key: cord-0718979-mmvakqfi
authors: Beig, Gufran; Jayachandran, K.S.; George, M.P.; Rathod, Aditi; Sobhana, S.B.; Sahu, S.K.; Shinde, R.; Jindal, V.
title: Process based Diagnostics of extreme pollution Trail using Numerical modelling during Fatal second COVID-19 wave in Indian capital
date: 2022-03-08
journal: Chemosphere
DOI: 10.1016/j.chemosphere.2022.134271
sha: d7dc35d2fbe5feac9c81bfc885390f827cf3a3b6
doc_id: 718979
cord_uid: mmvakqfi

The world's worst outbreak, the second COVID-19 wave, in Indian capital not only unleashed unprecedented devastation of human life, but also made impact of lockdown in particulate matter (PM: PM(2.5) and PM(10)) virtually ineffective during April to May 2021. The air quality not only remained unabated but also marred by some unusual extreme pollution events. SAFAR-framework model simulations with different sensitivity experiments were conducted using the newly developed lockdown emission inventory to understand various processes responsible for these anomalies in PM. Model results well captured the magnitude and variations of the observed PM before and after the lockdown but significantly underestimated their levels in the initial period of lockdown followed by the first high pollution event when the mortality counts were at their peak (∼400 deaths/day). It is believed that an unaccounted emission source was playing a leading role after balancing off the impact of curtailed lockdown emissions. Model suggest that the unprecedented surge in PM(10) (690 μg/m(3)) on 23rd May, while Delhi was still under lockdown, was associated with large scale dust transport originating from the north west part of India combined with thunderstorm. The rainfall and local dust lifting played decisive roles in other unusual events. Obtained results and the proposed interpretation are likely to enhance our understanding and envisaged to help policy makers to frame strategies in such kind of future emergencies.

results well captured the magnitude and variations of the observed PM before and after the lockdown 24 but significantly underestimated their levels in the initial period of lockdown followed by the first high 25 pollution event when the mortality counts were at their peak (~400 deaths/day). It is believed that an 26 unaccounted emission source was playing a leading role after balancing off the impact of curtailed 27 lockdown emissions. Model suggest that the unprecedented surge in PM10 (690 µg/m 3 ) on 23 rd May, 28 heart-wrenching tragedy. The speed and scale of the outbreak suggested that India probably has an 55 emerging variants of the virus. Genomic surveillance data show that the variant first identified in India, 56 known as Delta. The Delta variant contributed to an overwhelming surge in Delhi (Singh et al., 2021) . 57

To control the surge in cases, Delhi imposed a lockdown during 19 th March to 31 st May 2021. There 58 are many studies addressing air quality with COVID-19 during the first wave of 2020 (Beig et al., 59 2020a and refences therein) but work assessing the 2 nd wave of 2021 in India in terms of air quality is 60 sparce. Therefore, this study is an important contribution to the global knowledge of the impact of 61 local lockdowns on air quality. 62

As the second wave of the COVID-19 raged across Delhi, experts have warned that the COVID-19 is 64 airborne and the low temperature incomplete combustion of biofuel will have lethal effect as virus 65 piggyback on aged carbon particles (Rathod and Beig, 2021) and may further aggravate casualties. As 66 both COVID-19 and air pollution predominantly affect the upper respiratory tract and lungs, it has 67 become a matter of great concern for the Indian capital city Delhi transport sector (41%) followed by windblown resuspended dust (21%) whereas for PM10, major 191 source is windblown dust (46%) followed by transport sector (18%). The total emissions from all 192 sources under normal scenarios during the period from March to May were 6420 tons/month and 14849 193 tons/month for PM2.5 and PM10 respectively ( Table 1) normal case which is shown in the 3 rd and 6 th Column in Table- marking near the x-axis filled with dark yellow color represents the intense rainfall period whereas the 304 yellow dot at 23 rd May represent a dry day with sunshine as can also be confirmed from the rainfall 305 plot of Figure 2 . Model results were well in agreement with observations before and after the lockdown 306 periods (before 19 th March and after 31 st May) validating it well. However, model simulated a 307 significant reduction in both PM10 and PM2.5 when lockdown scenario was used (blue) and results 308 highly underestimated observations. Observed levels of PM during lockdown period were found to be 309 as high as model results with normal emission scenario in the initial period followed by the first 310 prolonged peak during 26 th April to 1 st May (event-1) when surprisingly observed values were found 311 to be higher than the model simulated values even with normal emission scenario, particularly of PM2.5. 312

