Health Risk Appraisal Associated with Air Quality over Coal-Fired Thermal Power Plants and Coalmine Complex Belts of Urban–Rural Agglomeration in the Eastern Coastal State of Odisha, India

Abstract

Manufacturing and mining sectors are serious pollution sources and risk factors that threaten air quality and human health. We analyzed pollutants at two study sites (Talcher and Brajrajnagar) in Odisha, an area exposed to industrial emissions, in the pre-COVID-19 year (2019) and consecutive pandemic years, including lockdowns (2020 and 2021). We observed that the annual data for pollutant concentration increased at Talcher: PM_2.5 (7–10%), CO (29–35%), NO₂ and NO_x (8–57% at Talcher and 14–19% at Brajrajnagar); while there was slight to substantial increase in PM₁₀ (up to 11%) and a significant increase in O₃ (41–88%) at both sites. At Brajrajnagar, there was a decrease in PM_2.5 (up to 15%) and CO (around half of pre-lockdown), and a decrease in SO₂ concentration was observed (30–86%) at both sites. Substantial premature mortality was recorded, which can be attributed to PM_2.5 (16–26%), PM₁₀ (31–43%), NO₂ (15–21%), SO₂ (4–7%), and O₃ (3–6%). This premature mortality caused an economic loss between 86–36 million USD to society. We found that although lockdown periods mitigated the losses, the balance of rest of the year was worse than in 2019. These findings are benchmarks to manage air quality over Asia’s largest coalmine fields and similar landscapes.

1. Introduction

Air pollution is a global problem and measures to fix it need to be identified [1]. Growth in population, urbanization, industrialization, motorization, and modernization has increased the pace of air pollution and has pushed it beyond a self-repairable limit [2]. The effects of air pollutants and their synergistic interactions are known to affect multiple resources, including life forms, biodiversity, heritage and property of national interest, and human well-being and life [2,3]. According to the World Health Organization (WHO) database on global air pollution, 14 out of 15 cities having the worst air pollution are in India [4]. The primary anthropogenic emission sources of particulate matter (PM_2.5 and PM₁₀) are households, industries, and transport sectors [5,6,7]. Among gaseous pollutants, oxides of nitrogen (NO_x) and sulphur dioxide (SO₂) emissions are dominated by transport [8], industry, and coal-based thermal power plants [9]; carbon monoxide (CO) emissions are the result of incomplete combustion of fuels used in residential biomass burning, the transport sector, and industry [10]. SO₂, nitrogen dioxide (NO₂), and PM are common, but more profoundly available in industrial and urban landscapes [11].

Several epidemiological studies have reported increased morbidity and mortality due to PM exposure [12]. Globally, an average increase in mortality per 10 mg/m³ changes in ambient PM concentrations at all low and high estimates were reported as 0.34% and 0.75%, respectively [13], whereas a combined analysis for four cities in Asia was estimated by the Public Health and Air Pollution in Asia (PAPA) program to have an average of 0.55% increase in mortality per 10 mg/m³ change in ambient PM₁₀ concentrations [14]. Most of the earlier studies described acute effects from exposure to PM₁₀, SO₂, and NO₂, whereas the recent studies focused on the acute effects of exposure to PM_2.5, ozone (O₃), and CO. A meta-analysis in China demonstrated that exposure to PM₁₀, PM_2.5, SO₂, NO₂, O₃, and CO is significantly associated with increased all-cause and cardio-respiratory mortality risks [15].

The widespread and rapid social and economic changes in response to COVID-19 were expected to bring improvements in air quality at large and for hotspots in particular. Scholars have reported significant improvement in air quality due to COVID-19 shutdowns in 2020 [16]. In 330 Chinese cities and New York (USA), air quality improved by 11% and 50%, respectively [17]. As reported in many towns and cities across India, around 20–60% reduction in PM_2.5, PM₁₀, SO₂, and NO₂ and 10–60% increase in daytime O₃ concentrations were witnessed due to lockdowns [17]. During COVID-19 lockdowns, significant reductions in pollutants were also observed over the megacities of India [18,19]. These reductions in air pollution provide an opportunity to re-draw maps of natural and reduced emissions in order to understand the baseline air pollution over different regions.

