Analysis of Air and Soil Quality around Thermal Power Plants and Coal Mines of Singrauli Region, India

Singrauli region is known as the energy capital of India, as it generates nearly 21 GW of electricity, supplied to various parts of the northern India. Many coal-based Thermal Power Plants (TPPs) using coal from several nearby coal mines, and numerous industries are set up in this region which has made it as one of the highly polluted regions of India. In the present study, detailed temporal analysis and forecast of carbon dioxide (CO2), nitrogen dioxide (NO2), sulfur dioxide (SO2), and methane (CH4) concentrations retrieved from satellite data have been carried out for the periods 2005–2020. Based on the classical multiplicative model and using linear regression, the maximum concentration of CO2, NO2, SO2, and CH4 in the year 2025 is found to be 422.59 ppm, 29.28 ppm, 0.23 DU, and 1901.35 ppbv, respectively. Detailed analysis shows that carbon dioxide has a 95% correlation with all other trace gases. We have also carried out the geo-accumulation index for the presence of various contaminants in the soil of this region. The geo-accumulation index shows that soil in and around thermal power plants and coal mines is contaminated by heavy metals. The cumulative index shows that soil around Hindalco industries, Bina coal mines, Khadia coal mines, and coal-based TPPs (Anpara and Vindhayachal) are highly polluted and a threat to human population living in the region.


Introduction
The growing anthropogenic activities are the main sources associated with the changes in land, atmosphere, biosphere, and cryosphere that have direct/indirect threats to our ecosystem, on a local, regional, and global scale. Before the industrial revolution, pollution was associated with natural causes, however, after the industrial revolution in the 19th century, increasing atmospheric pollution is impacting the ecosystem. Monsoon, drought, expansion of the desert, change in the genetics of forests, and melting of snow/glaciers are a few of the consequences of increasing pollution [1][2][3][4][5][6]. The growing population pressure, rapid industrialization, and development of megacities have made developing countries more vulnerable compared to developed nations. This has deteriorated the air, water, and soil quality, in addition to land degradation, thus impacting the surrounding environment, and causing various respiratory, gastro-intestinal and cardiovascular diseases [7][8][9]. According to WHO 2016, about a million people are killed every year in India due to air pollution, which is considered as one of the high health risks in the country [10] Likewise, indirect intake of heavy metals from soil contamination can cause liver, lung, neurological, and gastrointestinal diseases [11][12][13][14].
India's energy consumption has increased more than twice since 2000 and its energy demand is projected to increase 1.5 times by 2030. As a result, electricity generation has already increased by 84% in 2021 as compared to 2010 [15]. Till 31 May 2022, 50% of In view of the increasing pollution, understanding of long-term annual/seasonal variations and forecast of concentrations of different pollutants (CO 2 , NO x , SO 2, PM, and CH 4 ), the long-term trend and variability of pollutants provide important information to policymakers to find ways and means to control the emissions. The main objectives of the present study are to (a) analyze the long-term trend (15 years) and seasonal variations of pollutants, CO 2 , NO 2 , SO 2 , and CH 4 , with respect to TPPs in the Singrauli region using satellite data; (b) predict the concentration of these pollutants in the study area for the next 5 years, and (c) evaluate the geo-accumulation index for heavy metals using field observation of soil contamination in the study area.

Study Area
There are many coal-based, gas-based thermal power plants (TPPs) in India as shown in Figure 1a, out of which the Ministry of Environment and Forest (MoEF) has identified Singrauli as a "Critically Polluted Area". This study is carried out in Singrauli region also known as Urjanchal, "Energy Capital of India", which is situated on the border of Uttar Pradesh and Madhya Pradesh in Central India. Singrauli region is located around 24.1960 Table 1) and coal mines are located; all these activities impact the atmosphere and air quality in the areas (photo taken by Ramesh Singh).   Table 1) and coal mines are located; all these activities impact the atmosphere and air quality in the areas (photo taken by Ramesh Singh).
In the past, the area was covered with an inhabitable dense forest with abundant natural and mineral resources. During the 1800s to 1950s, the original inhabitants and tribal communities were dependent on agricultural activities [45]. Pre-industrialization studies show that the area was covered with forest (43.35%), cropland (38.56%), and culturable wasteland (17.44%) [46]. Industrialization started with the construction of Rihand dam creating the Govind Ballabh Pant Sagar reservoir in the late 1950s. It was mainly built for irrigation purposes and hydro-power generation of 400 MW capacity. However, diverse electricity-requiring industries were established in the region instantaneously. For example, the Hindalco aluminum industry was established in 1962 followed by Kanoria Chemicals in 1964, and UP state cement corporation in 1970 [47]. During the early 1970s, the largest coal deposit was found which is being mined currently in the Singrauli region.
We have considered an area of a 2 × 2 • box (23 N, 81 E, 25 N, 83 E) in the Singrauli region ( Figure 1a) for carrying out our analysis. There are many Thermal Power Plants (TPPs) in this region generating around 21 GW of electricity ( Figure 1; Table 1) (https: //endcoal.org/tracker/ accessed on 10 May 2022) which act as a stationary pollution source (Figure 1b). Details of TPPs installed in the Singrauli region since 2005 are given in Table 1. Numerous studies from the early 1990s have reported contamination in soil, water, and air in this region due to industrial development [48,49]. The climate of the study area can be characterized by hot summer and cold winter. The temperature ranges from a minimum of 2 • C during the month of January to a maximum of 47 • C during June. The average annual rainfall in the region is around 1133 mm, about 80% of which occurs during the south-west monsoon from June to September.

