Ozone and Particulate Matter Variations in Southwest Ohio, USA, during the COVID-19 Pandemic in 2020

In this study, ozone and particulate matter variations from four monitoring stations in the Southwest Ohio region were analyzed at different stages of the COVID-19 pandemic in 2020, and compared with those in 2019. These stations include a U


INTRODUCTION
On March 11, 2020, Coronavirus Disease 19 (COVID-19) was declared as a global health emergency by the World Health Organization (WHO, 2020).Restrictions were imposed around the world to stop COVID-19 transmission, such as stay-home orders (lockdown), reduced occupancy and social distancing, etc.Most non-essential operations were closed during the lockdown period.These drastic changes in people's lives also affected the surrounding environment such as air and water quality.There have been multiple studies and reports around the world on air pollution comparisons during the lockdown period with previous years, and reported significant decreases of criteria air pollutants, such as PM2.5 and NO2, from both ground and satellite monitoring (Huang et al., 2021;Ogen, 2020;Amouei Torkmahalleh et al., 2021).
Due to the severity of the pandemic, 42 out of 50 states in the USA issued stay-at-home orders i.e., lockdown.Several studies looked into the impact of the lockdown on air pollutant variations.At a near-road site in Seattle, WA, concentrations of PM2.5, NO2 and CO were 2-4% lower during COVID-19 lockdown than pre-COVID concentrations (Xiang et al., 2020).A study of the lockdown impact in Phoenix, AZ used CO, NO2 and PM10 data from 3 monitoring stations together with meteorological data, such as planetary boundary layer height.Results suggested that the concertation changes of CO and NO2 were not significantly related to the lockdown despite 38% traffic reduction, while PM10 reduction was due to the lockdown (Miech et al., 2021).A study of the lockdown impact in New York City used a time-lagged linear regression model to evaluate variations of NO2 and PM2.5 from 15 monitoring stations.It suggested that the decrease was not statistically significant compared with previous years (Zangari et al., 2020).
A study of air pollutant changes in Atlanta, GA was performed at a near road and an urban site.Comparing with the same time in 2019, NO2 reductions during the lockdown period (Mar.14-Apr.30, 2020) were 10.8% and 8.2% respectively at these two sites, with 15.7% ground level ozone and 5% PM2.5 reductions (Huang et al., 2021).These reductions were attributed to a 50% reduction in traffic.A study in Memphis, TN during the lockdown period (Mar.25 to May 4, 2020) evaluated data from 5 monitoring sites (one NCore, three background and one low-income community in downtown) by comparing with the same periods in 2017-2019.It indicated that the reductions of PM2.5, NO2 and ozone were not statistically significant in spite of 57% traffic reduction (Jia et al., 2020).These results suggested that traffic might not be an important source in the Memphis area.Lange et al. (2021) studied air pollutant variations in Apr.2020 (during the lockdown) compared with data for the past four years at 4 stations in Pittsburgh PA.Reductions of NO2 and PM10 were observed at most sites while PM2.5 reduction was not uniform.
A brief air pollutant variation study was also performed in Cincinnati, OH, a midsized city in the Midwest of the US, at a near-road site, which is located adjacent to Interstate 75 (I-75) south approaching the Hopple Street exit (SWOAQA, 2020).From Mar. 23 through Jun. 30, 2020, criteria pollutant concentration reduction at the near-road site comparing with the same time period during 2017-2019 were reported as follows, NO2 by 18.6%, CO by 9.1% and PM2.5 by 14.1%.The total traffic counts from Mar. 24 through May 19, 2020 were down 42% at this site as compared to 2019 (SWOAQA, 2020).
Later on, some COVIID-19 impact studies went beyond the lockdown period.Steffeck (2021) evaluated PM2.5 and ozone monitoring data at three sites, an urban site with heavy traffic, a suburban site and a rural site at the Chicago, IL metropolitan area, during the lockdown and reopening periods of 2020.Compared to 2019, PM2.5 concentrations during the lockdown period decreased at the suburban site, but increased in the urban site.Ozone increased during Jun.-Aug.for all three sites.These has been attributed to truck traffic, which actually increased during the pandemic, ostensibly due to online shopping.There are also national level studies, which indicated that impacts of COVID-19 lockdown on air pollutant concentrations are not uniform across the US (Chen et al., 2020;Berman and Ebisu, 2020;Bekbulat et al., 2021).Chen et al. (2020) used data from 28 NCore stations from Mar. 15 to Apr. 25, 2020 and found that air pollutant reductions tend to be more significant in areas with higher population densities in the Northeast or California, where reductions of traffic and energy consumption were more significant.Bekbulat et al. (2021) analyzed the national trends of criteria air pollutant changes at NCore sites from Jan. 1 to Sep. 1, 2020 and developed a statistical algorithm to evaluate whether pollutant concentrations were higher or lower than expected as compared with 2010-2019.Data from two Ohio NCore sites were used, one of which is the Taft site located in Southwest Ohio.Results indicated that PM2.5 and ozone decreased during the lockdown period in the state of Ohio.
There are state level studies on air quality variations during the lockdown period, such as one for the State of California (Pan et al., 2020) and another one covering 5 states in the US (Alabama, California, Florida, Louisiana and North Carolina) and 5 provinces in China (Shakoor et al., 2020).These studies all reported non-uniform particulate matter reductions during the lockdown.
A global scale study also included select cities from the US.It suggested that NO2 globally declined due to COVID-19 lockdown, changes were non-uniform for PM2.5 while O3 increase was reported from the US cities selected (Amouei Torkmahalleh et al., 2021).
Results from US cities suggested that there were significant reductions of PM2.5 and NO2 emissions at sites strongly impacted by traffic.In the meantime, data from multiple sites indicated COVID-19 impact can be regionally non-uniform, which warrants evaluation of region-specific characteristics.In addition, most of these studies are only during the lockdown period whereas much of the world did not resume business as usual until well after the lockdown ended.Therefore, the goal of this paper is to study COVID-19 impact on the air quality of the Southwest Ohio, which is subject to multiple air pollution sources and with ozone and PM2.5 concentrations peaking in the summer rather than the lockdown period.Ozone and PM2.5 are selected as these pollutants are of regulatory interests in this area.The region is still in marginal non-attainment for the 8-Hour Ozone (2015) National Ambient Air Quality Standards (NAAQS) (U.S. EPA, 2022).In addition, a longer time period during the pandemic was studied which included both the lockdown and also the gradual reopening periods of 2020.Four monitoring sites were investigated, one urban, two suburban and one industrial, representative of this region.

