The concentration data of PM_2.5 and PM₁₀ as measured in six cities, namely, Jaipur, Jodhpur, Kota, Udaipur, Ajmer, and Alwar, in the state of Rajasthan were obtained from the Central Pollution Control Board (CPCB) for the years 2019 and 2020. The PM_2.5 and PM₁₀ data in these cities located in the northwest part of India were analyzed to investigate the pollution levels during the long-term COVID-19 lockdown period. The locations of the study sites in the maps of India and Rajasthan are shown in Figure 1.

The data studied include during the pre-lockdown period (PRELD) (January 1 to March 4, 2020), partial lockdown (PLDN) (March 5 to 24, 2020), COVID-19 lockdown (LDN) (March 25 to May 31, 2020), and unlocking (ULC) period (June 1 to August 31, 2020). The average matching periods of the year 2019 were considered as the references to study the change in air quality during the COVID-19 lockdown period. We have used the National Centre for Environmental Prediction reanalysis datasets to derive the wind streamlines and temperature at a spatial resolution of 2.5° × 2.5° during the entire study period for 2019 and 2020, as shown in Figure 2. The precipitation rate (mm day⁻¹) data from the CPC Merged Analysis of Precipitation (CMAP), at a spatial resolution of 2.5° × 2.5°, was used. The Moderate Resolution Imaging Spectroradiometer (MODIS) active fire count maps over India during the entire study period for 2019 and 2020 are plotted in Figure 3. Rajasthan is the largest state of India, located in the north-western region with a population of ~6.8 million (http://census2011.co.in). However, being located near most cities, the state has various types of industrial units, including those of marble, steel, cement, handicrafts, textiles, chemicals, IT, and thermal plants (Report 2018). The primary sources of air pollutants are vehicular, industrial, agriculture residual burning, municipal solid waste treatment, diesel generator, biofuel and LPG (residential cooking), and construction activities. The common sources of PM_2.5 and PM₁₀ include biofuel and fossil fuel combustion, industrial processes, and biomass burning. However, for PM₁₀, wind-blown dust, road dust suspension, building construction, lime kilns, and slab polishing are important non-combustion sources (Yadav et al., 2017). Daily mean temperature ranges of 10–27, 32–45, and 35–40°C are recorded during winter (January–February), pre-monsoon (March–May), and monsoon (June–September) seasons, respectively. The strong hot winds called “Loo” blow from the west during pre-monsoon, while relatively low winds blow from the north and northeast during the winter (Yadav et al., 2014a). Rainfall values as high as 819 mm in the eastern part and as low as 170 mm in the western part of the state were recorded (Maanju and Saha, 2013). In the monsoon season, the inter-tropical convergence zone moves northward across India, and strong southwest (SW) winds prevail over the study region. Therefore, the measurements in this season are influenced by the transport of cleaner air masses from the Arabian Sea (see Supplementary Figure S1).

