 2017).

Hangzhou, the capital of Zhengjiang Province situated in the Yangtze River Delta and one of the most famous and prosperous cities in China (Fig. 1), has a population of 7.03 million and covers an area over 16,000 km2 (http://xzqh.mca.gov.cn/). Influenced by subtropical monsoon climate, Hangzhou has four distinctive seasons. Surrounded by mountains on three sides, this city belongs to basin geomorphology which does not favor pollution dispersal. Due to the implementation of pollution control strategies such as restructuring of industry and energy, coal consumption decreased from 14.6 million tons in 2005 to 13.4 million tons in 2013, despite doubling the numbers of vehicles from 1.1 to 2.5 million (http://www.hangzhou.gov.cn/col/col805867/).

Fig. 1
figure 1

a Distributions of 11 air quality monitoring station locations in Hangzhou. b Time series of observed hourly concentrations of PM2.5, PM10, SO2, CO, NO2, O3 at urban and rural sites in Hangzhou city from January 1 to February 28, 2020. Comparing pollutant levels before lockdown and during lockdown, concentrations of PM2.5, PM10, CO, NO2 displayed significantly decreases, while overall augmentations were observed for O3 in both urban and rural areas. The patterns began to reverse after February 15 due to the resumption of work and production activities

Full size image

Heavy photochemical pollution and haze episodes occur frequently with exceedances in 2013 of 38 days for O3 and 62 days for PM2.5 for the secondary Chinese Ambient Air Quality Standards (CAAQS): PM2.5: 75 μg m−3 for 24 h average and O3: 75 ppbv for daily maximum 8 h (Zhang et al. 2018). During a severe haze episode from December 3–9, 2013, PM2.5 mean concentrations reached as high as 293.4 ± 103.2 μg m−3 (Yu et al. 2014). Local emissions, complex atmospheric conditions, and regional transport exert combined effects on air pollution occurrence in Hangzhou (Yu et al. 2014; Li et al. 2015; Wu et al. 2018; Chen et al. 2020). On the other hand, Zhejiang, known for its thriving private sector, was the first province both to declare a highest level response to coronavirus on January 23, 2020 and to relax control measures and support work resumption (https://www.scmp.com/economy/chinaeconomy/article/3049821/coronavirus-chinese-province-orders-relaxation-excessive). As the capital of Zhejiang province, Hangzhou city declared the first-level public health emergency response on January 23, 2020 (http://www.zj.gov.cn/art/2020/1/26/art_1228996604_41859419.html).

A recent editorial underlines the importance and relevance of research in environmental chemistry to understand pandemics (Sharma et al. 2020). Here we study the impacts of the city lockdowns due to the COVID-19 outbreak on air quality in Hangzhou from January 1 to February 28, 2020. Our hypothesis was that air pollutants should have decreased during lockdown due to the absence of activities and traffic. This report reveals that, on the contrary, the concentrations of some air pollutants increased in both urban and rural areas.

Experimental

Hourly observational data of the six pollutants, PM2.5, PM10, SO2, CO, NO2, O3, at the monitoring sites from January 1 to February 28, 2020, were collected from the website of China National Environmental Monitoring Center (CNEMC) (http://www.mee.gov.cn/hjzl/dqhj/). In addition data were acquired at 11 air quality monitoring stations are set up in Hangzhou, as shown in Fig. 1a, including 10 urban sites (Binjiang, Xiasha, Wolongqiao, Zhejiangnongda, Zhaohuiwuqu, Hemuxiaoxue, Linpingzhen, Xixi, Yunqi and Chengxiangzhen), and one rural background site (Qiandaohu), a lake park located about 170 km away from Hangzhou center. Average concentrations at the 10 urban sites were estimated to represent urban conditions, while the concentrations at the rural site were used to represent rural conditions. According to the China National Ambient Air Quality Standard (CNAAQS, GB3095-2012), the 24-hour average levels of each pollutant are regarded as valid only under if at least 20 h of hourly data are acquired. The 8 h-O3 concentrations could be used only with more than 6 h available data for each 8 h period. In this study, the data in 2016, 2017, 2018 and 2019 were also used for comparison purpose.