Study Area
Four different sites in the Southwest Ohio region were selected reflective of the area's industrial and residential activities (Fig. 1, Table 1).The Taft site is one of the 78 NCore sites in the US that monitors multiple types of air pollutants and precursors (Chen et al., 2020).It is an urban site and is also close (about 3 km) to both interstate highways I-75 and I-71.There are many commute destinations in the vicinity, such as universities, schools, several hospitals, research institutions and various businesses.The Taft site is within the City of Cincinnati, with a population density of 1500 km -2 and the air pollutant monitoring instruments are located on the roof of a two-story building where the local air agency is located.
Sycamore is a suburban monitoring site with traffic, industrial and residential sources nearby.It is part of the state and local monitoring (SLAM) network, which is typically for areas with multiple sources and high population density (777 km -2 ).Although much less studied, it has the highest ozone concentrations in the Southwest Ohio region as it is downwind of the City of Cincinnati and is often used for air quality forecasting.This site is 2 meters above ground and 1.12 km from highway I-275.• 3 Day PM2.5 • 6 Day PM10 • 3 Day PM2.5 Amanda and Yankee monitoring sites are both about 45 km north of the Taft site.They are special purpose monitoring sites (SPMS) installed due to a new facility permit near Yankee and reported high particulate matter concentrations from time to time.Amanda is situated in a school amongst a residential area close to various industries.Yankee is located in a heavily industrial area.The population density in this area is about 325 km -2 .Both sites are 2 meters above ground.