Materials
We have carried out an air pollution analysis and soil quality index for heavy metals. For air pollution analysis, we have selected SO 2 , NO 2 , CO 2, and methane (CH 4 ) pollutants. The daily data for SO 2 and NO 2 is taken from ozone monitoring and instrument (OMI) [50,51] between 2005 and 2020 with a spatial resolution of 0.25 × 0.25 • . Similarly, for CO 2 , monthly data is taken from Atmospheric Infrared Sounder (AIRS) [52] between 2003 and 2016 with a spatial resolution of 2 × 2.5 • . In the case of CH 4 , daily data is taken from AIRS [53] between 2003 and 2020 with a spatial resolution of 1 × 1 • (Table 2). The soil samples were collected around coal mines and TPPs in the Singrauli region in polyethylene collection bags from a depth of 15-30 cm. The soil samples were taken from this depth to remove the contamination from the top surface due to anthropogenic activities. These samples were analyzed for arsenic, fluoride, iron, copper, chromium, manganese, zinc, and titanium [25].

Air Quality Analysis
We have carried out the descriptive analysis and evaluated the minimum, maximum, mean, standard deviation, and variance of pollutants to attribute distribution of data, range of data, outliers, and errors. After the descriptive analysis, with the help of a box and whiskers plot, outliers are identified using the following conditions [54]: where IQR = Inter-quartile Range, Q 1 = First Quartile of the data, Q 3 = Third Quartile of the data. These outliers are selected and removed from the data. As the IQR of the data will change, the box plot is again plotted and most of the outliers were removed except a few ( Figure 2). The median value for CO 2 and CH 4 is lower than their respective mean values whereas, the median value is higher than the mean value for NO 2 and same for SO 2 This pre-processed data is further used for time series, linear regression analysis, and henceforth for forecasting pollutant concentration. Figure 3 shows a detailed flowchart showing the methodology and analysis steps. We have used the classical multiplicative model for time series analysis using the following equation [55].
where Yt = original data/ predicted data, St = Seasonal component, It = Irregularity component, Tt = Trend component. Seasonal, trend, and irregularity components are required to predict the pollutant concentration as expressed in Equation (3). Seasonality (St) and irregularity (It) components are extracted by smoothening the data using a 12-month moving average and calculating the center moving average as the seasonal data is even numbered. The mathematical equation used to calculate St It is [55]: where Yt = Original Data/ Predicted Data, St = Seasonal Component, It = Irregularity Component, CMA = Centre Moving Average.
Next, the irregularity component is removed to obtain the seasonal component that is used to deseasonalize the data. It is conducted by averaging the seasonality and irregularity of individual months for the entire data set. The result thus obtained is a 12month cyclic trend. Using this seasonal component, the data is deseasonalized using the following equation [55]. This pre-processed data is further used for time series, linear regression analysis, and henceforth for forecasting pollutant concentration. Figure 3 shows a detailed flowchart showing the methodology and analysis steps. We have used the classical multiplicative model for time series analysis using the following equation [55].
where Y t = original data/predicted data, S t = Seasonal component, I t = Irregularity component, T t = Trend component.
where Yt = Original Data, St = Seasonal Component. This deseasonalized data is used in linear regression to extract trend components using a simple linear equation [55].
Y= mx + c (6) where m is the slope and c is the intercept that is obtained from linear regression and x is the time component. Further, the trend and seasonal components are used in Equation (3) to forecast pollutants until 2025. While obtaining the trend component, three iterations with different training periods are used. The first iteration uses an initial 10 years of data for training and the next 5 years of data is used to test the success of regression followed by a forecast for the next 5 years. In the second iteration, first, 12 years of data is used for training, the next 3 years for testing, and then forecasted for 5 years. In the third iteration, first, 8 years of data is used for training and after this, the next 7 years of data is used for testing, and further forecasted for 5 years. Root mean square error and r 2 were calculated to attribute the best-suited iteration for the forecast. Additionally, Spearman rank correlation was performed to understand the statistical significance between pollutants under observation to obtain the correlation matrix among the pollutants. It is the measure of the strength of a monotonic relationship between paired data, mathematically described as [56]: where ρ = Spearman rank correlation, di = the difference between the ranks of corresponding variables, n = number of observations.

