Insight into festival ushered pollutants in Indian metro cities: concerns, differences and similarities

Multi-fold increase in particulate matter (PM) and trace gases in connection with celebrations associated with rework displays is a matter of concern all over the world. The current study, commenced as an assessment of pollutant escalation due to the key Indian festival event through observations online and cumulative sampling along with chemical speciation of particulates and trace gases over four SAFAR networked mega and metro cities of India viz. Delhi, Ahmedabad, Mumbai and Pune. It is seen that the amplitude and duration of the event is largely dependent on the attitude and culture of the inhabitants of each city; PM2.5 and PM10 in Delhi are observed to increase by 353% and 213% respectively. In Pune and Ahmedabad, enhancement in PM2.5 is half of that of Delhi while the effect in Mumbai is just 1/7th of Pune, where variation is atypical to other cities. The enhanced pollution levels may lead to chronic or acute health emergency if they persist for long hours under unfavourable weather conditions. Metal content (K, Mg, Na, Mn and Pb) in PM 2.5 has nearly doubled in all the cities; huge increase in pulmonary range particulates and the steep elevation in their heavy metal content is a matter of concern due to their toxicity and acute health effects. Trace gases, NOx and CO which are also a health worry indicate a continuing increase due to the festival episode that happened in a non-conducive weather. Online and gravimetric samplings are compared for the rst time and they agree well.


Introduction
Diwali is celebrated all over India on various scales depending on the local cultures and precedence and for many areas it is a ve day event in which bursting crackers could be a small part on any day. However, in the night of the 'Laxmi Pooja' day it is at its peak going on for 2-3 hours. The concern related to health effects on account of re displays has started since long (Bach et al. 1975), while the atmosphere gets more and more stable, concentration of pollutants also increase (eg. Murthy Parkhi et al., 2016 attribute the sharp differences between two Diwali periods to atmospheric stability and other meteorological variables. However, the inter-annual range of the difference presented in their study in the particulates is di cult to be explained only in terms of meteorology. AP (attributable-risk proportion) is a concept mooted by the World Meteorological Organization (WMO) which says about the health effects due to pollutant inhalation that gets worse with spike in pollution. Methodology to calculate AP is done by Douwes et al., (2002); Rothman and Greenland (2008). Xing et al., (2016) presented a detailed review of the particulate matter in the size less than 2.5 µm which can penetrate deep into the lungs and cause lung diseases, advocating population to restrict exposure to PM2.5 and exhorting authorities to create a health index relating to the same.
Adverse health effects like lung disease, neurological and haematological disease are attributed to sudden spike in air pollution due to reworks which has been getting some serious attention since Hirai et al., (2000) and still is not fully understood for various reasons (Greven et  The concern in India about increase in air pollution during Diwali festival gave rise to a court appeal for court to intervene in controlling the reworks by law in 2018, in response to which a direction was given in the mode of use of green crackers with a prescribed time frame instead of blanket ban all over the country. They release water vapour and don't allow the dust particles to rise. They are designed to have 30% less particulate matter pollution. They also have a small shell size compared to traditional crackers and are produced using less harmful raw materials. The court held that though there were many studies (Barman et  Based on continuous AQMS (Air Quality Monitoring Station) and speci c eld sample collection (campaign) at four locations under SAFAR network; the study describes how the nature and timing of Diwali celebrations are different in each metropolis. Chemical characterization of the same is done to see the incremental increase in metals, carbonaceous matter and trace gas variations are also looked into. An effort is also made to look into any characteristic signature for each locality in Diwali celebrations, along with meteorological background. Spatial distribution of anomalous change in PM2.5 due to Diwali is studied to know the hot spot localities in each city. Finally, unlike the earlier work the time taken for the Diwali -induced pollution to retract to the pre-Diwali levels at each station is assessed. Possible control measures, suggestions, and the importance of considering the dispersive capacity of the atmosphere are also presented.

2.1Study Area
In this campaign, the study area includes four Metropolitan cities which are Delhi, Pune, Mumbai and Ahmedabad ( Figure 1). The System of Air Quality and Weather Forecasting and Research (SAFAR, www.safar.tropmet.res.in) networks were established at these four cities for better air quality forecasting during the period from 2010-2017. Delhi, capital of India, with a current domain of 65 x 70 sq km for the SAFAR network (NCR region) is situated in northern India at an elevation of 216 m above sea level. Pune at 559 m AMSL, ~100 km from west coast of India has a network domain of 50x 50 km which is a fast-growing urban city that includes Pune as well as industrial twin city Pimpri-Chinchwad. Coastal SAFAR network city, Mumbai at 14 m AMSL is the most populous city and has a domain of 35 x 35 km. The fourth city where SAFAR network is established is Ahmedabad in a domain of 30 x 35 km which covers Ahmedabad, Gandhinagar city area and surrounding villages.
Among the SAFAR monitoring network stations in each city the station with considerable residential activity was chosen for lter collection sampling through Low volume sampler so that we get good chemical representation of rework episodes. Malad in Mumbai has a population of 9.3 lakhs whereas Noida in NCR Delhi is with 6.3 lakhs Chandkheda in Ahmedabad and Hadapsar in Pune is about 1 lakh as per the 2011 census are the sampling stations chosen for the special campaign during Diwali. The newly acquired low volume (16.7 litres per minute, LPM) samplers, model APM 550 (Envirotech, India make with US-EPA equivalent inlet) were used for the Diwali campaign to collect PM2.5 with an accuracy better than 2µgm -3 similar to the one used by (Perrino et al. 2011). The sampling period was xed as 10AM to next day 10AM using quartz ber lter of 47mm with a loss about 15 minutes for the lter changing and other regular checks. The sample collection period was xed from 03/11/2018 to 12/11/2018 with 7th November as the Diwali day. Decision to sample only PM2.5 is based on the RR nding based on Beig et al., 2013.
The sampler was leak checked every day and other regular maintenance were rigorously done. Pre-treated, baked at 900 0 C for 4hours and pre-weighed 47mm quartz paper lter (Pall make) was used for sample collection. After collection of PM2.5 each lter was packed back in its respective petri-dish and sent to Pune for chemical analysis. Filters were post weighed for gravimetric determination of PM2.5 using ultra-accurate microbalance; model CPA26P, Sartorious make with an accuracy of 1µg after acclimatizing them to the room temperature in a temperature-controlled condition. Post gravimetric determination, the lter is used for various chemical analyses.

