*2.4. Method of Analysis and Validation*

The data collected were divided into two groups. One is the data in 2020 representing the pandemic period, and the other is data in 2018 and 2019 representing the historical period. Each group was divided into four subgroups according to the PPC level, that is, no PPC period, first-level PPC period, second-level PPC period and third-level PPC period. No PPC period is 1–21 January 2020. In order to reduce the impact of the first reported case on residents' lifestyles, the data from 22–24 January were excluded. The end of the statistical date for the third-level response was set to 30 April to ensure that the numbers of days in periods under different response levels are similar, so as to avoid the interference brought by the time length. Table 1 lists the time ranges for different response levels (http://www.nanjing.gov.cn/zt/yqfk/zccs/202001/t20200127\_1782811.html, accessed on 1 March 2022).

**Table 1.** Time ranges for response levels of pandemic prevention and control of COVID-19 in Jiangsu Province.


The daily average AQI value of the whole province was used to calculate the grade of daily air quality, and the grades I–VI correspond to the air quality of very good, good, slightly polluted, moderately polluted, heavily polluted and severely polluted. The days with air quality of grades I–II are good air days, and the days with air quality of grades III–VI are polluted air days.

Based on the time-series analysis method, the air quality and concentrations of six atmospheric pollutants in different PPC periods in 2020 were compared with those during the same periods in 2018–2019 to explore the variations of air quality under different PPC levels in Jiangsu. The rate of change (CR) in the period without PPC in 2020 over the same period in previous two years was regarded as the natural change rate (NCR) in 2020. During the periods in first-level (L1), second-level (L2) and third-level (L3) responses to PPC, CR minus the NCR was regarded as the change rate under the PPC conditions (PCR). The formulas of CR, NCR and PCR are as follows.

$$\text{CR} = \frac{X\_{\text{rec}} - X\_{\text{his}}}{X\_{\text{his}}} \times 100\%\_{\text{r}}$$

NCR = CR, in the period without PPC,

PCR = CR − NCR, in the period with PPC,

where *Xrec* is the air quality or concentration of six atmospheric pollutants in 2020, and *Xhis* is the air quality or concentration of six atmospheric pollutants during the same periods in 2018–2019. The PCR of the mean and extreme values were analyzed separately to study the impact of the implementation of PPC policies on air quality. The analysis of mean values used the daily mean values of AQI, pollutant concentrations and meteorological elements, and the analysis of extreme value used the maximum daily mean values of AQI, pollutant concentrations, the minimum daily mean value of visibility, and the maximum daily mean values of other meteorological elements. In order to verify the significance of the differences in air quality before and after the outbreak of COVID-19 pandemic and

among different levels of PPC measures, we first used the *F*-test to test the variance of samples in two different periods, and then carried out the Student's *t*-test of equal variance or heteroscedasticity according to the results of the *F*-test. If *p* < 0.05 in the two tailed *t*-test, it was regarded as a significant difference. The variations of air quality in 13 prefecture-level cities of Jiangsu under different PPC levels were compared, so as to explore the differences in the response of urban air quality variations to the PPC levels under different economic development levels.

#### **3. Results**

## *3.1. Differences in Air Quality in Different Scenarios*

Figure 2 shows the number of days for each grade of air quality at the four PPC stages from 2018 to 2020. It can be seen that on the no PPC stage (Figure 2a), the number of days with air quality of grade I in 2020 increased over previous years, and the number of days with air quality of grade V decreased. Overall, there were few changes in the past 3 years. In the first-level PPC period (Figure 2b), the number of days with air quality of grades I and II in 2020 increased over the same period in the previous 2 years, and the number of days with air quality of grade III were significantly reduced, and there was no moderate or severe pollution. There were 9, 11 and 2 polluted days in 2018, 2019 and 2020, respectively, suggesting that the pollution time was significantly reduced. In the second-level PPC period (Figure 2c), the number of days with air quality of grades I and II in 2020 also increased over the previous 2 years, and there were no polluted days. In the third-level PPC period (Figure 2d), there were still no polluted days in 2020, and the number of good air days was the same as that in 2019. In general, the air quality in 2020 was significantly improved compared with the same period in previous years, which means the implementation of PPC policies had a certain impact on air quality. It is worth noting that whether the epidemic occurred or not, the number of polluted days was decreasing from January to April. This is due to the obvious cooling of ground radiation at night in winter, and the "temperature inversion layer" is easy to appear in the low altitude of the atmosphere, resulting in the accumulation of pollutants and thus the poor air quality. With the increase of temperature in spring, the atmospheric stability decreases and the diffusion conditions become better, so the air quality is improved.