Soil Quality Analysis
The soil samples were collected and analyzed for arsenic, fluoride, titanium, iron, chromium, lead, copper, zinc, and manganese using the inductively coupled plasmaatomic emission spectrometry (ICP-AES) at Anacon Laboratories, Nagpur, recognized by Seasonal, trend, and irregularity components are required to predict the pollutant concentration as expressed in Equation (3). Seasonality (S t ) and irregularity (I t ) components are extracted by smoothening the data using a 12-month moving average and calculating the center moving average as the seasonal data is even numbered. The mathematical equation used to calculate S t I t is [55]: where Y t = Original Data/Predicted Data, S t = Seasonal Component, I t = Irregularity Component, CMA = Centre Moving Average. Next, the irregularity component is removed to obtain the seasonal component that is used to deseasonalize the data. It is conducted by averaging the seasonality and irregularity of individual months for the entire data set. The result thus obtained is a 12-month cyclic trend. Using this seasonal component, the data is deseasonalized using the following equation [55].
where Y t = Original Data, S t = Seasonal Component. This deseasonalized data is used in linear regression to extract trend components using a simple linear equation [55].
where m is the slope and c is the intercept that is obtained from linear regression and x is the time component. Further, the trend and seasonal components are used in Equation (3) to forecast pollutants until 2025. While obtaining the trend component, three iterations with different training periods are used. The first iteration uses an initial 10 years of data for training and the next 5 years of data is used to test the success of regression followed by a forecast for the next 5 years. In the second iteration, first, 12 years of data is used for training, the next 3 years for testing, and then forecasted for 5 years. In the third iteration, first, 8 years of data is used for training and after this, the next 7 years of data is used for testing, and further forecasted for 5 years. Root mean square error and r 2 were calculated to attribute the best-suited iteration for the forecast. Additionally, Spearman rank correlation was performed to understand the statistical significance between pollutants under observation to obtain the correlation matrix among the pollutants. It is the measure of the strength of a monotonic relationship between paired data, mathematically described as [56]: where ρ = Spearman rank correlation, d i = the difference between the ranks of corresponding variables, n = number of observations.

Soil Quality Analysis
The soil samples were collected and analyzed for arsenic, fluoride, titanium, iron, chromium, lead, copper, zinc, and manganese using the inductively coupled plasma-atomic emission spectrometry (ICP-AES) at Anacon Laboratories, Nagpur, recognized by the Ministry of Environment & Forests (MoEF) which are given in Table 3 [25]. This data of heavy metal concentration in soil samples is used to estimate the geo-accumulation index. The geo-accumulation index of heavy metals in soil samples is evaluated using the following equation as described by Muller [57]: I geo = log 2 (Cn/1.5Bn) (8) where Cn is the measured concentration of metal and Bn is the geochemical background values of metals as explained by Muller [57]. The background reference values are taken based on the lithology where the sites are situated. All the sites have major lithology of sandstone with minor shale so the background concentration of sandstone is used. However, Rihand dam and Hindalco TPP are situated on granitic-gneissic complex. A factor of 1.5 is used to include a possible variation of background values due to lithogenic effects [57]. Soil quality is classified according to geo-accumulation index (I geo ) values, i.e., unpolluted, moderately polluted, and extremely polluted ( Table 4). Table 4. Division of class for Geo-accumulation index according to Muller, 1979 [57].

Results
In the Singrauli region, coal mining activities are prevalent, and the coal from these mines are used in the TPPs. The transport of coal from mining areas to coal-based power plants is the source of pollution along the roads that affect the people living in the region. The people in this region use coal for cooking and heating purposes during the winter season [22] thus elevating the levels of PM2.5 to very high. Furthermore, Indian coal has high ash content and low calorific value. Sulfur content is also less compared to coal found in the United States and China. The low calorific value and high ash content increase the emissions per kWh electricity generated. In addition, the coal mined from opencast mines such as in the study area has more ash content. Indian coal has another problem as its silica and alumina content is high, which reduces the ash collection efficiency at electrostatic precipitators (ESPs). The Indian government has mandated the use of coal after the reduction of ash content by at least 34% in critically polluted and ecologically sensitive areas. However, due to a lack of access to continuous monitoring data, compliance is uncertain [22]. We have analyzed long-term variations of air quality and have discussed long-and short-term variations of pollutants in the subsequent sections.

