A Spatio-temporal Study of Changes in Air Quality from Pre-COVID Era to Post-COVID Era in Chicago, USA

Received: January 31, 2022
Revised: April 11, 2022
Accepted: May 24, 2022

 Copyright The Author(s). This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are cited.

Cite this article:

Khan, S., Dahu, B.M., Scott, G.J. (2022). A Spatio-temporal Study of Changes in Air Quality from Pre-COVID Era to Post-COVID Era in Chicago, USA. Aerosol Air Qual. Res. 22, 220053. https://doi.org/10.4209/aaqr.220053

HIGHLIGHTS

• PM_2.5, NO₂, and CO have been reduced during COVID-19 stay-at-home in 2020.
• Air quality improvement during COVID-19 was transitory.
• On average, all air pollutants increased in the post-COVID era except CO.

Keywords: COVID-19, Air pollution, Urban air quality, NO2, Particulate matter

1 INTRODUCTION

The unprecedented coronavirus disease 2019 (COVID-19) pandemic has drastically changed the world’s human life and environment. The pandemic has affected more than 351 million people and killed more than 5.6 million lives worldwide as of January 2022 (CRC, 2022). It has also caused enormous economic, public health, and social damage. The World Health Organization (WHO) announced COVID-19 as a pandemic on March 11, 2020 (Archer et al., 2020). The pandemic has created many challenges for the United States (US) government to balance the economy and public safety. As a result, the USA declared a national emergency on March 13, 2020. Most of the major cities of the USA went under full or partial lockdown. Thus, the human movement was limited at that time. It also caused a reduction in road traffic in that period (Hudda et al., 2020; Parker et al., 2020; Tokey, 2021).

While the world population were under lockdown and experiencing health risk, one of the rare benefits that the world has experienced is the improvement in air quality globally due to societal efforts and policies to mitigate COVID-19 by different measures, including the prohibition on mass gathering, social distancing, shutting-down of non-essential industries and the restriction of public transportation (Liu et al., 2021). As a result, the COVID-19 restrictions have reduced the emission of the primary air pollutants worldwide in general and particularly in the United States, due to the decrease in industrial activities and transportation (Taquechel et al., 2020). For example, the US National Oceanic and Atmospheric Administration stated an almost 30% decrease in nitrogen dioxides (NO₂) emissions in the urban northeast during April 2020 (Zhang et al., 2021).

In addition, there was a massive reduction in the volatile organic compound concentrations (Zhang et al., 2021).

Air pollution has a detrimental effect on human health, the climate, and sustainable development (Guo et al., 2021). Research studies have proven that exposure to the fine particulate matter (PM), especially PM_2.5 (particulate matter with aerodynamic diameters below 2.5 µm), may cause significant harmful health effects, including cardiovascular, respiratory, diabetes, kidney disease mortality, and morbidity (Bowe et al., 2021; Chen et al., 2021) that lead to millions of premature deaths across the world every year (Karaer et al., 2020). The WHO estimates that 4.2 million people die each year due to exposure to air pollution (Archer et al., 2020; WHO, 2020). Additionally, ecological studies indicate that living in areas with high levels of ambient fine particulate matter increases the risk of adverse COVID-19 outcomes (Bowe et al., 2021). Though COVID-19 restrictions have improved air quality globally by reducing the emission of air pollutants, this effect might be transitory. Though coronavirus continues to affect human beings badly, the US has achieved a significant improvement in controlling COVID-19 by providing vaccines and other mitigatory measures (Diesel et al., 2021). Thus, almost everything went back to normal life. This return might cause air quality to decline and match the pre-COVID-19 era. As a result, the evaluation of air quality in the post-COVID era is necessary for policymakers.

Many studies attempted to assess the change in the air quality of the US during the pandemic. Though some studies reported a decline in air pollutants and significant improvement in air quality (Berman and Ebisu, 2020; Chen et al., 2020; Liu et al., 2021; Parker et al., 2020), some studies also reported no significant improvement in air quality (Adhikari and Yin, 2020; Archer et al., 2020; Pata, 2020; Razzaq et al., 2020), even increase of air pollutant in some cases (Bekbulat et al., 2021; Perera et al., 2021; Shakoor et al., 2020). In addition, the range of change in air quality in different studies is also very wide. While analysis at the finer spatial level can give more insight into this issue, few studies conducted spatio-temporal analysis over the whole US or at the state level. None of the studies explored microclimate within a finer spatial extent, i.e., within city level (community area within a city). Moreover, those few spatio-temporal studies used larger spatial extents with low-resolution remote sensing images (Bar et al., 2021; Kerr et al., 2021; Naeger and Murphy, 2020). Again, to the best of our knowledge, no studies in the US attempted to examine the air quality using all six parameters of air quality by the U.S. Environmental Protection Agency (U.S. EPA) (Tables 1 and 2).