Industrial and mining sector development is the engine of economic growth in developing nations. In India, the eastern coastal state of Odisha is known for intensive manufacturing and mining. Talcher is a town in the Angul district and Brajrajnagar is a town in the Jharsuguda district, the industrial areas of the state. The Angul–Talcher industrial area has been identified as one of the world’s largest emissions hotspots [20]. With the increasing mechanization of mines, more fine dusts hazardous to human health were generated [21]. Cropper et al. [22] reported that coal field sites areas’ average damage per ton are greater for directly emitted PM_2.5 (23 deaths per 1000 t of PM_2.5) than for SO₂ (10 deaths per 1000 t of SO₂) or NO_x (9 deaths per 1000 t of NO_x). In 2017, Talcher was identified as one of the critically polluted industrial areas in the CPCB report ‘Comprehensive Environmental Pollution Index (CEPI)’ [23]. A plethora of research articles reported and represented the impact of lockdowns on air quality in several aspects [16,17,18,19]. However, there are few scientific reports that consider coal-fired thermal power plants and coal-mining clusters [22] to elucidate the air quality during lockdown and pandemic years [23]. Accordingly, we analyzed the air quality of two towns recognized as constituting the hub of thermal power plants and coalmine clusters. The objectives of this study were to estimate (i) the trends in air pollutants parallel to mortality and economic loss, and (ii) to perform a health risk analysis in the eastern coastal state Odisha, India. Our analysis will be important and represent the area as well as the intense mining, manufacture, and industrial sectors, and will enable public health authorities and policymakers to take the necessary preventative actions to combat air pollution.

2. Research Design

2.1. Study Area

The Talcher coalmine belts and Brajrajnagar open-cast mines are identified as the major hotspot of poor air quality in Odisha, India [23]. Talcher (site #1, 20.95° N, 85.23° E), a town in the Angul district, is well known for coal mines and huge deposits of coal. The area is situated on the right bank of the Brahmani River and is one of the fastest-growing industrial and mining complexes in the country. The climate is tropical, with summers having higher rainfall (330.20 mm) than the winters (12.70 mm). The average annual temperature is 26.8 °C. A large portion of the district is covered by forests (42.16%), with a variety of minor and major forest produce [24]. There are 11 coal mines covering 10,474.3 ha, 1 sand (stow) mine (17.5 ha), and 1 quartz mine (10.7 ha) [25]. The large industrial clusters include the Mahanadi Coal Field Ltd. (MCL), Odisha, India, the Talcher Thermal Power Station (TTPS), Odisha, India and the National Aluminum Company Ltd. (NALCO) plant, Bhubaneswar, India. In addition, coal-based thermal power plants, heavy industries, coal washeries, and subsidiary industrial units are situated in the area [23].

Brajrajnagar (site #2) is a city and municipality in the Jharsuguda district known for its coal mines. It has numerous open-cast and underground coal mines in the locality of the Ib valley coalfield region [26]. The town is small, located on rocky terrain on the banks of the Ib River (21.82° N, 83.92° E). The CPCB [27] reported that 78% of the cities in the Air Quality Index (AQI) bulletin fall into the good and satisfactory categories, and the remaining 12% of cities still had poor air quality during the lockdown phase, although only 44% cities had good air quality in the pre-lockdown phase. The geo-locations of the study area are presented in Figure 1 (Supplementary Table S1).

2.2. Data Resources

Air quality data were extracted for the period of 1 January 2019 to 31 May 2021 from the CPCB stations installed at Talcher and Brajrajnagar. We analyzed the spatiotemporal variation in criteria pollutants from 2019 to 2021 and the COVID-19 interventions phase between 25 March–31 May 2020 and 5 May–31 May 2021. For comparative analysis of a whole year, 2019 was taken as the base year in order to understand the comprehensive distribution of air pollutants at site #1 and site #2. The seven criteria pollutants (PM_2.5, PM₁₀, NO_x, NO₂, SO₂, O₃, and CO) were analyzed over the sites. The quality of data was ensured by filtering zeroes, negative values, erroneous data, and outliers. The greatest amount of missing data was observed for NO₂ and NO_x at site #2 and SO₂ at both of the sites. The hourly pollutant data were collected for the diurnal variation, while data were converted to 24 h averages for time series analysis. The daily averaged meteorological parameters, namely relative humidity (RH), temperature (°C), wind speed (WS), and precipitation (PCPN) at a spatial resolution of 0.5° × 0.5° (MERRA-2 model) were collected from the NASA (National Aeronautics and Space Administration, Washington, DC, USA) POWER website. Population and death rates for three years (2019–2021) were estimated using census data for the year 2011. The health risk analysis was performed using epidemiological equations. The datasets were filtered in order to have a more consistent time series. Both sites had around 75% of valid data points that were used in analysis.

