Abstract

The purpose of this study is to compare the Air Quality Index (AQI) in India before (1^st January 2020 - 16^th March 2020) and during the lockdown period (26^th March 2020 - 4^th June 2020). The AQI improved significantly throughout the lockdown resulted through spatio-temporal fluctuations based on Geographic Information System (GIS) mapping and multivariate statistical analysis using Correlation Matrix Analysis and Principal Component Analysis (PCA). The AQI declined by 34.51 percent in Northern India, 22.14 percent in Central India, 48.84 percent in Eastern India, 42.75 percent in Western India, 32.64 percent in Southern India, and 54.46 percent in Northeastern India during the lockdown. The findings would prompt the governments across the world to consider ways to reduce vehicle and industrial pollution to enhance air quality.

1. Introduction

COVID - 19, a disease first reported in December 2019 from Wuhan city (China), caused by β-Coronavirus (family Coroniviridae), has created enormous stress and anxiety amongst people around the globe. It has affected health care and has also impacted social relationships and the economy ¹. Approximately 19.8 crores confirmed cases with 42.2 lakh deaths worldwide, and 3.17 crores confirmed cases with 4.24 lakh in India had been recorded until 1^st August 2021 (WHO, 2021). In India, the first case of COVID – 19 was reported from Kerala in January 2020. To control the disease spread, a nationwide public curfew was imposed in India on 22^nd March, followed by a lockdown on 25^th March 2020 (lockdown 1.0, 21 days). Subsequently, the lockdown was extended to lockdown 2.0 (19 days), lockdown 3.0 (14 days), lockdown 4.0 (14 days). The lockdown enforced restrictions on public transport, construction work, industrial and services sector, mass gatherings at various places etc., with self-quarantine measures. Similarly, shopping malls, places for recreational activities, gymnasiums and theatres etc., were closed all over the country. A study from 12 countries suggests the lockdown measures, social distancing and restricted anthropogenic activities decelerate the spread of COVID-19. It has resulted in the decrease of public transport by up to 90% from March 2020, which remained significantly less till May, 2020 ². In many countries such as Barcelona, Brazil, and China, there was a decrease in particulate matter due to lockdown, resulting in low levels of air pollution ^{3, 4, 5}. In the Indian context, the improvement in air quality has been observed during lockdown ^{6, 7, 8}. Restricted human activities played a significant role in the revival of the environment by improving air quality ⁹.

The air quality is expressed as Air Quality Index (AQI) or Air Pollution Index, a standard used to indicate the severity of air pollution ¹⁰. AQI represents the cumulative effect of the pollutants showing overall air quality status comprehensively and directly by categorising it into good, moderate, poor, and severe ¹¹. The augmented concentration of air pollutants corresponds to high AQI levels revealing deteriorating air quality as per WHO standards. Several Indian cities for many years have been in the top 20 most polluted cities of the world ¹². Poor air quality affects masses, makes them vulnerable to various airborne ailments such as respiratory diseases, lung cancer, cardiovascular diseases and Chronic Obstructive Pulmonary Disease (COPD), especially people with weak immunities ^{13, 14, 15}.

In some cases, the Reactive Oxygen Species (ROS) generated by water-soluble particles produce hydroxyl radical (OH^˚) by activating metals, that causes oxidative damage of DNA and have increased the risk of infections ¹⁶. Industrial and transport sectors that together form a significant chunk of anthropogenic emissions of pollutants, specifically in developing countries, impact air quality and impose significant health risks ¹⁷. Pollution is the primary environmental cause of disease and premature deaths worldwide, and it has been estimated that in 2015, nearly 4.2 and 2.9 million people died due to ambient and indoor air pollution, respectively, across the world ¹⁸.

Pollution has no political boundaries and cannot be controlled unless governments all across the world take strong steps and rigorously execute preventative measures to reduce pollution. It is undeniable that technological advancements have overtaken environmental aims. A healthy environment and a healthy economy go simultaneously in nature, and sustainability is impossible to attain without considering the environment. Improving air quality while still looking after the economy is a difficult task for any country, and it necessitates substantial measures. The aforementioned intriguing trend prompted us to investigate several aspects of the air quality index, air pollution, multiple contributors, and AQI before and during the lockdown. Thus, the main purpose of this article is to discuss the improvement in the air quality index in India's major cities during COVID-19 pandemic due to restricted anthropogenic activities in the lockdown period. Also, to discover a remedy or potential approaches to maintain the current AQI level for future strategic planning and management to enhance the air quality.