Data of population (7.03 millions) and area (16,596 km2) of Hangzhou city were acquired from the Civil Affairs Ministry of the People’s Republic of China (http://xzqh.mca.gov.cn/). The change of the Baidu Migration Scale Index to characterize the daily travel intensity index inside Hangzhou was downloaded from the public Baidu Maps website (http://qianxi.baidu.com/). Daily traffic density was calculated using the travel intensity index multiplied by population, and then divided by area.

Results and discussion

Trends of air pollutants in early 2020

Overall trends

Figure 1b shows the time series of hourly concentrations of six pollutants at the urban and rural sites from January 1 to February 28, 2020, with the corresponding statistical values summarized in Table S1. To analyze the temporal variations of concentration and impacts of the lockdown, the entire study period was divided into three periods: before lockdown (BL, January 1–23), during lockdown (DL, January 24–February 15), and resumption (February 16–28). Results show that when comparing pollutant levels before and during lockdown, concentrations of PM2.5, PM10, CO, NO2 decreased significantly, as expected. However, results also show an overall increase in ozone (O3) levels both in both urban and rural areas. After February 15, these trends began to reverse due to the resumption of work and production activities.

Particulate matter

In the urban area, hourly average concentrations of PM2.5 decreased from 45.9 ± 31.3 μg m−3 before lockdown to 29.6 ± 19.5 μg m−3 during lockdown (Table S1). Similarly, at the rural site hourly PM2.5 levels decreased from 25.0 ± 24.0 to 20.1 ± 16.9 μg m−3. In the same way, mean PM10 decreased significantly from 70.5 to 48.4 μg m−3 in the urban area and from 31.0 to 24.1 μg m−3 at the rural site. Then, both PM2.5 and PM10 levels rose to higher values after resumption. It is known that coal combustion, biomass burning, vehicle emissions, and industrial emissions are the main anthropogenic sources of PM2.5 (Chow and Watson 2002; Cheng et al. 2014; Wu et al. 2016; Wang et al. 2016a; Mehmood et al. 2018, 2020). Therefore, our observed reduction in PM2.5 concentrations in Hangzhou city during lockdown is mainly due to limited industrial activity and traffic (Wang et al. 2012; Lyu et al. 2016; Zhang et al. 2017). In the rural area, reduction in PM2.5 and PM10 can be ascribed to restricted vehicle travel and lower impacts of regional transports. The consistent patterns of PM2.5 and PM10 in both urban and rural areas suggested the important contributions of anthropogenic emission sources and impacts of regional transport during the lockdown.

Carbon monoxide

CO decreased in both urban areas from 1.0 to 0.59 mg m−3, and in rural areas from 0.78 to 0.60 mg m−3. Since CO mainly originates from fossil fuel combustion and biomass burning (Zhang et al. 2009; Chai et al. 2014), decreasing anthropogenic emissions after reduction in industrial activities explain at least partly the CO patterns in both urban and rural areas.

Sulfur dioxide

SO2 mean concentrations in the urban area decreased from 6.3 to 5.3 μg m−3 after lockdown, then increased to 5.6 μg m−3 in the resumption period. However, surprisingly, SO2 mean concentrations at the rural site increased from 4.7 to 6.1 μg m−3 continuously during the entire period (Table S1). The unexpected increase in SO2 at the rural site is explained by the increase in coal consumption for heating coupled with inputs from the traditional fireworks during Spring Festival Eve and the Lantern Festival. Overall, SO2 concentrations remained low, which resulted from the implementation of control of SO2 emission (Fang et al. 2009), as observed in the Yangtze River Delta (Huang et al. 2012).

Nitrogen dioxide

Concentrations of NO2 at the urban sites dropped from 44.2 to 13.0 μg m−3 after lockdown, then slightly increased to 21.8 μg m−3 during resumption (Fig. 1, Table S1). At the rural site, NO2 levels decreased sharply from 21.0 μg m−3 to 6.8 μg m−3 during lockdown, then remained at very low levels of 6.4 μg m−3 during the resumption period. These findings are explained by lower NOx emissions from local vehicle emissions, power plants and industrial activities (Wang et al. 2014b, 2016b) during the lockdown, whereas rural decreases in NO2 concentrations might result from less vehicle emissions and less pollution transport from urban areas.