2.3. Health Risk, Mortality, and Economic Loss

We analyzed the health risks associated with the criteria pollutants (PM₁₀, PM_2.5, NO₂, SO₂, CO and O₃). The mortality or health endpoint is taken into account for health risk evaluation studies, since death is the most clearly defined health endpoint. It assumes that all incidences are the result of the exposure concentration in any area (Table S2). We determined the relative risk (RRi) for each pollutant, estimated based on health effect studies (Equation (1)); the excess risk (ER) (Equation (2)); and the attribution factor (AF_i) to estimate the impact of exposure variations (Equation (3)) [28]:

$$R{R}_i=\exp \left[{\beta }_i\times \left(C_i-C_{i,0}\right)\right],C_i>0$$

(1)

$$E{R}_i=R{R}_i-1$$

(2)

$$A{F}_i=(R{R}_i-1)/R{R}_i$$

(3)

where ${\beta }_i$ was the exposure–response relationship coefficient, representing the ER of a health effect (such as mortality) per unit increase of pollutant (i.e., 1 μg/m³ of PM_2.5); $C_i$ is the absolute concentration of pollutant i, and C₀ is the threshold concentration below which a pollutant demonstrates no obvious adverse health effects (i.e., RR_i = 1). C₀ was considered 0 for all pollutants except CO, whose threshold value was taken as 4 mg/m³. Therefore, air pollutants exert ER of mortality only when they are above the given thresholds, that is, RR_i−1. For this study, β values were chosen from daily mortality for all ages: 0.38% (PM_2.5), 0.32% (PM₁₀), 0.81% (SO₂), 1.30% (NO₂), and 0.48% (O₃) per 10 μg/m³, and the β value was 3.7% per 1 mg/m³ increase in CO [28]. The premature deaths (PD) due to different criteria pollutant exposure was calculated using Equation (4):

$$P{D}_i=n_i\times A{F}_i$$

(4)

where $P{D}_i$ is the premature death from pollutant i; $n_i$ is the total number of deaths due to pollutant i; and $A{F}_i$ is the attribution factor of pollutant i.

Due to the unavailability of a market value for human lives, the monetary burden due to health risks was calculated using the value of statistical life (VSL) method. We estimated that PM_2.5, PM₁₀, NO₂, SO₂, and O₃ were responsible for economic costs in Talcher and Brajrajnagar due to the reported health endpoints, using the following equation (Equation (5)):

$$E{C}_{mort}=n_i\times VS{L}_i$$

(5)

where EC_mort is the economic cost per year of PM_2.5-, PM₁₀-, NO₂-, SO₂-, and O₃-related premature mortality at site #1 and site #2 and $n_i$ is the number of premature deaths due to particular pollutants. $VS{L}_i$ is the Indian value of statistical life. The OECD (Organization for Economic Cooperation and Development, 2014) gave an Indian $VS{L}_i$ value of 0.60 million USD for 2010, while [29] reported 1.10 million USD for 2011. Shanmugam [30] carried out a VSL study in India during the year 1996–1997 and the value ranged from 0.17–0.23 million USD. Furthermore, Cropper et al. [31] estimated a VSL for India in the range of 0.08–0.18 million USD. In the current study, we used the most recent value of VSL, i.e., 0.59 million USD estimated in the year 2016 [32].

3. Results

3.1. Spatiotemporal Variation of Air Pollutants

We found positive and negative fluctuations in air pollutants in two consecutive pandemic years as well as during two lockdown periods at both sites in comparison to the base year 2019 (the average absolute values are reported in

Table 1. Average concentration and standard deviation of pollutants during two successive lockdowns (25 March–31 May 2020 and 5 May–31 May 2021) and the base year (period from 25 March–31 May 2019).

Year	PM2.5	PM10	CO	SO2	NO2	NOx	O3
	(µg/m3)		(ppm)	(ppb)
Site #1
2019	38.80 ± 13.54	159.36 ± 53.28	1.23 ± 0.29	25.98 ± 2.96	13.06 ± 3.48	25.33 ± 4.49	5.41 ± 1.83
LD2020	41.66 ± 22.11	165.15 ± 92.33	1.88 ± 0.31	19.88 ± 6.69	12.76 ± 9.45	24.94 ± 18.68	44.87 ± 9.29
LD2021	37.92 ± 10.12	162.14 ± 55.89	1.73 ± 0.27	13.84 ± 4.96	28.22 ± 5.78	49.00 ± 7.78	25.33 ± 5.04
Site #2
2019	69.88 ± 31.51	152.88 ± 54.76	2.65 ± 0.50	7.51 ± 1.26	19.66 ± 4.25	29.85 ± 9.56	10.96 ± 3.61
LD2020	78.35 ± 30.61	171.94 ± 64.57	1.11 ± 0.28	4.92 ± 1.05	3.35 ± 1.64	20.42 ± 5.10	54.91 ± 15.08
LD2021	61.09 ± 11.56	143.29 ± 28.32	2.26 ± 0.13	4.83 ± 0.49	22.02 ± 10.07	37.55 ± 9.66	21.79 ± 16.69