This study conducts a spatio-temporal analysis of the change of air quality, i.e., six parameters of air quality of Chicago city in 2020, and compares the spatio-temporal pattern in previous (2019) and next (2021) years. The aim was to understand the change in air quality during and after the COVID-19 era. We hypothesized that the improvement of air quality during the COVID-19 pandemic corresponding with reduced human activity was followed by the decline of air quality again due to the normalization of human activity in 2021. In brief, the novelties of this study are as follows:

• Spatio-temporal study in finer spatial level, i.e., within city level (community area level) in Chicago, Illinois using in-situ ground sensor data.
  
  
• Assessing the air quality change during (2020) and after the COVID-19 era (2021).
  
  
• Considering all six air quality parameters.
  
  

The remainder of this paper is organized as follows: a review of the previous related studies is provided in Section 2; the study area, datasets, and statistical methods are introduced in Section 3; the analysis and results are discussed in Section 4, and conclusions are given in Section 5.

 
2 REVIEW OF PREVIOUS RELATED STUDIES

There were several studies conducted around the globe in 2020 to evaluate and measure the effectiveness of the COVID-19 lockdown in different countries, cities, and towns on urban air quality. Some of these findings primarily described changes in the concentrations of particulate matters such as PM_2.5, and PM₁₀ (PM with diameters below 10 µm), whereas other studies focused on gaseous components like NO₂, CO, SO₂, and ozone (O₃) in several countries around the globe (Adam et al., 2021). Previous studies have shown a decrease in the concentrations of CO, SO₂, and the oxides of nitrogen (NO_x) worldwide (Adam et al., 2021).

Specifically, a huge decrease in NO₂ (approximately 70%) was founded in India and Spain (Sicard et al., 2020). Also, such huge decreases were described for SO₂ in Singapore (approximately 52%) and India (approximately 62%) as well as for CO in Brazil (> 30% and up to 100%) and India (approximately 70%) (Gautam, 2020; Resmi et al., 2020). The concentrations of PM₁₀, PM_2.5, NO₂, CO, and SO₂ reduced by 19%, 23%, 29%, 54%, 6%, and 52%, respectively in Singapore (Li and Tartarini, 2020). On the other hand, ozone increased by 18% during the lockdown in 2020 (April–May) compared to the concentrations detected in the similar exact months from 2016 to 2019 (Li and Tartarini, 2020). The decline in the concentration of the major air pollutants has been recognized by the lowered emissions of air pollution from non-essential industries and on-road vehicles (Hudda et al., 2020; Rossi et al., 2020).

Other research studies discovered a strong association between the observed reductions in human mobility and reductions in NO₂ concentrations (Archer et al., 2020). The NO₂ levels were decreased by more than 4 ppb on a monthly average, where the mobility was decreased to almost 0, and around 1 ppb decline where mobility was decreased to 20% of normal or less (Archer et al., 2020). In addition, no noticeable pattern was identified between mobility and PM_2.5 concentrations. This difference suggests that decreases in personal-vehicle traffic alone could not be efficient at lowering PM_2.5 pollution (Archer et al., 2020).

Research studies have discovered that the mean of NO₂ concentrations significantly decreased in 2020 due to the lockdown by 18–40% over the main urban areas located in Europe, such as Paris, Madrid, and Milan, and the USA, such as Boston, Springfield, and New York (Bar et al., 2021). However, other research findings showed that the enormous changes from the COVID-19 response have not decreased PM_2.5 levels across the United States beyond their normal range. In addition, the CO, NO₂, O₃, and PM₁₀ concentrations were decreased, but the reduction was transient and modest (Bekbulat et al., 2021).

