Air pollution and its associated health risks before and after COVID-19 in Shaanxi Province, China

Abstract

Air pollution is a serious environmental problem that damages public health. In the present study, we used the segmentation function to improve the health risk–based air quality index (HAQI) and named it new HAQI (NHAQI). To investigate the spatiotemporal distribution characteristics of air pollutants and the associated health risks in Shaanxi Province before (Period I, 2015–2019) and after (Period II, 2020–2021) COVID-19. The six criteria pollutants were analyzed between January 1, 2015, and December 31, 2021, using the air quality index (AQI), aggregate AQI (AAQI), and NHAQI. The results showed that compared with AAQI and NHAQI, AQI underestimated the combined effects of multiple pollutants. The average concentrations of the six criteria pollutants were lower in Period II than in Period I due to reductions in anthropogenic emissions, with the concentrations of PM_2.5 (particulate matter ≤2.5 μm diameter), PM₁₀ (PM ≤ 10 μm diameter) SO₂, NO₂, O₃, and CO decreased by 23.5%, 22.5%, 45.7%, 17.6%, 2.9%, and 41.6%, respectively. In Period II, the excess risk and the number of air pollution–related deaths decreased considerably by 46.5% and 49%, respectively. The cumulative population distribution estimated using the NHAQI revealed that 61% of the total number of individuals in Shaanxi Province were exposed to unhealthy air during Period I, whereas this proportion decreased to 16% during Period II. Although overall air quality exhibited substantial improvements, the associated health risks in winter remained high.

Graphical abstract

Capsule abstract: The excess risk and the number of air pollution–related deaths decreased by 46.5% and 49% during Period II.

1. Introduction

With the rapid development of the economy and the acceleration of urbanization in China, air pollution has become a key environmental problem, exerting substantial negative effects on human health (Yin et al., 2020). The levels of air pollution in China are considerably higher than the standards outlined by the World Health Organization (WHO); accordingly, the number of air pollution–related annual deaths in China are more than 2 million per year (Maji et al., 2018; World Health Organization, 2021). In recent decades, the effect of air pollution on public health has garnered widespread attention (Goudarzi et al., 2017; Huang et al., 2018). Epidemiological studies have indicated that continual exposure to air pollutants increases morbidity and mortality due to cardiovascular, circulatory, and respiratory diseases and shortens life expectancy by severely damaging human health (Lelieveld et al., 2015; Ma et al., 2018; Maji et al., 2018). Therefore, the characteristics of air pollution and its associated health risks must be investigated to pursue environmental improvement and sustainable development (Zhao et al., 2021).

Since the outbreak of the SARS-CoV-2 (COVID-19) pandemic, several studies have indicated an association between air pollutants and the pandemic (Filonchyk et al., 2020; Ma et al., 2022; Wang et al., 2021). The central government of China imposed stringent nationwide lockdown measures to control the pandemic. Reduced factory and traffic activities resulted in decreased pollutant emissions, thus improving the overall air quality (Miyazaki et al., 2020). A study of a total of 31 Chinese provinces (Nie et al., 2021) revealed that varying degrees of decreases in the concentrations of NO₂, PM_2.5 (particulate matter ≤2.5 μm diameter), PM₁₀ (PM ≤ 10 μm diameter), and CO outweighed the increases in the concentration of O₃, leading to an overall improvement in air quality. Travel restriction policies during COVID-19 impacted the levels of traffic volume on the National and Provincial Trunk Highway Network and Provincial Expressway Network in Shaanxi, China (Ma et al., 2022a; Ma et al., 2022b). Tian et al. (2020) reported that the COVID-19 lockdown in China caused low productive activity and traffic across the country for more than a month. The pandemic harmed economies and pressured health-care systems around the globe; nevertheless, the only silver lining is that it provided a valuable opportunity to study the effects of human activities on air pollution and its characteristics (Huang et al., 2021).

The six criteria pollutants (PM_2.5, PM₁₀, SO₂, NO₂, O₃, and CO) adversely affect human health and increase various health risks (Gao et al., 2022). Particulate matters (PM_2.5 and PM₁₀) are easily absorbed into the respiratory system and reduce life expectancy (Lelieveld et al., 2015). PM_2.5 has been associated with a high number of deaths in China and China accounted for >25% of the total worldwide PM_2.5-related deaths in 2015 (Cohen et al., 2017). Gaseous pollutants (e.g., SO₂, NO₂, O₃, and CO) also pose major health risks and may even cause death (Luo et al., 2020; Zhang et al., 2021). The air quality index (AQI) based on the maximum concentrations of the aforementioned six criteria pollutants is often used to assess the effects of pollutants on public health; however, a single pollutant index inadequately reflects the actual levels of air pollution. To comprehensively consider the health effects of exposure to multiple air pollutants, Kyrkilis et al. (2007) and Wang et al. (2021) proposed the aggregate AQI (AAQI) and health risk–based AQI (HAQI), respectively. Compared with the AAQI and HAQI, the AQI reportedly underestimates the combined effects of multiple pollutants (Hu et al., 2015; Mao et al., 2020). Therefore, new indices may help us better assess the health risks of air pollution.

