Introduction

The novel coronavirus disease (COVID-19), which is caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), was first reported in the Wuhan city of China on 1st December 2019 (WHO 2020). This virus is extremely infectious and quickly spreads to several other countries. Thus, WHO has officially declared this as a global pandemic on 11 March 2020 (WHO 2020; Cucinotta and Vanelli 2020). As of 9 July 2020, the coronavirus has infected more than 12 million people and over 550,159 deaths have been reported from 184 countries (WHO 2020; Johns Hopkins University 2020). Therefore, it has resulted in a global health crisis and a huge challenge for the world in our time (United Nations Development Programme, 2020). India is also facing similar challenges as this virus is spreading exponentially to all corners of the country with the total number has gone up to 767,296 confirmed cases and 21,129 deaths until 9th July 2020 (COVID19INDIA 2020); thereby, it has created an emergency situation in several states. Maharashtra is one of the worse affected states in India with a fatality rate of 4%, which is significantly higher than the rest of the countries with a confirmed total of 223,724 cases, including 123,192 recoveries and 1700 deaths as of 9th July 2020 (MoHFW 2020). The State and Central Government are taking several actions to manage this critical situation. After a self-quarantine curfew, called as ‘Janta Curfew’, the Government of India has imposed a complete lockdown from March 25 to April 14, 2020, for 21 days (phase 1) and subsequently this has been extended to phases 2, 3, and 4 to avoid further spread of the virus. This entailed self-isolation, a significant reduction of travel both on the road and air, and a halt to various other outdoor activities, including industrial, commercial, construction, etc. (MHA 2020). As a consequence, this lockdown has led to a very rare positive impact on the natural environment, especially a significant improvement in the air quality in various parts of the world. This could be a blessing in disguise (Muhmmad et al. 2020) as air pollution is one of the major life threats that results in numerous health diseases including cardiovascular and respiratory illness, asthma, bronchitis, chronic obstructive pulmonary disease, stress to lung and heart, and shortened life span (Gorai et al. 2016; Li et al. 2018; Pani et al. 2019, 2020a). The first lockdown impact on air quality has been reported for NO2 from Wuhan, China (NASA 2020). Thereafter, other studies have reported a drastic reduction of PM2.5, PM10, NOx, and CO during the lockdown period from several cities of China, New York, São Paulo, Barcelona, and others (Xu et al. 2020; Tobías et al. 2020; Dantas et al. 2020; Krecl 2020). In India, this has been studied in some major cities like Delhi, Mumbai, Hyderabad, Kolkata, etc (Jain and Sharma 2020; Bera et al. 2020; Sharma et al. 2020; Mahato et al. 2020). However, these studies examined the early trend only and it is still not clear what has happened after the lockdown is lifted. Also, it has been mentioned that change in air quality due to COVID-19 has been different in different regions/places and may not follow a similar trend (Masum and Pal 2020). Therefore, it is important to have a local study to evaluate the pattern of changes.

Furthermore, since the beginning of this outbreak, numerous researches have been going on to understand the factor affecting the transmission of SARC-CoV-2, in particular trying to find the link of host immunity, population density, and climate/environmental conditions (Pequeno et al. 2020; Wu et al. 2020a, b; Li et al. 2020). Air pollution is known to cause numerous respiratory diseases and it is the longer the exposure, the higher is the risk theory (Gorai et al. 2016); therefore, there has been considerable interest in understanding whether ambient air pollutants such as PM2.5, PM10, NOx, and CO are associated with the incidence of the COVID-19 (Zoran et al. 2020; Li et al. 2020). Some studies reported that poor air quality is linked with the severity of COVID-19 outcomes (Ogen 2020; Wu et al. 2020b; Zoran et al. 2020; Li et al. 2020). Furthermore, it has been documented that variation in temperature and humidity leads to influenza and other respiratory viruses such as severe acute respiratory syndrome (SARS), etc. (Tan et al. 2005). Considering this logic, several studies aimed to find the role of meteorological parameters like temperature, relative humidity, dew point, wind speed, and rainfall in the spread of COVID-19 (Sahoo et al. 2020, 2021; Pani et al. 2020b; Wu et al. 2020b; Ma et al. 2020). Other factors including demographic factors such as population density, household age profiles, and sex-ratios are also reported to be linked with this pandemic (Cruz et al. 2020; Hamidi et al. 2020; Babbitt et al. 2020). However, the conclusions of these studies are still controversial. Furthermore, this topic is still at an early stage and there is insufficient data at this moment. Moreover, the spread of COVID-19 can vary from one region to another depending upon the local weather and environmental conditions as well as government guidelines, etc. Therefore, more careful and vigorous studies are required for any region considering these factors for an accurate assessment of this epidemic.

