Air Quality Change in Seoul, South Korea Under COVID-19 Social Distancing: Focusing on PM2.5

International Journal of Environmental Research and Public Health

Article Air Quality Change in Seoul, South Korea under COVID-19 Social Distancing: Focusing on PM2.5

Beom-Soon Han 1, Kyeongjoo Park 1 , Kyung-Hwan Kwak 2,* , Seung-Bu Park 1 , Han-Gyul Jin 1, Sungju Moon 1 , Jong-Won Kim 1 and Jong-Jin Baik 1

1 School of Earth and Environmental Sciences, Seoul National University, Seoul 08826, Korea; [email protected] (B.-S.H.); [email protected] (K.P.); [email protected] (S.-B.P.); [email protected] (H.-G.J.); [email protected] (S.M.); [email protected] (J.-W.K.); [email protected] (J.-J.B.) 2 School of Natural Resources and Environmental Science, Kangwon National University, Chuncheon 24341, Korea * Correspondence: [email protected]

Received: 29 July 2020; Accepted: 24 August 2020; Published: 27 August 2020

Abstract: Seoul, the most populous city in South Korea, has been practicing social distancing to slow down the spread of coronavirus disease 2019 (COVID-19). Fine particulate matter (PM2.5) and other air pollutants measured in Seoul over the two 30 day periods before and after the start of social distancing are analyzed to assess the change in air quality during the period of social distancing. The 30 day mean PM2.5 concentration decreased by 10.4% in 2020, which is contrasted with an average increase of 23.7% over the corresponding periods in the previous 5 years. The PM2.5 concentration decrease was city-wide and more prominent during daytime than at nighttime. The concentrations of carbon monoxide (CO) and nitrogen dioxide (NO2) decreased by 16.9% and 16.4%, respectively. These results show that social distancing, a weaker forcing toward reduced human activity than a strict lockdown, can help lower pollutant emissions. At the same time, synoptic conditions and the decrease in aerosol optical depth over the regions to the west of Seoul support that the change in Seoul’s air quality during the COVID-19 social distancing can be interpreted as having been aﬀected by reductions in the long-range transport of air pollutants as well as local emission reductions.

Keywords: COVID-19; social distancing; PM2.5; urban air quality; Seoul; air quality monitoring station

1. Introduction The rapid spread of coronavirus disease 2019 (COVID-19), now declared a pandemic by the World Health Organization (WHO), prompted the national and municipal governments around the world to take unprecedented measures, such as lockdowns, strict travel bans, and other restrictions on human activity. In South Korea, the first COVID-19 case was confirmed on 20 January 2020, and there was a dramatic rise in new cases, starting from 18 February 2020, following several clusters of mass infections around the country. On 29 February 2020, the term “social distancing” was first used at a press briefing conducted by the Korea Centers for Disease Control and Prevention (KCDC) in reference to a set of recommended practices including staying away from crowded spaces, avoiding non-essential travel, and keeping a safe interpersonal distance of at least 2 m. The global responses to COVID-19, ranging from lockdowns to social distancing guidelines, offer a rare opportunity for assessing, through real-world events, the environmental scenarios conditioned by reductions in local air pollutant emissions. For this reason, scientists began looking into air pollution mitigation as a side-effect of the COVID-19 induced restrictions [1,2]. Clearer signs of PM2.5 and NO2 air quality improvements were seen at a city-scale rather than at a national-scale [3–5], suggesting that the reductions in air pollutant emissions were mainly concentrated in densely populated areas or

Int. J. Environ. Res. Public Health 2020, 17, 6208; doi:10.3390/ijerph17176208 www.mdpi.com/journal/ijerph Int. J. Environ. Res. Public Health 2020, 17, 6208 2 of 12 areas with large industrial complexes. Large cities around the world reported considerable reductions in air pollutant emissions, particularly from traffic and industrial activities [6–8]. Reductions in primary air pollutant emissions under COVID-19 led to improved air quality in terms of decreased concentrations of carbon monoxide (CO), sulfur dioxide (SO2), nitrogen dioxide (NO2), and particulate matter with a diameter of 2.5 µm or less (PM2.5) in Brazil [9,10], China [11–14], India [15], and Italy [16]. Similar reductions in PM2.5 concentration in three Asian cities under COVID-19 and its positive effects on human health are reported in [17]. Adversely, such reductions may have contributed to the elevated ozone (O3) concentrations in Wuhan and cities in Southern Europe [18]. On the other hand, social restrictions might not lead to a significant difference in air quality in cities with low baseline concentrations of pollutants such as New York [19]. Owing to the apparent diversity in the air quality and COVID-19 situations among different regions, a specific case study that takes into account the unique conditions in each city can be helpful. Seoul is the capital of South Korea with a population of approximately 10 million. Between 30 January to 29 March 2020, the cumulative total number of COVID-19 cases in Seoul rose from four to 410, over a 100-fold increase, in the span of 60 days. It is probable that, faced with the nation-wide rise in the number of new cases, the residents of Seoul cooperated with social distancing guidelines, particularly following the government announcement on 29 February 2020. Such a decline in human activity may have led to reductions in local air pollutant emissions. Note that this is not the first time Seoul has experienced an improvement in air quality in terms of decreased concentrations of primary pollutants owing to emission reductions from societal causes. During the economic recession 3 of 1998, Seoul saw a 9 µg m− (13%) decrease in annual PM10 concentration compared to 1997 [20]. According to the National Air Pollutants Emission Service of Korea [21], as of 2017, the annual PM2.5 and PM10 emissions in Seoul, excluding resuspension sources, are largely comprised of construction and on-road traffic sectors. Therefore, the restrictions to human activity in Seoul by social distancing can still have a significant impact on the overall air quality in the city. This study is a report on the air quality change in Seoul during the period of COVID-19 social distancing, with emphasis on PM2.5 concentration. Using air quality monitoring data, we first make comparisons between the periods before and after the start of social distancing in 2020, as well as comparisons against the climatological air quality in the past 5 years, and discuss the possible contributors to such changes in air quality.