Although model results show an increasing tendency at around 28 th April during event-1, but the 313 magnitude of prolonged peak could not be captured by the model. The back trajectory ending on 28 th 314 towards Delhi but the wind speed was quite high which could have prevented particles to get 316 accumulated. 317 J o u r n a l P r e -p r o o f sufficient enough to fully explain the prolonged peak observed in particulate matter during event-1 as 320 evident from model results in Figure 3 . This tend suggest that a strong unaccounted source was not 321 only offsetting the declined emissions of lockdown in the initial phase but also adding additional 322 emissions during the entire period of event-1 as model accounted for dust storm related transport. This 323 led us to believe that an unaccounted additional emissions source, rich in producing finer particles like 324 biofuel or fossil fuel rather than the dust was active because the growth rate in PM2.5 levels was much 325

higher with respect to PM10 as also confirmed from the PM ratio (~40-55%) in Figure 2 (d) which is 326 relatively higher than the normal. During lockdown, as traffic flow reduced, the transport related fossil 327 fuel emissions may not have played a significant role. Hence, this additional significantly high 328 emission may be related to low temperature combustion associated with biomass burning in 329 crematories which were at peak during this period. Many unconfirmed news articles reported that the 330 situation was so dire during almost same period that Delhi crematories are overwhelmed which have 331 never seen such a never-ending assembly line of death. The high mortality, very high mortality and 332 peak mortality period is marked in Figure 3 (a and b) . As evident, the peak mortality period was 333 directly coinciding with peak levels of PM2. 5 The study found that the total amount of wood required for cremation is around 300-400 kg/pyre for 339 open pyre (Kumar et al., 2019) . The crematoria flue gases contains higher percentage of organic, 340 inorganic matter and particulate dust material. This additional biomass emission was so high and 341 multiple pyres were used 24x7 for several days during peak that it superseded the impact of lockdown. 342 description in text. Also, due to high number of patients needing hospitalization, many ambulances 345 and private vehicles must be on road. Additionally, surplus vehicles due to heavy demand of oxygen 346 cylinder were on service as an emergency measure. Hence this surge in PM levels might be related to 347 a combination of biomass burning and fossil fuel emissions besides other possible emission source. 348

However, in absence of any reliable additional biomass burning emission data, we are unable to further 349 investigate. 350

The event-2 was due to consistent rainfall related washout which is quite obvious and trend is well 352 simulated by the model in both the scenarios. It is noteworthy to mention that during the lockdown 353 period, impact of dominant natural events in PM was relatively well captured by the model during 354 both event-2 and event-3 albeit varying magnitudes. The peak in event-3 was unprecedented during 355 this time of the year wherein PM10 peaked much faster than PM2.5 unlike event-1. We have investigated 356 this event using the model in detail to understand the underlined processes. 357 358 Figure 4 shows the model simulated circulation pattern and the processes governing the dust particle 359 movement during the event-3 (22 nd -25 th May) when the lockdown was in force. The period before 360 and after 23 rd May was marked by significant rainfall but there was sunshine for a day on 23 rd May 361 with heavy thunderstorm and winds started to blow from North-North-West part of India with a rapid 362 and heavy influx of dust leading to event-3. The synoptic analysis revealed that under the influence of Figure S2 ). To understand the rapid buildup of 368 additional PM10 particles overnight, the model simulated spatial distribution and pathways of dust rapidly from less than 65± 32 µg/m3 on 22 nd May to 684 ± 524 µg/m 3 overnight ( Figure 2 ). There 371 was an increase in PM2.5 levels but the magnitude of increase was not as high as PM10. 372 373 During this period, the PM ratio declined significantly (25%), indicating the highly dominant role of 374 coarser particles. Thereafter, the impact of dust inflow has reduced significantly in Delhi region and 375 under the influence of Westerly disturbances, widespread rainfall started again from 24 th May onwards 376 washing away the accumulated mass and levels of PM rapidly decline within 24hr. Thereafter as rain 377 continued, substantial improvement in PM levels was observed which continued even after the 378 lockdown was lifted on 31 st May. The last event-4 (depicted with light Indian red shaded region) in 379 Figure 3 have shown a different characteristic than that of event-1 and 2. In this case, PM10 level has 380 increased significantly but level of PM2.5 remained almost unchanged. This peak is reasonably well 381 captured by the model when the dominance of coarser particles played a major role. The long range 382 transport of dust is ruled out as the back trajectory ending on 8 th June ( Figure S2 , last panel) indicates 383 that flow was from southeast and not from the desert region as was the case in previous event. This 384 peak is attributed mainly due to lifting of local dust due to very high local wind during a break spell 385 of rainfall under bright sunshine and warmer temperature. It dried out the surface dust quickly. Also 386 the moister supply from the southeast, triggered thunderstorms in Delhi locally to lift the dust on a 387 local scale. With an increase in traffic on roads after the lockdown was lifted, high winds and dry 388 atmosphere also started to lift the local dust leading to a peak in PM10 on 8 th -9 th June 2021 but this 389 local scale event could not affect abundance of finer particles (PM2.5). 390 