Keeping this in view, the present study aims to evaluate the lockdown and unlock impact on air quality and explore the association between environmental conditions (including meteorological parameters and ambient air pollutants), demographic factors, and COVID-19 incidence/mortality in the state of Maharashtra, India.

Materials and methods

Study area

Maharashtra is the third most urbanized and highly developed state in India, located in the western part of the country (Fig. 1). The state has 36 districts and it is the most industrialised state in India. The total population of this state is 112 million with approx. 50% population urban and 46% less than 24 years of age (Maji et al. 2016). Mumbai is the capital of this state and is the highest populated city (with a population of 13 million) in India. The average rainfall ranges between 650 and 750 mm. The region has a tropical wet and dry climate. The rainy season lasts from June to September followed by winter from November to February during which the minimum temperature ranges between 9 and 10 °C. The summer ranges from March to June in which the temperature shoots up to 45 °C and humidity being approximately 70 to 85%.

Fig. 1
figure 1

Map showing the outline of the study area and the location of 22 air quality monitoring stations in Maharashtra, India

Full size image

Data collection

The daily average concentrations (all stations) of air pollutants such as PM2.5, PM10, CO, NO, NOx, NO2, SO2, ozone, toluene, benzene, and the basic meteorological parameters such as ambient temperature (AT), rainfall (RF), and wind speed (WS) have been considered to evaluate the deviation of air quality in pre-, during, and post-lockdown period from 1 January to 3 July 2020. Dew point temperature (DP) was calculated using the following equation: DP = T – (100–RH)/5), where 'T' is atmosperic temperature and 'RH' is relative humidity (Lawrence 2005). A total of 22 monitoring stations which are comprised of several major cities such as Mumbai, Pune, Thane, Navi Mumbai, Nagpur, Solapur, Chandrapur, Kalyan, and Aurangabad are considered to cover the entire state of Maharashtra (Fig. 1). The data related to these parameters were obtained from the Central Pollution Control Board (CPCB, India) online portal (https://app.cpcbccr.com/ccr/#/caaqm-dashboard-all/caaqm-landing). The daily averaged air quality index (AQI) data were retrieved from the World’s Air Pollution online platform (https://waqi.info/). The AQI was calculated following the method (Eq. 1) prescribed by CPCB (CPCB 2014; Sharma et al. 2020). This index uses the concentration of PM10, PM2.5, SO2, NO2, O3, NH3, and CO (up to 24-hourly averaging period), of which a minimum of three pollutants must be considered with at least one of being PM2.5 or PM10 (CPCB 2014).

$$ {AQI}_i=\frac{IN_{HI}-{IN}_{LO}}{B_{HI}-{B}_{LO}}\ x\ \left({C}_i-{B}_{LO}\right)+{IN}_{LO} $$
(1)

where, BHI and BLO as breakpoint concentrations greater and smaller to Ci; INHI Ci as the concentration of pollutant i; INHI and INLO are corresponding AQI values. The AQI is followed as good (0–50), satisfactory (51–100), moderate (101–200), poor (201–300), very poor (301–400), and severe (401–500) (CPCB 2014; Sharma et al. 2020).

Daily new cases of COVID-19 data (total infection, recovery, and death cases) were collected from the government of Maharashtra COVID-19 portal and from the website (i.e. https://api.covid19india.org/csv/) between 14 March 2020 and 3 July 2020. Other parameters related to this study are taken from the government’s official sources. The data representing as pre (PL1: 1 January to 31 January; PL2: 1 February to 29 February; PL3: 1 March to 24 March), during (LD1: 25 March to 14 April; LD2: 15 April to 3 May; LD3: 4 May to 17 May; LD4: 18 May to 31 May) and unlock periods (UL1: 1 June to 30 June; UL2: > 1 July).

Statistical and spatial analysis

The Kendall rank correlation, a non-parametric test, was used to estimate the ordinal association between variables such as air pollutants, weather parameters, and COVID-19 incidences. The p values of < 0.01 are considered statistically significant. Kendall correlation is preferred over Spearman’s correlation as the former is more robust and efficient because of a smaller gross error sensitivity (GES) (more robust) and a smaller asymptotic variance (AV) (more efficient). Kendall’s tau also produces smaller and more realistic values than Spearman’s rho. Thus, in our study, the Kendall’s tau correlation matrix was considered (Howell 2012). Normality of the dataset was checked by using a combination of EDA plot and Shapiro-Wilk’s (S-W) test. The Kruskal-Wallis test, a non-parametric version of the ANOVA, was performed on air quality and meteorological data. The null hypothesis for this test is that the mean concentrations of each variable between different periods are not significantly different at a level of significance p > 0.05. Furthermore based on city-wise data, the relationship of pre-lockdown air quality data (the maximum average value was taken between PL1-3) and the demographic variables such as population size, density, and literacy with COVID-19 data was established using correlation analysis based on symmetric pivot coordinates, which solves the issue of data collinearity. All statistical treatments were performed by using R statistical software version 3.5 (R Core Team 2018) and MS-Excel 2016. Spatial distribution analysis of air pollutants was carried out using Geographic Information System (GIS) environment (ArcGIS 10.6) under the World Geodetic System 1984 (WGS84) datum.