2. Data and Methods In this study, we recognize 29 February 2020 as the start of social distancing based on the government announcement made on that date. Accordingly, we deﬁne the periods of 30 days before and after the start of social distancing as pre-SD (from 30 January to 28 February 2020) and SD (from 29 February to 29 March 2020), respectively. Regarding the data from the past 5 years from 2015 to 2019, the same dates were compared to the pre-SD of 2020. For comparisons against SD of 2020, the 30 days from 1 to 30 March in the years 2015, 2017, 2018, and 2019 and those from 29 February to 29 March in the leap year (2016) were used. The hourly-averaged PM2.5, PM10, NO2,O3, CO, and SO2 concentrations measured at 25 air quality monitoring stations (AQMSs) located in each district of Seoul were provided by Korea Environment Corporation [22]. Figure1 shows the locations of the AQMSs and terrain height of Seoul and surrounding areas. Note that these AQMSs are distinguished from roadside monitoring stations, and their locations tend to avoid being too close to major avenues and highways to capture representative measurements of urban air quality in Seoul. From the monitoring station data for the analysis periods in 2015–2020, a total of 1,263,418 pollutant concentration measurements were used in this study. This excludes the 32,582 missing values, which account for 2.5% of the total data. Int. J. Environ. Res. Public Health 2020, 17, 6208 3 of 12 Int. J. Environ. Res. Public Health 2020, 17, x FOR PEER REVIEW 3 of 13

Figure 1. Terrain height of Seoul and surrounding areas. The black and blue solid lines represent the city limit of Seoul and the Han river, respectively.respectively. TheThe 2525 AQMSsAQMSs areare indicatedindicated byby thethe blackblack dots.dots.

TraTrafficffic volumevolume data data were were acquired acquired from from the the Seoul Seoul Transport Transport Operation Operation and and Information Information Service Service [23] to[23] estimate to estimate the reduction the reduction in human in human activity activity due to duesocial to distancing. social distancing. Out of the Out 145 of vehicle-counting the 145 vehicle- sites,counting 53 sites sites, without 53 sites missing without values missing during values the during 60 day the periods 60 day corresponding periods corresponding to pre-SD andto pre SD-SD in 2020and SD were in 2020 selected were in selected this study. in this study. To examine examine air air quality quality in in wider wider areas are aroundas around Seoul Seoul and the and transport the transport of air pollutants, of air pollutants, the aerosol the opticalaerosol depthoptical (AOD) depth from(AOD) the from MERRA-2 the MERRA data- [224 data] produced [24] produced by the Nationalby the National Aeronautics Aeronautics and Space and AdministrationSpace Administration (NASA) (NASA) [25] were [25] analyzed. were analyzed. The spatial The and spatial temporal and resolutionstemporal resolutions of the MERRA-2 of the dataMERRA are- 0.52 data0.625 are 0.5°and × 30.625° h, respectively. and 3 h, respectively. In addition, In the addition, zonal and the meridional zonal and wind meridional components wind ◦ × ◦ andcomponents geopotential and height geopotential at the 900 height hPa level at the from 900 the hPa ERA5 level data from operated the ERA5 by the data European operated Centre by the for Medium-RangeEuropean Centre Weather for Medium Forecasts-Range (ECMWF) Weather [26 ] Forecasts were used (ECMWF) to investigate [26] meteorologicalwere used to in conditionsvestigate beforemeteorological and after the conditions start of social before distancing. and after We the analyzed start the of meteorologicalsocial distancing. variables We analyzeat the 900d hPathe levelmeteorological to illustrate variables the transport at the of900 air hPa pollutants level to by illustrate air flows the in transport the lower of atmosphere. air pollutants The by ERA5 air flows data havein the a lower spatial atmosphere. resolution of The 0.25 ERA50.25 dataand ha ave temporal a spatial resolution resolution of of 1 h.0.25° × 0.25° and a temporal ◦ × ◦ resolution of 1 h. 3. Results and Discussion 3. Results and Discussion The daily mean PM2.5 concentration averaged over all stations in Seoul decreased by 10.4% from pre-SDThe to daily SD (Table mean1 PM). This2.5 concentration is a notable averaged decrease comparedover all stations to the in 23.7% Seoul increase decreased on by average 10.4% overfrom thepre- sameSD to timeSD (Table period 1). in This the is years a notable 2015 decrease through compared to 2019. Figure to the2 23a shows.7% increase that, fromon average pre-SD over to SD, the thesame daily time mean period PM in2.5 theconcentration years 2015 through experienced to 2019. decreases Figure in 2a the shows maximum that, from and thepre upper-SD to quartile.SD, the Thedaily di meanfference PM between2.5 concentration the upper experienced and lower quartilesdecreases also in the decreased, maximum indicating and the aupper smaller quartile. variability The duringdifference SD ofbetween 2020. On the the upper contrary, and thelower upper quartiles quartiles also of decreased, the daily mean indicating PM2.5 concentrations a smaller variability in the previousduring SD years of 2020. from On 2016 the and contrary, onward the increased upper quartiles over the of same the daily time mean periods. PM2.5 concentrations in the previousThe reductionyears from in 2016 variability and onward appears increased to be related over the to same the number time periods. of days with pollution levels reachingThe reduction “high” in Figurein variability2b. The appears “high” daysto be inrelated this study to the are number defined of asdays the with days pollution with the dailylevels 3 meanreaching PM “high”2.5 concentration in Figure exceeding2b. The “high” 35.0 µ daysg m− inin this alignment study are with defined the guidelines as the days set with by the the Korean daily government.mean PM2.5 concentration Out of the 30 exceeding days of pre-SD 35.0 μg and m SD,−3 in the alignment number ofwith “high” the guidelines days went set down by the from Korean 9 to 3, agovernment. dramatic decrease Out of the not 30 found days in of the pre previous-SD and SD, years. the In number fact, the of years“high” since days 2016 went through down from 2019 saw9 to increasing3, a dramatic trends decrease in the not number found of in “high” the previous days over years. the In same fact, time the periodyears since (Figure 20162b). through This appears 2019 saw to beincreasing a clear indication trends in the that number social distancing of “high” exerteddays over a downward the same time pressure period on (Figure air pollution 2b). This levels. appears to beBy a clear comparison, indication there that havesocial been distancing reports exerted of substantial a downward decreases pressure in PM on2.5 airconcentrations pollution levels. due to reducedBy comparison, human activity there have related been to reports COVID-19 of substantial (e.g., Collivignarelli decreases in et PM al. [2.516 concentrations]; Kerimray et due al. [ 7to]; Lireduced et al. [human13]; Lian activity et al. [ 4related]; Sharma to COVID et al. [-1519]; (e.g., Zangari Collivignarelli et al. [19]). et For al. example, [16]; Kerimray Wuhan, et al. which [7]; Li was et subjectedal. [13]; Lian to aet series al. [4]; of Sharma strict lockdown et al. [15]; orders, Zangari saw et al. a 36.9%[19]). For decrease example, in PM Wuhan,2.5 concentration which was [ subjected4]. Milan, anotherto a series city of hard-hitstrict lockdown by COVID-19, orders, experiencedsaw a 36.9% adecrease decrease in of PM 47.4%2.5 concentration in PM2.5 concentration [4]. Milan, another during city hard-hit by COVID-19, experienced a decrease of 47.4% in PM2.5 concentration during a lockdown [16]. Unlike other major cities around the world, a complete city-wide lockdown was never imposed