Long-Term Variations of Pollutants Associated with Thermal Power Plants
The natural and anthropogenic activities enhance CO 2 concentrations in the atmosphere that are responsible for climate change and global warming. The CO 2 concentrations in the atmosphere have a long-term residence time of about 300 to 1000 years. The descriptive analysis shows that monthly concentration of CO 2 ranges from 370.58 ppm to 403.69 ppm between 2003-2016. The average CO 2 concentration is 387.65 ppm with a standard deviation of ±8.70 and a variance of 75.52 ppm (Table 5). This shows that the minimum and maximum concentrations have a low deviation from the average concentration. It is observed that CH 4 Concentrations ranges from 1786.99 ppbv to 1895.47 ppbv between 2003-2020 ( Table 5). The average concentration of CH 4 is 1842.56 ppbv with a standard deviation of ±20.18 ppbv and a variance of 407.14 ppbv. Hence, between 2005-2017, the total power generation capacity installed in the study area is around 13,580 MW, which is almost double, as compared to capacity installed before 2005, i.e., 7584 MW. Thus overall, around 21 GW power generation capacity is installed in the study area. The increase in the number of TPPs, as well as the increase in the installed capacity of existing TPPs, has led to an increasing in the concentration of pollutants in the study area.
The long-term variation of each pollutant is plotted on an annual basis to see the effect of installation and expansion of TPPs in relation to pollutant concentrations. As the data range for each pollutant is different, a min-max normalization technique is applied for comparable results. The trend in variation of each pollutant is obtained by using the slope of the trend line of normalized data. It is observed that all the pollutants are showing an increasing trend.
For the whole time period, CO 2 has increased from 374.27 ppm in 2003 to 401.80 ppm in 2016 with a slope of 7.65 ( Figure 4). Similarly, SO 2 increased from 0.15 DU in 2005 to 0.20 DU in 2020 with a slope of 6.20 and NO 2 increased from 20.00 ppm in 2005 to 22.00 ppm in 2020 with a slope of 6.05. Also, CH 4 increased from 1818.64 ppbv in 2003 to 1865.34 ppbv in 2020 with a slope of 5.74. Thus, the concentration of CO 2 increased at the fastest rate as its slope is maximum followed by SO 2 and NO 2 , whereas, CH 4 increased at the least rate among all four pollutants.
Furthermore, the data is segregated for different time periods to relate the variation in pollutant concentration with the installation or expansion of TPPs. The installation or expansion of TPPs are not at a uniform rate. Hence, the time period for pollutant concentration was chosen on the basis of increased power generation capacity. Three different time periods from 2003-2006, 2006-2015, and 2015-2018 are selected, during which the cumulative increase in power generation capacity in the study area is 1500 MW, 9920 MW, and 13,580 MW respectively. Thus, after 2015, as the cumulative power generation has increased to around 21 GW, it is anticipated that the concentration of pollutants will enhance in the future will further degrade air quality and associated impacts such as the formation of haze and fogs etc. It is attributed that in 1st time period CO 2 concentration is increased with a slope of 7.43 and CH 4 concentration is increased with a slope of 4.70. In the 2nd time period, CO 2 concentration increased with a slope of 7.70, CH 4 increased with a slope of 6.17, NO 2 concentration increased with a slope of 4.53, and SO 2 concentration is increased with a slope of 2.97. Lastly in the 3rd time period, NO 2 concentration increased with a slope of 7.48, SO 2 concentration increased with a slope of 10.42 and CH 4 concentration increased with a slope of 5.70. Hence, CO 2 increased at the fastest rate in 1st-time period and 2nd-time period, however, SO 2 increased at the fastest rate in the 3rd-time period followed by NO 2 .
7.70, CH4 increased with a slope of 6.17, NO2 concentration increased with a slope of 4.53, and SO2 concentration is increased with a slope of 2.97. Lastly in the 3rd time period, NO2 concentration increased with a slope of 7.48, SO2 concentration increased with a slope of 10.42 and CH4 concentration increased with a slope of 5.70. Hence, CO2 increased at the fastest rate in 1st-time period and 2nd-time period, however, SO2 increased at the fastest rate in the 3rd-time period followed by NO2. It can be observed that CO2 and CH4 have been on a continuous rise since 2003. On the other hand, NO2 and SO2 have some sinks in concentrations. This is because rainfall does not have an immediate, but a long-term effect on CO2 and CH4 [58]. Both the pollutants decrease with an increase in plant or forest cover in the area. In other words, they are not directly mixed with rainfall to decrease their concentration in the surrounding air. These pollutants are decreased as rainfall or monsoon increase the green cover in the area, which acts as a sink for their concentration. However, NO2 and SO2 can be mixed with rainfall and result in the formation of acid rain [59]. The dip in SO2 and NO2 in the year 2012 may be caused due to a 61% higher than average rainfall received in the study area [60]. Additionally, the data for both these pollutants is total column data, any variation in the atmosphere can cause a change in concentration. This may be the reason for higher uncertainty in the data. Similarly, in the year 2014, the heavy rainfall and winds are caused by the Hudhud cyclone [61]. The strong winds caused the dispersion of pollutants, reducing the concentration of pollutants, and also causing strong mixing of the It can be observed that CO 2 and CH 4 have been on a continuous rise since 2003. On the other hand, NO 2 and SO 2 have some sinks in concentrations. This is because rainfall does not have an immediate, but a long-term effect on CO 2 and CH 4 [58]. Both the pollutants decrease with an increase in plant or forest cover in the area. In other words, they are not directly mixed with rainfall to decrease their concentration in the surrounding air. These pollutants are decreased as rainfall or monsoon increase the green cover in the area, which acts as a sink for their concentration. However, NO 2 and SO 2 can be mixed with rainfall and result in the formation of acid rain [59]. The dip in SO 2 and NO 2 in the year 2012 may be caused due to a 61% higher than average rainfall received in the study area [60]. Additionally, the data for both these pollutants is total column data, any variation in the atmosphere can cause a change in concentration. This may be the reason for higher uncertainty in the data. Similarly, in the year 2014, the heavy rainfall and winds are caused by the Hudhud cyclone [61]. The strong winds caused the dispersion of pollutants, reducing the concentration of pollutants, and also causing strong mixing of the pollutants which could be the cause of acid rains in the study area, showing a scavenging effect that result in the declining of pollutants in the atmosphere.