More than 25 studies were conducted in the United States about the effect of the COVID-19 lockdown on the concentration of NO₂, CO, SO₂, PM_2.5, PM₁₀, and O₃. The major US states such as Texas, California, Illinois, and New York clearly showed a reduction in the concentration of NO₂ and CO during the COVID-19 lockdown (Elshorbany et al., 2021). These reductions were, in many cases, affected by meteorological conditions or compensated by local emissions (Elshorbany et al., 2021). Some research analyses in the US showed a statistically significant reduction in the concentration of particulate matter across most of the regions during the COVID-19 lockdown (Ghosal and Saha, 2021). For example, in California, the ground-based observations showed a 31%, 38%, and 49% drop in the concentration of PM_2.5, NO₂, and carbon monoxide (CO) during the lockdown period between March 19 and May 7 compared to the period of January 26 until March 19 in 2020 (Liu et al., 2021). These are 19%, 16%, and 25% greater than the means of the previous five years in the same time periods, respectively (Liu et al., 2021).

A study conducted in New York indicated an almost 23% improvement in PM_2.5 levels during the COVID-19 shutdown compared to the average level for the same months period in 2015–2018 (Perera et al., 2021). In addition, some research studies indicated that the overall concentrations of NO₂, PM_2.5, and CO were decreased by 37%, 1%, and 19%, respectively, while SO₂ and PM₁₀ were increased by 4% and 28%, respectively, in five states of the USA during the lockdown (Shakoor et al., 2020). On the other hand, the overall concentration was reduced by 38% for PM₁₀, 18% for PM_2.5, 18% for SO₂, and 39% for CO in all the selected provinces of China (Shakoor et al., 2020).

In Pittsburgh, Pennsylvania, studies showed no significant change in the industry-related intraday adaptability of PM_2.5 and CO in the industrial areas after the COVID-19 related closures (Tanzer-Gruener et al., 2020). In addition, in New York City, some studies have indicated a significant decrease in NO₂ (51%) and PM_2.5 (36%) concentrations were observed after an abbreviated period right after the shutdown took place, but there was no significant difference compared with the same span of time in 2015–2019 (Zangari et al., 2020).

From the results of the previous studies, we noticed that some studies reported an increase in air pollutants and a major improvement in air quality. On the other hand, some studies reported no significant improvement in air quality or even an increase in the air pollutants in some cities and states. It is important to mention that there are no studies in the US that measure the effectiveness of the COVID-19 lockdown on air quality at a finer spatial level within a city using all six air pollutants set by EPA: CO, SO₂, NO₂, O₃, PM_2.5, and PM₁₀.

 
3 METHODOLOGY AND MATERIALS

 
3.1 Study Area

Chicago city of Illinois, USA, was selected as the study area to evaluate the air quality at a finer spatial level. With its 2.67 million people, Chicago is the third-most populous city in the US and the most populous city in Illinois state (World Population Review, 2022). In addition, the city is a major national hub of transport at the crossroads of the USA’s rail, road, and air traffic, with one of the largest international airports. In recent years, transport emissions have been Chicago’s largest emission source (IQAir, 2022). With the exacerbating condition of COVID-19, the governor of Illinois declared the stay-at-home order on March 21, 2020. However, after gaining initial control of the spread of coronavirus, it started to reopen on May 29, 2020.

 
3.2 Data

The EPA open-source ground sensor-based data were downloaded for three years (from 2019 to 2021) from the AirNow website on the first week of January 2022. Data were collected for six air quality parameters set by EPA: daily maximum 1-hour NO₂ and SO₂, daily maximum 8-hours for CO and O₃, and daily 24-hour average concentrations for PM_2.5 and PM₁₀ for all available monitoring stations. Then, the daily average was calculated based on data of all available stations on that date. Georeferencing has been done using stations’ latitude and longitude to convert the dataframe into geospatial point dataframe. Finally, spatial filtering was used to keep the Chicago city area data only.

Three years were treated as three eras, i.e., 2019, 2020, and 2021 for pre, peri, and post COVID-19 eras, respectively. As we mentioned earlier, the state of Illinois declared stay-at-home order on March 21, 2020, and reopened on May 29, 2020; each era (i.e., each year) was classified in three timespans: from January 1 to March 20 as “before”, from March 21 to May 29 as “during” and May 30 to June 30 as “after” the stay-at-home period. The “before”, “during”, and “after stay-at-home orders” were used as generalized terms, though the stay-at-home order was only observed in 2020. These terms will refer to the respective period of each year as stated above. In the post-COVID era air quality assessment, the monthly average for the whole year in each era also has been calculated. In addition, population demographic data was downloaded from the website of the US census. This data was pre-processed using the R script provided by the census and was merged with the census TIGER/Line shapefiles to georeferenced it. Finally, it was exported to geodatabase using SQL.