Results

The daily variation averaged over all stations of various air pollutants (PM2.5, PM10, NOx, NO, NO2, CO, SO2, ozone, toluene, benzene) and air quality index (AQI) during the different time periods, including pre-lockdown (PL1–3), lockdown (LD1–4), and unlock (UL1–2) from 1st January to 3rd July, are shown in Figs. 2 and 3. The number of incidents of each pollutant exceeding the National Ambient Air Quality Standards (NAAQS) defined by (CPCB) and percentage change in average pollutant levels during the different periods of LDs and ULs with respect to the immediate pre-lockdown (PL3) are given in Table 1; the station-wise percentage changes are given in Table 1SM. It has been observed that in most of the pollutants like PM2.5, PM10, NOx, NO2, CO, benzene, AQI show a sharp declining trend with up and down spikes (Fig. 2), and a significant reduction during the lockdown periods. For pollutants like PM2.5 and PM10 and AQI, this reduction was even greater during the ULs, but for some gaseous pollutants like CO, NO, and NOx, this trend is slightly increased in the ULs (Fig. 2). The Kruskal-Wallis test (Table 2) shows that most of the air quality parameters had p values < 0.05 pollutants, which indicates that these pollutants were significantly varied among the different phases of lockdown and unlock. When compared with the NAAQS limits, average concentrations of PM10 in PL1–3 phases were mostly exceeding its permissible limit, i.e. 100 μg/m3 (Table 1); however, these levels were drastically reduced well below its permissible limit in both LDs and ULs. The PM10 level (averaged of all stations) was reduced up to 51% in the LDs and further declined up to 80% in the ULs compared with PL3. Similarly, PM2.5 was dropped significantly up to 46% in LDs and up to 78% in ULs, and in both cases, it was noticed that the reduction in PM2.5 levels was well below it permissible limits, i.e. 60 μg/m3. Compared with the PLs, NOx (which consists of NO and NO2) is significantly declined up to 62% in LDs and a little lower up to 50% in the ULs. Similarly, NO was dropped significantly up to 63% in LDs but this trend was comparatively lower in ULs (up to 44%). Moreover, the NO2 levels were lesser than the permissible limit (80 μg/m3) even before the COVID-19 lockdown. The average concentration of CO was also significantly reduced in the LD phases compared with PDs; the lowest was in LD4, but the level was slightly increased in ULs in a similar pattern like NOx. The CO concentration dropped up to 54% and 35%, respectively, in the LD and ULs. Moreover, the CO levels never exceeded the permissible limit, i.e. 2.0 mg/m3 in the entire studied period. Among the gaseous pollutants, the reduction of ozone was much lower in LDs (up to 18%), but it has significantly dropped in the ULs (up to 66%), although irrespective of pre-lockdown and lockdown, ozone was mostly found to be within the standard limit (100 μg/m3). Similarly, the reduction of SO2 was lower in the LDs and ULs and was not prominent compared with other pollutants, but it was well below the permissible limit in all periods. Among organic pollutants, benzene was found to be exceeding the standard limit, i.e. 5 μg/m3 as per CPCB. It was also drastically reduced in the LDs as well as in the ULs in a similar way like toluene, benzene, and oxylene. AQI was an overall assessment of air quality based on the criteria pollutants. During the PLs, the average AQI was mostly between 100 and 200 with the category of moderate quality; however, this level was significantly dropped to below 100 during the lockdown and it has been reduced further to below 50 during the unlock phases.