). The two successive phases of lockdown can be assumed as two sets of experiments that represent conditions in the absence of human activities (i.e., the forced shutdown offers zero background concentration). The diurnal temporal variation and pollutant correlation with meteorology are reported in detail in Section 3.1, Section 3.2 and Section 3.3 (Table S3).

3.1.1. Particulate Matter

The analysis of pandemic years 2020 and 2021 revealed that the average concentration of PM_2.5 marginally varied between 2–10 µg/m³ in 2020–2021, but the daily fluctuations were equal to or even higher than the pre-lockdown year (2019) at both sites. The spatiotemporal variation of PM_2.5 had a 7.1–10.3% increase during LD2020 and a 2.6–14.7% decrease during LD2021. Comparatively, moderate to higher continuous precipitation during LD2021 led to washout of pollutants. The drop in PM_2.5 concentrations was not noteworthy when compared to other cities (such as Delhi, Chennai, Kolkata, Bangalore, Hyderabad, and Mumbai, etc.), particularly during LD2020. This is important as site #1 and site #2 have numerous open-cast and underground coal mines and four units of thermal power plants that were operational during the LD period. The PM₁₀ distribution showed similar trends to PM_2.5. PM₁₀ concentrations varied between 1.8–11.0% during LD2020 and LD2021 at both sites compared to 2019 (Figure 2 and Figure 3 and Table 1). Scholars have reported that surface- and satellite-based data both showed a 42–60% and 24–62% reduction in PM_2.5 and PM₁₀, respectively, in Indian megacities during LD2020 [33]. Other cities in Odisha also reported a significant drop in particulate matter concentration, except Talcher and Brajrajnagar. For example, in the capital city Bhubaneswar, a 10–42% drop in surface-based concentrations of PM_2.5 [34] and 33% in PM₁₀ was reported during LD2020 as compared to the base year. Contrary to thermal power plant cities, semi- and moderately industrialized regions of India also reported a significant reduction in PM_2.5 and PM₁₀ concentrations during LD2020. Thomas et al. [35] reported that concentrations of PM_2.5 (24–47%) and PM₁₀ (17–20%) decreased, and the overall air quality of a less-industrialized region, Kerala (a southern coastal state of India), significantly improved. The PM₁₀ concentration at the semi-industrialized town of Kannur, Kerala, declined by 61% [36].

3.1.2. CO

We found fluctuations in CO concentration at site #1 (Figure 2 and Figure 3) during LD2020 and LD2021 compared to the base year. A decrease in CO concentrations at site #2 was especially observed during LD2020 and LD2021 as compared to the base year. The difference in the behavior of the two sites is the attribute of local anthropogenic emissions (coal burning in homes and small-scale commercial businesses like street-food cooking, agricultural waste burning, and biofuel burning). At site #1, a 29–35% increase in concentration was found during both consecutive lockdowns. The higher concentration at site #1 was possibly due to the forest fires and agricultural burning that are predominant during this season (March–May, pre-monsoon) [37,38]. The reduction in CO at site #2 (around 50%) may be attributed to a limitation of combustion activities in different sectors such as transport, industry, and residential.

3.1.3. SO₂

A significant daily fluctuation (both positive and negative) in SO₂ concentration was observed at both sites in 2020–2021 as compared to 2019 (Figure 2 and Figure 3 and Table 1). The overall SO₂ at both sites was found to have decreased (30–86%) as compared to the base year, which indicates a reduction of coal burning in small-scale commercial businesses like street-food cooking and biofuel burning, since utilization of sulphur-containing fuel or other coal-combustion activities were sluggish during lockdown [39]. The primary source of SO₂ is the combustion of sulphur-containing fuels such as coal and diesel used in thermal power plants, industries, and transport [39]. India is one of the highest emitters of anthropogenic SO₂.

