COVID-19 and Air Quality
Case Study: Investigating the Effects of the COVID-19 Lockdown on Air Quality in Toronto
This is an edited version of what was included in one of my chapters of my PhD thesis. I have successfully defended my dissertation in April, 2021, and the Thesis was accepted in the same month. The Thesis is available via the University of Toronto Library Archive, here.
In late 2019, several cases of respiratory illness were reported in China. The culprit was soon identified to be a strain of coronavirus, with some similarities to the virus responsible for the SARS epidemic in 2002-2004. The virus was named Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) (van Doremalen et al., 2020), and the disease it brings about, COVID-19. The virus quickly spread and was deemed to be an outbreak by the World Health Organization (WHO) in January 2020 and a pandemic in March 2020 (https://www.who.int/emergencies/ diseases/novel-coronavirus-2019/situation-reports/). The pandemic caused a global social and economic disruption, including a global recession comparable to the Great Depression, according to the International Monetary Fund (https: //blogs.imf.org/2020/04/14/). Over 1 million deaths have been attributed to the virus worldwide, including over 10,000 in Canada (as of November, 2020) (latest updates can be found on a COVID-19 tracking website hosted by Johns Hopkins University: https://systems.jhu.edu/).
The virus first landed in Canada in January 22, 2020, and has since had profound social and economic effects. On March 16 2020, Prime Minister Justin Trudeau recommended closure of all non-essential services (i.e., all recreation programs, libraries, private schools, daycares, and churches and other faith settings, as well as bars and restaurants, except those that offer takeout or delivery), and in Ontario, a state of emergency was declared a day later, resulting in closures of daycares, bars and restaurants, theaters and private schools (https://www.canada.ca/coronavirus).
Ontario Premier Doug Ford released a ”Framework for Reopening our Province” in late April, which described the gradual lifting of economic restrictions, divided into three ”Stages.” Stage 1 would allow the reopening of outdoor spaces, and select businesses. Stage 2 would allow for more businesses to reopen, and the limit on the participants would be increased. Stage 3 would further relax the restrictions, aside from restrictions on large public gatherings. The reopening processes began in mid-May. On July 24, several parts of Ontario, including Hamilton, the regions of York, Durham, Halton and Niagara as well as the counties of Haldimand, Horfolk and Lambton were permitted to enter Stage 3 of the reopening plan set forth by the province, and a week later on July 31, Toronto and Peel Region was also permitted to enter Stage 3 (https://www.ontario.ca/page/reopening-ontario-stages). The pandemic is still ongoing, however, and Premier Ford declared the province to officially be in the ”second wave” of the coronavirus pandemic on September 28. The Ontario government implemented additional public health measures (”modified” Stage 2) in Ottawa, Peel, and Toronto on October 10, and a province-wide shutdown on December 26, 2020 (https://www.ontario.ca/page/covid-19-provincewide-shutdown). Similar preventative measures, often called lockdowns, were taken in many parts of the world.
These lockdowns have had significant effects on air quality. Emissions of NOx in China were estimated to be down by 36 % from early January to mid-February, with more than 80 % of reductions occurring after their respective lockdown in most provinces (Miyazaki et al., 2020), and average NO2 columns, measured by space-borne instruments (TROPOMI) decreased by 40% in all Chinese cities (Bauwens et al., 2020). Reduction in surface PM2.5, NO2, CO, and SO2 concentrations were also observed in northern China, although the reduction in O3 precursor (NOx) emissions increased surface O3 northern China (Shi and Brasseur, 2020).
Here, we present preliminary findings indicating reduced total columns of CO from the FTIR at the University of Toronto Atmospheric Observatory (TAO) (an instrument used to obtain column abundance of various atmospheric species) during the lockdown. Figure 1 shows total column CO observations from 2020 in comparison to 2002-2019. Table 1 compares the monthly mean values of CO from 2002-2019 and 2020, as well as the 2020 monthly means extrapolated from 2002-2019 data, using the fit described in Yamanouchi et al., (2020) (my paper), which accounts for trends and seasonalities in the data (trended Fourier series of several orders were used for this analysis; see Figure 2 for the time series and fit). The lockdown in Ontario (and other parts of Canada) began in March, with some restrictions in Ontario being lifted towards the end of July. 2020 monthly averages of CO total columns in April, May and June are more than 1σ lower than the 2002-2019 counterparts (with respect to the 2020 standard deviation). When comparing with the extrapolated 2020 values (i.e., accounting for trends and seasonalities in the data; see Figure 2), the observed mean value of April is more than 1σ lower. It should be noted here that the lifetime of CO is about 1 to 2 months, so reduced emissions may not be immediately evident.