Statistics of atmospheric pollutant data in 13 cities show that except for SO2, the levels of pollutants and air quality indexes are significantly different during the periods before and after the implementation of PPC policies (*p* < 0.05, Table 2), indicating that the social restrictions implemented in the PPC period had a direct impact on atmospheric pollutants. As long as PPC measures are taken, the concentration of pollutants could be affected regardless of the level of PPC. From the perspective of the PPC levels, the concentrations of SO2, NO2 and CO were significantly different between the periods with the first- and second-level PPC measures, and the concentrations of SO2, NO2, CO and O3 were significantly different between the periods with the first- and third-level PPC measures, while there was little difference in pollutant concentrations between the periods with the second- and third-level PPC measures.

## *3.2. Impacts of COVID-19 on Air Quality*

Compared with the same period in 2018 and 2019, the values of AQI, SO2, NO2, CO, PM10 and PM2.5 in 1–21 January of 2020 decreased by 9.75%, 45.36%, 21.01%, 13.00%, 19.29% and 9.33%, respectively (Table 3), which are the NCRs in air quality in 2020 as defined above. This indicates that without the influence of PPC, the concentrations of air pollutants were also gradually decreasing, which is consistent with the results in other parts of China. The change in air quality is mainly attributed to the "Three-Year Action Plan for Cleaner Air" to win the battle for a blue sky released by the State Council of China in 2018. This action plan aims at significantly reducing the total emissions of major air pollutants and greenhouse gases, further lowering the concentrations of fine particulate matter (PM2.5), significantly reducing the number of heavily polluted days and thus improving the air quality.

**Figure 2.** The number of days for each grade of air quality under different pandemic prevention and control levels in Jiangsu Province ((**a**), Non; (**b**), L1; (**c**), L2; (**d**), L3).

**Table 2.** Significance levels for periods with different levels of PPC measures by Student's *t*-test (α = 0.05).



**Table 3.** Variations of AQI, atmospheric pollutants and meteorological conditions during the pandemic period and the same period of previous years.

After the outbreak of COVID-19, the AQI and concentrations of NO2, CO, PM10 and PM2.5 decreased with a magnitude larger than those before the outbreak, displaying the characteristics of varied degrees of declines under different levels of PPC measures. Specifically, the AQI and pollutant concentrations decreased the most under the first- and second-level PPC, and decreased slightly under the third-level PPC. The NCR of SO2 concentration in 2020 was −45.36%, which was significantly lower than the value in the same period in historical years, indicating that the control effect of SO2 in Jiangsu Province is obvious. But the decreasing trend of SO2 weakened after the outbreak of COVID-19, and similar results have been obtained in other parts of China [21]. We speculate that on one hand, it is related to the industrial production activities that have not been interrupted during the epidemic period, such as increasing the emission of coal-fired pollution from coal-fired power plants and coal-fired heating boilers. To some extent, epidemic control has increased the demand for household electricity, heating and cooking. From the perspective of provincial distribution, cities with a high proportion of secondary industry have a relatively high SO2 PCR, such as Changzhou, Zhenjiang and Taizhou (data omitted). On the other hand, sulfur dioxide is easily soluble in water, and the reduction of precipitation during the epidemic increased the content of sulfur dioxide in the atmosphere. The O3 concentration in 2020 increased. In particular, the O3 concentration increased by 20.8% under the first-level PPC measures compared with that during the same period in previous years. This result is similar to the results of studies focusing on the Chinese mainland, the Guangdong–Hong Kong–Macao Greater Bay Area [9,24], Europe [25] and India [18]. This phenomenon is mainly due to the particularity and complexity of ozone formation and depletion mechanisms. Ozone is formed by photochemical reactions between nitrogen oxides and volatile organic compounds emitted from natural sources and human activities [26]. Air pollution is also somewhat related to meteorological conditions [27,28]. Generally speaking, high temperature, low relative humidity and high solar radiation are conducive to the formation of ozone [29–31]. During the period with first-level PPC measures, the daily average temperature in Jiangsu was 3.43 ◦C higher than that in the same period of historical years, with an increase of 129.07%, while precipitation decreased by 29.88%. Brighter weather and lower concentrations of particulate matters allow more sunlight to pass through. Higher temperatures and stronger light, along with increased photochemical activity, lead to higher ozone concentrations. Changes in visibility also reflect changes in air quality to some extent [32]. During the periods in the first-, second- and third-level responses to PPC, the average values of atmospheric visibility respectively increased by 24.25%, 45.80% and 48.60% compared with those in the same period of previous years, and the daily minimum value of visibility also increased. This indicates that after the outbreak of COVID-19, the concentrations of atmospheric particulate matters decreased and thus the atmospheric transparency gradually increased.