3.1.4. NO₂ and NO_X

At site #1, during 2020 and 2021 both NO₂ (7.69–57%) and NO_x (8–45%) emissions varied when compared to the base year (Table 1 and Figure 2 and Figure 3). It was observed that during LD2020, a large amount of data for NO₂ and NO_x was missing for site #2; therefore, average concentration was not depicting a representative variation. During LD2021, a 14–19% increase in NO₂ and NO_x concentrations was found in comparison to the base year. The oscillating peak of NO_x was showing a very similar follow-up trend to NO₂ within both site #1 and site #2 during LD2021. Ghude et al. [40] reported that the industrial and thermal power plant regions in India are major sources of nitrogenous emissions. The annual variations depict that throughout the years 2020 and 2021, NO₂ and NO_x concentrations were higher as compared to 2019, and this difference was very pronounced at site #1 as compared to site #2 (Figure 2). Some very sharp and high emission peaks were also observed at site #1, indicating local anthropogenic activity burning coal in the home, agricultural waste burning, and biofuel burning [41]. The reported concentration of NO₂ over a less-industrialized region (Kerala) showed a reduction of 48%, and NO_x 53–90% [35].

3.1.5. O₃

The daily time series plot (Figure 2 and Figure 3) depicts significantly higher concentrations during LD2020 and LD2021 as compared to the base year. The O₃ concentration showed a uniformly increasing trend throughout the year (an increase of around 41–88% in 2020 and 2021; Table 1) and during the lockdown period. O₃ is a secondary pollutant that is formed in the presence of sunlight and its precursors, viz., NO_x and volatile organic compounds (VOCs). The rising CO concentration in 2020 and 2021 also indicates significant biomass burning, which possibly contributed to higher O₃ formation during lockdown periods as compared to the base year. It has been previously reported that among all the regions of India, biomass burning associated with O₃ production was observed to be highest over central Indian states like Odisha [38,39].

3.2. Air Pollutants and Meteorological Parameters

At site #1, RH was in weak correlation with pollutants NO₂ (r = 0.25), PM_2.5 (r = −0.25) at p (≤0.01) and NO_x and PM₁₀ (p ≤ 0.001); temperature had correlation with PM_2.5 (r = −0.19) at p (≤0.05), and WS with PM_2.5 and PM₁₀ (p ≤ 0.001). PM₁₀ and PM_2.5 had strong correlation (r = 0.82), as did NO₂ and NO_x (r = 0.60); RH had a positive correlation (r = 0.48), and O₃ and NO₂ and NO_x also had positive correlation at different levels of significance (Figure 4a). At site #2, temperature, PCPN, and RH had positive correlation (r = 0.40), as did PM₁₀ and PM_2.5 (r =0.94), PM_2.5 had positive correlation with NO₂, NO_x, and PM₁₀, whereas CO had positive correlation with NO₂ (r = 0.67). WS had positive correlation with temperature and PCPN. O₃ had negative correlation with PCPN and RH and was also negatively correlated with temperature and WS at p (≤0.001). At site #2, air pollutants had good correlation with meteorological parameters as compared to site #1 (Figure 4 and Table S4). The correlation matrix of pollutants and meteorology did not depict any common behavior between the two sites. However, site #2 had comparatively better meteorology correlation with pollutants. At site #1, WS played a significant role in the distribution dynamic of PM.

3.3. Diurnal Variation of Pollutants

Diurnal variation of a pollutant was predominantly governed by local emissions, meteorological conditions, and daytime and nighttime chemistry. The diurnal variation of some criteria pollutants (O₃, NO_x, and NO₂) was found to be higher than the base year, whereas SO₂ concentrations were found to have decreased (Table S5a,b and Figure S1a–c). Some pollutants showed variation (PM_2.5, PM₁₀, and CO) across sites, which may be attributed to local meteorology and anthropogenic activities (coal burning in homes and small-scale commercial businesses like street-food cooking, agricultural waste burning, and biofuel burning).

3.3.1. Diurnal Variation of Particulate Matter

The diurnal variation of PM_2.5 and PM₁₀ (Figure 5) followed a bimodal distribution. The first peak developed between 4 a.m. and 12 noon, and the second peak developed around 18 p.m. and 22 p.m. These peaks ranged from blunt (PM_2.5 and PM₁₀ at site #1) to moderate (PM_2.5 and PM₁₀ at site #2). We found both sites’ PM concentration was more or less the same, particularly during LD2020 and compared to the base year. The diurnal trends also adhered to a daily variation pattern for particulate matter. During the LD2021 period, PM_2.5 and PM₁₀ concentrations were found to be slightly lower than LD2020 and the base year concentration, especially from 4 a.m. to 12 noon and around 18 p.m. to 22 p.m. This drop in concentration may be due to reduced data points in LD2021 and washout of PM concentration. During 5−10 May 2021 there was heavy rainfall reported due to cyclone “Yaas” and frequent rains were observed throughout May 2021. The reduction in primary anthropogenic emissions may be the cause of variation in pollutant concentrations in different hours of the day. Mahato and Ghosh [42] reported a drastic drop (70%) in diurnal concentrations of PMs in LD2020.