Int. J. Environ. Res. Public Health 2020, 17, 6208 4 of 12 a lockdown [16]. Unlike other major cities around the world, a complete city-wide lockdown was never imposed in Seoul, and social distancing was announced only as a recommendation. It is surprising that even this comparatively weaker forcing toward reduced human activity might have resulted in such noticeable improvements in the PM2.5 air quality in Seoul.

Table 1. Thirty day mean concentrations of air pollutants for pre-SD and SD in 2020 and the corresponding periods in 2015–2019, the relative changes from pre-SD to SD in 2020 and the corresponding periods in 2015–2019, and the relative changes from the periods corresponding to SD in 2015–2019 to SD in 2020.

Relative Change (%) Relative Change (%) for SD * Pollutant/Year Pre-SD * SD * Pre-SD * vs. SD * 2015–2019 vs. 2020 3 PM2.5 (µg m− ) 2020 27.6 24.7 10.4 − 31.2 2015–2019 29.1 35.9 23.7 − 3 PM10 (µg m− ) 2020 40.4 44.1 9.1 29.9 2015–2019 56.0 62.9 12.3 −

NO2 (ppbv) 2020 32.0 26.7 16.4 − 26.9 2015–2019 34.9 36.5 4.7 −

O3 (ppbv) 2020 18.0 26.4 47.0 5.7 2015–2019 17.5 25.0 42.5 CO (ppbv) 2020 588.3 489.1 16.9 − 16.1 2015–2019 633.4 583.3 7.9 − − SO2 (ppbv) 2020 3.3 3.2 3.3 − 41.5 2015–2019 5.6 5.4 2.5 − − * For 2015–2019, the periods corresponding to pre-SD and SD in 2020 were used for the calculation.

Figure3 shows that major peaks in PM 2.5 concentration appeared three times during each 30 day period: 2, 14, and 21 February during pre-SD, and 8, 18, and 24 March during SD. These peak PM2.5 concentrations can be attributed to factors other than local emission such as meteorological conditions associated with stagnant air in the region or the long-range transport of pollutants. That the PM2.5 concentration peaks reached higher levels pre-SD than during SD, with smaller fluctuations during SD, indicates that those other factors had a stronger influence in pre-SD than in SD. To take a closer look at the effects of social distancing on PM2.5 concentration, workday–holiday comparisons are also considered. In this study, weekends and national holidays are classified as the holidays and the rest of the days are classified as the workdays. The daily mean PM2.5 concentrations 3 3 averaged over the workdays were 25.0 µg m− and 25.5 µg m− in pre-SD and SD, respectively. 3 3 In contrast, the holiday average saw a dramatic decrease from 34.8 µg m− in pre-SD to 23.2 µg m− in SD. Even considering the possibility that the results may have been skewed by a few high pollution events that coincidentally happened to occur on holidays during pre-SD, there were sizable decreases in holiday PM2.5 concentrations, which must at least partly reflect the suppressed emission due to social distancing during SD. Traffic data show an 8.4% decrease in the number of vehicles per day from pre-SD to SD on holidays compared to a 3.0% decrease on workdays. The 30 day mean diurnal variations in the PM2.5 concentrations in Figure4 also confirm that, overall, the PM2.5 concentrations were lower during SD than pre-SD. The maximum concentrations occur at the same time of the day (11 LST) for both pre-SD and SD, which can be attributed to the increased local emissions during the morning rush hour. The difference in PM2.5 concentration between SD and pre-SD reaches its peak at 13 LST during the hours when human activity levels are high, but is smaller during nighttime when human activity levels are thought to be low, even in the absence of Int. J. Environ. Res. Public Health 2020, 17, 6208 5 of 12

Int. J. Environ. Res. Public Health 2020, 17, x FOR PEER REVIEW 5 of 13 social distancing. This might be an indication that, compared to nighttime, social distancing had absence of social distancing. This might be an indication that, compared to nighttime, social a stronger inﬂuence on PM2.5 concentrations during high human activity hours. distancing had a stronger influence on PM2.5 concentrations during high human activity hours.