 
3.3 Methods of Statistical Analysis

To compare the air quality in three periods of three years, along with time series and trend analysis, locally estimated scatterplot smoothing (LOESS) was employed to smooth out daily fluctuations in air quality monitoring. However, enough ground observation points for spatial variation analysis were unavailable for all six air pollutants. Therefore, we only analyzed spatial variation in PM_2.5, NO₂, and O₃.

As the air quality observation stations are spatially discrete points, to formulate continuous spatial surfaces for each air quality parameter, we used spatial interpolation, more specifically inverse distance weighting (IDW) methods. Spatial interpolation is the process of using points with known values to estimate values at other points. IDW is one of the most popular exact spatial interpolation methods that estimate an unknown value as the weighted average of its neighboring points, in which the weight is the inverse of the distance raised to a power (Chang, 2004). The IDW is expressed as below,

where Z_u is the unknown value to be estimated at u, Z_i is the attribute value at control point i, d_iu is the distance between points i and u, s is the number of control points used in estimation, and k is the power (Wang, 2015).

 
4 RESULTS AND DISCUSSION

 
4.1 Temporal Change in Air Pollutants

4.1.1 Change in particulate matter concentration

Analysis of PM_2.5 revealed that the daily mean concentration of PM_2.5 in three periods of 2020 was less than respective periods of both 2019 and 2021, except after the stay-at-home period in 2021 (Fig. 2). In addition, PM_2.5 concentration during the stay-at-home period was less than before and after (5% less than before) in that year (2020). This result indicates that due to the COVID-19 pandemic, PM_2.5 concentration was lower than the previous year. Again, after the stay-at-home period, PM_2.5 concentration increased. This change might result from increased human activities after the reopening on May 29. In 2021, PM_2.5 concentration increased again, and it was greater than the concentration in 2020. Thus, the beneficial effect of COVID-19 lockdown in PM_2.5 concentration might be transitory. One common thing in all three years is that PM_2.5 concentration was lower in stay-at-home periods in all three years. Therefore, there might be a seasonality effect, which can be explained by trend analysis. Also, a decrease in PM_2.5 concentration in the stay-at-home period cannot solely be contributed by reduced human activity if there are similar seasonal trends in other years.

Fig. 2. Mean concentration of each air pollutant, (A) PM_2.5, (B) PM₁₀, (C) NO₂, (D) CO, (E) SO₂, and (F) O₃ in three periods (before, during, and after stay-at-home) for three years.

Now, from the analysis of the trend of daily mean PM_2.5 concentration, it is seen that there was always a peak and fall in all three eras (years) (Fig. 3), and the slope was not consistent all over the years. Still, the pattern in 2019 and 2021 is nearly similar. In 2020, it started to decline in mid-February, which was the opposite of 2019 and 2021. Though the stay-at-home period started on March 21, 2020, there were other mitigatory measures before stay-at-home. For instance, the governor of Illinois declared Illinois a disaster area on March 9, prohibiting mass gatherings on March 13, and ordered the closure of restaurants and March 16. It is worth mentioning that the initial stay-at-home order was up to April 30, 2020. However, it extended to May 29, 2020, reducing some restrictions.

Analysis of PM₁₀ concentration showed a different result than PM_2.5. The mean concentration of PM₁₀ in 2020 was slightly higher both before and during the stay-at-home period in 2020 with respect to the equivalent periods of 2019 (Fig. 2). However, it was much higher after the stay-at-home period than during the stay-at-home period in 2020 (50%). In 2021, the mean concentration was greater than in the previous two years. These findings are also evident in trend analysis of daily mean PM₁₀ concentrations (Fig. 3). After the stay-at-home period, PM₁₀ concentration got a sudden spike. This spike could be a direct result of increasing activities after the reopening. Additionally, it is also apparent that, though the trend of PM₁₀ concentration is much higher in 2021 than in the previous two years, it follows a nearly similar rising and fall as of 2019. However, the trend in 2020 was different. In addition, it is seen that there is an increasing trend from 2019 to 2021. Therefore, it can be said that, though the PM₁₀ concentration was a little bit higher before and during stay-at-home in 2020 than the respective period in 2019, it could be much higher without stay-at-home order.