Fig. 2
figure 2

Temporal variation of daily average level of PM2.5, PM10, NOx, NO, NO2, NH3, CO, SO2, ozone, toluene, and benzene from 1 Jan to 3 Jul 2020 in Maharashtra

Full size image
Fig. 3
figure 3

Temporal variation of daily average level of Air Quality Index (AQI), Ambient Temperature (AT), Temperature (Temp), dew point (DP), windspeed (WS), and rainfall (RF) in Maharashtra from 1st Jan to 3rd Jul 2020

Full size image
Table 1 (a) Number of incidents of daily average pollutants exceeding the NAAQS limit during the pre-lockdown (PL1–3), lockdown (LD1–4) and unlock (UL1–2) periods in Maharashtra from 1st January to 3rd July. (b) Percentage change in average pollutants and meterological parameters during the LD1-4 and UL1-2 phases with respect to the immediate pre-lockdown period (PL3)
Full size table
Table 2 Kruskal-Wallis test results of the chemical parameters among grouping variables (lockdown and unlock periods)
Full size table

The spatio-temporal variation of selective parameters in immediate pre-lockdown (PL3), lockdown (LD1,4), and unlock (UL2) periods is shown in Fig. 4 (other parameters can be found in Fig. 1SM). Also, the station-wise variations are given in Table 1SM to quantify the impact of lockdown over individual stations/cities. It can be noticed that most of the air pollutants were widely varied over the study area with more pollution levels in the western part than other regions. During the immediate pre-lockdown (PL3), the concentrations of PM2.5 and PM10 were higher in Mumbai (Airport (T2), Powai, Sion, Vile Parle, Kurla), Navi Mumbai (Mahape, Nerul), Kalyan, and Solapur, similar with AQI. All these stations witnessed a significant reduction of these pollutants during the lockdown and unlock phases. Similarly, NO2 was reduced in most of the stations during LDs and ULs, unlike CO, NOx, benzene, which show an increasing trend in some stations. SO2 shows a little variation between stations during PLs, LDs, and ULs, with the highest values recorded in the Kalyan region. Ozone was also less varied during the lockdown, and particularly shows an increasing pattern in some stations during LD1, but in stations like More Chowk Waluj in Aurangabad, it continued to increase in all the phases of LD and UL.

Fig. 4
figure 4figure 4

Spatial distribution of PM2.5 and PM10, NO2, NOx, CO, and SO2 in pre-lockdown (PL3), lockdown (LD1–4) and unlock (UL1–2) periods in Maharashtra.

Full size image

The daily averaged variation over all stations of meteorological parameters is shown in Fig. 3. The temperature shows a moderately increasing trend from 1 January to 3 July, similar with DP. Overall, WS and RF show a slight variation in the study period with abnormal fluctuations in April and May. The Kruskal-Wallis test (Table 2) shows a p value < 0.05 for most of the meterological parameters, although rainfall had a relatively low Z score as compared with others.

Daily COVID-19 incidences including confirmed, recovered and death cases are presented in Fig. 5. This shows a speedy growth of the COVID-19 cases for Maharashtra starting from 14 confirmed cases on March 14, 2020, which rose to 125 until the March 26 (PL), 2916 until 15 April and then spiked rapidly to 67,655 cumulative cases until 31 May (the end of LD4) and 192,990 cumulative cases until 3 July 2020 (beginning of UL1). The highest and lowest daily COVID-19 incidences in Maharashtra were 6364 cases on 1st July and 3 cases on 17 March, respectively, as of 3 July. At the same pace, the number of recoveries and death cases are also increasing. Until 1 July, the total number of death cases were 6014, and the highest (1409 cases) was on 15 June.

Fig. 5
figure 5

Daily confirmed, recovered and dealth cases of COVID-19 in Maharashtra from 14 Mar 2020 to 3 July 2020

Full size image

Table 3 presents the result of Kendall’s tau correlation analysis of air pollutants and meteorological variables and daily COVID-19 incidences. This has been calculated based on the data from 1 Jan 2020 to 3 July 2020. The Kendall’s correlation matrix showed that most of the ambient air pollutants particularly PM2.5, PM10, CO, and NOx were moderately negatively correlated with atmospheric temperature (r = − 0.32 to 0.57. Similarly, air pollutants are weakly negatively correlated with wind speed (r = − 0.02 to − 0.29), and weakly positively correlated with rainfall (r = 0.2 to 0.36). Among the air pollutants, PM2.5 and PM10 show a strong positive correlation with each other (r = 0.82) and they are mostly moderately positively correlated with other pollutants. Furthermore, we studied the correlation between environmental parameters and COVID-19 incidences (Table 2 and Fig. 5), which is calculated from 1 January 2020 to 3 July 2020 (Table 3). This shows that temperature (r = 0.58 to 0.62) is predominantly positively correlated with COVID-19 incidence and death cases. Dew point also shows a strong positive correlation with COVID-19 incidences (r = 0.71 to 0.73). Most of the air pollutants are significantly negatively correlated (r < − 0.67) with COVID-19 (Table 3). Rainfall and wind speed are weekly and negatively correlated with COVID-19 cases (Table 3). The relationship between daily COVID-19 confirmed cases and meteorological parameters is shown in Fig. 6. Furthermore, heat-plot (Fig. 7) shows the correlation between demographic factors, air pollutants (this is based on the maximum average concentration during the pre-lockdown phases—PL1–3) and COVID-19 incidence. This indicates that COVID-19 cases and deaths are more closely related to population density and PM2.5 and PM10.