Another point worthy of discussion is the spike in CO in 2020 September. This is likely due to a biomass burning (e.g., wildfire) plume traveling over Toronto. An analysis using the FLEXPART Lagrangian dispersion model (v8.2) (Pisso et al., 2019) in its back-trajectory mode, coupled with satellite-based Burned Area Data Product (Giglio et al., 2015) shows that TAO measurements made between 13 September and 17 September (peak of the enhancement event) are sensitive to regions with fires (see Figure 3). Details of the source attribution methodology can be found in e.g., Lutsch et al., (2016) and Yamanouchi et al., (2020) (my paper). This enhancement likely led to the high monthly average in CO for September 2020.
Similar analyses were done for NH3 and tropospheric O3 (see Figures 4, 5, and Tables 2, 3, respectively). NH3 monthly mean variability is too large for meaningful comparison. The tropospheric O3 in March and July, 2020 showed a reduction by more than 1σ (with respect to the 2020 standard deviation). Comparisons of the 2020 observations with extrapolated values also show >1σ reductions in March and July. This is comparable to findings by Steinbrecht et al., (2020), who observed a 7 % reduction in northern extratropical free tropospheric (1 to 8 km) O3 in late spring to summer of 2020, using multiple sites, including TAO. The analysis of O3 presented in this study were done with tropospheric (0 to 12 km) column, and resulted in 10.8 % reduction when comparing 2002-2019 and 2020 data.
The effects of COVID-19 lockdowns are still under investigation in many parts of the world, and the data from TAO provide an urban dataset. The preliminary results shown here can be expanded upon to shed insight into the effects of the lockdown on urban air quality.
Figure 1: Toronto CO total columns from 2002 to 2020 October (from the FTIR at TAO) (all data points on the left, and monthly means ±1σ (shading) (2002-2019 data) on the right). Data from 2002 to 2019 are shown as grey points and 2020 are shown by red points (for both plots). Note the spike in CO in September, 2020; this may be a biomass burning enhancement event (see discussion in-text, and Figure 3).
Figure 2: Time series and fitted trends of Toronto (TAO) total columns of CO (2002-2019). Trended Fourier series of order 3 was used. This was chosen after analyzing residuals of the fit of various Fourier series with statistical tests (e.g., examining the RMSE, using the Kolmogorov-Smirnoff test for normality). Enhancement events (purple points) were obtained using a methodology described by Zellweger et al., (2009). Observed trend was −0.90 ± 0.07 %/year (−0.87 ± 0.05 %/year after removal of enhancement events). Error analysis was done by bootstrap resampling.
Figure 3: FLEXPART seven-day back-trajectory sensitivity plot for the CO enhancement event in September 2020. Tracer particles were released between 13 and 17 September. The red dots indicate fires (based on the MODIS Burned Area Product, a dataset using satellite-based thermal imaging).
Tables 1, 2, and 3: Comparison of 2002-2019 and 2020 monthly mean (1σ in parenthesis) TAO FTIR CO and NH3 total columns, and O3 tropospheric columns (Tables 1, 2, and 3, respectively), and 2020 monthly mean extrapolated from 2002-2019 data (using the fit discussed in-text, also see Figure 2).
Figure 4: Toronto NH3 total columns from 2002 to 2020 October (from the FTIR at TAO) (all data points on the left, and monthly means ±1σ (shading) (2002-2019 data) on the right). Data from 2002 to 2019 are shown as grey points and 2020 are shown by red points (for both plots).
Figure 5: Toronto O3 tropospheric columns from 2002 to 2020 October (from the FTIR at TAO) (all data points on the left, and monthly means ±1σ (shading) (2002-2019 data) on the right). Data from 2002 to 2019 are shown as grey points and 2020 are shown by red points (for both plots).