Due to COVID-19 pandemic, air pollution markedly improved in Shaanxi Province, China (Shen et al., 2020). In the present study, we sought to improve the HAQI to make it more realistic and accurate; the improved version of this index was named new HAQI (NHAQI). Based on the AQI, AAQI, and NHAQI, we evaluated the spatiotemporal distribution characteristics of air pollutants and the associated health risks in Shaanxi Province before (Period I, 2015–2019) and after (Period II, 2020–2021) COVID-19.

2. Data and methods

2.1. Study areas

Shaanxi Province is located in the northwestern part of China. The topography is low in the middle but high in the northern and southern parts, with a clear slope from the west to the east, which is not conducive to the diffusion of air pollutants (Dong et al., 2013). In addition, the main industries in the study area—nonferrous metallurgy, equipment manufacturing, and energy and chemical industries—are all heavily polluting industries, leading to considerable pollution in most parts of Shaanxi Province (Mestl and Edwards, 2011; H. Zhang et al., 2020). Fig. S1 depicts a map of Shaanxi Province and the 10 cities located in this province (Ankang [AK], Baoji [BJ], Hanzhong [HZ], Shangluo [SL], Tongchuan [TC], Weinan [WN], Xi'an [XA], Xianyang [XY], Yan'an [YA], and Yulin [YL]).

2.2. Data collection

For the aforementioned 10 cities, daily data of PM_2.5, PM₁₀, SO₂, and NO₂ and 8-h-averaged data for O₃ and CO were obtained from the National Environmental Monitoring stations established in these cities. Pollutant data between January 1, 2015, and December 31, 2021, for each city were calculated by averaging the pollutants’ concentration data obtained from all stations in that city. The population and mortality data (2015–2021) for each city were obtained from the Statistical Yearbook of Shaanxi Province (http://tjj.shaanxi.gov.cn/tjsj/ndsj/tjnj/). China has adopted several control policies for COVID-19, which have markedly influenced pollutant concentrations.

2.3. Calculation of air quality indices

2.3.1. AQI

The Ministry of Environmental Protection (MEP) calculates AQI with reference to the Chinese Ambient Air Quality Standards (CAAQS). The individual AQI for each criteria pollutant (AQI_i) was calculated using Eq. (1) (range, 0–500; MEP, 2012a). The total AQI was calculated using the maximum of the individual AQI_i values of the six criteria pollutants, as expressed in Eq. (2).

$${AQI}_i=\frac{\left({AQI}_{i,j}-{AQI}_{i,j-1}\right)}{\left(C_{i,j}-C_{i,j-1}\right)}\times \left(C_{i,m}-C_{i,j-1}\right)+{AQI}_{i,j-1},j>1{AQI}_i={AQI}_{i,1}\times \frac{C_{i,m}}{C_{i,1}},j=1$$ (1)

$$AQI=\max \left\{{AQI}_1,{AQI}_2,\dots \dots ,{AQI}_n\right\},n=\mathrm{1,2},\dots \dots ,6$$ (2)

where AQI_i represents the index of pollutant i, C_i,m is the measured concentration of i, j refers to the health category index, C_i,j and C_i,j−1 represent the concentrations of pollutant i corresponding to the jth and j−1st health categories, respectively, and AQI_i,j and AQI_i,j−1 represent the AQI values corresponding to the jth and j−1st health categories, respectively. Table S1 presents the six health categories and their corresponding AQI values and pollutant concentration ranges obtained from the Chinese MEP. The upper limit of the CAAQS 24-h Grade II standards corresponds to an AQI_i value of 100 (MEP, 2012b); air quality is regarded as unhealthy when the total AQI value is > 100 (Hu et al., 2015).

2.3.2. AAQI

AQI only considers the pollution status of the predominant pollutant, while AAQI also considers the comprehensive effects of the six criteria pollutants (Swamee and Tyagi, 1999; Kyrkilis et al., 2007). The calculation formula is as follows:

$$AAQI={\left({\sum }_{i=1}^n{\left({AQI}_i\right)}^{\rho }\right)}^{\frac{1}{\rho }}$$ (3)

where ρ is an empirical constant, and the selection range of the optimal ρ value remains obscure. Previous studies suggested that the selection range of ρ values was between 2 and 3 (Khanna, 2000; Cairncross et al., 2007). In a multicity study (Hu et al., 2015), four different values of ρ (i.e., 1.5, 2.0, 2.5, and 3.0) were used to calculate the mean and standard deviation values of the AAQI/AQI ratio. These ratios were found to be less sensitive to variations in ρ; ultimately, using 2.0 as the value. Therefore, in the present study, the 2.0 value of ρ was used. For better comparison with AQI, the same health categories and a scale range of 0–500 (similar to that of the AQI) were adopted for the AAQI. The upper limit of AAQI was set at 500 when the value exceeded 500.