The time series analysis of PM_2.5 and PM₁₀ over a period of 8 months (January 1–August 31, 2020) reveals interesting trends. The daily means of PM_2.5 and PM₁₀ concentrations measured in Jaipur, Jodhpur, Kota, Udaipur, Ajmer, and Alwar are shown in Figure 4. The daily mean time series concentrations of PM_2.5 and PM₁₀ for all cities are divided into the four different periods of PRELD (January 1–March 4), PLDN (March 5–24), COVID-19 LDN (March 25–May 31), and ULC (June 1–August 31). During the business-as-usual period, many studies have reported a clear seasonality in the daily mean concentrations of PM_2.5 and PM₁₀, with the highest in the dry period (winter to pre-monsoon) and the lowest in the monsoon season, in different cities of India (Beig et al., 2007; Yadav et al., 2017; Hama et al., 2020). However, the findings become effectively more crucial in 2020 to researchers and policymakers because of the COVID-19-related consecutive LDN and ULC situations in which anthropogenic activities were restricted. The daily mean concentrations of PM_2.5 were in the ranges of 7–81, 18–156, 12–91, 14–93, 10–114, and 15–63 μg m⁻³, while those of PM₁₀ varied in the ranges of 20–256, 39–282, 23–185, 26–202, 15–266, and 30–121 μg m⁻³, respectively, in Jaipur, Jodhpur, Kota, Udaipur, Ajmer, and Alwar during the entire study period. The mean concentrations of PM_2.5 were 51 ± 10, 78 ± 21, 45 ± 13, 49 ± 12, 47.3 ± 13, and 47.2 ± 8 μg m⁻³ in the PRELD period, which reduced to 34 ± 09, 58 ± 16, 29 ± 3, 29 ± 2, 32 ± 11, and 24 ± 4.4 μg m⁻³ during the LDN period in Jaipur, Jodhpur, Kota, Udaipur, Ajmer, and Alwar, respectively. On the other hand, the mean concentrations of PM₁₀ were 115 ± 23, 162 ± 41, 97 ± 25, 104 ± 18, 100 ± 25, and 85 ± 9 μg m⁻³ in the PRELD period, which decreased by 90 ± 27, 116 ± 29, 67 ± 8, 65 ± 14, 80 ± 35, and 53 ± 10 μg m⁻³ during the LDN periods in Jaipur, Jodhpur, Kota, Udaipur, Ajmer, and Alwar, respectively. The daily mean concentrations of PM_2.5 and PM₁₀ during PRELD, PLDN, various phases of LDN, and UCL situations are shown in Table 2. The National Ambient Air Quality Standards (NAAQS) recommend the daily permissible PM_2.5 of 60 μg m⁻³ and PM₁₀ of 100 μg m⁻³ for India (MoEFCC, 2015). About 3, 33, and 6% of PM_2.5 daily data exceeded the permissible limit of NAAQS during the COVID-19 LDN period in Jaipur, Jodhpur, and Ajmer, while 35, 46, 4, 7, and 19% of PM₁₀ exceeded in Jaipur, Jodhpur, Kota, Udaipur, and Ajmer, respectively. It is suggested that the concentrations of PM_2.5 significantly decreased in all cities that met the NAAQS limits. However, in a few cities, like Jodhpur, Jaipur, and Ajmer, the PM₁₀ levels did not fall below the standard limit during most of the COVID-19 LDN period due to non-combustion-related sources like wind-blown dust. The highest reduction by 47% in PM_2.5 was observed in Alwar, followed by Udaipur (40%), Kota (36%), Jaipur (32%), and Ajmer (32%), while Jodhpur showed the lowest reduction of about 25% during the COVID-19 LDN, respectively. PM₁₀ was reduced by 37% in Alwar and 36% in Udaipur. However, the lowest reductions of 19–30% were estimated for Jaipur, Jodhpur, Kota, and Ajmer cities (Figure 5). Additionally, we have compared the PM_2.5 and PM₁₀ concentrations measured in 2020 with those during the previous year (2019) for the COVID-19 LDN and ULC periods (see Figure 6). The differences are, more or less, the same during PLDN, but significantly lower concentrations of both pollutants were estimated during the COVID-19 LDN and ULC periods. Wilcoxon test has been performed, and P-value < 0.05 was found, which was considered statistically significant in the LDN periods. The reductions in PM_2.5 and PM₁₀ varied in the ranges of 16–50 and 22–47% during the COVID-19 LDN period with respect to the same period of the year 2019 (without meteorological normalization), as shown in Figure 7. The reductions of 6–46% in PM_2.5 and 28–49% in PM₁₀ were estimated during the overall ULC period.