Table 3 Nonlinear correlation (Kendall’s correlation matrix) between meteorological, air quality, and COVID-19 parameters from 1st Jan to 3rd July 2020 in Maharashtra
Full size table
Fig. 6
figure 6

Relationship between daily COVID-19 confirmed cases and wind speed (a), dew point (b), temperature (c), and rainfall (d) in Maharastra from 1 January to 3 July 2020

Full size image
Fig. 7
figure 7

Heat-plot showing the relation between different demographic and air quality parameters and COVID-19 incidences in Maharashtra. This is made based on city wise demographic, air quality, and COVID-19 data. For air quality data, maximum average concentration between the three pre-lockdown phases (PL1, PL2, and PL3: Jan to March 2020) was taken. COVID-19 data was taken as on 3 July 2020

Full size image

Discussion

Impact of lockdown and unlock phases and the meteorological factors on air quality

A sharp decline in most of the major air pollutants like PM2.5, PM10, NO2, and CO along with AQI in most of the cities during lockdown (Fig. 2, 3) clearly illustrates a rare benefit and a blessing of forced lockdown that shut down most of the anthropogenic activities such as industries, heavy traffic, road, and infrastructure construction activities, which are substantial contributors to both fractions of particulate matter and other air pollutants (MPCB 2019; Masum and Pal 2020). Furthermore, the greater reduction of these pollutants in major stations like Khadakpada (Kalyan), Kurla (Mumbai), Nerul (Navi Mumbai), Chhatrapati Shivaji Int. Airport (Mumbai) (Table 1SM), which displayed trends of higher PMs level among other regions during pre-lockdown, further evidenced that larger cities have a higher improvement in air quality and vehicular emission along with industrial establishments and a high level of urbanization are the major contributor of PMs. Similarly, a significant fall of NO2 levels in lockdown phases and a maximum decline in densely populated and busy traffic regions, like Chhatrapati Shivaji Intl. Airport (Mumbai), Mahape (Navi Mumbai), Sion (Mumbai), and Vasai West (Mumbai), indicate motor vehicle exhaust and industrial activities are the major sources of NOx. SO2 was the least impacted pollutant, which did not show a clear reduction trend in the LDs. In fact, this is increased in some stations during the lockdown as well as unlock periods. This is possibly due to some small/medium to heavy industries including the coal-based thermal power plants which were likely to be operating during these periods. Maximum SO2 content in the Kalyan region is due to the presence of industrial setups and having numerous manufacturing units (MPCB 2018, 2019). The major toxic organic pollutants such as benzene, toluene, xylene, and ethyl benzene were also significantly reduced in both lockdown and unlock phases. Benzene is one of the ingredients of petroleum products and it is mainly released to the atmosphere via petrol and diesel exhausts and petrol evaporation from vehicles as well as from industrial activities such as refinery and petrochemicals, etc. (CPCB 2012). Toluene and xylene are also present in the above-mentioned source along with benzene (CPCB 2012). Thus, shutting down of the transport (including road, rail, aviation) and industrial sectors is the major reason for the sharp decline in these pollutants during the lockdown phases. In India, a significant reduction of criteria air pollutants and a drastic improvement in AQI was also witnessed from major cities like Delhi, Mumbai, Bangalore, Kolkata, Chennai, Punjab, and Chandigarh (Jain and Sharma 2020; Sharma et al. 2020; Mahato et al. 2020; CPCB 2020; Sahoo et al. 2021). Particulate matters have significantly reduced all over India with a maximum decline in northern India, while the average decrease for PM10, PM2.5, NO2, and CO across India during lockdown was 40.5%, 35%, 27.9%, and 13.9%, respectively (Navinya et al. 2020). In another study, Ghosh et al. (2020) reported a significant drop of PM10 (64.6%), PM2.5 (60.9%), NO2 (76.8%), and VOC (69%) from different monitoring stations located across Kolkata. Several other studies worldwide from major cities of Brazil, China, Spain, Italy, Kazakhstan, Germany, France, etc. also reported a sharp decline in air pollutants like PM2.5, PM10, CO, NO2, and SO2 during COVID-19 strict quarantine periods (Xu et al. 2020; Tobías et al. 2020; Dantas et al. 2020; Otmani et al. 2020; Kerimray et al. 2020). ESA (2020) reported almost 20–30% reduction of NO2 in France, Spain, Italy, and Germany during the lockdown phase. Reduction of PMs was over 40% reported in urban areas in China, while it was dropped by 19–40% in the USA (Xu et al. 2020; Tobías et al. 2020; Berman and Ebisu 2020). The strong positive correlation between PM2.5, PM10, NO2, and CO suggests a common source possibly from industrial sources and automobiles combustion which are common to both fractions of PMs, NOx, and CO (Table 3). The poor relationship between SO2 and NO2 could be due to different sources particularly from diesel engines of buses and lorries and industrial emission. Ozone is a secondary pollutant, which is produced chemically by the photolysis of nitrogen compounds (NOx) and volatile organic compounds (VOCs) under sunlight (Fang et al. 2020). Despite the reduction in traffic pollution during the lockdown phases, the increasing level of O3 in the lower atmosphere mainly in lockdown phase 1 and its weak correlation with NO2 is possibly due to low consumption of O3 as a result of less emission of NO2, which is due to no industrial activities and vehicles were off the road during the lockdown (Sharma et al. 2020; Tobías et al. 2020). A similar findings of an increase of O3 in the lower atmosphere during the COVID-19 lockdown days also reported in other studies (Sahoo et al. 2021; Bera et al. 2020; Shi and Brasseur 2020). As the lockdown eases and the vehicles returned on roads road in the unlock phases, the pollution started to increase. This is evident from the increasing concentrations of CO, NO2, and NOx from the cleanest lockdown phase. However, PM10 and PM2.5 have been dropping continuously even during the unlock period, which is due to the reduction of construction and industrial activities as well as intermittent rainfall. The rain helped wash away the particulate matters, which further helped to decrease AQI well below 50 ‘good’ categories during the unlock periods. Overall, Maharasthra has witnessed nearly a 45-70% decline in air pollution during the lockdown and unlock phases. The contribution of PM2.5 and PM10 was maximum to the AQI variability, thus these parameters are crucial to maintaining the air quality level. Thus, in the coming years curbing vehicular and industrial emissions and temporary lockdown mechanism could be used as an essential measure to decrease the pollution (Bera et al. 2020). Thus, the lesson learnt in this lockdown due to COVID-19 could be thought-provoking for policymakers to be used as a marker or baseline level to develop strategies for better air quality or manage air pollutants in urban cities that are often overlooked directly or indirectly (Masum and Pal 2020).