2.3.3. HAQI

To better illustrate the exposure–response relationships associated with various pollutants, several studies have proposed the use of health risk–based indices (Hu et al., 2015; Shen et al., 2020). To define HAQI, Cairncross et al. (2007) proposed the concept of total excess risk (ER) of exposure to multiple pollutants. The relative risk (RR_i) of pollutant i is calculated as follows:

$${RR}_i=\mathit{exp}\left[{\beta }_i\left(C_{i,m}-C_{i,0}\right)\right],C_{i,m}>C_{i,0}$$ (4)

where β_i represents the exposure–response relationship coefficient, indicating the excess health risk for per unit increase in the concentration of pollutant i. The β values, referenced from a meta-analysis conducted in China (Shang et al., 2013), were 0.038%, 0.032%, 0.081%, 0.13%, 0.048%, and 3.7%, respectively, per a 1 μg/m³ increase in the concentrations of PM_2.5, PM₁₀, SO₂, NO₂, and O₃ and a 1 mg/m³ increase in that of CO. C_i,m is the concentration of pollutant i, and C_i,0 is the risk limit of pollutant i, indicating that if the value of C_i,m is lower than that of C_i,0, it is considered to have no excess health effect (RR = 1).

The ER_i of pollutant i is calculated as follows:

$${ER}_i={RR}_i-1$$ (5)

The total ER of the simultaneous exposure to the six criteria pollutants was calculated by summing the ER of each pollutant:

$${ER}_{total}={\sum }_{i=1}^n{ER}_i={\sum }_{i=1}^n({RR}_i-1)$$ (6)

Studies (Cairncross et al., 2007; Stieb et al., 2008) have transformed ER_total into an arbitrary index between 0 and 10 to indicate the health risk of air pollution. Using the same scale range as that of the AQI and AAQI (0–500), Hu et al. (2015) defined the HAQI and the equivalent concentration of i (C*_i,m). C*_i,m indicates the equivalent concentration of pollutant i when ER_i is equal to ER_total. C*_i,m was calculated directly using Eq. (8), which might have resulted in higher HAQI values. We used the segmentation function in the present study. C*_i,m remained the observed concentration when pollutant i imposed no excess health risks and was calculated using Eq. (8) when it imposed excess health risks. Thus, we use the new-HAQI (NHAQI) to distinguish it from the previous HAQI. The equivalent RR_i relative risk (RR_i*) can be calculated as follows:

$${RR}_i^{*}={ER}_{total}+1=\exp \left[{\beta }_i\left(\left(C_{i,m}-C_{i,0}\right)\right)\right]$$ (7)

C*_i,m is calculated as follows:

$$C_{i,m}^{*}=\ln \left({RR}_i^{*}\right)/{\beta }_i+C_{i,0},C_{i,m}>C_{i,0}$$ (8)

$$C_{i,m}^{*}=C_{i,m},C_{i,m}<C_{i,0}$$ (9)

Using the equivalent concentration (C*_i,m) of the ith criteria pollutant instead of the actual concentration (C_i,m), NHAQI is calculated as follows:

$${NHAQI}_i=\frac{\left({AQI}_{i,j}-{AQI}_{i,j-1}\right)}{\left(C_{i,j}-C_{i,j-1}\right)}\times \left(C_{i,m}^{*}-C_{i,j-1}\right)+{AQI}_{i,j-1},j>1$$

$${NHAQI}_i={AQI}_{i,1}\times \frac{C_{i,m}^{*}}{C_{i,1}},j=1$$ (10)

$$NHAQI=\max \left\{{NHAQI}_1,NH{AQI}_2,\dots \dots ,{NHAQI}_n\right\},n=\mathrm{1,2},\dots \dots ,6$$ (11)

2.4. Mortality burden of the pollutants

We estimated the mortality burden attributable to the six criteria pollutants during the two periods in each of the 10 cities located in the study province. The calculation formula is as follows (Guo et al., 2016):

$$\Delta \mathrm{M}\mathrm{o}\mathrm{r}\mathrm{t}\mathrm{a}\mathrm{l}\mathrm{i}\mathrm{t}\mathrm{y}=Y_b\times ER\times POP$$

where ΔMortality represents pollutant-related additional deaths, Y_b is the baseline mortality values of total mortality, ER is the percent change of the excess mortality risk, and POP is the population of exposure. Baseline mortality and population data for the period 2015–2019 were obtained from the Statistical Yearbook of Shaanxi Province for 2016–2020; however, the Statistical Yearbook for 2021 lacked baseline mortality data. Thus, we used the average baseline mortality rate of 2015–2019 as an alternative to calculate the additional deaths during Period II.

3. Results

Table 1 presents the average concentrations of the six criteria pollutants during Periods I and II in the 10 cities located in Shaanxi Province. In terms of average concentrations, air quality appears to have improved substantially in Period II compared with the quality in Period I. The concentrations of PM_2.5, PM₁₀, SO₂, NO₂, O₃, and CO decreased by 23.5% (11.9 μg/m³), 22.5% (21.4 μg/m³), 45.7% (7.5 μg/m³), 17.6% (6.4 μg/m³), 2.9% (2.7 μg/m³), and 41.6% (0.5 mg/m³), respectively. The concentration of O₃ exhibited the least decrease; moreover, an increasing trend was noted in the O₃ concentration of WN. The concentrations of SO₂ and CO were relatively moderate in Shaanxi Province and reached the CAAQS Grade I standards. BJ, WN, XA, and XY (in the central part of Shaanxi Province) were identified to be highly polluted cities of Shaanxi Province.