3.3.2. Diurnal Variation of NO₂, NO_x, and O₃

The diurnal variation of NO₂ and NO_x showed an increasing concentration in the morning hours from 5 a.m. to 9 a.m., then in the proceeding hours it decreased, again increasing between 16 p.m. and 22 p.m. (Figure 5). This trend was clearly observed during LD2020 at site #1 for NO₂ and NO_x. At site #2, the NO₂ and NO_x concentration was found to be uniformly lower in the base year as compared to the LD2020 and LD2021. Due to missing data for NO₂ and NO_x at site #2 during LD2020, a representative trend could not be observed (Figure 5).

O₃ concentrations were found to be higher throughout day and night in LD2020 and LD2021 as compared to the pre-lockdown period of 2019. In LD2020, O₃ concentration was greatest, and clear emission peaks were observed between 8 a.m. and 8 p.m. at both sites. The LD2021 diurnal trend showed slight bulging peaks from 8 a.m. to 18 p.m., as in LD2020 (Figure 5). The diurnal O₃ variation profile shows similarity with the daily O₃ profile.

3.3.3. Diurnal Variation of CO

We observed that both sites had different diurnal trends for CO. At site #1 during LD2020 and LD2021, CO concentration remained higher compared to the base year (Figure 5), whereas at site #2 during LD2020 and LD2021, CO concentration remained lower as compared to the base year and LD2020 showed the lowest concentration. CO is a biomass- and traffic-related pollutant, owing to biomass activities [37] at site #1, whereas at site #2 lower concentration during the lockdown period was due to reduction in biomass burning as well as transport activity. The spatiotemporal and diurnal variability trend of CO suggests a diversity in sources.

3.3.4. Diurnal Variation of SO₂

SO₂ concentration was higher at both sites during the base year as compared to LD2020 and LD2021 (Figure 5). SO₂ concentration was observed to be greatest from 8 a.m. to 14 p.m. in 2019 at both sites. LD2021 showed a bulging peak from 8 a.m. to 14 p.m. at site #1 but the overall magnitude was low as observed in the daily and temporal variation profile. At site #2, no clear peak was observed. Diurnal variation indicated that there were some anthropogenic activities during LD2021 at both sites that involved coal burning. Unlike other urban regions in India, the diurnal variation of SO₂ over the eastern coastal state of Odisha showed elevated daytime values due to the predominant influence of regional emissions [43].

3.4. Comparative Fundamental Statistics of Pollutants

Overall, the PM_2.5 and PM₁₀ for both sites showed a similar pattern of distribution. In the case of PM_2.5 most data were within the first and third quartile at both sites, except for LD2021 (Figure 6). In the case of NO₂ and NO_x pollutants, the distribution was entirely different for both sites. Site #1 violin plots depicted distinct fine bimodal distribution during LD2020 and skinnier and elevated outliers in the LD2021 violin. At site #1, higher concentrations during LD2020 and LD2021 were also observed as compared to the base year, and most observations were gathered around the first and third quartile ranges for NO₂ and NO_x. On the other hand, site #2 depicted very skinny violins with more outliers, particularly for LD2020, suggesting more missing data points. The SO₂ concentration was found to have decreased during LD2020 and LD2021 as compared the base year, and a smaller and skinnier violin shape indicates a deficiency of data points at both sites. Contrary to that of SO₂, O₃ concentration during LD2020 and LD2021 increased as compared to the base year; data were highly skewed with skinnier violins and more outliers (Figure 6). For LD2021, a skinnier shape is indicative of missing data points. The CO data distribution was found to be less skewed and showed higher probability of data around the first and third quartile. At site #1, CO showed a bimodal distribution and a rise in concentration during the lockdown period as compared to the base year. On the other hand, at site #2 more data were gathered around the mean, and the base year concentration was higher as compared to the lockdown period. Overall, the LD2021 violins are very skinny and smaller in shape for all pollutants because of fewer data points—only 25 data points (5 May 2021 to 31 May 2021) as compared to the LD2020 and base year (68 data points).