Meteorological impacts such as a change in summer temperature, relative humidity, rainfall, and wind speed have an important role in controlling the air pollutants (Sharma et al. 2020; Navinya et al. 2020). The lower temperature in pre-lockdown from January to March 2020 could lower mixing height of air and cold conditions trap air pollutants and may be one of the causes of increasing concentrations of all the criteria pollutants during winters, while higher temperature during the lockdown phase 4 and unlock phases may have contributed in the dispersion of pollutants. This is clearly evident in the significant negative correlation between temperature and most of the air pollutants (Table 3). These findings indicate that the influence of temperature on gaseous pollutants is much more effective in summer than other seasons due to the higher temperature range. This is corroborated with the previous studies (Sharma et al. 2020), while relatively small changes in humidity and wind speed during pre-lockdown, and lockdown indicate little impact on air quality. Moreover, although rainfall shows a weak relationship with most of the air pollutants, scattered rainfall during the lockdown and unlock periods could help washing away particulate pollutants and result in further improvement in air quality. Ozone is also greatly influenced by weather conditions, in particular O3 concentration increases with an increase of ambient temperature and solar radiation (Ghosh et al. 2020), in this study we have not observed any significant relationship. Overall, it can be observed that the changes in the meteorological conditions play a role in influencing air quality; but, these changes are less pronounced to induce such drastic reductions of air pollution well below the permissible limit, and thus the sharp decline in pollution levels during lockdown is predominantly influenced by the shutting down of anthropogenic activities during the lockdown (Navinya et al. 2020; Dutheil et al. 2020). However, during the unlock period an additional factor such as intermittent rainfall could lead to the removal of airborne particulate matters, as a result the AQI was further improved.