Table 1.

Average concentrations of six criteria pollutants during Periods I (2015–2019) and II (2020–2021) in 10 cities located in Shaanxi Province.

City	PM2.5		PM10		SO2		NO2		O3		CO
	(μg/m3)		(μg/m3)		(μg/m3)		(μg/m3)		(μg/m3)		(mg/m3)
	I	II	I	II	I	II	I	II	I	II	I	II
Ankang (AK)	41.5	29.5	66.3	48.2	14.9	8.7	19.1	16.4	84.2	77.0	0.9	0.6
Baoji (BJ)	54.8	43.0	100.6	74.6	11.0	7.3	36.0	26.9	87.9	87.3	1.0	0.6
Hanzhong (HZ)	49.0	36.8	76.4	58.0	12.4	10.6	26.1	20.8	83.2	79.6	1.2	0.8
Shangluo(SL)	37.5	27.6	67.6	49.2	16.2	8.0	25.2	20.3	94.1	91.3	1.0	0.5
Tongchuan (TC)	51.6	39.3	94.7	72.3	19.1	10.1	34.1	29.4	101.1	95.1	1.1	0.7
Weinan (WN)	63.2	49.3	128.1	96.8	16.9	11.1	44.6	36.1	92.8	94.2	1.0	0.8
Xi'an (XA)	64.4	47.1	124.1	94.0	17.1	8.1	50.7	41.4	89.1	86.3	1.4	0.8
Xianyang (XY)	70.4	52.0	125.6	97.5	17.4	9.2	46.0	40.5	92.6	90.4	1.2	0.8
Yan'an (YA)	39.0	30.9	87.6	71.6	21.7	5.0	43.0	33.0	93.6	92.8	1.4	0.7
Yulin (YL)	35.4	32.5	81.6	76.7	17.2	10.3	38.9	34.9	101.8	98.5	1.3	0.7
Average	50.7	38.8	95.3	73.9	16.4	8.9	36.4	30.0	92.0	89.3	1.2	0.7
CAAQS Grade I/II	35/75		50/150		50/150		40/80		100/160		2/4

Fig. 1 presents the cumulative number of days for the six AQI categories during Periods I and II. The number of days with excellent (AQI <50) and good (50 < AQI <100) air quality increased in these 10 cities from an average of 265.2 days during Period I to 297.8 days during Period II. In particular, the air quality of TC, WN, XA, and XY exhibited the greatest improvements, with an increase of >40 days in the number of days with excellent and good air quality. Although the air quality of WN, XA, and XY improved considerably, these remained the three most polluted cities in Shaanxi Province. On average, the numbers of days with excellent and good air quality for these three cities were 203.5 and 247.0 in Periods I and II, respectively. The number of days with moderate (150 < AQI <200), serious (200 < AQI <300), and severe pollution (AQI >300) levels in these three cities were higher than those of the other cities. In most cities, the number of days with serious and severe pollution levels decreased from an average of 15.3 days during Period I to 10.8 days during Period II. However, the number of days with serious and severe pollution levels increased in YA and YL, particularly the severe pollution days, which increased from an average of 1.8 days during Period I to 7.5 days during Period II.

Fig. 1.

Cumulative number of days for the six AQI categories during Periods I (2015–2019) and II (2020–2021) in 10 cities of Shaanxi Province.

Fig. 2 presents the proportions of the predominant pollutants—the pollutants with the largest AQI calculated using Eq. (1) when the AQI value is > 50—in all cities during Periods I and II. PM_2.5, PM₁₀, and O₃ were the frequent predominant pollutants in Shaanxi Province, together accounting for >90% of the total pollutants. In the two periods, the proportions of PM₁₀ and NO₂ were basically the same. The proportion of PM_2.5 decreased from 30.1% to 28.5%, whereas that of O₃ increased from 33.4% to 36.3%. SO₂ and CO occasionally became the predominant pollutants in a few days, and together accounted for <2% of the total pollutants. During Period II, these two pollutants were not predominant. The level of NO₂ pollution was relatively serious in YL, YA, and XA, which indicated a specific air pollution characteristic of these cities. Moreover, the proportions of the predominant pollutants exhibited prominent seasonal differences (Fig. S2). During spring, PM₁₀ was the predominant pollutant in all cities, followed by O₃, which accounted for more than 45% and 30%, respectively. In particular, PM₁₀ accounted for more than 50% in the cities located in central Shaanxi Province (e.g., XA, XY, and BJ). During summer, O₃ was the predominant pollutant in all cities, accounting for more than 80%, whereas the other pollutants had low proportions. During autumn, the frequent predominant pollutants were PM_2.5, PM₁₀, and O₃; their proportions varied across the cities. The types and proportions of the pollutants were complex. The level of CO pollution was serious in SL during Period I, whereas it was well controlled during Period II. In winter, PM_2.5 was the predominant pollutant accounting for more than 60% of the total proportions of predominant pollutants, particularly in AK and HZ, where PM_2.5 accounted for more than 85%. O₃ fluctuated markedly across the seasons and rarely became the predominant pollutant in winter.