Short-Term Variation of Pollutants Associated with the Thermal Power Plants
The short-term variation of each pollutant is plotted on a monthly basis (monthly average of pollutant concentration) to see the effect of installation and expansion of TPPs in relation to pollutant concentrations.
The growing anthropogenic activities are associated with the increasing population, urbanization, biomass burning, traffic, and coal burning in TPPs and households which are the main sources of the increasing CO 2 concentration and its various adverse impacts. pollutants which could be the cause of acid rains in the study area, showing a scavenging effect that result in the declining of pollutants in the atmosphere.

Short-Term Variation of Pollutants Associated with the Thermal Power Plants
The short-term variation of each pollutant is plotted on a monthly basis (monthly average of pollutant concentration) to see the effect of installation and expansion of TPPs in relation to pollutant concentrations.
The growing anthropogenic activities are associated with the increasing population, urbanization, biomass burning, traffic, and coal burning in TPPs and households which are the main sources of the increasing CO2 concentration and its various adverse impacts.   CH 4 gas emission is one of the second-highest contributors to atmospheric warming after the CO 2 , being 28 times more effective at trapping radiation and warming the planet. The net increase in CH 4 concentration in the atmosphere is mainly due to high anthropogenic emissions such as coal mining, coal burning in TPPs, etc. The high CH 4 concentrations can be observed in the study region. For instance, the capacity increase of NTPC Shaktinagar by 500 MW in the month of June 2012 increases the total power generation to 11,784 MW. This has resulted in the rise of CH 4 concentration from 1820. 60