Ozone

Contrary to the dramatic decreases in NO2 during the lockdown, relatively low O3 levels in the urban area occurred before lockdown with almost all the values below 100 μg m−3, then began to reach higher concentrations during lockdown with the maximum over 139 μg m−3 as shown in Fig. 1. Mean concentrations of O3 increased from 24.6 μg m−3 before lockdown to 60.6 μg m−3 during lockdown in the urban area, whereas a lower increase in O3 levels from 42.0 to 62.9 μg m−3 occurred in the rural area (Table S1). Then, during the resumption, the mean O3 concentrations increased slightly in both urban and rural areas. As a secondary pollutant formed by photochemical reactions of nitrogen oxides (NOx) and volatile organic compounds (VOC) (Seinfeld and Pandis 2016), O3 will decrease with decrease in NOx in the NOx sensitive regime. The unexpected increase in O3 levels during lockdown with the decreases in NO2 is explained by the weakening of chemical titration of O3 by NO during the nighttime, resulting in increases in background O3 concentrations, as shown in Fig. 2.

Fig. 2
figure 2

a Diurnal variations of the mean O3 concentrations for the periods of before (BL) and during (DL) lockdowns in the urban and rural areas in Hangzhou in 2020 and 2019; b Comparisons of the weekly mean values of NO2 and PM2.5 concentrations in the years of 2016, 2017, 2018, 2019 and 2020. Before Lockdown (BL), During Lockdown (DL), and Resumption denote the periods of January 1–January 23, January 24–February 15, and February 16–February 28, respectively. The diurnal variations of mean O3 concentrations for the periods of BL and DL in (a, b) clearly indicate that the mean O3 concentrations for the non-photochemical reaction times from 20:00 PM to 9:00 AM (local time) during lockdown in 2020 were 53.7 and 58.0 μg m−3 in the urban and rural areas, respectively, much higher than the corresponding values (i.e., 18.8 and 35.6 μg m−3 in the urban and rural areas, respectively) before lockdown

Full size image

In 2020, from 20:00 p.m. to 9:00 a.m. local time, the mean O3 concentrations for the non-photochemical reaction time period increased from 18.8 μg m−3 before lockdown to 53.7 μg m−3 during lockdown in urban areas, and from 35.6 to 58.0 μg m−3 in rural areas (Fig. 2). By comparison, in 2019, the mean O3 concentrations for the same time periods increased from 10.0 to 35.9 μg m−3 in urban areas, and from 30.6 to 47.8 μg m−3 in rural areas. The higher levels during the lockdown time period are partly explained by higher atmosphere temperatures for late January/early February versus January 1–23. Apart from higher levels in 2020 for the whole time period, Fig. 2 shows meaningful higher O3 concentrations at night in 2020 relative to those in 2019. This confirms indirectly the fact that the O3 increase in 2020 during lockdown is due to the weakening of chemical titration of O3 by NO during the nighttime, resulting in increases in background O3 concentrations.

Comparisons of air quality between 2019 and 2020

Daily average concentrations of ambient PM2.5, PM10, CO, NO2, SO2 and daily maximum 8 h-O3 concentrations in 2020 relative to those in the same period of 2019 in the urban and rural areas are displayed in Fig. 3 and Table S1. PM2.5 and PM10 concentrations in the urban areas decrease sharply on January 24, and then remained below 75% during a few days January 24–27 (Fig. 3a,b). Similar drops of 96% for PM2.5 and 95.6% for PM10 were observed at the rural site on January 26. CO and SO2 exhibited no significant change in the urban areas, whereas rising trends were observed at the rural site. This is explained by increasing biomass burning for residential heating or cooking in the rural area. At both urban and rural sites, NO2 decreased and the maximum drop appeared during January 24–30, as expected. Daily maximum 8 h-O3 concentrations were enhanced during almost all days in January and February in 2020 relative to 2019, although the larger enhancements existed before lockdown in both urban and rural areas (Fig. 3).