Fig. 2.

The proportions of predominant pollutants in Shaanxi Province during Periods I (a) and II (b).

Fig. 3 presents the scatter plots comparing the AAQI, HAQI, NHAQI, and AQI values for all cities. When the concentrations of the pollutants were below the CAAQS Grade II upper limits, the air quality was determined to be excellent or good, imposing no excess health risks. Thus, the HAQI and NHAQI values were equal to those of the AQI when the AQI value was <100; the plots shown in Fig. 3 did not include the aforementioned data. When air quality was unhealthy (AQI value > 100), the AAQI, HAQI, and NHAQI values were all higher than the corresponding AQI values, indicating that the cumulative health risk of multiple pollutants was higher than that of a single predominant pollutant and that the AQI underestimated the air pollution–associated health risks. The slopes of the AQI with the AAQI, HAQI, and NHAQI were 1.29 (r² = 0.98), 1.22 (r² = 0.97), and 1.18 (r² = 0.97), respectively. Compared with the NHAQI, the HAQI exhibited a higher slope and might have overestimated the air pollution–associated health risks. The HAQI value calculated directly using Eq. (8) skewed the results and increased the number of days when PM_2.5 was the predominant pollutant. The main reason was that the β value of PM_2.5 was lower and the AQI values were higher when the concentration of PM_2.5 was lower than that of the other pollutants. For instance, between June 11 and 13, 2016, in XA, the concentration of O₃ was high, whereas those of the other pollutants were below the CAAQS Grade II upper limits (Table S2); however, the HAQI identified PM_2.5 to be the predominant pollutant and the result was relatively high. Table S3 presents the number of days (2015–2021) for dominant pollutants calculated using the AQI, HAQI, and NHAQI when the AQI value was >100. The HAQI indicated a total of 6239 days when PM_2.5 was the predominant pollutant, accounting for 98%. The number of days calculated using the NHAQI was similar to that calculated using the AQI. Therefore, the improved NHAQI was more realistic and accurate; thus, in the present study, the AAQI and NHAQI were used to analyze air pollution–associated health risks.

Fig. 3.

Correlations of the AQI with the AAQI (a), HAQI (b), and NHAQI (c) (AQI value > 100) (The red line is y = x and the black line is the trend line.). (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Fig. 4 shows the AAQI- and NHAQI-mediated reclassification of the five health risk categories of the AQI and the classifications of the average number of days for all cities. Data distribution varied across classification types. On days with healthy air quality based on AQI (AQI value < 100), the NHAQI were equivalent to the AQI data because of no ERs; however, 51% and 4% of the total number of days had light and moderate levels of pollution, as evident from the AAQI. For the light pollution days, based on AQI (100 < AQI value < 150), 7% and 2% of the total number of days had moderate and serious levels of pollution based on NHAQI; the percentages of days with moderate and serious levels of pollution were 65% and 9% based on AAQI. For days with moderate levels of pollution based on AQI (150 < AQI value < 200), 46% and 3% of the total number of days were classified as serious and severe levels of pollution based on NHAQI; whereas 85% of the total number of days had serious levels of pollution based on AAQI. For days with serious levels of pollution based on AQI (200 < AQI value < 300), 56% and 44% of the total number of days were classified as severe levels pollution based on NHAQI and AAQI, respectively. In general, there was an underestimation of the health risk class classification based on AQI, and it might cause serious effects on human health when the pollution was moderate based on AQI classification.

Fig. 4.

AAQI-classified (a) and NHAQI-classified (b) health risks based on the health categories of the AQI (2015–2021) in Shaanxi Province.

Fig. 5 depicts the spatial distributions of the average NHAQI and AAQI values in Shaanxi Province during Periods I and II. The central part of this province had the highest level of pollution; nonetheless, air quality in this area exhibited the greatest improvements in Period II. On the basis of the NHAQI, four cities in central Shaanxi Province (i.e., BJ, WN, XA, and XY) were under light levels of pollution (100< NHAQI <150) during Period I, with XY being the most polluted city. During Period II, three cities (i.e., BJ, WN, and XA) were identified as having healthy air quality (NHAQI <100) with NHAQI values for all of these cities decreased by > 20; however, XY was still under light pollution. The cities in southern Shaanxi Province (i.e., AK, HZ, and SL) had the best air quality, exhibiting the lowest NHAQI values in both Periods. Two cities in northern Shaanxi Province (i.e., YL and YA) exhibited the least improvement in air quality. The NHAQI values of most cities decreased by > 10, whereas those of YL and YA decreased by < 10, particularly YL, for which the value decreased by only 3. According to the AAQI, all cities were under air pollution during Period I; WN, XA, and XY were even under moderate levels of pollution. During Period II, the air quality of AK, HZ, and SL was in healthy level (AAQI <100), whereas the other cities were in light pollution (100< AAQI <150).