Overall, the declining trends of PM_2.5 and PM₁₀ observed in all cities during the COVID-19 LDN period are attributed to the reduction in combustion-related emissions (NASA, 2020; Navinya et al., 2020). Conversely, to some extent, elevated levels of PM_2.5 and PM₁₀ were observed during the LDN3 and LDN4 periods (see Figure 4). During all COVID-19 LDN phases, the restrictions were diverse and resulted in variations of PM_2.5 and PM₁₀ in each LDN phase—for example, major anthropogenic and industrial activities were stopped, except for emergency services like the health sector (Yadav et al., 2020). Later, some relaxations with limitations were allowed in the subsequent LDN periods (MHA, 2020). Sahu et al. (2011) have reported that the emissions from transport and industrial sectors have significant contributions to the ambient concentrations of both PM_2.5 and PM₁₀. In addition to this, wind-blown dust is a more substantial source of coarser particles (PM₁₀) in the northern and western regions of India (Sahu et al., 2011; Yadav et al., 2014b). During the ULC period, only Alwar City exhibited higher values of PM_2.5 and PM₁₀ compared with the overall LDN phases in the same year; this could have happened due to limited local rainfall events. Interestingly, the highest plummet can be attributed to rainfall in the ULC period (monsoon) compared to 2019. In the monsoon period, episodes of rainfall can wash out the dust particles (PM₁₀) much faster than the finer particles (PM_2.5), unless there is frequent rainfall for many days (Yadav et al., 2014c). Figure 8 shows the ratios of PM_2.5/PM₁₀ during the study period in the year 2020 in six cities that were used to investigate aerosol combustion and non-combustion sources. The ratio of PM_2.5/PM₁₀ was observed to be <0.50 μg μg⁻¹ in all cities due to higher temperatures during the COVID-19 LDN period (pre-monsoon months), leading to dry and dusty conditions (Chan and Yao, 2008). The emissions from wind-blown dust are highly significant, mainly contributing to PM₁₀, and are probably the reason for the high levels of coarser particles. Interestingly, about 0.40 μg μg⁻¹ ratio of PM_2.5/PM₁₀ was observed during all phases of COVID-19 LDN in Jaipur City (nearby Delhi) due to the impact of dust storm. In the pre-monsoon season, dust storms originate from Arab countries, and the Thar Desert can influence most regions, like Delhi, Rajasthan, etc., of India (Anand et al., 2019), thus rapidly increasing the PM₁₀ levels. Overall, dust pollution is prominent in particle loads over the Rajasthan state. In addition to this, we have incorporated the plots of O₃/NO₂ and SO₂/NO₂ ratios during the study period in Figure 8. In the LDN periods (pre-monsoon months), the high O₃/NO₂ ratios were due to the transport of photochemical pollutants from higher altitudes. The efficient vertical mixing with free tropospheric air increases O₃ in the pre-monsoon months. The enhanced SO₂/NO₂ indicates that the predominant source of SO₂ is coal-based thermal power plants. In other words, the site receives reasonably aged SO₂ from distant regions, resulting in high ratios (Mor et al., 2021).

In addition to the local emissions, the variation in ambient air pollutants can also be influenced by the local meteorology, transport, transformation, and scavenging by the precipitation (Sahu et al., 2016a,b, 2020; Yadav et al., 2019a,b). Therefore, the mean wind streamlines along with temperature and precipitation at 925 mb pressure level are used in the study to investigate any unusual changes in meteorological and transport processes during the COVID-19 LDN period. The comparison of meteorological parameters during 2020 and 2019 for the PRELD, COVID-19 LDN, and UCL periods is shown in Figure 2. The trends in wind speed, temperature, and precipitation indicate the different features during the entire study period due to the change of seasons from winter to pre-monsoon and then monsoon (Yadav et al., 2014a; Sahu et al., 2017). Hence, the trends of PM_2.5 and PM₁₀ may also include a factor related to the change of meteorological conditions in the same year in comparison with the PRELD period. In the pre-monsoon season (LDN period), high temperatures, high wind speeds, and more profound boundary layer heights may lead to a higher dispersion of pollutants and lower concentrations (Sahu et al., 2020). However, the difference between 2019 and 2020 does not show significant fluctuations in wind speed, temperature, and precipitation in the western region of India during the PRELD to ULC periods. The mean surface temperatures and rainfall were in ranges of 30–32°C and 0–1.5 mm day⁻¹ during the COVID-19 LDN period in the years 2019 and 2020 over the study region. The strong