4. Health Risk, Premature Death, and Economic Cost Associated with Pollutants

The health risk from air pollution in India is the focus of an international comparison. The study sites have an emission-intensive source of air pollution in their close environment (

Table 2. Details of excessive risk, premature death attribution, and economic cost at site #1 and site #2.

Sites	Year	Annual/ Lockdown	Pollutant-Wise Death Attribution/100,000					Total Attributed Deaths/100,000	EC (USD in Million)
			PM2.5	PM10	NO2	SO2	O3
Site #1	2019	Annual Pre-lockdown	47	84	26	24	6	187 (54%)	109
			7	10	5	4	3	30 (52%)	17.6
	2020	annual	61	85	38	33	15	233 (63%)	137
		LD2020	7	11	9	6	4	37 (52%)	22
	2021	annual	62	113	32	17	8	232 (62%)	136
		LD2021	7	11	9	9	1	37 (60%)	22
Site #2	2019	annual	24	54	21	20	11	130 (58%)	76
		Pre-lockdown	5	10	2	2	0	19 (50%)	11
	2020	annual	31	56	27	25	8	147 (61%)	86
		LD2020	5	9	3	4	1	23 (58%)	14
	2021	annual	29	63	25	25	8	151 (62%)	89
		LD2021	5	10	4	4	1	24 (60%)	14

and Figure 7). The CO concentration was found to be less than the minimum (CO = 2 mg/m³) concentration; therefore, it was excluded from the health risk analysis in the present study. At both sites, the annual increase in total %ER ranged from 3−8% as compared to the base year. During LD, a 1-5% reduction in ER was found in 2020 and 2021. It can therefore be interpreted that the forced lockdown lowered the ER to 5% within the year, which was similar to the findings of [44]. Furthermore, at site #1 we found that atmospheric pollution caused 62−63% of annual deaths during the 2020−2021 period as compared to the base year. At site #2, atmospheric pollution caused 61−62% of total deaths annually in 2020 and 2021 as compared to 2019. We found that individually, particulate matters (PM_2.5: 16−26%; PM₁₀: 31−43%) and the gaseous pollutants O₃, SO₂, and NO₂ (O₃: 4−7%, SO₂: 11−25% and NO₂: 15−21%, Figure 7) were causing substantial premature mortality, similar to Cropper et al. [22].

Thermal power plant-having towns and cities always need extra care for their surrounding environment and at the same time need to attract the research community for local monitoring and analysis. Mishra (2015) [45] conducted 300 interviews in Angul−Talcher; five villages were 3.5 km or less from mines, and two were approximately 47 km away. From the villager interviews he concluded that 90% of households in the mining villages reported health problems, compared with 52% in non-mining villages in recent years. Another study reported that a quarter of the ambient PM_2.5 in southeastern coastal states like Tamil Nadu, Andhra Pradesh, and Odisha were coming from coal-fired power plants. This increased ambient PM_2.5 concentration from power plants led to 44% of deaths and attributable deaths. Our results were also close to previously reported results [46]. A forecast reported that Odisha’s coal-based thermal power plants led to 6100 deaths in 2017 and 7560 deaths in 2020 [46]. Our results were also close to the previously reported results [45,47].

At site #1, the annual EC of total mortality due to air pollution was 109 million USD in 2019 and it increased to 137 million USD in 2020 and 136 million USD in 2021. At site #1, lockdown period air pollution mortality costs slightly increased to 22 million in 2020 and 2021 from 17.6 million in the pre-lockdown period of base year 2019. The annual EC of air pollution-attributed mortality at site #2 was 76 and 86 million USD in year 2019 and 2020, respectively, and it increased to 89 million USD in 2021. For the pre-lockdown period at site #2, we found that the EC was 11 million USD in 2019 and increased to 14 million USD in 2020 and 2021.