Fig. 5.

Spatial distributions of average NHAQI and AAQI values in Shaanxi Province during Periods I (a) and II (b).

Fig. 6 illustrates the average total ER (ER_total) in Shaanxi Province during Periods I and II. The average ER_total value of the 10 cities was 0.79% during Period I; however, the ER_total fluctuated markedly across the cities. ER values were the largest (>1.5%) in WN, XA, and XY. Among the six criteria pollutants, PM_2.5 and PM₁₀ were the two primary pollutants contributing to the ER_total, accounting for 85%, followed by O₃. The overall contribution of NO₂ to ER_total was only 4%, but the pollution level was serious in WN, XA, and XY. During Period II, the average ER_total value decreased to 0.43%, representing a decline of 45.6%. All other cities exhibited markedly decreased ER_total values, except for YL and YA, where the ER_total values increased, probably due to the increase in the number of days with serious and extreme levels of pollution. During Period II, NO₂ and PM_2.5 pollution was better controlled; however, the contributions of PM₁₀ and O₃ to ER_total increased.

Fig. 6.

Average of total excess risk in Shaanxi Province during Periods I (a) and II (b).

To better assess the proportion of people exposed to air pollution, the NHAQI and population data for each city were combined to calculate the cumulative population distribution estimated using the NHAQI during Periods I and II (Fig. 7 ). On average, 61% of the total number of individuals were exposed to light pollution (NHAQI value > 100) during Period I. The proportion of the population exposed to light pollution was similar between spring (53%) and autumn (55%). The air quality in summer was better, with all people living in a healthy environment (NHAQI value < 100). The level of air pollution was most severe in winter, with 76% of the population exposed to polluted air (NHAQI value > 100), 64% to unhealthy air (NHAQI >150), and 43% to very unhealthy air (NHAQI value > 200). During Period II, air quality improved substantially, with only 16% (annual average) of the population exposed to polluted air. Air quality was under good level in spring, summer, and autumn, with no excess health risks. However, in winter, 72% of all individuals were exposed to polluted air but not to unhealthy or very unhealthy air.

Fig. 7.

Cumulative population distribution estimated using the NHAQI in Shaanxi Province during Periods I (a) and II (b).

We further evaluated the average annual number of air pollution–related deaths in each city during Periods I and II (Fig. 8 ). A total of 2540 average annual air pollution–related deaths occurred during Period I. The air pollution–related deaths mainly distributed in central Shaanxi Province. BJ, WN, XA, and XY had high numbers of air pollution–related deaths (annual average) due to the heavy pollution and dense population in these cities. During Period II, the total annual average number of air pollution–related deaths in Shaanxi decreased to 1,286, exhibiting a decrease of 49%. The number of air pollution–related deaths in other cities decreased considerably, except for YL and YA, where slight increases were noted in the annual average number of air pollution–related deaths. This increase was attributed to the increase in ER values rather than the increase in population or baseline mortality.

Fig. 8.

Spatial distribution of the annual average air pollution–related deaths in Shaanxi Province during Periods I (a) and II (b).

4. Discussion

Owing to the stringent lockdown measures imposed in China to control the COVID-19 pandemic, the annual average concentrations of the six criteria pollutants decreased during Period II. Air quality has great improvement in the cities with dense population, high abundances of secondary industries and high traffic intensity under the same COVID-19 lockdown measures. Thus, the air quality of TC, WN, XA, and XY exhibited the greatest improvements. The north and northwest of Shaanxi is Mu Us Desert, and the northern part of Shaanxi Province is the main part of the Loess Plateau in China (Wang et al., 2022; Tong et al., 2022). The northern cities of Shaanxi (YA and YL) are heavily influenced by natural sources, especially sandstorms. Although anthropogenic sources of emissions were reduced (Miyazaki et al., 2020), the strong dust storms led to great increase in spring concentrations of PM_2.5 and PM₁₀ in YA and YL, resulting in minimal air quality improvement. The concentration of O₃ exhibited the least decrease (only 2.9%). In some cities, O₃ concentration even demonstrated an increasing trend. This finding is consistent with those of other relevant studies (Filonchyk et al., 2020; Pei et al., 2020; Wang et al., 2021). Chauhan and Singh (2020) reported considerable decreases in the concentrations of PM_2.5 in several cities across the globe that were severely affected by COVID-19; these decreases resulted mainly from the measures implemented to control the COVID-19 outbreak. A study conducted in Beijing–Tianjin–Hebei (Wang et al., 2021) also indicated that the average annual concentrations of PM_2.5, PM₁₀, NO₂, SO₂, and CO decreased by 12%, 23%, 19%, 33%, and 11%, respectively, in 2020 compared with the values in 2019. Wang et al. (2020) identified that the transport sector was strongly associated with the reduction in NO₂ emissions, whereas lower emissions from secondary sectors were responsible for the reduction in the concentrations of PM_2.5 and CO; the reduction in the concentration of SO₂ was associated only with the industrial sector. The concentration of PM_2.5, PM₁₀, SO₂, NO₂ and CO decreased by 37%, 30%, 29%, 52% and 33%, respectively, during the strictest restrictions period in the Guanzhong Basin, the center of Shaanxi Province (K. Zhang et al., 2020). The reduction in the levels of traffic volume in Shaanxi (Ma et al., 2022a) might lead to a great decrease in the concentration of NO₂. The concentration of O₃ increased during the lockdown period (K. Li et al., 2021), possibly due to low fine-particle loadings, which resulted in the reduction of HO₂ scavenging (Wang et al., 2020), low nitrogen oxide (NOx) emissions, and nighttime NO₃ radical formation (Huang et al., 2021). In the present study, O₃ concentration increased in the spring of 2020 in some cities located in Shaanxi Province, but a prominent downward trend was noted in summer. Our findings highlight the considerable effects of human activity on air pollution and the interaction among various air pollutants.