Data Sources
Publicly available PM2.5, PM10 and ozone data from 2019 and 2020 for these sites were obtained from the U.S. EPA Airdata Center (U.S. EPA, 2021).Both Sycamore and Taft have daily data for PM2.5 and ozone.Ozone is measured from Mar. 1 to Oct. 31 annually in Southwest Ohio.Both Amanda and Yankee have filter based PM2.5 (every 3 days) and PM10 (every 6 days) data while the local agency also has data on PM2.5 daily measurements.While the main contributors for PM2.5 are complex, such as regional transport, meteorology and local sources, PM10 variations may be more indicative of local emission reductions (Bekbulat et al., 2021).
Air pollutant concentration data in 2020 were binned into the following time periods due to different stages of COVID-19 restrictions.The lockdown period in Ohio started on Mar.23 and lasted to May 30, 2020 (The New York Times, 2020).The re-opening period was sub-divided into two.Based on average temperatures, Jun. 1 st -Aug.31 st corresponds to the summer season in the study area (Xie et al., 2018).Most schools remained closed, while companies were partially reopened, and traffic was less than business as usual.PM2.5 tends to peak during summer in this region (Xie et al., 2018) and ground level ozone also tends to peak during the summer season in the midwestern USA (Jing and Goldberg, 2022).
The second part of the reopening period ranged from Sep. 1 st to Dec. 31 st when most schools and most businesses were reopened although at reduced capacities.This division also largely coincides with the four seasons in the local area.A pre COVID-19 period (Jan.1-Mar.22, 2020) was also analyzed at Yankee and Amanda to represent impacts of air pollution controls implemented prior to COVID-19.
Daily vehicle traffic counts were obtained from the Ohio Department of Transportation's Transportation Data Management System at the site ID 127431 (labelled as Sharonville in Fig. 1), which has traffic data for both 2019 and 2020.Data processed include total traffic volume, volumes of light-duty vehicle and heavy-duty vehicle on counts per day.
The average, standard deviation (which represents the temporal variations), maximum concentrations were analyzed for each air pollutant at each site.To ensure that concentration changes are statistically significant (p-value = 0.05), unpaired t-tests were performed between the 2019 and 2020 data for each time period.

PM2.5 Concentrations during the Pandemic
Fig. 2(a) shows PM2.5 data at Sycamore in both 2020 and 2019, together with select statistical results presented in Table 2.In Sycamore, PM2.5 concentrations are slightly lower during the lockdown period and slightly higher during both summer and fall.The changes of the four time periods are all less than 5% while the p-values of the t-test are all greater than 0.05, indicating that these changes are not statistically significant.The results suggested that COVID-19 restrictions did not reduce PM2.5 concentrations at the Sycamore site.This site might be more subjected to local residential activities and local traffic, which were not much affected by COVID-19 restrictions.
Fig. 2(b) shows PM2.5 trends at the Taft site with statistical summary in Table 2.The 15.6 % reduction during the lockdown period is consistent with that of the Near Road site (14.1% reduction) as reported by the Southwest Ohio Air Quality Agency (SWOAQA, 2020) due to the close vicinity of these two sites.For 2020 vs. 2019, PM2.5 reductions at Taft site are statistically significant except for the summertime.
At Taft, the PM2.5 variation trend during lockdown is consistent with Bekbulat et al. (2021), but is different from Chen et al. (2020).The increasing trend from Chen et al. (2020) was during a partial lockdown time frame (Mar. 15-Apr. 25, 2020), as opposed to the full lockdown period (Mar. 23-May 30, 2020) used in this study.The average concentration of PM2.5 at the Taft site was 8.21 µg m -3 during the lockdown period of this study, which is in agreement with the 8.5 µg m -3 reported by (Chen et al., 2020) for Mar.15-Apr.25, 2020.PM2.5 concentration changes at Taft may be related to traffic volume and pattern changes on the highways.At a highway traffic site in Southwest Ohio (Fig. 1), total traffic volume reduction during the lockdown period was about 40%, with a 43% reduction for light duty vehicles and a 12.5% reduction for semi-trucks.Total traffic volume reductions in the summer and fall were 13%, and 12% respectively with 16% reduction over the whole year of 2020.
As shown in Fig. 2(c) and Table 2, PM2.5 concentrations at Yankee are lower in 2020 than 2019 for all the 4 time periods studied.Yankee has the highest PM2.5 concentrations among the 4 sites.For 2020 vs. 2019, more than 10% decreases in average concentrations were reported for all the time periods, although the reductions were not statistically significant.This is consistent with a study Pittsburgh, PA, where the largest PM2.5 reduction was also observed at an industrial site during the lockdown (Lange et al., 2021).The Southwest Ohio Air Quality Agency attributed the reduction at Yankee to both air pollution control measures continuously implemented over the years and COVID-19 restrictions.The 19.30% PM2.5 reduction prior to COVID-19 (p-value of 0.06) served as an indication of control measures.Fig. 2(d) shows Amanda PM fine in 2019 and 2020 with statistical analysis in Table 2.Although Amanda is only 0.77 miles west of Yankee, the PM2.5 data seemed independent.The correlation   between Amanda and Yankee PM2.5 data are low, with R-squares of 0.023 in 2019 and 0.0065 in 2020.PM10 correlations between Amanda and Yankee have R-squares of 0.008 in 2019 and 0.50 in 2020.
As shown in Table 2, the average PM2.5 is higher in Amanda during the lockdown period but are lower in the rest of 2020 compared to the same time periods in 2019.However, these changes are not statistically significant, which is likely due to limited data points in each time period as well as large data variations.
The temperature variations between 2020 and 2019 were also analyzed as shown in Table 3 (Tumbleson and Balachandran, 2022).In 2020, the average temperatures were 2 and 2.8°C lower during the lockdown and the fall respectively.While the decrease did not correlated well with PM2.5 variations, it may contribute to lower ozone levels.
To further evaluate the impact of local source reduction, PM10 data for both Yankee and Amanda were analyzed with the statistical summary presented in Table 4. T-tests were not performed due to limited number of data points (1 in every 6 days).The PM10 annual average at Yankee is about 22% lower in 2020 than that of 2019, with a 26.6% reduction during the lockdown, and the reduction trends seem consistent with those for PM2.5.PM10 variation in Amanda did not correlate with those of Yankee despite their close vicinity.5.The trends of ozone variations are in agreement at the two stations studied, largely due to the regional nature of ozone (Jing and Goldberg, 2022).