Fig. 3. The trend of the daily mean concentration of each air pollutant in three years, (A) PM_2.5, (B) PM₁₀, (C) NO₂, (D) CO, (E) SO₂, and (F) O₃. Locally estimated scatterplot smoothing was used to smooth out daily fluctuations. The shadows represent standard errors.

The difference in the effect of COVID-19 on PM_2.5 and PM₁₀ concentrations can be explained by the source of particulate matter. The main sources of particulate matter are the combustion of fossil fuels (both vehicle and residential fuel combustion), industrial processes, fugitive dust (wind and mechanical erosion of local soil), and photochemically produced particles (Illinois EPA, 2020). Combustion and photochemical products tend to be a smaller size (PM_2.5 - PM with diameters below 2.5 µm), while fugitive dust and industrial products are typically larger in size (PM₁₀ - PM with diameters below 10 µm) (Illinois EPA, 2020). Therefore, as the traffic volume was reduced due to COVID-19 and thus fuel combustion was also reduced. As a result, the smaller particles (PM_2.5) concentration continued to decrease. On the contrary, all industries were not shut down, especially food industries (Berman and Ebisu, 2020) and industries related to COVID-19 products like masks and hand sanitizers. Thus, the contribution of these industries and the fugitive dust can still contribute to an increase in the concentration of larger-sized particles.

 
4.1.2 Change in concentration of gaseous compounds in air

Similar to the particulate matters, all gaseous components of air did not show the same response in COVID-19 stay-at-home periods. In 2020, the mean concentration of NO₂ before the stay-at-home period was 28.17 ppb which was reduced during stay-at-home periods to 26.79 ppb (decrease by 5%) and increased to 27.93 ppb (increase by 4%) after the stay-at-home period (Fig. 2). If we compare the NO₂ concentration in these periods in three years, we see that before and during the stay-at-home period, the concentration of NO₂ was lower than in the same periods in both 2019 and 2021 but higher after the stay-at-home period. This result indicates that due to the stay-at-home order and other mitigatory measures before the stay-at-home order, the concentration of NO₂ was reduced. However, with the end of the stay-at-home period, it increases again. In addition, it increases in post COVID era (in 2021) and is nearly similar to 2019. Now, from the trend of NO₂ concentration, it is seen that at the beginning of January, all three years experienced an uprising of NO₂ concentrations (Fig. 3). In mid-February of 2020, it started to decrease and was relatively stable up to the end of April. Then, it spiked after the stay-at-home period. As discussed earlier, the early decrease in NO₂ concentrations can be attributed to the early COVID-19 response before the stay-at-home order.

Compared to NO₂, no immediate drops were observed during the stay-at-home in case CO, SO₂, and O₃ (Fig. 2) in 2020. The mean concentration of these three air pollutants increased by 3%, 34%, and 32%, respectively. These even further increase after the stay-at-home period. This analysis might not be a fair comparison due to seasonality and trend. However, CO and O₃ concentrations were lower before and during the stay-at-home periods in 2020 than in equivalent periods of the previous year. On the contrary, this scenario is the opposite in the case of SO₂. Though the trend of the daily mean concentration of CO, SO₂, and O₃ was not always consistent, there is still a similar upward trend of these three air pollutants in all three years (Fig. 3). Thus, the increase in CO, SO₂, and O₃ after stay-at-home in 2020 cannot be entirely attributed to reopening after stay-at-home. In general, CO concentrations decreased in the post COVID era (2021). On the contrary, both SO₂ and O₃ have increased in post COVID era.

The source of these four gaseous components can explain the difference in the impact of COVID-19 on the concentration of these pollutants. NO₂ comes from fossil-burning sources such as vehicles, power plants, and industrial emissions (Liu et al., 2021). As road traffic was reduced and industries were shut down, NO₂ declined. The major source of CO is motor vehicles (Illinois EPA, 2020). With the reduction of traffic volume, CO concentration should decline. However, CO has another type of source, that is indoor fuel-burning appliances (Liu et al., 2021). As more people were at home, there was a chance of more use of these appliances, which can cause an increase in CO concentration. O₃ is a photochemical secondary oxidation product created when nitrogen oxides (NO_x) and volatile organic compounds (VOCs) react in warmer temperatures and sunlight. Since it occurs at a warmer temperature, it is much more prevalent in the summer than in winter (IQAir, 2022). Thus, all three years showed an uprising trend up to June. O₃ can be more limited by VOCs than NO_x in the NO_x-rich urban air, and thus reducing NO_x emissions would increase O₃ unless VOCs are reduced at a higher rate at the same time (Chen et al., 2020; Kleinman et al., 2005).