Correlation between weather conditions and COVID-19 infections

Meteorological parameters such as temperature and relative humidity are reported to be important factors in influencing the transmission of respiratory-borne infectious diseases such as SARS and influenza (Tan et al. 2005; Chan et al. 2011; Vandini et al. 2013; Ma et al. 2020; Davis et al. 2016). These viruses are often more active when the temperatures drop in cold weather and reduce their activity when temperature increases considering the fact that higher temperature damages the lipid layer of the virus (Chan et al. 2011). Similarly, it has been reported that higher RH will attenuate viral activity, and when the humidity is lower, the infected aerosol particles can survive for longer in the air, thereby greater chances of infecting other people (Ma et al. 2020; Ward et al. 2020; Avhad et al. 2020). However, the use of daily (24 hr) average RH is problematic for outdoor, as the daily averaged RH as an independent variable for an analysis of virus spread (Davis et al. 2016). In such scenario, David et al. (2016) sugegsted to consider dew point (DP) instead of RH. Dew point is the temperature at which air must be cooled to become saturated without changing the pressure and is highly associated with the human comfort. Higher the DP, more discomfort outside because the air’s humidity slows the evaporation of perspiration from human body, which prevents the nature’s way of cooling (Pani et al. 2020b). In our study, average DP values varies from 18.23 to 33.25 oC. This level is quite uncomfortable and it indicates heat stress issues for human outdoor. However, despite being this heat stress issues, the positive correlation between daily COVID-19 cases and DP (Table 3; Fig. 6) indicates that increase of dew point can not weaken the tranmission of this vírus. Furtheremore, it has been expected that COVID-19 cases will slow down in summer when the weather becomes too hot and warmer and will rise in winter. In the line of this evidence, some studies reported that temperature is as an environmental driver of the COVID-19 outbreak (Shi et al. 2020; Wang et al. 2020; Sajadi et al. 2020; Wu et al. 2020b; Ma et al. 2020). These studies reported that the temperature is negatively related to the daily new cases of COVID-19. Qi et al. (2020) also find that every 1 oC rise of daily average temperature can reduce 36 to 57% daily COVID-19 new cases. Gupta et al. (2020a) also reported that the majority of the COVID-19 cases were found in lower range of temperature 4 to 11 °C. This is very similar with Wu et al. (2020b) who find that 1 oC rise of temperature can reduce 3.08% and 1.19% daily COVID-19 and daily new deaths, respectively, while a 1% increase in relative humidity can decrease 0.85% and 0.51% daily new cases and daily new deaths, respectively. In another study Holtmann et al. (2020) also reported that a rapid spread of COVID-19 is linked with lower ambient temperature. However, our results didn’t support this evidence as we observed the COVID-19 pandemic is significantly positively correlated with temperatue (Table 3). A similar positive association with temperature and humidity was reported in other studies in Singapore (Pani et al. 2020b), Punjab, India (Sahoo et al. 2021), Jakarta, Indonesia (Tosepu et al. 2020), and Oslo, Norway (Menebo 2020). In fact, this virus even comfortably survived at high temperatures as recently this pandemic is sharply rising in summer climate in India and in other tropical and sub-tropical countries like Pakistan, Singapore, Brazil. Also, in our study, during the unlock period when the COVID-19 cases were sharply rising, the relationship with temperature was very week (Fig. 6). Thus, our results demonstate that higher temperatures do not decrease the transmission of the COVID-19 significantly. Xie and Zhu (2020) in a study from 122 cities in China also revealed that no evidence of supporting that COVID-19 infections might decline when the weather becomes warmer. Brassey et al. (2020) also stated that weather alone will not be effective to slow the transmission of COVID-19 cases. Other studies also reported that no evidence for COVID-19 transmission with temperature (Gupta et al. 2020b; Sahoo et al. 2020; Yao et al. 2020; Jamil et al. 2020). Furthermore, although COVID-19 cases are weakly correlated with wind speed, the higher cases at lower WS (Fig. 6) indicate the unventilated air may contribute to the transmission. Previous studies also indicated that windy weather could also help predict the growth of the pandemic (Coşkun et al. 2020; Thomas 2020). In addition, the region receives low precipitation (less than 400 mm) could be one of the reasons to be more vulnerable to the virus (Thomas 2020), but in this study, we have not found any significant relationship with rainfall (Table 3, Fig. 6).

Relationship between demographic factors, air pollutions, and COVID-19 incidences

In spite of the weather, population density is one of the major factors that may facilitate the faster spread of COVID-19 cases as this virus spreads through close proximity between individuals (Babbitt et al. 2020; Dalziel et al. 2018). Figure 7 shows COVID-19 daily new cases are strongly positively correlated with population size and population density. In Maharashtra, the cities having higher population density are Mumbai (20980) followed by Thane (956), Pune (603), Nagpur (470), etc, which reported the total COVID-19 cases 77,658, 37,629, 22,327, 1468, respectively, as of 1st July 2020. Mumbai has the highest population density and has the highest COVID-19 cases. Bherwani et al. (2020) also reported that the western region of Maharashtra which has a high population has recorded high COVID-19 confirmed cases, while the south region which has low population has evidently reported low infected cases. A similar picture was reported from Algeria (Kadi and Khelfaoui 2020), Turkey (Şahin 2020; Coşkun et al. 2020), Wuhan city of China (Pansini and Fornacca 2020), the US state and some Europian countries (Babbitt et al. 2020) where their population density creates the potential for higher person-to-person contagion rate, thus the higher rate of spread of this virus. Thus, this indicates that in densely populated cities, proximity living results in a lack of social distancing, which ultimately triggers the outbreak of this pandemic despite higher humidity and temperature which is unfavourable for the spreading of SARS-CoV-2 (Avhad et al. 2020; Coşkun et al. 2020). Similary, Gupta et al. (2020a) reported that past exposures to a high level of PM2.5 over a long period are more prone to COVID-19 mortality (p < 0.05).