Fig. 3
figure 3

Change percentages of the daily average concentrations of ambient PM2.5, PM10, CO, NO2, SO2 and daily maximum 8 h-O3 concentration during January and February in 2020 relative to those in 2019 in Hangzhou in (a) the urban area and (b) the rural area. As indicated in Fig. 3a and Fig. 3b, PM2.5 and PM10 concentrations in the urban areas experienced a sharp decrease on January 24, and then remained relatively larger negative percentages (below − 75%) during a few days from January 24–27. The simultaneous quick drops after January 24 also occurred for PM2.5 and PM10 levels at the rural site with the minimum values as low as − 96% and − 95.6% on January 26, respectively

Full size image

For the urban area, before the lockdown period, January 1–23, 2020/2019 decreases amounted to 24.8% for PM2.5, 19.8% for PM10, 29.2% for SO2, 14.1% for CO and 13.7% for NO2. By comparison, the decreases were much higher during the lockdown period, January 24–February 15: 42.7% for PM2.5, 47.9% for PM10, 22.3% CO and 58.4% for NO2; except for SO2 (28.6%) (Table S1). Higher decreases after January 24 reveal the reductions in industrial activities and traffic in Hangzhou. The similar SO2 changes are in agreement with the contribution of coal combustion for heating and firework emissions during lockdown.

For the rural area, only PM2.5, PM10, and NO2 showed significantly higher decreases of − 18.5%, − 39.6%, − 48.0% during the lockdown period, respectively, compared with the corresponding 4.4%, − 27.0%, − 4.8% before lockdown (Table S1). Whereas SO2 and CO exhibited weak decreases during lockdown, O3 concentrations exhibited an increase during the whole period at all sites relative to those in 2019. With similar decrease percentages of NO2 at the urban sites (− 58.4%) and rural site (− 48.0%) during January 24–February 15, larger enhancement percentages of O3 were observed at the urban (22.2%) relative to rural (15.7%) sites in 2020 in comparison to those in 2019. Moreover, the weekly mean concentrations of NO2 and PM2.5 in 2020 were consistently lower than those in the same periods of 2016, 2017, 2018 and 2019, due to the emission reductions because of the city lockdown (see Fig. 3b) in addition to the possible effects from the differences in meteorological conditions for each year.

Correlations between NO2 and city traffic index in Hangzhou

NOx, including NO and NO2, are major precursors of PM2.5 and O3, and are primarily emitted as NO from anthropogenic combustion sources including transportation, power plants, industries and residential combustion, and from natural sources such as soil, lightning, and wildfire. NOx is an excellent tracer of human activity (Chen et al. 2013; Yu et al. 2014). Our results show that NO2 concentrations in both urban and rural areas in Hangzhou dropped sharply during the lockdown (Figs. 1, 2). Higher temperatures, stronger sunlight and increasing precipitation in February can decrease NO2 relative to those in January for fixed emissions in Hangzhou.

To understand the impact of traffic flows on the NO2 in Hangzhou, Fig. 4a displays changes in daily traffic density against the changes in daily NO2 mean concentrations in 2020 relative to those in 2019. We found a significant positive correlation (correlation coefficient (r) = 0.51) between the change in traffic density and NO2 in the urban area. This finding reinforces the hypothesis by which traffic reduction is at least partly responsible for NO2 decreases. By contrast, traffic density and NO2 is very weakly correlated at the rural site (r = 0.18). This suggests that NO2 does not mainly originate from local traffic in rural areas, but may come from long-distance transport. It has been shown that regional transport of NOx and VOCs from urbanized upwind areas can lead to high O3 levels (Sikder et al. 2011; Monteiro et al. 2012; Yu et al. 2006; Li et al., 2015). Moreover, NO2 exhibits a strong correlation with PM2.5 in the urban area of Hangzhou (r = 0.51, P < 0.05) (Fig. 3d), indicating local vehicle emissions are an important source of PM2.5.