 
4.2 Spatio-temporal Change in Air Pollutants

Analysis of the spatio-temporal change revealed a similar result to what we found in the previous section, with some exceptions. For instance, while the average concentration of PM_2.5 during the stay-at-home period in both 2019 and 2021 was higher than the respective period in 2020, spatio-temporal analysis shows a slightly opposite relation. The range of spatial variation was larger in 2020 compared to the respective periods of previous and next year. This variation implies that, though, on average, PM_2.5 concentration was lower in the stay-at-home period in 2020, there were few point locations where PM_2.5 was increased. In addition, NO₂ concentration in 2021 after the stay-at-home period is greater than the respective period of 2020, though it was lower in temporal analysis. In general, all three air pollutants show a reduction in concentration during the stay-at-home period in 2020 compared to the respective period of 2019.

There was notable spatial variation in PM_2.5 concentrations in all three periods in all three years (Fig. 4). The southwest industrial part of the city was always a high PM_2.5 concentration area in all periods in all three years except the stay-at-home period in 2020. This variation indicates that the closure of industries contributed to reducing PM_2.5 concentration. In addition, the Chicago airport area was always a high concentration area in all three years compared to the concentration in other parts of the city in any period. While the middle part of the city was a low PM_2.5 concentration area before the stay-at-home period, it turned into a high concentration area during the stay-at-home period in 2020.

Fig. 4. Spatio-temporal variations in PM_2.5 in three periods of three eras: (A-1) before, (A-2) during, and (A-3) after stay-at-home periods in the pre-COVID era (2019), (B-1) before, (B-2) during and (B-3) after stay-at-home periods in peri-COVID era (2020), (C-1) before, (C-2) during and (C-3) after stay-at-home periods in post-COVID era (2021).

High O₃ concentration was predominantly observed in the eastern part of the city (Fig. 5). We mentioned earlier that O₃ formation occurred at higher temperatures. Some studies found that the eastern part of Chicago city is usually exposed to higher temperatures due to the warming effect of Lake Michigan in colder weather (English et al., 2020). The O₃ concentration was also higher in the southwest industrial area.

 Fig. 5. Spatio-temporal variations in O₃ in three periods of three eras: (A-1) before, (A-2) during, and (A-3) after stay-at-home periods in the pre-COVID era (2019), (B-1) before, (B-2) during and (B-3) after stay-at-home periods in peri-COVID era (2020), (C-1) before, (C-2) during and (C-3) after stay-at-home periods in post-COVID era (2021).

The concentration of NO₂ was always higher in the Chicago airport area during all periods in all three years (Fig. 6). In addition, the downtown area where population density is higher, was also a high NO₂ concentration zone for 2020 and 2021. In general, NO₂ is lower in the lower middle part of the city, where population density is comparatively lower (Fig. 1).

Fig. 1. Map of Chicago city, USA showing Census tract wise population (left) and population density per sq. km (right) in 2020.

Fig. 6. Spatio-temporal variations in NO₂ in three periods of three eras: (A-1) before, (A-2) during, and (A-3) after stay-at-home periods in the pre-COVID era (2019), (B-1) before, (B-2) during and (B-3) after stay-at-home periods in peri-COVID era (2020), (C-1) before, (C-2) during and (C-3) after stay-at-home periods in post-COVID era (2021).

Due to the low number of monitoring stations, the spatial surface created by spatial interpolation might not resemble the actual pollution surface. Still, it is a reasonable estimation of spatial variation in air pollution. In addition, the spatial variation of these three air pollutants could be better explained by the land use, built environment, vegetation index, industrial presence, population demographic profile, weather, and environmental factors.

 
4.3 Monthly Change in Air Quality during and Post-COVID Era

The average concentration of PM_2.5 concentration was lower every month in 2020 compared to the respective month in 2019 (Fig. 7). It reduced from a range of 0.03% to 29.03% in different months. However, the concentration of PM_2.5 increased in 2021 almost every month except June, September, and December. While February experienced the highest reduction in 2020, it also showed the highest increase in 2021. In the case of PM₁₀ concentration, it increased in 2020 every month except February, March, and December (Fig. 7). Though it started to increase earlier in the year, after May, it shows a higher percentage increase than in 2019. It continued to increase in 2021. While it shows an increase in concentrations up to May, it started to decline after May.