Furthermore, cities with high population density and size demand high energy, a large number of industries, and a huge number of vehicles. As a consequence, those cities will face most environmental challenges, especially air pollution (Maji et al. 2016). Continuous exposure to dirty air is known to increase unfavourable health impacts such as acute respiratory distress syndrome, which may have compromised their immune system and may be more susceptible to increase death rates from the disease (Xing et al. 2016; Kowalski and Konior 2020). Furthermore, cities experiencing serious air pollutions are linked to the faster spread of the COVID-19 as PMs can act as a carrier to transport the virus through the air in the form of aerosol (Comunian et al. 2020; Kowalski and Konior 2020). With this premise, we first studied the relationship of COVID-19 incidences with the air quality data from 1 Jan 2020 to 3 July 2020, which shows that the air pollutants have a strong negative correlation with COVID-19 incidences (Table 3). This relationship could be due to the impact of lockdown that drastically reduced most of the pollutants during this period; thus, they have little role in spreading COVID-19 cases and other factors are more relevant for the spread of this virus. Secondly, when we evaluated the impact of past exposure to air pollutants (PM2.5, PM10, CO, and SO2) based on the pre-lockdown concentration of the respective cities, it is found that PM2.5 and PM10 are better correlated with the COVID-19 pandemic along with population density (Fig. 7), providing the fact that higher particulate matters in densely populated cities can lead to the spread of the virus as well as aggravate the chances of COVID-19 mortality. A similar association of PM2.5 and population density with COVID-19 infection and mortality is highlighted by Pansini and Fornacca (2020) based on the data from eight countries such as China, Iran, Italy, Spain, France, Germany, the UK, and the USA. In another study, Wu et al. (2020a) in a nationwide study from the USA mentioned that long-term average exposure to PM2.5 is a greater chance of increasing the risk of COVID-19 death. Gupta et al. (2020b) also reported the link between COVID-19 cases and past exposes to a high level of PM2.5. Other studies also reported a significant positive correlation between the COVID-19 infections and air quality variables in China, the USA, Italy (Pansini and Fornacca 2020), and England (Travaglio et al. 2020). Zhu et al. (2020) reported a positive association between PM2.5, PM10, CO, NO2, O3, and COVID-19 confirmed cases. In Northern Italy, Setti et al. (2020) reported that PM10 pollution could be related to the spread of COVID-19. A similar observation was reported from Kuala Lumpur, Malaysia (Suhaimi et al. 2020). Thus, these studies are providing preliminary evidence that highly polluted cities should be more vigilant and take the necessary steps to increase their immune system to prevent COVID-19 mortality. However, further in-depth research is needed to elucidate the underlying mechanism and validate this finding.

Conclusion

The sharp decline in levels of PM2.5, PM10, NO2, and CO during the COVID-19 lockdown in Maharashtra is due to strict restrictions on transportation, industries, and construction. Thus, this pandemic could be considered as a blessing in disguise. AQI remained ‘satisfactory’ during the lockdown phases and reduced further even post-lockdown due to the change in local weather patterns such as rainfall. Ground-level ozone and SO2 were the least affected pollutants during this lockdown. A slight increase in the concentrations of gaseous pollutants like CO, NO, and SO2 was again occured as the lockdown was relaxed. The AQI was continued to improve significantly during the unlock periods as a results of intermittent showers combined with the lockdown conditions. The relationship between meterological factors and COVID-19 incidences indicates that higher temperature and dew point cannot suppress the transmission of COVID-19. While, population density was found to be an important factor for the rapid spread of this virus in Maharashtra. Furthermore, past exposure to high levels of PM2.5 and PM10 may lead to a higher risk of COVID-19 mortality. These results show how multiple variables contribute to the spread of viruses and should be used to inform individuals about the proper use of safety measures when visiting more populated/industrial areas. The most important outcome of this study is to highlight to policymakers and the private sectors that purposeful actions to limit atmospheric pollution and even to attenuate population density in cities can have important implications for human survival. If we can alter our behaviour during a pandemic, and if that change in behaviour has such a positive result, then why cannot we formulate policy towards long-term survival and a better quality of life—let us use our knowledge wisely.