westerly wind prevails over the area during both years in the COVID-19 LDN period. In general, all meteorological parameters were, more or less, similar during the study period, especially in the COVID-19 lockdown period in 2019 and 2020. Several studies have reported that the meteorological parameters were identical in the previous years and 2020 during the COVID-19 LDN over the Indian region (Navinya et al., 2020; Sharma et al., 2020; Singh et al., 2020; Yadav et al., 2020). Additionally, in the pre-monsoon season, the local emissions become lesser, but outside contributions become more prominent due to the increased transport of regional air masses. Furthermore, ~7% open agricultural burning in this season is mainly due to the burning of residues from wheat crops. Figure 3 shows MODIS-active fire maps over India during the entire study period for the years 2019 and 2020. In addition to this, to some extent, the SW winds also influenced the region after passing through populated areas of central India; they may get polluted during the pre-monsoon-to-monsoon transition period (Sahu et al., 2020; Maji et al., 2021; Yadav et al., 2021). The maps clearly show that the pre-monsoon fire events in 2020 during COVID-19 LDN have somewhat decreased to a shallow level compared to those in 2019, especially in western, northern, and central India (Kant et al., bib2020), as biomass burning is a significant aerosol source. Overall, the reductions in fire events along with local emissions have also been deficient. Additionally, we have normalized the mass concentrations of PM_2.5 and PM₁₀ with meteorology to investigate the actual change in emissions due to COVID-19 LDN. In the method, ventilation coefficient, a product of wind speed (WS) and boundary layer height (BLH), is used to see the role of weather dynamics for all cities of Rajasthan. The WS (above 10 m) and BLH datasets used in the study are from Copernicus Emergency Management Service (Climate Data Store) (https://cds.climate.copernicus.eu/cdsapp#!/search?type=dataset). The extensive details of normalizing the meteorological effects on ambient pollutants are reported elsewhere (Dai et al., 2020; Falocchi et al., 2021; Mishra et al., 2021). The reductions in PM_2.5 and PM₁₀ (with meteorology-normalized data) were in the ranges of 20–55 and 22–49% during the COVID-19 LDN period compared to the same period of 2019, as shown in Figure 7. Enhancement (differences ~10%) was observed in the reduction of PM_2.5 and PM₁₀ (in meteorology-normalized data) compared to without meteorology-normalized data during the LDN period in all cities.

Air quality index is a tool for people to understand the air quality status in an easy way. This tool transforms the complex data of various pollutants into a single number (index value), classification, and color (Beig et al., 2010). AQI was calculated for all cities to identify the overall improvement in air quality, and the details of AQI are available elsewhere (CPCB, 2014). There are six AQI categories: good + satisfactory (0–100), moderately polluted (101–200), poor (201–200), very poor (301–400), and severe (401–500). We have converted the PM_2.5, PM₁₀, and SO₂ concentration values into AQI using typical values, as the minimum of three parameters was very much needed for the calculation. Therefore, only SO₂ concentration was used for the AQI analysis. Each category is decided based on the ambient concentration values of air pollutants and their likely health impacts known as health breakpoints. The AQ sub-index and breakpoint concentration of different pollutants are provided elsewhere (CPCB, 2014). Figure 9 indicates the AQI variation during the entire study period over six cities. Overall, the AQI was satisfactory to moderately polluted categories during the whole study period (PRELD to ULC). The AQI varied in the ranges of 85–161 during PRELD in all cities, with the highest in Jaipur, Jodhpur, and Udaipur cities, during which the traffic was regular. Later, the AQI declined during various LDN periods compared to during PRELD in all cities, except Jaipur, Jodhpur, and Ajmer. More or less, in all cities, increasing values were seen during the subsequent LDN periods but with the highest of 117 in Jaipur, 174 in Jodhpur, and 121 in Ajmer. Overall, in LDN1 to LDN3, all cities were in satisfactory ranges, but Jaipur, Jodhpur, and Ajmer showed moderate air quality conditions in LDN4. Interestingly, Jodhpur City gets good air quality during the third phase of ULC since the LDN periods. It suggests that more contributions to coarser particles (PM₁₀) are wind-blown dust and dust storms and, of course, anthropogenic activities, which have been allowed with limitations in the subsequent LDNs.