Figure 2. (a) Box plot of the daily mean PM2.5 concentrations in the 30 days of pre-SD in 2020 and the Figure 2. (a) Box plot of the daily mean PM2.5 concentrations in the 30 days of pre-SD in 2020 and the corresponding periods in 2015–2019 (blue) and SD in 2020 and the corresponding periods in 2015– corresponding periods in 2015–2019 (blue) and SD in 2020 and the corresponding periods in 2015–2019 2019 (red). The box indicates the lower and upper quartiles. The center line and black dot inside each (red).box The represent box indicates the median the lowerand 30 andday uppermean values, quartiles. respectively. The center The linewhiskers and blackabove dotand insidebelow each each box representbox indicate the median the maximum and 30 day and mean minimum values, values, respectively. respectively. The ( whiskersb) Numbers above of days and with below “high” each box

indicatelevels the of maximum PM2.5 concentration and minimum (>35.0 μg values, m−3) out respectively. of the 30 days ( bof) pre Numbers-SD in 2020 of daysand the with corresponding “high” levels of 3 PM2.5periodsconcentration in 2015–2019 (>35.0 (blue)µg mand− SD) out in of2020 the and 30 the days corresponding of pre-SD in periods 2020 and in 2015 the corresponding–2019 (red). periods in 2015–2019 (blue) and SD in 2020 and the corresponding periods in 2015–2019 (red). Although the lower levels of PM2.5 concentration during SD can be explained in part by the Althoughreduced emissions the lower due levelsto social of dis PMtancing,2.5 concentration there may also during be other SD complexities can be explained involved in, such part as by the reducedvariations emissions in local due weather to social conditions distancing, and the there long- mayrange also transport be other of PM complexities2.5. Further analyses involved, of these such as variations may help explain the unexpected results exhibited by the diurnal variations such as the 4 variations in local weather conditions and the long-range transport of PM2.5. Further analyses of these variationsh long maywindow help in explainearly morning the unexpected from 2 LST to results 6 LST exhibitedwhen SD had by higher the diurnal PM2.5 concentrations variations such than as the pre-SD. The diurnal variation in PM2.5 concentration for pre-SD also exhibits a somewhat unusual 4 h long window in early morning from 2 LST to 6 LST when SD had higher PM concentrations downward-sloped tail end, where the variation is typically known to have an uptick2.5 as midnight thanapproaches. pre-SD. The diurnal variation in PM2.5 concentration for pre-SD also exhibits a somewhat unusual downward-sloped tail end, where the variation is typically known to have an uptick as midnight approaches.

Int. J. Environ. Res. Public Health 2020, 17, 6208 6 of 12 Int. J. Environ. Res. Public Health 2020, 17, x FOR PEER REVIEW 6 of 13 Int. J. Environ. Res. Public Health 2020, 17, x FOR PEER REVIEW 6 of 13

FigureFigure 3. 3. TimeTime series series of of the the daily daily mean mean PM PM2.5 2.5concentrationconcentration during during pre- pre-SDSD and and SD inSD 2020. in 2020. The TheFigurehorizont horizontal 3.al Time dashed dashed series lines lines of indicate the indicate daily the themean 30 30day day PM mean mean2.5 concentration values values for for pre pre-SD during-SD (blue) (blue) pre and-SD and SD and SD (red), (red), SD inrespectively. respectively. 2020. The 3 −3 ThehorizontThe greengreenal horizontal dashedhorizontal lines solid solid indicate line line marks marks the 30 the theday threshold threshold mean values level level of for PMof pre PM2.5-SD2.5concentration concentration (blue) and (35.0SD (35.0 (red),µg μg m respectively.− m, “high”, “high” if 3 −3 −3 >Theif35.0 >35.0 greenµg μg m horizontal −m).). The The period period solid ofline of social social marks distancing distancing the threshold (SD) (SD) is level is indicated indicated of PM by2.5 by concentration the the shaded shaded area. area. (35.0 μg m , “high” if >35.0 μg m−3). The period of social distancing (SD) is indicated by the shaded area.

Figure 4. Diurnal variations of the 30 day mean PM2.5 concentrations averaged over all stations for

Figurepre-SD 4.4. and DiurnalDiurnal SD. variationsvariations of thethe 3030 dayday meanmean PMPM2.52.5 concentrations averaged over all stations for pre-SDpre-SD andand SD.SD. The spatial distributions of PM2.5 concentrations in Seoul show that there was an across-the- boardThe decrease spatialspatial distributions fromdistributions pre-SD ofto of PMSD PM2.5 in2.5 concentrationsalmost concentrations all stations in Seoulin (Figure Seoul show show5). that In particular,that there there was anwasthe across-the-board locationsan across of-the the- decreaseboardtop 10 decrease monitoring from pre-SD from stations pre to- SD in to in terms SD almost in of almost allthe stations magnitude all stations (Figure of (Figure PM5).2.5 In concentration5). particular, In particular, the decrease locations the locations are of clustered the of top the 10topamong monitoring 10 monitoring the northern stations stations and in termscentral in terms of stations, the of magnitude the including magnitude of PM the 2.5of station concentrationPM2.5 inconcentration Jung- decreasegu, the decrease traditional are clustered are city clustered among center theamongof Seoul. northern the Empirical northern and central correlationand stations,central coefficientsstations, including including the(R) stationbetween the in station Jung-gu,PM2.5 inconcentrations Jung the traditional-gu, the attraditional citythe Jung center -citygu of stationcenter Seoul. Empiricalofand Seoul. other Empirical correlation stations correlation were coeffi calculatedcients coefficients (R) betweenfor quantitative (R) PM be2.5tweenconcentrations analysis. PM2.5 concentrations From at pre the- Jung-guSD at to the SD, station Jung the - numbergu and station other of stationsandstations other were with stations calculated which were the for calculated Jung quantitative-gu station for quantitative analysis. had empirical From analysis. pre-SD correlation From to SD, pre thecoefficients-SD number to SD, ofexceeding the stations number with0.950 of whichstationsdecreased the with Jung-gu from which 19 stationto the 4. Although Jung had-gu empirical station other factors, correlation had empirical such coe as ffi meteorological correlationcients exceeding coefficients conditions, 0.950 exceeding decreased can affect 0.950from the 19decreasedempirical to 4. Although correlation from 1 other9 to coefficients, 4. factors, Although such the other as change meteorological factors, in the such empirical conditions, as meteorological correlation can affect coefficient conditions, the empirical from can correlation affectpre-SD the to coeempiricalSDffi impliescients, correlation that the, change to some coefficients, in extent the empirical, there the werechange correlation differences in the coe empiricalffi amongcient from the correlation local pre-SDities tocoefficient in SD responding implies from that, to pre the to- SD somesocial to extent,SDdistancing implies there thatguidelines were, to disomeff erencesduring extent SD among, there, leading were the localitiesto differences the spatial in respondingamong disparities the local to in the theities social changes in responding distancing in local to emission. guidelines the social duringdistancing SD, guidelines leading to theduring spatial SD, disparities leading to inthe the spatial changes disparities in local in emission. the changes in local emission.