Time Series Analysis of Pollutants
Natural occurring CO 2 is essential in warming the planet to make it habitable. However, anthropogenic activities have significantly increased their concentration contributing to global warming. The pre-industrial CO 2 level was at 280 ppm and the global average crossed 400 ppm in 2018 [62], whereas the maximum CO 2 concentration is observed to be 403.69 ppm in 2016 in the study area. The monthly CO 2 concentrations are on an increasing trend despite its natural sinks (Figure 6) that follow a "Keeling's Curve". Keeling's Curve, is a graph of CO 2 concentration, which shows that the concentration peaks during spring and sinks during fall (autumn). It shows that if no mitigation measures are taken, the CO2 will continue to increase and soon reach the upper limit of 430 ppm, which will overshoot the 1.5-degree global temperature rise goal, according to IPCC.
The SO2 associated with anthropogenic activities play a major role in climate change [63] which is also responsible for acid rain. Secondary aerosol particles are considered to be a successor of SO2 and are responsible for the formation of haze. As per the natural cycle of SO2, high concentrations can be observed during the winter season and low concentrations during the summer season. This is related to temperature, humidity, and wind speed [64]. The pre-processed data between 2005 to 2014 is used for training the forecast model and data between 2015-2020 is used to test the model. The test data is compared with the predicted data and a low r 2 value of 0.41 is obtained. This low r 2 value can be attributed to the high deviation in the data i.e., the average value is almost three times the standard deviation. This shows a high uncertainty in the SO2 concentration data. Also, it can be observed that after 2016 data shows sudden spikes, due to which r 2 value estimated is low (Figure 7). However, using the regression model, it is predicted that the maximum concentration can rise to 0.23 DU, and the minimum concentration can reach to 0.13 DU in 2025. It shows that if no mitigation measures are taken, the CO 2 will continue to increase and soon reach the upper limit of 430 ppm, which will overshoot the 1.5-degree global temperature rise goal, according to IPCC.
The SO 2 associated with anthropogenic activities play a major role in climate change [63] which is also responsible for acid rain. Secondary aerosol particles are considered to be a successor of SO 2 and are responsible for the formation of haze. As per the natural cycle of SO 2 , high concentrations can be observed during the winter season and low concentrations during the summer season. This is related to temperature, humidity, and wind speed [64]. The pre-processed data between 2005 to 2014 is used for training the forecast model and data between 2015-2020 is used to test the model. The test data is compared with the predicted data and a low r 2 value of 0.41 is obtained. This low r 2 value can be attributed to the high deviation in the data i.e., the average value is almost three times the standard deviation. This shows a high uncertainty in the SO 2 concentration data. Also, it can be observed that after 2016 data shows sudden spikes, due to which r 2 value estimated is low (Figure 7). However, using the regression model, it is predicted that the maximum concentration can rise to 0.23 DU, and the minimum concentration can reach to 0.13 DU in 2025. Under the influence of solar radiations, which results in ground-level ozone formation, the concentration of NO2 during summer is low [65]. On the other hand, due to low temperature, high humidity, and low wind speed, the NO2 concentration increases during the winter season (Figure 8). The pre-processed data between 2005 to 2014 is used for training the forecast model and data between 2015-2020 is used to test the model. The test data is compared with the predicted data and a low r 2 value of 0.55 is obtained. This Under the influence of solar radiations, which results in ground-level ozone formation, the concentration of NO 2 during summer is low [65]. On the other hand, due to low temperature, high humidity, and low wind speed, the NO 2 concentration increases during the winter season ( Figure 8). The pre-processed data between 2005 to 2014 is used for training the forecast model and data between 2015-2020 is used to test the model. The test data is compared with the predicted data and a low r 2 value of 0.55 is obtained. This low r 2 value can be attributed to the high deviation in the data, i.e., the average value is almost four times that of the standard deviation. This shows a high uncertainty in the NO 2 concentration data. However, this regression model predicts that the maximum concentration can rise to a value of 29.28 ppm, and the minimum concentration to 21.82 ppm in 2025. Under the influence of solar radiations, which results in ground-level ozone formation, the concentration of NO2 during summer is low [65]. On the other hand, due to low temperature, high humidity, and low wind speed, the NO2 concentration increases during the winter season (Figure 8). The pre-processed data between 2005 to 2014 is used for training the forecast model and data between 2015-2020 is used to test the model. The test data is compared with the predicted data and a low r 2 value of 0.55 is obtained. This low r 2 value can be attributed to the high deviation in the data, i.e., the average value is almost four times that of the standard deviation. This shows a high uncertainty in the NO2 concentration data. However, this regression model predicts that the maximum concentration can rise to a value of 29.28 ppm, and the minimum concentration to 21.82 ppm in 2025. CH4 is one of the short-lived climate pollutants with 28 times greater power than CO2 in warming the planet, nearly 60% of methane is produced by anthropogenic activities [66]. Additionally, it promotes the formation of ground-level ozone and smog. The burning of CH4 gas forms black carbon and volatile organic compounds [28]. Due to its catalytic nature, it is imperative to regulate its emission. An increasing trend of methane is CH 4 is one of the short-lived climate pollutants with 28 times greater power than CO 2 in warming the planet, nearly 60% of methane is produced by anthropogenic activities [66]. Additionally, it promotes the formation of ground-level ozone and smog. The burning of CH 4 gas forms black carbon and volatile organic compounds [28]. Due to its catalytic nature, it is imperative to regulate its emission. An increasing trend of methane is observed in the study area ( Figure 9). The pre-processed data between 2003 to 2014 is used for training the forecast model and data between 2015-2020 is used to test the model. The test data is compared with the predicted data and the r 2 value obtained is 0.66 due to the low value of standard deviation ±19.19 with respect to the mean value of 1840 ppbv. The predicted data shows that the maximum concentration can rise to a value of 1901.35 ppbv and the minimum concentration can increase to a value of 1859.81 ppbv in 2025. observed in the study area ( Figure 9). The pre-processed data between 2003 to 2014 is used for training the forecast model and data between 2015-2020 is used to test the model. The test data is compared with the predicted data and the r 2 value obtained is 0.66 due to the low value of standard deviation ±19.19 with respect to the mean value of 1840 ppbv. The predicted data shows that the maximum concentration can rise to a value of 1901.35 ppbv and the minimum concentration can increase to a value of 1859.81 ppbv in 2025.

Spearman's Rank Correlation
The pollutants in the environment follow a cycle, they complete their residence time and convert into other compounds through chemical reactions. For instance, CH4 oxidizes to CO2 and H2O, after its residence period; NO2 results in the formation of NOx and O3 (ozone); NO2 and SO2 form secondary aerosols, i.e., particulate matter (PM) [67]. These

Spearman's Rank Correlation
The pollutants in the environment follow a cycle, they complete their residence time and convert into other compounds through chemical reactions. For instance, CH 4 oxidizes to CO 2 and H 2 O, after its residence period; NO 2 results in the formation of NO x and O 3 (ozone); NO 2 and SO 2 form secondary aerosols, i.e., particulate matter (PM) [67]. These aerosols are responsible for direct solar radiation absorption, which results in the warming of the planet. Additionally, they also form smog, haze, etc., which reduces visibility, thus impacting the environment. With an increase in warming, the atmospheric circulation reduces, and the accumulation of pollutants in a region increases. This leads to an increase in natural emissions of CO 2 and CH 4 in the atmosphere. The enhanced concentrations of these pollutants in the warming climate create a positive feedback loop. Hence, these pollutants need to be studied in relation to aerosols present in the atmosphere.
Therefore, in this section, AOD (Aerosol Optical Depth), downloaded from the NASA Giovanni portal, PM 2.5 , and PM 10 , taken from CPCB ground observations, are analyzed with other pollutants. Spearman's rank correlation is used to correlate pollutants and aerosols in the study area. Spearman's correlation coefficients are depicted in the lower diagonal and p-values are shown in the upper diagonal of the correlation matrix of Table 6. The significant correlations, at a ≥95% significance level, are shown in bold in Table 6. NO 2 shows a high positive correlation with PM 2.5 (0.60) and PM 10 (0.66). Moreover, CO 2 is significantly positively correlated with NO 2 (0.47), CH 4 (0.48), and SO 2 (0.28). Similarly, CH 4 has a high positive correlation of 0.59 with SO 2 , while it has a significant negative correlation of −0.23 with AOD. Hence, it shows that an increase in CO 2 will result in an increase in NO 2 , CH 4 , and SO 2 . This is because the increase in CO 2 changes the composition of surrounding air and results in an increase of other pollutants in the atmosphere [68]. Since NO 2 is the precursor of secondary aerosols, NO 2 shows a high positive correlation with PM 2.5 and PM 10 . On the other hand, the negative correlation between CH 4 and AOD is attributed to the short life span of CH 4 and its early conversion to soot (BC).