Fig. 7. Monthly change in concentrations (%) of six air pollutants with respect to previous year (A) PM_2.5, (B) PM₁₀, (C) NO₂, (D) CO, (E) SO₂, and (F) O₃. The missing months of CO and SO₂ are due to the lack of data.

NO₂ concentration was lower in the first five months of 2020 than in the respective month of 2019 (Fig. 7). From June to August, it shows an increase in NO₂. However, the percentage increase was lower than the percentage of reduction that happened earlier in the year. In 2021, it showed the exact opposite scenario. The months that experienced a reduction of NO₂ concentration in 2020, show an increase in 2021. The concentration of CO in 2020 was lower in every month compared to the respective month of 2019 except for July (Fig. 7). However, these changes were higher in the last three months of the year. It also shows a reduction in the first half of 2021 and increases with a very high percentage in the last three months.

SO₂ experienced an increase in the concentration in 2020 up to August except for February and May (Fig. 7). It shows the decline in the later part of the year. The month which experienced an increase in 2020 shows a decline in 2021. It is also partially true for O₃ concentrations. O₃ concentrations were lower in the first four months of 2020.

In general, all six air pollutants show a change in the opposite direction in most months in the peri-COVID (2020) and post-COVID era (2021). It indicates that the post-COVID era experienced a reverse direction change than what happened in the COVID era. Most likely, normal activities that were almost restored in 2021 caused air pollutants to return to the normal cycle of the pre-COVID era. It can be concluded that the concentration of PM_2.5, NO₂, and CO has been reduced in 2020 and increases again in 2021 except for CO. The concentration of SO₂ increased during the COVID era and in the post-COVID era.

 
5 CONCLUSIONS

This study evaluated the effect of COVID-19 on air quality by analyzing the spatio-temporal changes in the concentrations of six air pollutants from the pre-COVID era (2019) to post-COVID (2021). The study found that PM_2.5 and NO₂ concentration was reduced during the stay-at-home period in 2020, and it increased after the stay-at-home due to the resumption of activities. In general, these two pollutants were lower than the previous year and increased in the post-COVID era. On the other hand, the mean concentration of CO, SO₂, and O₃ increased during the stay-at-home period. These even further increase after the stay-at-home period. These three pollutants, especially O₃, show an increasing trend in all three years. Thus, the increase in CO, SO₂, and O₃ after stay-at-home in 2020 cannot be solely attributed to reopening after stay-at-home. In general, CO concentrations decreased in the post COVID era (2021). On the contrary, both SO₂ and O₃ have increased in post COVID era. PM₁₀ concentration increases slightly during the stay-at-home period and becomes higher after the stay-at-home period. In 2020, PM₁₀ concentration was much higher than in 2019 and continued to increase in the post-COVID era. Considering the increasing trend from 2019 to 2021, it can be said that, though the PM₁₀ concentration in the stay-at-home period in 2020 is higher than the respective period in 2019, it could be much higher without stay-at-home order. The difference in the impact of COVID-19 on the concentration of these six air pollutants may occur due to these pollutants' source differences. Temperature and other weather conditions may also be responsible for this discrepancy.

The spatio-temporal analysis also revealed nearly similar results with very few exceptions. There was notable spatial variation in PM_2.5, NO₂, and O₃ concentrations in all three periods in all three years. The Chicago airport area and the densely populated downtown of the city were predominately higher concentration zones for NO₂. The Eastern part of the city was observed as a high concentration area for O₃. The southwest industrial part of the city was always high PM_2.5 concentration area in all periods in all three years except the stay-at-home period in 2020. This variation indicates that the closure of industries contributed to reducing PM_2.5 concentration. In general, NO₂ was lower in the lower middle part of the city, where population density is comparatively lower.

In conclusion, it can be said that reduced human activities contributed to improving air quality by reducing some air pollutants, especially PM_2.5, NO₂, and CO. However, this improvement was transitory, and it reverted in the post-COVID era. Therefore, policies should be formulated and practiced to improve air quality for the long term, especially when it is proven that exposure to air pollution may cause major harmful health effects.

 
ACKNOWLEDGMENTS

We thank Justin Krohn, a Ph.D. student at the Institute of Data Science and Informatics of the University of Missouri, for providing access to the geodatabase of the US population demographics.