5. Discussion

The forced lockdown gave an opportunity for nature to sprout after hundreds of years and provided an opportunity for specialists and policymakers as well to conduct a close and comprehensive study of the effect of anthropogenic emissions during switch-off situations in various areas of the earth–atmosphere system. Contrary to cities with thermal power plants, semi- and moderately industrialized regions of India reported significant reduction in PM_2.5 and PM₁₀ concentrations during LD2020 [35,36]. Most of the cities of Odisha also reported a significant drop in particulate matter concentrations [34] except Talcher and Brajrajnagar. We found trivial spatiotemporal deviation in PM_2.5 (7−10% increase at site #1 and up to a 15% drop at site #2) and a slight to substantial increase in PM₁₀ (up to 11%) as compared to the base year. CO is a biomass- and traffic-related pollutant, attributable to biomass use [37] at site#1, though at site #2 lower concentrations during the lockdown timeframe were attributable to a decrease in biomass use as well as transport movement. The substantial drop in SO₂ (28% to 53%) demonstrated a decrease in nearby anthropogenic action [39], since use of sulphur-containing fuels or other coal-ignition sources was idle during 2020 and 2021 when contrasted with 2019. However, in the state capital city of Bhubaneswar, SO₂ was overwhelmed by provincial emissions as opposed to neighborhood sources [34,43]. O₃ was found to be higher over the course of LD2020 and LD2021 when contrasted with the pre-lockdown period of 2019. O₃ concentration is governed by several bio-geophysical elements [48,49,50]. We found a significant increase in O₃ (41–88%). The industrial and thermal power stations in India are significant wellsprings of nitrogenous discharges [40]. Some extremely sharp and high discharges were found at site #1, demonstrating neighborhood anthropogenic action [41]. The decrease in essential anthropogenic outflows might be a reason for the variability in nitrogenous oxide fixation in several hours of the day.

To comprehend the financial effect of the thermal power plant and coal-mine complex belt, we further assessed excessive health risk, pollutant-specific mortality attribution, probability of death, and associated economic loss. The effect of health risk, pollutant mortality attribution, probability of death, and economic loss are closely related. The coalmine complex belt and thermal power station-related uses were functional during 2020 and 2021, however everyday scheduled business was halted and both large- and limited-scope shops were shut, so it tends to be construed that the constrained lockdown brought ER down 5% in contrast to the base year. At an individual level, particulate matters (PM_2.5: 16−26%; PM₁₀: 31−43%) cause significantly higher mortality attribution than gaseous pollutants (4−25%), eventually causing substantial premature mortality. This led to a minor increase in societal cost of around 136 million USD in 2020 to 2021, which was 109 million USD in 2019. All the quantified results indicate that technical, economic, environmental, and social possibilities should be adopted in processing and mining coal for a possible sustainable solution [51].

6. Conclusions

We quantified variation in criteria pollutants at temporal scales (diurnal, daily, LD2020, and LD2021) and at a spatial scale (Talcher and Brajrajnagar sites). The coalmine complex clusters and coal-fired thermal power plants were continuously operational during the lockdown. Therefore, the findings elucidate the impact of thermal power plants and coalmine complex sites on air quality in an eastern coastal state of India. The PM_2.5 increased by about 5−12% and PM₁₀ by 3−13% during LD2020 and LD2021 at both sites. CO showed spatial variation at site #2, around a 17-54% drop, and at site #1, a 52−54% increase was observed, indicating the contribution of local activities such as forest fires, agricultural waste burning, and biofuel. Meanwhile, SO₂ and O₃ depicted contrasting trends to each other, with SO₂ illustrating a uniformly decreasing trend (28−53%) and O₃ portraying a uniformly increasing trend (41−88%) during both the pandemic year and lockdown periods. NO₂ and NO_x concentration presented random behavior, either vaguely increasing or remaining the same as the base year (8−57% temporal deviation at site #1 and 14−19% rise at site #2). The current study results conflict with other cities and industrial sites in India and around the world in reference to the pandemic year and COVID-19 lockdowns.

These pollution sources pose a significant health risk, as well as mortality and economic loss at a regional and state level. The results suggest that irrespective of pandemic restrictions excessive risk increased 3−8% and premature death increased 58−62% compared to the base year 2019. Except for CO and O₃, all criteria pollutants caused substantial attribution of premature mortality (PM_2.5: 16−26%; PM₁₀: 31−43%; NO₂: 15−21% and SO₂: 4−7%). This cost society 86−136 million USD. The results also advocate the need for improved emission regulations in coal clusters and coal-based thermal power plants. The results are significant as well as indicative that in order to clean the environment we should shift toward green energy sources for power and energy production. In parallel, governments must take the pollution issue seriously in these coal clusters and industrial belts, and environmental standards must be enforced and strictly maintained. The results demonstrate the need for replacement of fossil fuels by renewable energy sources, and the results should put pressure on local bodies toward meeting 2030 carbon emission reduction targets in India. The results may be handy for decision makers and local engineers for environmental management and for new insights in terms of air quality modelling, since several models are now used, as a real-world tool for improved and efficient management of the environment.