2. Materials and Methods

The air quality data was collected for 93 cities from 18 States and NCT of Delhi from India (Figure 1) at an interval of 10 days starting from 6^th January 2020 to 4^th June 2020, and dates were classified as before lockdown (6^th January to 16^th March 2020) and during lockdown (26^th March to 4^th June 2020). The study area was selected based on the data availability for the year 2020 at the National Air Quality Index website of Central Pollution Control Board, India (CPCB, 2020), AQI Bulletin Archive (CPCB, 2020), SAFAR India website (SAFAR, 2020) and Chhattisgarh Environment Conservation Board (CECB, 2020).

CPCB (CPCB, 2020), SAFAR ¹⁹, and CECB ²⁰ follow the standard protocol for sub-indices and breakpoint concentrations of pollutants (PM10, PM2.5, NO₂, SO₂, CO, O₃, NH₃, and Pb), which are made according to Indian National Ambient Air Quality Standard under the National Air Monitoring Program (NAMP). These agencies cover different areas for monitoring, SAFAR forecast air quality for 4 cities, i.e. Delhi (8 different locations), Pune (10 different locations), Mumbai (10 different locations) and Ahmedabad (8 different locations) under the authority of the Ministry of Earth Science, Govt. of India and Indian Institute of Tropical Meteorology, Pune; CECB covers Chhattisgarh state by 4 regional offices located at Raipur, Bilaspur, Korba, Durg-Bhilai and Raigarh; whereas CPCB monitor AQI for the almost entire country with 793 operating stations covering 344 cities/towns in 29 states and 6 Union Territories of India.

The spatial and temporal variations of air pollution during COVID-19 were analysed using Geographic Information System (GIS) ²¹. The Spatio-temporal variability of the AQI dataset for various locations within the study area (spatial analysis) and multi-dates (temporal analysis) in the year 2020 was carried out using interpolation method using ArcGIS 10.1, GIS software ²². Interpolation method offers the continuous dataset for the locations having no dataset, through predicting the values from the locations having desired dataset ²³. In the present investigation, Topo to Raster interpolation method was carried out using spatial analyst tool of ArcGIS 10.1 software ^{24, 25}. The output dataset was classified further in different classes of AQI Values, and colour-coding was applied following the CPCB AQI colour code guide (CPCB, 2020) for a better visual interpretation of the dataset.

The time-series based AQI comparison was carried out using multivariate statistical analysis techniques through Correlation analysis and Principal Component Analysis (PCA). Correlation analysis and PCA were done to evaluate the association between AQI of different dates and principal components explaining the information of the data matrix, respectively ²⁶. The mathematical derivatives and the correlation matrix and PCA graphical representation were generated using SPSS 16.0 and R- software version 3.6.3 ²⁷. Principal Components (PC's) were extracted from the original dataset following eigenvalue >1. Kaiser-Meyer-Olkin (KMO) Bartlett's sphericity test has been performed to validate the PCA test ²⁸. PCA is the dimension reduction method applied to a large dataset to extract eigenvalues from the covariance matrix of the original dataset ²⁹. In the present study, PCA was applied on AQI values for 2020 following the Varimax rotation method with Kaiser Normalization ³⁰.

3. Results and Discussion

A significant improvement in air quality as an impact of nationwide lockdown has been observed across India (Table 1 a-f). In the present study, Spatio-temporal variation of AQI using visual interpretation of interpolation maps for AQI values for different dates were prepared (Figure 3a, b). The Spatio-temporal analysis helped identify pollution hotspots, particularly states and relative change in AQI in the year 2020 on selected dates. The comparison between different dates of 2020 highlights the various air pollution hotspots across the country. With the visual interpretation of Spatio-temporal distribution maps of AQI, it becomes evident that the northern and eastern zones comprise more pollution hotspots than other zones of India (Figure 3 a, b). These areas were designated hotspots with high pollution levels because of a dense population with more anthropogenic activities.

In the present study, entire India has been divided into 6 zones, i.e. Northern zone (49 cities in 5 states, i.e. Delhi, Haryana, Punjab, Rajasthan, Uttar Pradesh), Central Zone (10 cities in 2 states, i.e. Chhattisgarh, Madhya Pradesh), Western zone (14 cities in 2 states, i.e. Gujarat, Maharashtra), Southern zone (9 cities in 5 states, i.e. Andhra Pradesh, Karnataka, Kerala, Telangana, Tamil Nadu), Eastern Zone (10 cities in 4 states, i.e. Bihar, Jharkhand, Odisha, West Bengal), and North-eastern zone (1 city, i.e. Assam). Significant changes in AQI have been observed over almost the entire nation due to restricted human activities and shutting down almost all the services except essential ones (Figure 4). The decline in AQI can also be associated with different climatic conditions in these zones. Amongst different zones of India, the maximum decline in AQI, i.e., 54.46%, has been observed in Northeastern India, followed by 48.84% in Eastern India, 42.75% in Western India, 34.51% in Northern India, 32.64% in Southern India and 22.14% in Central India. The average AQI level reached the lowest value of 100 during lockdown compared to 220 before lockdown in 2020 (Figure 4). The southern zone shows the lowest AQI value in 2020 across India. The southern part of the nation exhibits low, moderate to satisfactory range AQI compared to other parts.