The predominant pollutants in Shaanxi Province exhibited considerable seasonal variations in their characteristics. The predominant pollutants in spring, summer, autumn, and winter were PM₁₀, O₃, composite pollutants, and PM_2.5, respectively. Air pollution in many northern Chinese cities exhibits similar seasonal variations (Liu and Ren, 2020; Yin et al., 2020). For instance, in Lanzhou (Luo et al., 2020) and Zhengzhou (Shen et al., 2017), the highest concentration of PM_2.5 was noted in winter and that of O₃ was observed in summer. There were several deserts and gobi in northwest China with low vegetation coverage (Guan et al., 2019). Frequent sandstorms in spring make PM₁₀ a predominant pollutant in northwest China (Song et al., 2017). In the present study, O₃ was the predominant pollutant in summer and had low concentrations in winter. Due to the intense solar radiation and high temperature in summer, photochemical reactions of precursors gases, such as NOx and volatile organic compounds (VOCs) occur proactively, resulting in high concentrations of O₃ (Wang et al., 2017; Kuerban et al., 2020). In winter, combined coal and biomass burning for residential heating, stable weather conditions, and a low boundary layer result in high PM_2.5 concentrations (Li et al., 2019; Zeng et al., 2019). The concentrations of PM_2.5, PM₁₀, SO₂, NO₂ and CO decreased significantly in most cities during the four seasons in period II due to the restrictions on business, less traffic and human activities. The concentration of O₃ in winter increased by 12.9%. A study in Nepal (Dhital et al., 2022) indicated that the mean aerosol optical depth (AOD), NO₂ and CO decreased by 27.7%, 12.7%, and 5.12% during the lockdown in the dry season, respectively. A study in India (Ratnam et al., 2021) showed that natural sources were the main contribution to air pollutants during the dry season. A study in Delhi NCR, India (Siddiqui et al., 2022) demonstrated that the concentration of NO₂ decreased by ∼ 48% in summer season and the magnitude of variation caused by COVID-19 lockdown was much higher than seasonal variation (Olusola et al., 2021). The concentration of O₃ increased significantly in winter of 2020 in the North China Plain due to a rapid reduction in NO_x emissions following the COVID-19 lockdown (M. Li et al., 2021).

It is important to assess and quantify the effects of air quality on public health (Yin et al., 2020). The AQI is based on the maximum concentrations of the six criteria pollutants; however, a single pollution index may not adequately reflect the actual levels of air pollution and thus may underestimate the severity of health risks due to the exposure to multiple air pollutants (Xu et al., 2022). Therefore, various air quality indices have been defined to evaluate the associated health risks (Li et al., 2017; Hossain et al., 2021). The HAQI and AAQI are based on the combined effects of the six criteria pollutants and the exposure–response relationships associated with multiple pollutants, thus facilitating highly accurate evaluations of health risks (Luo et al., 2020). Hu et al. (2015) suggested that for the unhealthy risk classification based on the AQI, 80% and 96% of the total number of days were identified as having very unhealthy or hazardous air quality when applying the HAQI and AAQI classifications, respectively. A study conducted in Henan Province, China (Shen et al., 2017) also revealed that for days with serious levels of pollution based on AQI, 63% and 66% of the total number of days classified as severe levels of pollution based on HAQI and AAQI. In the present study, we used the segmentation function to improve the HAQI. The NHAQI, improved based on HAQI, is more realistic and accurate and may be used in future studies to better assess air pollution–related health risks and to form a scientific basis for disease prevention and control.