Ozone Concentrations during the Pandemic
The average ozone concentration during the lockdown period was 43 ppb at Sycamore as (µg m -3 ) Jun.-Aug.
(µg m -3 ) The Whole Year (µg m -3 )  2020 and 2019 (Tumbleson and Balachandran, 2022).BLH has been used as an indicator of air quality, but its impact can be inconclusive for some urban locations (Geiß et al., 2017).Precipitation was not studied in this work, which can be a potential contributing factor.Huang et al. (2021) suggested that increased precipitation during the lockdown could contribute to PM2.5 reductions in Atlanta, GA.However, precipitation is less of an impact for summer time ozone increase in a study in Chicago, IL, which was attributed to the increased number of dry, sunny, hot and yet stagnant days (Jing and Goldberg, 2022).Care must be taken to identify the determinant meteorological factors.Air quality can also be impacted by the topography and source characteristics of the study area, which should also be studied in the future.

CONCLUSIONS
The impact of COVID-19 on air quality was evaluated with data from four sites in southwestern Ohio: an urban site (Taft), an industrial site (Yankee) and two suburban sites (Sycamore and Amanda).Ozone and particulate matter variations were studied during both the lockdown period and reopening from Jun. to Dec. 2020, and compared with those in 2019.PM2.5 reductions were not uniform across the region, and displayed both spatial and temporal variations.Concentrations of PM2.5 in Sycamore remained almost unchanged (less than 5%), while those at Taft decreased more than 15% during the lockdown and the fall except summertime.At Yankee, the average PM2.5 during the lock down was 11.6% lower, but it was 11.2% higher at Amanda.Reduction at Taft may be related to traffic restrictions while that at Yankee may be due to both reduced industrial activities and source control measures.PM10 decreased at Yankee during the 2020 pandemic periods.Ozone concentration reductions were observed in both Sycamore and Taft with uniform trends.Although air quality improvements were reported to the public during the pandemic year of 2020 in the form of AQI, the actual air pollutant variations are not uniform, especially for particulate matter.

Fig. 3
Fig.3shows ozone variations at (a) Sycamore and (b) Taft sites in both year 2020 and 2019, with statistical results in Table5.The trends of ozone variations are in agreement at the two stations studied, largely due to the regional nature of ozone(Jing and Goldberg, 2022).The average ozone concentration during the lockdown period was 43 ppb at Sycamore as

Table 1 .
Summary of monitoring sites studied.

Table 2 .
Statistical results of PM2.5 at four sites of Southwest OH.

Table 3 .
Temperature variations at the Taft site (°C).

Table 4 .
Statistical results of PM10 at two sites of Southwest OH.