On processes controlling fine particulate matter in 418 four Indian megacities

Physical characterization of particulate matter emitted from 421 wood combustion in improved and traditional cookstoves

System of Air Quality Forecasting and Research 424 (SAFARIndia)

Towards baseline air pollution under COVID-19: 426 implication for chronic health and policy research for Delhi

On modelling growing menace of household 429 emissions under COVID-19 in Indian metros

SAFAR-High Resolution Emission Inventory of Mega

Ministry of Earth Sciences (Govt. of India)

India's Maiden air quality forecasting framework for 435 megacities of divergent environments: The SAFAR-project

Objective evaluation of stubble emission of North 438

India and quantifying its impact on air quality of Delhi

Local characteristics of and exposure to fine 444 particulate matter (PM2.5) in four indian megacities

Mitigation of PM2.5 and ozone pollution in Delhi: A 447 sensitivity study during the pre-monsoon period

Gridded emissions of air pollutants for the 450 period 1970--2012 within EDGAR v4.3.2

HYSPLIT (HYbrid Single-Particle Lagrangian Integrated

NOAA Air Resources Laboratory

Sources and distributions of dust aerosols simulated 456 with the GOCART model

Fully coupled "online" chemistry within the 459 WRF model

Establishing Integrated Weather , Climate , Water 462 and Related Environmental Services for Megacities and Large Urban Complexes

Emission estimates and trends (1990-464 2000) for megacity Delhi and implications

and others in Delhi, Delhi Pollution Control Committee (DPCC) Government of Delhi

Temporary reduction in daily global CO2 469 emissions during the COVID-19 forced confinement

The AFWA dust emission scheme for the 472 GOCART aerosol model in WRF-Chem v3

Impacts of Megacities on Air Quality: Challenges and 475 Opportunities

Health and economic impact of air pollution in 477 the states of India: the Global Burden of Disease Study

The Weather Research and Forecasting 480 Model: Overview, System Efforts, and Future Directions

Impact of biomass induced black carbon particles in cascading COVID-483 19. Urban Clim. 38, 100913

Emissions inventory of anthropogenic PM2

Delhi during Commonwealth Games

Comparison of emissions inventories of 488 anthropogenic air pollutants and greenhouse gases in China

SARS-CoV-2 variants of concern are 491 emerging in India

Household Stoves in India

COVID-19 lockdown and air quality of SAFAR

Ambient particulate matter and carbon monoxide at an 496 urban site of India: Influence of anthropogenic emissions and dust storms

Investigation of emission characteristics of 499 NMVOCs over urban site of western India

Acknowledgments: This work has been supported by the Indian Institute of Tropical Meteorology 503 (IITM) and authors extend their gratitude to Director

MoES and Program Director of SAFAR at MoES for their cooperation and support. We 505 also acknowledge Chairman, DPCC with gratitude

Delhi Pollution Control Committee (DPCC), Govt. of Delhi

Indraprastha Institute of Information Technology, Delhi *Corresponding Author; E-mail: beig@nias.res

Lockdown Emission inventory developed and used in SAFAR-framework model

This work investigated the variability in particulate matter to understand the processes responsible for 394 the same using the SAFAR-Framework model which accounted for lockdown emissions observed 395 during the deadly 2 nd wave of pandemic during April and May 2021 in Delhi. We developed the 396 emission inventory of lockdown emissions. The overall estimated net emission present during the 397 lockdown period for PM2.5 was 3238 tons/month (51 % of normal) and that of PM10 was 8599 398 tons/month (58% of normal). However, these estimated emissions are informative in multiple ways 399 and not free from uncertainty which may range between 20-35 %. Model reproduced well the impact 400 of extreme pollution events in the trend of PM caused due to natural processes (event 2 to 4) but failed 401 to reproduce some observed unusual features mainly related to unaccounted hidden source of emission 402 which was related to biomass burning at crematorium in all likelihood. All 4 events discussed here had 403 different characteristics and processes which has been explained in this work. The model 404 underestimated the continued elevated levels of PM in the initial week after the lockdown followed by 405 prolonged peak of event-1, a period coincided well with the peak mortality period. The hidden source 406 of emission is believed to be associated with additional biofuel burning related to crematories whose 407 emissions could not be accounted in the model in absence of reliable source specific data. The model 408 also established that the North-westerly winds often brought dust particles from the desert region to 409Delhi leading to peaks in PM even during lockdown. Both COVID-19 and the dust can cause 410 overlapping respiratory symptoms and hence a suitable strategy needs to be worked out during such 411 emergencies. That can include the science based coordinated effort to prioritize source based 412 mitigation planning. The modelling effort to understand the nexus between the COVID-19, additional 413 biomass emissions and natural episodes can enhance the current state of knowledge that could provide 414 directions to future research benefitting both environmentalists and epidemiologists. 415