Int. J. Environ. Res. Public Health 2020, 17, 6208 7 of 12

Int. J. Environ. Res. Public Health 2020, 17, x FOR PEER REVIEW 7 of 13

Figure 5. Spatial variations in the 30 day mean PM2.5 concentration for (a) pre-SD and (b) SD, and (c) Figure 5. Spatialrelative variationschange in the 30 in day the mean 30 PM day2.5 concentration mean PM from2.5 concentration pre-SD to SD. The AQMS for (a in) Jung pre-SD-gu is and (b) SD, and (c) relativeindicated change by J. in the 30 day mean PM2.5 concentration from pre-SD to SD. The AQMS in Jung-gu is indicated by J. Beyond PM2.5, the concentrations of other pollutants, PM10, NO2, O3, CO, and SO2, were examined for changes from pre-SD to SD. The daily mean concentrations of CO (Figure 6d), NO2 (Figure 6b), Beyond PMand, 2.5to ,a thelesser concentrations extent, SO2 (Figure of 6e) other decreased pollutants, on average PM by10 16.9%,, NO 16.4%,2,O3 ,and CO, 3.3%, and respectively SO2, were examined for changes from(Table pre-SD1). Furthermore, to SD. clear The downward daily mean trends are concentrations seen in the time series of CO for NO (Figure2 and CO6d), butNO not so2 (Figure6b), and, to a lessermuch extent, for SO2 SO. It is speculated(Figure6 thate), decreased unlike complete on lockdowns, average bysocial 16.9%, distancing 16.4%, in Seoul and mainly 3.3%, had respectively effects on commuting2 and daytime social activities rather than industries, leading to a decreased level (Table1). Furthermore,of traffic. Note clear that NO downward2 and CO are trends known areto have seen direct in thelinks timeto vehicle series emission for NOs, whereas2 and SO CO2 but not so much for SO2emission. It is speculated is largely associated that, unlikewith industrial complete activities lockdowns, [4]. According social to the distancing National Air Pollutants in Seoul mainly had eﬀects on commutingEmission Service and of daytime Korea [21], social Seoul does activities not have rather a significant than local industries, source of SO leading2. That the to relative a decreased level decreases in NO2 and CO were larger than that in SO2, therefore, indicates reductions in local of traﬃc. Noteemissions that NO due 2 toand decreased CO are traffic known volume to as have more people direct began links following to vehicle the social emissions, distancing whereas SO2 emission is largely associated with industrial activities [4]. According to the National Air Pollutants

Emission Service of Korea [21], Seoul does not have a significant local source of SO2. That the relative decreases in NO2 and CO were larger than that in SO2, therefore, indicates reductions in local emissions due to decreased traffic volume as more people began following the social distancing guidelines over time since the initial announcement on 29 February 2020. The 30 day mean traffic volume in Seoul showed a 5.8% decrease from pre-SD to SD. The large increase in O3 concentration (47.0%) is also expected due to the less active O3 titration by NO as well as seasonal effects with the days getting longer into spring. This was also observed in other cities that experienced reduced levels of human activity due to COVID-19 (e.g., Collivignarelli et al. [16]; Kerimray et al. [7]; Li et al. [13]; Lian et al. [4]; Sicard et al. [18]). Int. J. Environ. Res. Public Health 2020, 17, x FOR PEER REVIEW 8 of 13

guidelines over time since the initial announcement on 29 February 2020. The 30 day mean traffic volume in Seoul showed a 5.8% decrease from pre-SD to SD. The large increase in O3 concentration (47.0%) is also expected due to the less active O3 titration by NO as well as seasonal effects with the days getting longer into spring. This was also observed in other cities that experienced reduced levels Int. J. Environ. Res. Public Health 2020, 17, 6208 8 of 12 of human activity due to COVID-19 (e.g., Collivignarelli et al. [16]; Kerimray et al. [7]; Li et al. [13]; Lian et al. [4]; Sicard et al. [18]).

