Air Pollutant Levels during the Large-scale Social Restriction Period and its Association with Case Fatality Rate of COVID-19

The COVID-19 outbreak has caused millions of deaths in all over the world since it was declared by the World Health Organization (WHO) as a pandemic in March 2020. To stop the deadly spread of the virus, many countries, including Indonesia, have applied the ‘Large-scale Social Restriction’ (LSSR) policy. Numerous studies have reported positive impacts of air quality due to this policy. However, in Indonesia, data on the impacts of LSSR on air quality are still sparse. Therefore, this study aims to analyze changes in air quality at before and during the LSSR periods in the South Sumatera Province, Indonesia using the satellite-based observations of particulate matter (PM10), sulfur dioxide (SO2), nitrogen dioxide (NO2) and carbon monoxide (CO). The results showed that the concentrations of the measured pollutants markedly declined during the LSSR period from the highest was SO2 (98.90%) and followed by NO2 (34.79%), CO (12.70%) and PM10 (11.54%), respectively. The emissions from biomass burning activities were expected as a major source of air pollutant during the LSSR. Furthermore, we found a positive association between PM10 and the case fatality rate of COVID-19 in the study area (r = 0.514, p < 0.05). Finally, this study concluded that the implementation of LSSR could reduce air pollutants concentration in the study area while a higher PM10 exposure could increase the risk of death from COVID-19. The output of the study can be used to arrange air quality management practice and COVID-19 transmission control in Indonesia.


A C C E P T E D M A N U S C R I P T
6 Tropospheric amount of nitrogen dioxide (NO2) retrieved as a daily with 0.25˚ × 0.25˚ 113 resolution and measured by the Ozone Monitoring Instrument (OMI) onboard the NASA's Aura 114 Satellite. Column amount of sulfur dioxide (SO2) retrieved as a daily with 0.25˚ × 0.25˚ resolution, 115 it was measured by the OMI, an instrument aboard the NASA's Aura Satellite. Carbon monoxide 116 (CO) mole fraction in air was measured daily by Atmospheric Infrared Sounder (AIRS) on NASA's 117 Aqua Satellite at 1˚ × 1˚ resolution. The AIRS applied 3D analysis of the atmospheric column 118 along with trace gases, cloud and surface properties. Overall, the data were compared between 119 before the LSSR (April 1 -14, 2020), during the LSSR phase 1 (May 20 -June 2, 2020) and the 120 LSSR phase 2 (June 3 -16, 2020). 121 Additionally, ground-level measurements of NO2, SO2, CO and PM10 concentrations were 122 collected from air quality monitoring station (AQMS) in the South Sumatera Province operated 123 by the Ministry of Environment and Forestry of The Republic of Indonesia. All satellite data were 124 directly compared with the AQMS. In order to match with the satellite data, the AQMS data were 125 arranged into a daily basis just like the satellite data acquisition time. A single satellite pixel 126 which located at the nearest point to the AQMS station was then used for the validation, this 127 technique was the same as shown by the other studies (Kanniah et al., 2020;Yang et al., 2020). 128

Spatial distribution of air pollutants 154
In this study, the satellite data were validated against the AQMS in the South Sumatera region. 155 The validation outputs indicated a good accuracy between the satellite data and the AQMS with 156 R 2 = 0.85, RMSE = 0.11 and MAE = 0.08. The remarkable agreement between the satellite data 157 and the AQMS enabled for applying the satellite data to analyze the air pollutants concentrations 158 and their distribution in the study area before and during the LSSR phases. 159 The PM10 concentration was analyzed in study area before and during the LSSR period ( Fig.  160 2). The period of LSSR had two phases which were May 20 -June 2, 2020 (1 st phase) and June 3 161 -16, 2020 (2 nd phase). The average PM10 concentration at before the LSSR (April 1 -14, 2020), 162 was 28.50 µg m -3 and decreased by 7.69% (LSSR phase 1) and 11.54% (LSSR phase 2) ( Table 1). 163

A C C E P T E D M A N U S C R I P T
9 this result, we conducted the back trajectories in the study area (Fig. 6). By using the back 166 trajectory we would like to know where is the possible sources of PM10 during the LSSR, 167 because based on the LSSR policy, industries and factories operation were restricted. The period 168 of the back trajectories was selected at the same duration with the LSSR implementation (May 20 169 -June 2, 2020). Figure 6 revealed the hotspots of burning activities in the South Sumatera region 170 were near to the Palembang city but they were not the major sources of emission. This study 171 found the primary sources of the PM10 were from the northeast area. The increase in PM10 172 might be elucidated because of the high concentration of biomass burning activities in the area, 173 especially in the land use conversion from peatland to agricultural sectors that occurred during 174 the LSSR period. Although, we found a significant reduction of PM10 level during the LSSR 175 phase 2, but when the LSSR phase 1, the PM10 level was just partly reduced, it was because 176 industrial activities was still in operation. The same situation also found at coal mining areas in 177 India which recorded a positive anomaly (+11 to 40%) during the lockdown period because of 178 continuing mining operation (Ranjan et al., 2020). 179 The average column SO 2 concentration in study area showed high reductions, 84.61% during 180 LSSR phase 1 and 98.90% during LSSR phase 2 ( Table 1)