Being one of India's significant densely populated and polluted cities, Delhi generally has higher AQI values throughout the year due to heavy traffic, construction work, and industrial activities ^{31, 32}. During the COVID-19 pandemic lockdown, Delhi witnessed a remarkable decline in AQI as an impact of lockdown (Table 1 a, Figure 5). In Delhi, the AQI level reached its minimum, i.e., 92 in March 2020, from a maximum value of 325 in January 2020. According to the CPCB AQI database, air pollution reduction occurred merely in four days since the lockdown ⁸. Out of seven pollutant parameters (PM₁₀, PM_2.5, SO₂, NO₂, CO, O₃ and NH₃), Delhi has witnessed a significant reduction (> 50%) in PM₁₀ and PM_2.5 pollutants during the lockdown phase ⁸. The maximum decline in NO₂ concentration with an increase in O₃ concentration has been observed during lockdown from a global perspective ¹⁴. In the present study, the data was available for 23 cities in Haryana, showing wide variation in AQI values. Cities like Sonipat, Panipat, Rohtak and Ambala have shown moderate changes in AQI when compared before lockdown and during the lockdown in 2020 (Table 1 a). After lockdown, there was a sharp decline in the AQI value from 100-250 to 100-200. The AQI data shows the variation from an average value of 150.9 before lockdown to 100.67 during the lockdown. Cities like Gurugram, Bahadurgarh, Faridabad, Fatehabad and Ballabgarh have witnessed the inferior air quality before lockdown (Table 1 a). There is no such substantial change in the air quality in these areas, even during the lockdown. This might be because various factories and industrial sites working for essential services are present in these areas, one of the significant sources of air pollution. In Punjab, the AQI shows a variation from an average of 98.05 AQI before lockdown to 69.8 during lockdown (Figure 5). In Uttar Pradesh, 10 different cities have been analysed, wherein Noida, Ghaziabad, and Greater Noida always have high AQI levels. These cities are corporate hubs, and a lot of polluting activities such as the construction of residential and office complexes are observed round the clock. In such cities also, the variation in AQI ranged from 191 (pre-lockdown) to 132 (during lockdown) (Figure 5). Ajmer, Alwar, Bhiwadi, Jaipur, Jodhpur, Kota and Udaipur of Rajasthan, were evaluated for AQI. Cities like Jaipur, Udaipur, Jodhpur and Bhiwadi are famous for their tourism, cultural and heritage industries. In 2020, the index value showed a variation in AQI before and during the lockdown phase from 130 to 92, respectively (Figure 5).

In central India (Table 1 b), Raipur (Chattisgarh) showed a variation from an average of 65.75 index value before lockdown to 62.25 index value during the lockdown. In Madhya Pradesh (Figure 5), Mandideep, the average AQI declines to a difference in average AQI of 70.75 before and during the lockdown, whereas the lowest difference between before and during lockdown average AQI was found in Ujjain, i.e., 38.75. Significant changes have been observed in the AQI of central India. It was always either moderate (101-200) or, if changed during the lockdown, it turned to satisfactory (51-100) or good (0-50).

Western Zone has India's two most important states, Maharashtra and Gujarat. Mumbai is India's financial capital and the largest city ³³. AQI has changed to a suitable category in index value during lockdown (Figure 5). The AQI average value was 144.38 before lockdown and resulted in a gradual decrease during lockdown (58) by the difference of 86.375. The steep decrease of AQI value during lockdown suggests a slowdown of production at factories and restricted traffic movement during lockdown (Table 1 c). For the last 2 years, Mumbai has experienced moderate to unhealthy air pollution, but due to lockdown, there is a significant lowering of AQI making air quality improved. The most negligible impact on the index value of lockdown has been observed in Nagpur, in western India, as the average value of AQI is 83.25 before lockdown and is 74.25 during the lockdown. The air quality of the remaining seven cities of Maharashtra also showed a similar trend of changing AQI of moderate level to good with varying dates from January to June (Table 1 c). In Gujarat, all 4 cities showed low AQI during lockdown except its largest city, Ahmedabad, with an average AQI of 104.63. During the lockdown, the AQI of Ahmedabad increased to 125.5 though both values were found in moderate air quality. Ankleshwar and Vapi are considered the important industrial clusters of Gujarat even though their AQI reached satisfactory levels. Gandhinagar had witnessed the slightest variation, as an average of AQI before and during lockdown were more or less the same and at a satisfactory level. Ankleshwar, Gandhinagar, Vapi and Vatva almost had satisfactory levels during the lockdown.