Figure 6. Time series of the daily mean values of (a) PM10, (b) NO2, (c) O3, (d) CO, and (e) SO2 Figure 6. concentrationsTime series during of the pre daily-SD and mean SD in values 2020. The of horizontal (a) PM10 dashed,(b) NOlines2 indicate,(c)O the3,( 30d) day CO, mean and (e) SO2 concentrationsvalues duringfor pre-SD pre-SD (blue) and and SD SD (red), in 2020.respectively. The horizontal The period of dashed social distancing lines indicate (SD) is indicated the 30 day mean values for pre-SDby the shaded (blue) area. and SD (red), respectively. The period of social distancing (SD) is indicated by the shaded area. PM2.5 shows a high empirical correlation with CO (R = 0.754). Since CO is more closely linked to local air pollutant emissions, this suggests that the decrease in PM2.5 concentration during SD can PM shows a high empirical correlation with CO (R = 0.754). Since CO is more closely linked 2.5mainly be attributed to changes in local emissions [27,28]. The time series for PM10 in Figure 6a shows to local aira pollutanthigh empirical emissions, correlation this (R = suggests0.893) with thatthat of the PM decrease2.5 in Figure in 3. PMSimilar2.5 toconcentration PM2.5, the daily duringmean SD can mainly bePM attributed10 concentrations to changes exhibit in a localsmaller emissions variation during [27,28 SD]. than The during time series pre-SD. for Although PM10 inthere Figure is an 6a shows a high empirical correlation (R = 0.893) with that of PM2.5 in Figure3. Similar to PM 2.5, the daily mean PM10 concentrations exhibit a smaller variation during SD than during pre-SD. Although there is an increase (9.1%; Table1) in the 30 day mean PM 10 concentration, this is in the context of an even larger average increase (12.3%; Table1) over the same time periods in the past 5 years. The ratio between the 30 day mean PM2.5 and PM10 concentrations decreased from 0.68 in pre-SD to 0.56 in SD. Although the ratio between PM2.5 and PM10 concentrations is affected by many factors, such as weather conditions and long-range transport of particulate matter, local emissions of anthropogenic pollutants tend to raise this ratio [29–31]. Therefore, the decrease in the ratio between PM2.5 and PM10 concentrations from pre-SD to SD can be explained in part by reduced anthropogenic emissions due to social distancing. Since the change in the air quality of Seoul may be associated with the air quality of its neighboring regions, the AOD field over the broad region encompassing the Korean Peninsula was analyzed. Note that the PM2.5 concentrations at the surface level may not have direct relations with the AOD distributions, which is a column integrated index that also includes information about aerosols at Int. J. Environ. Res. Public Health 2020, 17, 6208 9 of 12 the upper levels as well as at the surface level. The 30 day mean AOD and the synoptic patterns at the 900 hPa level for pre-SD and SD are shown in Figure7. The 30 day mean AOD for pre-SD has a steep eastward decline toward the Korean Peninsula. The AOD over Seoul (0.368) is only slightly higher than those over the other regions of South Korea along the same longitude despite Seoul being much more urbanized than other regions. This implies that the AOD over Seoul during pre-SD is partially explained by the transport of the aerosols with external origins. Figure7b shows the average northwesterly flow over the Yellow Sea transporting aerosols from the west during pre-SD. During SD, the 30 day mean AOD decreased over parts of China, the Yellow Sea, and the west coast of the Korean Peninsula (~33–41◦ N, ~113–128◦ E) (Figure7c,e). Across these regions, the circulation di fference from pre-SD is anti-cyclonic, and there exists a small relative divergence (Figure7f), suggesting that the eastward aerosol transport to Seoul may have been slightly reduced during SD. The 30 day mean AOD over Seoul decreased by 16.5% from pre-SD to SD. According to the Korea Ministry of Environment (2019), the annual contribution to the PM2.5 concentration in Seoul during 2013–2017 was comprised of domestic sources (45%) and transboundary transport from China (34%), Japan (1%), and other countries (20%) [32]. In another study, the long-range transport of pollutants was estimated to have raised PM10, NO2, CO, and SO2 concentrations from 2001 to 2014 in Seoul by 21%, 9%, 6%, and 14%, respectively [33]. Considering the above analyses, our results imply that the reduction in the eastward transport of pollutants to Seoul, in addition to the reduced local emissions in Seoul, contributed to the decrease in PM2.5 concentration during SD. Int. J. Environ. Res. Public Health 2020, 17, x FOR PEER REVIEW 10 of 13

Figure 7. The 30 day mean AOD for (a) pre-SD and (c) SD, and (e) relative change in the 30 day mean Figure 7. The 30 day mean AOD for (a) pre-SD and (c) SD, and (e) relative change in the 30 day mean AOD from pre-SD to SD. The 30 day mean geopotential height and wind vector at the 900 hPa level AOD fromfor pre-SD (b) pre to-SD SD. and The(d) SD, 30 and day (f) mean differen geopotentialces in the 30 day height mean geopotential and wind vectorheight and at thewind 900 vector hPa level for (b) pre-SDat and the 900 (d) hPa SD, level and from (f) dipreﬀ-SDerences to SD. inThe the cross 30 symbol day mean indicates geopotential the location of height Seoul. and wind vector at the 900 hPa level from pre-SD to SD. The cross symbol indicates the location of Seoul. 4. Conclusions

PM2.5 and the five other air pollutant concentrations, measured in Seoul over the 30 days before and after the start of social distancing, were analyzed to illustrate the impact of COVID-19 social distancing on the air quality of Seoul. The 30 day mean PM2.5 concentration averaged over all stations in Seoul decreased by 10.4% since the start of social distancing, while it increased by an average of 23.7% over the same time periods in the years 2015–2019. Between pre-SD and SD, there was a dramatic drop in the number of “high” days exceeding the PM2.5 concentration of 35.0 μg m−3 from 9 to 3 days out 30, while this number went up over the same time periods every year from 2016 to 2019. This study shows that social distancing, a much weaker response to COVID-19 than a lockdown, can lead to decreases in air pollutant concentrations in megacities. The diurnal variations in PM2.5 concentration during pre-SD and SD confirm that PM2.5 concentration especially decreased during high human activity hours. The spatial distribution of the PM2.5 concentration difference between pre- SD and SD illustrates that the decrease in PM2.5 concentration occurred throughout the whole city, except for a few stations. Other pollutants such as NO2, CO, and SO2 also experienced decreases in concentration, which are partly attributed to the reduced local emissions during SD. On the other hand, O3 concentration increased by 47.0% because of the less active O3 titration by NO. The