Accumulation Index of Soil in Singrauli Region
A typical power plant uses 12,000 tons of coal per day and drains 1 Mt of waste per year. During this process, a significant number of byproducts are transferred to the surrounding environment. These byproducts have two essential routes to be released into the environment: atmospheric emission and leaching of dumped byproducts such as ash ponds, ash dumps, etc., in the surrounding soil.
It was observed that for arsenic contamination, Obra TPP, Bina coal mines, and Khadia coal mines are categorized in class 2 of geo-accumulation index, i.e., moderately polluted. For lead, Bina coal mines' sampling site is categorized in class 4 of geo-accumulation index, i.e., highly polluted, while, Obra TPP and Khadia coal mines are categorized in class 3 of geo-accumulation index, i.e., moderate to highly polluted. It is observed that for chromium Hindalco industries is classified in class 5, i.e., highly to extremely polluted, Rihand dam is categorized in class 4, i.e., highly polluted and Lanco TPP and Obra TPP are classified into class 3, i.e., moderately to highly polluted. In the case of zinc, the Anpara TPP site is classified in class 5 of geo-accumulation index, i.e., highly to extremely polluted, while the Bina coal mines and Vindhayachal TPP are classified in class 3 of geo-accumulation index, i.e., moderately to highly polluted.
Furthermore, we have added the geo-accumulation index for all the heavy metals to obtain the cumulative index. This shows that Hindalco industries, Anpara TPP, Bina coal mines, Khadia coal mines, and Vindhayachal TPP are the most polluted, followed by Lanco TPP, NTPC Shaktinagar, Near Vindhayanagar, and Rihand dam. On the other hand, Singrauli reservoir is least polluted as compared to other sampling sites which are moderately to extremely highly polluted. It is observed that the cumulative index is highly dependent on the variation of lead followed by zinc and chromium (Table 7). Table 7. Accumulation index calculated using Muller formula and divided into classes as described in Table 7 in Singrauli region. The background values used are As (1.5), Ti (3400), Fe (14,200), Cr (4.1), Pb (19), Cu (10), Zn (39), and Mn (390) for Rihand dam and Hindalco TPP. For all the other sites, background values taken are As (1), Ti (1500), Fe (9800), Cr (35), Pb (7), Cu (4), Zn (16), and Mn (850) [69].

Discussion
Singrauli region has one of the most important coal mines in the country and TPPs installed surrounding the mining area generate around 21 GW of electricity, making it an "Energy Hub". The main air pollutants emitted by coal mining and TPPs are PM, SO 2 , NO 2 , and CO 2 [70]. It is documented that for every kWh of coal-based electricity generation, 0.8-0.9 kg of carbon dioxide is emitted into the surrounding air [1]. The establishment of Rihand Dam which created Govind Ballabh Pant Sagar reservoir, TPPs, cement plant, and aluminum and chemical industries in the late 1950s has changed the land cover and land use pattern of the region drastically. Within a period of 1978-2010, the mining area has increased by 590%, built-up area by 350%, and cropland by 71.50%. On the other hand, open forest and dense forest has decreased by 25% and 56% [19]. Further, between 2000 and 2016, mining activities increased threefold [20]. The change in LULC such as a decrease in open and dense forests has increased the surface temperature by 6 10 , Total Suspended Particles (TSP), NO 2, and SO 2 concentrations present in the study area were higher than the National Ambient Air Quality Standards of India [38].
During the COVID-19 pandemic (2019-2020), the lockdown was imposed in many countries resulting in reduced air pollution levels [24]. Essential services such as electricity generation from TPPs and mining activities were also reduced during lockdown [73]. However, during this time, the air quality of Singrauli did not improve much, whereas other locations such as Delhi, Mumbai, etc. experienced reduced air pollution levels [74]. The same can be observed in our study, where, SO 2 has increased from 0.19 ppm in 2019 to 0.20 DU in 2020 and CH 4 has increased from 1862.44 ppbv in 2019 to 1865.37 ppbv in 2020.
In a work carried out by Guttikunda and Jawahar [75], it was forecasted that by 2030, the power generation capacity of the study area will increase by 170% which will result in a coal consumption increase by 170%. This project showed an increase in SO 2 , NO 2 , and CO 2 by 169.2%, 132.38%, and 169.10%, respectively. In our study, on the basis of past trends, we predicted pollutant concentration in the study area from 2021-2025. This forecast shows that CO 2 increases from 374. 27  The soil analysis of our study conducted on samples collected in the year 2015 [25] shows heavy metal contamination in and around TPPs and mining areas. We found that the soil samples of the region are highly to extremely polluted in the case of Cr, Pb, and As, and unpolluted to moderately polluted for Mn, Ti, and Fe. Similarly, Agrawal et al. [35] conducted a quantitative estimation of heavy metals in the soil around TPPs from March 2005 to February 2008. The 256 samples collected during pre-monsoon and post-monsoon reported a high concentration of heavy metals in the area. The average maximum concentration of cadmium, lead, arsenic, and nickel in soil was observed to be 0.69, 13.69, 17.76, and 3.51 mg/kg, respectively [35].