PM₁₀ and PM_2.5 were the predominant pollutants contributing the most to the ER_total (>85%). Although air quality has recently improved in China, PM pollution remains a major problem. PM_2.5 pollution worsens in winter, particularly in northern China (Song et al., 2017). A study on air pollution–related health risks in China between 2015 and 2018 (Shen et al., 2020) indicated that PM₁₀ and PM_2.5 were the primary contributors to ER, with health risks from PM₁₀ predominating. Xu et al. (2022) indicated that PM_2.5 and PM₁₀ are the primary indicators of air quality in northern China and contribute the most to ER, accounting for 41% and 40%, respectively. In the present study, during Period II, the ER_total decreased by 45.6%, but the contribution of PM₁₀ and O₃ to the ER increased. Wang et al. (2021) suggested that due to the reduction of anthropogenic emissions, the HAQI values decreased in the Yangtze River Delta, where PM_2.5 is a predominant pollutant associated with considerable health risks. Miyazaki et al. (2020) demonstrated that the incidences of morbidity primarily due to the exposure of people with asthma to O₃ increased by approximately 2100 cases and due to exposure to PM_2.5 decreased by at least 60,000 cases because of the COVID-19 lockdown. The reduction of anthropogenic emissions plays a key role in alleviating PM_2.5 pollution; continual air pollution control measures are required to alleviate O₃ pollution (Maji et al., 2020).

In the present study, the assessment of health risks due to human exposure to pollutants revealed that 61% and 16% of the total number of individuals in Shaanxi Province were exposed to unhealthy air during Periods I and II, respectively. The exposure level of pollutants varies across cities and years (Maji et al., 2020). The population in Henan Province was exposed to unhealthy air between 2014 and 2015, as indicated by population-weighted average data (Shen et al., 2017). Mao et al. (2020) demonstrated that 50%, 70%, and 80% of the total number of individuals in the upper, middle, and lower reaches of the Yangtze River were exposed to unhealthy air during the winter of 2017. Xu et al. (2022) reported that approximately 55% of the total number of individuals in northern China were exposed to polluted air from 2016 to 2019, and the highest levels of pollution were noted in winter. In spring, the population of Lanzhou was exposed to unhealthy air due to severe dust storms (Luo et al., 2020). Li et al. (2019) indicated that meteorological elements and topography substantially affect the concentrations of local pollutants. Air quality is also strongly associated with various socioeconomic factors. Shen et al. (2020) demonstrated that there was an inverse U-shape curve among total population, population density, developed areas, and air pollution. Increased population density would reduce the average costs of electricity, coal gas, natural gas, and public transportation, thereby increasing the per-household consumption of clean energy and use of public transport services and reducing the emission of pollutant gases (Chen et al., 2020). However, pollution has been worsening in most cities of China with the country's improving economy (Cheng et al., 2017). Our results identified that the densely populated cities located in central Shaanxi Province (i.e., BJ, WN, XA, and XY) were the most polluted cities with the highest number of air pollution–related deaths. Air pollution in these cities can be attributed to the high abundances of secondary industries, high traffic intensity, and coal-dominated energy structures. Recently, China has implemented air pollution prevention and control strategy. Anthropogenic emissions have a significant impact on the concentration of pollutants. Industrial restructuring and upgrading and clean energy use may serve as effective measures for reducing air pollution–related health risks in Shaanxi Province.

Our study has some limitations. First, we directly used the national β values as the exposure relationship coefficient for Shaanxi Province. In future studies, we would combine real epidemiological studies and clinical data to select appropriate β and C_i,0 values, which would enhance the accuracy and reliability of the study findings. Second, the HAQI values were obtained by summing the health risks of multiple pollutants on the basis of the individual health risk of each pollutant. However, the possible interactions among various pollutants were not accounted for, which warrants further investigation. Third, the second period does not include data for this year (2022) and we do not consider the effect of meteorological parameters on air pollution. Finally, the ER associated with the six criteria pollutants is related to not only mass loadings but also pollutant compositions; the effects of their compositions on human health must be investigated further.

5. Conclusions

In the present study, air quality exhibited substantial improvements during Period II because of the lockdown imposed to control the pandemic. PM_2.5, PM₁₀, and O₃ were identified to be the predominant pollutants in Shaanxi Province; the contribution of these pollutants on health accounted for >90% of the total pollutants. The densely populated cities had the highest number of air pollution–related deaths. On the basis of the cumulative population distribution estimated using the NHAQI, 61% and 16% of the total number of individuals were exposed to unhealthy air during Periods I and II, respectively. Compared with the AAQI and NHAQI, the AQI underestimated the combined effects of multiple pollutants, which highlights the need for improved health protection measures for at-risk populations. Our study may serve as a reference for future studies aiming at controlling air pollution.

Credit author statement

Yuxia Ma organized the research; Bowen Cheng performed the model and analyzed the results; Yuxia Ma and Bowen Cheng wrote and reviewed the manuscript; Heping Li, Fengliu Feng and Yifan Zhang analyzed data; Wanci Wang and Pengpeng Qin assembled data.

Funding

This research was supported by the National Natural Science Foundation of China (Grant Nos. 41975141). Part of the work was funded by a scholarship awarded to Yuxia Ma (File No. 20206185010) by the China Scholarship Council.

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.

Appendix A. Supplementary data

The following is the Supplementary data to this article.

Data availability

Data will be made available on request.