Int. J. Environ. Res. Public Health 2020, 17, 6208 10 of 12

4. Conclusions

PM2.5 and the five other air pollutant concentrations, measured in Seoul over the 30 days before and after the start of social distancing, were analyzed to illustrate the impact of COVID-19 social distancing on the air quality of Seoul. The 30 day mean PM2.5 concentration averaged over all stations in Seoul decreased by 10.4% since the start of social distancing, while it increased by an average of 23.7% over the same time periods in the years 2015–2019. Between pre-SD and SD, there was a dramatic drop 3 in the number of “high” days exceeding the PM2.5 concentration of 35.0 µg m− from 9 to 3 days out 30, while this number went up over the same time periods every year from 2016 to 2019. This study shows that social distancing, a much weaker response to COVID-19 than a lockdown, can lead to decreases in air pollutant concentrations in megacities. The diurnal variations in PM2.5 concentration during pre-SD and SD confirm that PM2.5 concentration especially decreased during high human activity hours. The spatial distribution of the PM2.5 concentration difference between pre-SD and SD illustrates that the decrease in PM2.5 concentration occurred throughout the whole city, except for a few stations. Other pollutants such as NO2, CO, and SO2 also experienced decreases in concentration, which are partly attributed to the reduced local emissions during SD. On the other hand, O3 concentration increased by 47.0% because of the less active O3 titration by NO. The reductions in air pollutants from pre-SD to SD are confirmed again through the analysis of the AOD. The AOD over Seoul decreased by 16.5% from pre-SD to SD, and this decrease can be attributed to the change in synoptic circulation and long-range transport during SD and the reductions in local emissions under social distancing. Further investigation is needed to better understand the spatial variability of the PM2.5 concentration decrease, which may have been affected by a variety of influencing factors, such as topography and sociological characteristics. In addition to the impact of emission reduction on air quality, the impact of meteorological conditions also deserves an investigation. Estimating the relative contributions of local emissions and long-range transport to the air quality change under social distancing is a challenging problem because of the difficulties in properly separating the two contributors using measurement data alone. Numerical simulations of air quality using a high-resolution chemistry-coupled atmospheric model can help estimate the relative contributions of local emissions and long-range transport, which deserves investigation.

Author Contributions: Conceptualization, K.P. and J.-J.B.; formal analysis, B.-S.H., K.P., K.-H.K., and H.-G.J.; validation, K.P., J.-W.K., and S.M.; investigation, B.-S.H., K.P., K.-H.K., H.-G.J., S.M., and J.-W.K.; data curation, B.-S.H., K.P., and J.-W.K.; writing—original draft preparation, B.-S.H., K.-H.K., S.-B.P., H.-G.J., and S.M.; writing—review and editing, K.-H.K., S.-B.P.,S.M., and J.-J.B.; visualization, B.-S.H., S.-B.P.,and H.-G.J.; supervision, K.-H.K. and J.-J.B. All authors have read and agreed to the published version of the manuscript. Funding: This work was partly supported by the National Research Foundation of Korea (NRF-2019R1A6A1A10073437). Acknowledgments: The authors are grateful to the two anonymous reviewers for providing valuable comments on this study. The authors would like to thank Korea Environment Corporation for providing the air pollutant measurement data from the air quality monitoring stations and the Seoul Transport Operation and Information Service for providing the traffic volume data. The authors also acknowledge the National Aeronautics and Space Administration and the European Centre for Medium-Range Weather Forecasts for making the satellite reanalysis data and meteorological data freely available. Conflicts of Interest: The authors declare no conflict of interest.

References

1. Arora, S.; Bhaukhandi, K.D.; Mishra, P.K. Coronavirus lockdown helped the environment to bounce back. Sci. Total Environ. 2020, 742, 140573. [CrossRef] 2. Zambrano-Monserrate, M.A.; Ruano, M.A.; Sanchez-Alcalde, L. Indirect eﬀects of COVID-19 on the environment. Sci. Total Environ. 2020, 728, 138813. [CrossRef]

3. Chauhan, A.; Singh, R.P. Decline in PM2.5 concentrations over major cities around the world associated with COVID-19. Environ. Res. 2020, 187, 109634. [CrossRef] Int. J. Environ. Res. Public Health 2020, 17, 6208 11 of 12

4. Lian, X.; Huang, J.; Huang, R.; Liu, C.; Wang, L.; Zhang, T. Impact of city lockdown on the air quality of COVID-19-hit of Wuhan city. Sci. Total Environ. 2020, 742, 140556. [CrossRef]