In Bihar, amongst three major cities, i.e. Gaya, Patna and Muzaffarpur, showed a variation from 220 (before lockdown) to 107 (during lockdown) (Table 1 d). Patna and Muzaffarpur have witnessed a similar trend in AQI. In Jharkhand, coal-based power plant emissions are the major sources of air pollutants, especially NO₂ ³⁴. In Jharkhand, Jorapokhar witnessed an increase in AQI value from 128 to 134 before and during the lockdown period in 2020 (Figure 5). In Odisha, the data was available for only two cities named Talcher and Brajrajnagar. The AQI variation is 156 (before lockdown) to 112 (during lockdown) (Figure 5). Four cities of West Bengal, i.e. Asansol, Howrah, Kolkata and Siliguri, also showed a significant drop in the AQI. Among these cities, Howrah and Kolkata have shown more air polluted areas than others due to densely populated cities with extensive human activities (Figure 5).

In Andhra Pradesh, in four cities, 12 AQI was observed in Amaravati on 5^th May 2020 during the lockdown. In Visakhapatnam, maximum AQI (142) was observed in January 2020 and a minimum of 30 during the lockdown. Similarly, Chikkaballapur, Karnataka, showed a minimum AQI of 16 on 16^th March 2020 (Table 1 e). This might be because, in southern India, the first few cases of India were observed, and people started reducing activities, travel etc., due to self-awareness from COVID-19 spread very early compared to other zones of India. Maximum AQI (123) was reported in January 2020 in Bengaluru. In Thiruvananthapuram (Kerala) minimum, AQI (31) was observed during lockdown and maximum (96) before lockdown. A similar pattern was observed in Chennai, i.e. minimum AQI (42) and maximum (139). In Telangana, Hyderabad showed a minimum (32) maximum (142) following a similar trend (Figure 3 f). The AQI all over southern India declined and improved AQI, resulting in improved air quality (Figure 5). In the North-eastern zone (Table 1 f), the data was available for only one city of Assam, i.e. Guwahati. It showed a variation in the average index value from 221(before lockdown) to 100 (during lockdown) (Figure 5).

Correlation analysis and Principal Component Analysis (PCA) statistical methods are used for a comprehensive understanding of the degree of association between different variables and to extract factors explaining the total variance of data, respectively. In the present study, a correlation matrix analysis table (Figure 6) is generated using 'R' software version 3.6.3. The table shows the distribution of each variable on the diagonal whereas, the bi-variate scatter plot with a fitted line are located on the bottom of the diagonal. The top of the diagonal explains the value of the correlation plus the significance levels. The present analysis depicts the correlation values (r) ranging between +1 and -1, where +1, -1 and 0 are associated with perfectly positive linear correlation, perfectly negative linear correlation and no linear correlation between two variables, respectively. The significance levels (p) of 0.05, 0.01, 0.001 are expressed as star '*', '**’, ‘***’, respectively. The analysis shows a strong positive correlation between different dates in 2020, i.e., before and during the lockdown, at a significance level of 0.01. As a part of before lockdown, the dates 06^th, 16^th, 26^th January and 05^th, 15^th, 25^th February have resulted in linear correlation values between 0.55 to 0.80 showing a strong positive correlation. The linear correlation (0.71) between 26^th March and 5^th April revealed the dates of the lockdown being enforced upon. The dates of 25^th April, 05^th, 15^th, 25^th May and 4^th June have shown a positive correlation between the AQI values on respective dates explaining the decline in AQI during the lockdown.

With the detailed analysis of the association between variables in 2020, the original data matrix also needs to be understood. PCA is a data reduction technique that extracts Principal Components [PCs] from the original variables to account for as much of the original total variance as possible ²⁹. Principal components show a linear combination with the original variables (AQI value for a particular location for a specific date) and are uncorrelated to each other ³⁵. The PCA results revealed a significant decrease in air pollutants concentrations in 93 cities before and during the lockdown. By following the Varimax rotation method, PCA extracted 4 principal components during 2020. The principal components have eigenvalues of more than one referred by scree plot. The PC I has the maximum possible variance and reveals the most information from the original dataset. The subsequent PCs explain the remaining information and variability, one by one. Four principal components during 2020 explained 73.09% of the total variance in the original dataset, as shown in (Figure 7, Table 2). Component loading and communalities for each variable in four selected components have followed the Varimax rotation method. The communalities or loading may provide an index for the efficiency of the set reduced components and the degree of contribution to each of the variables selected in the four principal components. The method was also tested by Kaiser-Meyer-Olkin (KMO) and Bartlett's Test, giving 0.856.