A C C E P T E D M A N U S C R I P T
10 outbreak. In general, the greatest source of SO2 in air was from the burning of fossil fuels by 184 power plants and industrial activities. Spatial distribution map of SO2 concentration indicated that 185 the pollutants were more concentrated in the northwest part of study area (Fig. 2a). This area that 186 were along rice mills and food processing plants each recorded high amounts of SO2 (5.43 -7.89 187 DU). Since SO2 could be produced by burning coal, the high SO2 level were observed in the 188 center region of study area (above 5.43 DU), where there were coal industry and power plants.

A C C E P T E D M A N U S C R I P T
11 Tartarini (2020) also found 50% reduction in SO2 concentrations during the COVID-19 lockdown 202 in Singapore. 203 Furthermore, the average tropospheric NO2 concentration in study area revealed a significant 204 reduction during LSSR phase 1 and LSSR phase 2 that were 34.76% and 19.30%, respectively 205 (Table 1). Also, CO pollutants with 7.51% and 12.70% reductions during the subsequent LSSR 206 phases (Table 1). Before the LSSR period, the center part of study area was recorded the highest 207 amount of NO2 (205 -263 × 10 13 molecule cm -2 ) and CO (95 -100 ppbv) (Fig. 4a, Fig. 5a). NO2 208 and CO pollutants were primarily spread in the atmosphere from emissions from cars, trucks and

A C C E P T E D M A N U S C R I P T
12 However, NO2 and CO pollutant levels recorded dropped after the LSSR implementation (Fig.  219 4b-c, and Fig. 5b-c). During the LSSR, transportation and industrial activities were restricted, 220 although power plants and biomass burning were still remained active in some areas, which then 221 slightly increasing NO2 and CO pollutants in air. The decrease in air pollutant concentration was 222 related to better air quality index (AQI), thus we compared our result with other provinces in 223 Indonesia like Jakarta and West Java that both also had high COVID-19 cases and applied LSSR 224 to avoid a larger scale transmission around the area. According to study by Pramana et al. (2020), 225 they revealed that the air quality index from Jakarta, Banten and West Java provinces also 226 showed better changes during the LSSR period. 227 228

The association between air pollutants and case fatality rate of COVID-19 229
To obtain the case fatality rate of COVID-19 over study area, we calculated the ratio between 230 confirmed deaths (182 deaths) and confirmed cases (3724 cases) in study area, thus the case 231 fatality rate of COVID-19 was 4.89%. The case fatality rate value was higher as compared with 232 other cities in China. The case fatality rate has been used to evaluate and compare the severity of 233 COVID-19 epidemic between cities or countries (Kim et al., 2020). Our result found that the case 234 fatality rate of COVID-19 was positively associated with PM10 concentration (r = 0.514, p < However, there was weak correlation between the case fatality rate of COVID-19 and SO2, 247 NO2 and CO (r < 0.1, p > 0.05) ( Table 2). This study suggested that these gaseous had more 248 significant relationship with incidence and transmissibility of COVID-19. Hou et al. (2020) also 249 noted the same result that they found weak correlation between the case fatality rate of COVID-250 19 and SO 2 , NO 2 and O 3 . As a whole, our findings obtained a positive association between PM10 251 level and the case fatality rate of COVID-19, that suggesting that long term PM10 exposure that 252 prior to the LSSR period might increase sensitivity of resident to COVID-19. Limited by ground M A N U S C R I P T 14 PM10, SO2, NO2 and CO levels using high spatial resolution satellite (e.g. 1 km × 1 km). 255

A C C E P T E D M A N U S C R I P T
Therefore, for future studies, it needs to use hyperspectral images to obtain air pollutant data for 256 getting better accuracy. 257 258 CONCLUSIONS 259 260 The average SO2, NO2, CO and PM10 concentrations during the LSSR indicated a notable 261 reduction over study area such as 98.90%, 34.79%, 12.70% and 11.54%, respectively. The 262 highest reduction of pollutants was at the second phase of LSSR. The reduction was due to a 263 policy during the LSSR that commanded industrial operations either shut down or worked with 264 limited capacity. Furthermore, this social distance policy recommended the people to stay at 265 home thus it decreased vehicles emissions. The back trajectory analysis found in the limited of 266 major industrial and transportation activities, biomass burning activities contributed as a primary 267 source of particulate matter emission during the LSSR period in study area. The study also 268 highlighted the PM10 concentration was associated with the case fatality rate of COVID-19 (r = 269 0.514, p < 0.05). Thus, the higher PM10 concentration, the higher case fatality rate of  This was because the PM10 increased a risk of mortality due to pulmonary diseases. These results 271 of study could assist in arranging better air pollution management and COVID-19 transmission 272 control. 273

A C C E P T E D M A N U S C R I P T
21 Tables and Figures   382 383