Limitations and Recommendation
The previous studies conducted in the Singrauli region are either short-termed or are focused on fewer pollutants. Our long-term study analyzes the variation of pollutants with respect to TPPs and forecasts their concentration on the basis of past trends using satellite-derived data. However, the limitations of this study are twofold. Firstly, while the satellite data provide broad spatial and temporal coverage, the ground-observation data are more accurate. Additionally, satellite data obtained is difficult to validate in case of a lack of ground monitoring stations. So, for high accuracy, a dense network of ground monitoring stations is required. However, in developing countries such as India, dense monitoring grid is not economical. Secondly, the satellite data available for CO 2 has a spatial resolution of 2 × 2.5 • , for CH 4 , the spatial resolution is 1 × 1 • . This means that a single grid will cover a very large area of the order of thousands of sq. km. This might not be accurate for point location studies. Thus, the concentration of the pollutants may not represent a very distinct and clear picture. Hence, better resolution satellite data will provide a more comprehensive and rigorous analysis of the concentration of pollutants.
The observed trend and forecast of our study show a continuous rise in pollutant concentration in the region. Their long-term exposure can cause an adverse impact on the environment and various health impairments in humans. For instance, if more and more pollutants are loaded into the atmosphere, the global temperature will continue to rise. This can cause frequent extreme events, the rise of sea level, drought, change in rainfall patterns, etc. Our study shows that if no mitigation measures are taken, it will be difficult for India to fulfill the Paris agreement goals to curb emissions by 2030. It is also articulated that CO 2 concentration has already reached 400 ppm, which is the limit defined by IPCC to control global temperature rise below 1.5 • and it can rise to 422 ppm in 2025. Hence, it is high time to take mitigation measures and control pollution levels at the regional and national levels. For instance, the short-lived pollutant CH 4 , if controlled, can show a near-term curb in temperature rise, thus helping in achieving Paris Agreement goals. Additionally, at present, non-renewable energy sources account for more than 50% of electricity production. The shift toward renewable sources such as solar, wind, and hydropower will reduce greenhouse gases in the environment. Zero waste approaches in TPPs and efficient production equipment for minimum emissions need to be implemented. Last, strict emission rules for industries and TPPs will help reduce the pollution and, by extension, climate change and deaths caused by ambient air pollution.

Conclusions
Singrauli is one of the most highly polluted regions, owing to high emissions from TPPs, coal mining, and numerous industries. In this study, the long-term and short-term variations of air pollutants with respect to TPPs in the region are analyzed using satellite data. This study shows that annual average concentration of CO 2 has increased from 374. 27  The long and short-term variation concludes that pollutants' concentration suddenly increased in 2007, 2014, 2015, and 2017, which coincided with the installation or expansion of TPPs in the study area. This concludes that increased TPPs and coal mining in the area are increasing the pollutant concentration in the atmosphere. The forecast of the study states that the concentration of CO 2 , NO 2 , SO 2 , and CH 4 in the year 2025 will rise to 422.59 ppm, 29.28 ppm, 0.23 DU, and 1901.35 ppbv, respectively, in the Singrauli region. It was observed that CO 2 is significantly correlated to all other pollutants using Spearman's rank correlation test, while CH 4 and SO 2 have a strong correlation with each other. Additionally, NO 2 is significantly related to PM 2.5 and PM 10 . Furthermore, this study concludes that the soil of the Hindalco industries, Anpara TPP, Bina coal mines, Khadia coal mines, and Vindhayachal TPP are the most highly polluted with heavy metals, while Singrauli reservoir is the least polluted. Thus, the air and soil of the Singrauli region is highly polluted. After spending some time in the area during the field visit, we are really concerned about how so many people live in such a highly polluted region. There are no records of patients suffering from different kinds of diseases, but early morning and in the evening, there is a huge lineup of patients seen at the health clinic and in the hospitals. The present study will attract the attention of Government to take steps to save the lives of people living in the surrounding areas by taking proper mitigation measures to alter the course of the pollution.

Data Availability Statement:
The data used in this work is freely available from the sources as mentioned in the acknowledgements.