5. Rodríguez-Urrego, D.; Rodríguez-Urrego, L. Air quality during COVID-19: PM2.5 analysis in the 50 most polluted capital cities in the world. Environ. Pollut. 2020, 266, 115042. [CrossRef] 6. Adams, M.D. Air pollution in Ontario, Canada during the COVID-19 State of Emergency. Sci. Total Environ. 2020, 742, 140516. [CrossRef] 7. Kerimray, A.; Baimatova, N.; Ibragimova, O.P.; Bukenov, B.; Kenessov, B.; Plotitsyn, P.; Karaca, F. Assessing air quality changes in large cities during COVID-19 lockdowns: The impacts of traffic-free urban conditions in Almaty, Kazakhstan. Sci. Total Environ. 2020, 730, 139179. [CrossRef] 8. Le, T.; Wang, Y.; Liu, L.; Yang, J.; Yung, Y.L.; Li, G.; Seinfeld, J.H. Unexpected air pollution with marked emission reductions during the COVID-19 outbreak in China. Science 2020, 369, 702–706. [CrossRef] 9. Dantas, G.; Siciliano, B.; França, B.B.; da Silva, C.M.; Arbilla, G. The impact of COVID-19 partial lockdown on the air quality of the city of Rio de Janeiro, Brazil. Sci. Total Environ. 2020, 729, 139085. [CrossRef] 10. Nakada, L.Y.K.; Urban, R.C. COVID-19 pandemic: Impacts on the air quality during the partial lockdown in São Paulo state, Brazil. Sci. Total Environ. 2020, 730, 139087. [CrossRef] 11. Bao, R.; Zhang, A. Does lockdown reduce air pollution? Evidence from 44 cities in northern China. Sci. Total Environ. 2020, 731, 139052. [CrossRef] 12. He, G.; Pan, Y.; Tanaka, T. The short-term impacts of COVID-19 lockdown on urban air pollution in China. Nat. Sustain. 2020.[CrossRef] 13. Li, L.; Li, Q.; Huang, L.; Wang, Q.; Zhu, A.; Xu, J.; Liu, Z.; Li, H.; Shi, L.; Li, R.; et al. Air quality changes during the COVID-19 lockdown over the Yangtze River Delta Region: An insight into the impact of human activity pattern changes on air pollution variation. Sci. Total Environ. 2020, 732, 139282. [CrossRef] 14. Wang, Y.; Yuan, Y.; Wang, Q.; Liu, C.G.; Zhi, Q.; Cao, J. Changes in air quality related to the control of coronavirus in China: Implications for traffic and industrial emissions. Sci. Total Environ. 2020, 731, 139133. [CrossRef] 15. Sharma, S.; Zhang, M.; Anshika; Gao, J.; Zhang, H.; Kota, S.H. Effect of restricted emissions during COVID-19 on air quality in India. Sci. Total Environ. 2020, 728, 138878. [CrossRef] 16. Collivignarelli, M.C.; Abbà, A.; Bertanza, G.; Pedrazzani, R.; Ricciardi, P.; Carnevale Miino, M. Lockdown for CoViD-2019 in Milan: What are the effects on air quality? Sci. Total Environ. 2020, 732, 139280. [CrossRef] 17. Ma, C.-J.; Kang, G.-U. Air quality variation in Wuhan, Daegu, and Tokyo during the explosive outbreak of COVID-19 and its health effects. Int. J. Envrion. Res. Public Health 2020, 17, 4119. [CrossRef] 18. Sicard, P.; De Marco, A.; Agathokleous, E.; Feng, Z.; Xu, X.; Paoletti, E.; Rodriguez, J.J.D.; Calatayud, V. Amplified ozone pollution in cities during the COVID-19 lockdown. Sci. Total Environ. 2020, 735, 139542. [CrossRef] 19. Zangari, S.; Hill, D.T.; Charette, A.T.; Mirowsky, J.E. Air quality changes in New York City during the COVID-19 pandemic. Sci. Total Environ. 2020, 742, 140496. [CrossRef] 20. Sharma, A.P.; Kim, K.-H.; Ahn, J.-W.; Shon, Z.-H.; Sohn, J.-R.; Lee, J.-H.; Ma, C.-J.; Brown, R.J.C. Ambient

particulate matter (PM10) concentrations in major urban areas of Korea during 1996–2010. Atmos. Pollut. Res. 2014, 5, 161–169. [CrossRef] 21. National Air Pollutants Emission Service of Korea. Emission Rate Inquiry. Available online: https: //airemiss.nier.go.kr (accessed on 22 July 2020). 22. Korea Environment Corporation. Airkorea. Available online: https://www.airkorea.or.kr (accessed on 19 June 2020). 23. Seoul Transport Operation and Information Service. Traﬃc Volume Information. Available online: https: //topis.seoul.go.kr (accessed on 19 June 2020). 24. Gelaro, R.; McCarty, W.; Suárez, M.J.; Todling, R.; Molod, A.; Takacs, L.; Randles, C.A.; Darmenov, A.; Bosilovich, M.G.; Reichle, R.; et al. The modern-era retrospective analysis for research and applications, version 2 (MERRA-2). J. Clim. 2017, 30, 5419–5454. [CrossRef] 25. National Aeronautics and Space Administration (NASA). Data Collections. Available online: https: //disc.gsfc.nasa.gov/datasets (accessed on 7 July 2020). 26. Copernicus Climate Change Service (C3S). ERA5: Fifth Generation of ECMWF Atmospheric Reanalysis of the Global Climate. Copernicus Climate Change Service Climate Data Store (CDS). Available online: https://cds.climate.copernicus.eu/cdsapp#!/home (accessed on 15 July 2020). Int. J. Environ. Res. Public Health 2020, 17, 6208 12 of 12

27. Nguyen, H.T.; Kim, K.-H.; Ma, C.-J.; Cho, S.-J.; Sohn, J.R. A dramatic shift in CO and CH4 levels at urban locations in Korea after the implementation of the Natural Gas Vehicle Supply (NGVS) program. Environ. Res. 2010, 110, 396–409. [CrossRef] 28. Kim, K.-H.; Sul, K.-H.; Szulejko, J.E.; Chambers, S.D.; Feng, X.; Lee, M.-H. Progress in the reduction of carbon monoxide levels in major urban areas in Korea. Environ. Pollut. 2015, 207, 420–428. [CrossRef] 29. Lee, M. An analysis on the concentration characteristics of PM2.5 in Seoul, Korea from 2005 to 2012. Asia-Pac. J. Atmos. Sci. 2014, 50, 585–594. [CrossRef] 30. Ahmed, E.; Shon, K.-H.; Song, Z.-H.; Song, S.-K. Long-term trend of airborne particulate matter in Seoul, Korea from 2004 to 2013. Atmos. Environ. 2015, 101, 125–133. [CrossRef] 31. Xu, G.; Jiao, L.; Zhang, B.; Zhao, S.; Yuan, M.; Gu, Y.; Liu, J.; Tang, X. Spatial and temporal variability of the

PM2.5/PM10 ratio in Wuhan, Central China. Aerosol Air Qual. Res. 2017, 17, 741–751. [CrossRef] 32. Korea Ministry of Environment (MoE). Summary Report of the 4th Stage (2013–2017) LTP Project. Available online: https://www.me.go.kr/home/file/readDownloadFile.do?fileId=184686&fileSeq=1 (accessed on 24 July 2020). 33. Jeong, U.; Kim, J.; Lee, H.; Lee, Y.G. Assessing the effect of long-range pollutant transportation on air quality in Seoul using the conditional potential source contribution function method. Atmos. Environ. 2020, 150, 33–44. [CrossRef]

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).