The principal components extracted from the PCA suggest the significant impact of lockdown on AQI parameters and identify the variables that have greater significance to the study. A significant loading of dates in the year 2020 for January 6^th, 26^th and 5^th, 25^th of February, was represented as PC I. The dominance of loading may be attributed to high AQI values because these dates are before lockdown showing similar conditions of anthropogenic activities, transportation, industries etc. The dates 5^th, 15^th, 25^th May with 4^th June and 25^th April, have strongly participated in PC II. These are the dates during the lockdown, which suggest less pollution level in the air due to restricted human activities and transportation. 26^th March, 5^th April, 15^th January and 16^th January 2020 have shown as PC III and resulted in lowering down of AQI and particulate matter due to heavy rainfall, wind flow or rise in humidity. 6^th March and 16^th March 2020 are represented in the PC IV as these are the dates of just before lockdown when people reduced their movements during the initial stage of COVID-19 in the country due to being conscious and aware of consequences (Figure 7).

The overall effect of lockdown can be observed across India because there was a reduction in power demand in the industrial sector, construction work, and residential sector apart from other restricted activities. Coal-based power generation was reduced by 26% during the lockdown (EM power web, 2020). In east India slight increase in AQI during the middle of lockdown might be due to immigration of labours from megacities to their native place or resumption of mining and factories after the declaration of no containment zone in Jorapokhar, Jharkhand. The meteorological variables are also the major factors for AQI level, resulting in good air quality throughout India. Another major factor for reducing AQI is fewer emissions from both industrial and transportation sectors, which emit a considerable amount of particulate matter into the atmosphere, thus increasing AQI.

4. Conclusion

During the outbreak of the COVID-19 pandemic in India, the lockdown was implemented to control the spread of disease across the country. The lockdown has witnessed restricted human activities resulting in the air quality improvement due to inhibited emission of toxic air pollutants. The present study focused on the evaluation of change in air quality during lockdown based on statistical analysis of the AQI dataset and GIS mapping. The spatiotemporal evaluation of GIS maps has revealed the decline in AQI during the lockdown in 2020. It has also suggested the overall improvement of air quality across the country. Multivariate statistical approach of Principal Component Analysis (PCA) has extracted 4 principal components in terms of significant variability in AQI levels with respective dates. It has resulted in the dates of before and during lockdown as different principal components indicating a significant change in AQI.

The revival of ecosystems during lockdown has given an overview for adopting planning strategies to combat the devastating problem of air pollution. The decrease in air pollution induced by the regulated emission of important air pollutants has the potential to significantly reduce a wide range of health issues, including respiratory illnesses, cardiovascular disease, asthma, and premature mortality. These positive consequences of air pollution regulations may offer governments and authorities confidence that strict air quality laws and emission control measures may significantly improve environmental and human health.

Acknowledgements

The authors wish to express their sincere gratitude to the Centre for Research, Maitreyi College, University of Delhi for providing an opportunity and necessary facilities to accomplish the research work. The authors also duly acknowledge Professor Saumitra Mukherjee, Head of Remote Sensing Applications Laboratory, School of Environmental Sciences, Jawaharlal Nehru University, for necessary statistical and GIS work assistance.

Funding

The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.

Competing Interests

The authors have no relevant financial or non-financial interests to disclose.

Author Contributions

All authors contributed to the study conception and design. Pooja Baweja conceptualised and supervised the work. Haritma Chopra, Pinkey B. Gandhi and Sandhya Gupta have helped in writing and editing part of the manuscript. Material preparation and data collection were performed by Nisha Poddar and Sakshi Suman. GIS interpolation maps, statistical analysis and results-discussion were prepared by Vikas Rena. All authors read and approved the final manuscript.

Data Availability

The datasets generated during and analysed during the current study are available in the National Air Quality Index site of Central Pollution Control Board, India (CPCB, 2020) National Air Quality Index (cpcbccr.com), AQI Bulletin Archive (CPCB, 2020) CPCB Central Pollution Control Board, SAFAR India site (SAFAR, 2020) SAFAR - India (tropmet.res.in) and Chhattisgarh Environment Conservation Board (CECB, 2020) Chhattisgarh Environment Conservation Board, Raipur (C.G.) (enviscecb.org).

References