Fig. 4
figure 4

a Scatter plots and linear regressions of the changes in daily traffic density against changes in daily concentrations of NO2 during January and February in 2020 relative to 2019 at urban and rural sites. b Scatter plots and linear regression of daily (24-h) NO2 against maximum 8-h O3 (MDA8 O3) during January 1–February 28, 2020, at urban and rural sites. c Scatter plots of 24-h PM2.5 against MDA8 O3 and boxplots of MDA8 O3 binned by daily (24-h) PM2.5 (at least three data points required) during January 1–February 28, 2020 at urban and rural sites. The boxes represent the 75th, 50th, and 25th percentiles from top to bottom, the whiskers and solid squares denote 1.5 times of IQR and mean values, respectively. d Scatter plots and linear regression of 24-h PM2.5 against 24-h NO2 during January 1–February 28, 2020 at urban and rural sites. The significant positive correlation (r = 0.51) between the change in traffic density and NO2 in the urban area of Hangzhou in Fig. 4a further explains NO2 decreases during lockdown in 2020 due to direct traffic emission reductions. A weak positive correlation between the change in traffic density and NO2 at the rural site in Fig. 4a indicates that NO2 in the rural area did not arise mainly from direct traffic emissions rather from regional transport

Full size image

Inter-species correlations in the observations

Negative correlations are found between daily maximum 8-h O3 and mean NO2 levels in both urban and rural areas (Fig. 4b). This suggests a VOC-limited regime of O3 formation prevailing during winter time periods in Hangzhou, in agreement with conclusions reported in Shanghai and Nanjing (Ding et al. 2013; Gu et al. 2020). Notably, a significantly weaker positive response is observed under low NO2 levels, below 20 μg m−3, showing the shift of O3 production to the transition regime or even NOx-limited regime, especially in the rural area. At low VOC/NOx ratios, chain termination reaction of OH and NO2 produces a VOC-limited regime. Conversely, at high VOC/NOx ratios, the primary chain termination and propagation reactions involve OH and VOCs, and oxidation of NO by HO2 and RO2, respectively, producing an NOx-limited regime (Seinfeld and Pandis 2016).

Little correlation was observed between daily mean PM2.5 and maximum 8-h O3 concentrations (Fig. 4c). Here, we displayed boxplots of maximum 8-h O3 concentrations versus daily mean PM2.5 concentrations. The initial PM2.5 bin was 5 μg m−3 and rose to 10 or 15 μg m−3 under high PM2.5 concentrations. Overall, median maximum 8-h O3 concentrations increased along with daily mean PM2.5 at low to moderate PM2.5 levels, then declined for daily PM2.5 concentrations > 40 μg m−3 as shown in Fig. 4c. The positive correlation between PM2.5 and O3 is associated with the conversion of gaseous organics to PM2.5 under OH and O3 oxidation, forming secondary organic aerosols (Seinfeld and Pandis, 2016). The observed decline in O3 driven by high PM2.5 concentrations could be attributed to reduced solar radiation due to extinction by particulate matter, depositional sink of HO2 and NOx radicals, and chemical titration of NO (Buysse et al. 2019; Li et al. 2019; Yu et al., 2014).

Conclusion

We investigated the effects of city lockdowns on air quality based on ground-based observations in both urban and rural areas in Hangzhou. Results indicate that during lockdown relative to the period before lockdown, concentrations of PM2.5, PM10, CO, and NO2 displayed significantly decreasing trends, as expected, while an overall rising trend was observed for O3 in both urban and rural areas, which is explained by the ‘absence’ of NOx. The sharp decreases in the NO2 concentrations in the urban area during the lockdown were mainly a result of direct traffic emission reductions, whereas in the rural area decreases were mainly from less regional transport. SO2 mean concentrations in the urban area decreased from 6.3 to 5.3 μg m−3, as expected, but increased in rural areas from 4.7 to 5.8 μg m−3, a consequence of increases in coal combustion for heating and firework emissions. Mean O3 concentrations increased significantly from 24.6 before lockdown to 60.6 μg m−3 during lockdown in the urban area, whereas the corresponding rural O3 levels modestly increased from 42.0 to 62.9 μg m−3. Here the O3 increase in the urban areas is explained by the reduction in NOx emissions. Overall, the city lockdown improved air quality by reducing PM2.5, PM10, CO, and NO2, but also decreased air quality by augmenting O3 in both urban and rural areas and SO2 in the rural area in Hangzhou.