Atomic Absorption Spectrometer (AAS)
AAS works on the principle that liquid sample when atomized in an atomizer; the free atoms in the gas phase absorb its characteristic wavelength which aids the detection of the element. The absorption path length as per the Beer-Lambert's law is proportional to its concentration. This gives the method excellent speci city and detection limit. High Resolution -Continuum Source AAS, model ContraAA-800D (Analytik Jena, Germany make) uses ame technique with C 2 H 2 /air to achieve 2100-2300 0 C, and C 2 H 2 /N 2 O for 2600-2900 0 C for concentrations expected above 25ppb and graphite technique with graphite tubes to detect concentrations further lower. Ca, K, Mg, Fe, Zn available in higher quantities were analyzed by ame technique while Cu, Cr, Cd, Ni, Mn, Pb were analyzed by graphite technique.
Quartz lter samples (3 punches each of 0.5 cm 2 ) were mixed with 3 ml of HNO 3 , 2 ml of HCl and 6 drops of HF and then topped with deionized water to make 25 ml sample. This was microwaved (Analytik Jena topwave) digested under high pressure (50 bar) and temperature (200 0 C) for an hour for complete dissolution of analyte in solvent. This uniformly mixed solution was used for analysis. Calibration standards are obtained from ISO guide 34 accredited company and is traceable to NIST SRM 3131a. Blank lters with the same digestion procedure were analyzed to eliminate matrix effect of quartz lter. Recovery of 100±10% from standard and relative standard deviation of <10% among sample duplicates was ensured.
The samples are analyzed through EC-OC analyzer as explained in (Ali et al. 2015) to derive OC (organic carbon) and EC (elemental carbon).

Data analysis
The data is processed from the AQMS for PM2.5/10, NOx, Ozone, CO, BTX, NH 3 and surface meteorological parameters depending on the availability at 15 minutes interval and used as hourly averages. Further, the ARC-GIS software is used for IDW overlay for obtaining area averaged plots for nding city scenario and marking of hot spots using the networked data. LVS and online data at a speci c location are compared for 24 h duration (10 am to next day 10 am). Generally daily averages for online samplers are averaged starting from 12am to 11pm; howver in this study 10am to 10am time period is used. LVS sampled lters are chemically analyzed for OC/EC analyzer, AAS and ion chromatography analysis (m/s Metrohm make). However, only AAS analysis results are presented here. To assess meteorological/dispersion effect meteorological parameters along with Ventilation Coe cient (V) derived from radiosonde ights were made use of at all stations. VC is calculated as the product of mixed layer height and mean wind in that layer; derived from the Wyoming University Radiosonde data ([CSL STYLE ERROR: reference with no printed form.]) for each city. Due to some synoptic system, Pune had rains during this period. Hence VC cannot be deciphered for some days and hence not included in the analysis.

Particulate matter -Physical
The speci c increase on Diwali day is the highest over Delhi and the lowest in Mumbai while Ahmedabad and Pune are in the second and third place after Delhi (Fig. 2). The time and magnitude of peak indicate the trend of activities at each location ( Table 1). The difference in daily maxima between pre-Diwali day (PD) and Diwali day (DD) in Delhi for PM10 is 1624 µgm − 3 and for PM2.5, it is 1356 µgm − 3 . In Mumbai the daily maxima difference for PM10 and PM2.5 are 114 µgm − 3 and 100 µgm − 3 respectively.
Time of peak for PM2.5 changed from PD to DD in Delhi, from 5 am to 6 am; for Mumbai, 12 to 1 am; the peaks for Ahmedabad at 2 am and for Pune at 12 am remained the same; PM10 peak in Mumbai and Delhi occurred an hour later than PM2.5 which may be due to reworks continuing late or reworks types that emit bigger sizes or due to growth of particle size as well. The percentage increase at Delhi was 353% for PM2.5 and 213% for PM10. In the absence of other weather systems, the entire additional particulates emitted may be related to Diwali episode. In this study PM2.5 is elaborated more as its sub-index always decided AQI and also as this portion of PM is particularly detrimental to health. Pune and Ahmedabad experienced about 1.7 times ampli cation in PM2.5 which was less than half times of Delhi. Diwali effect in Mumbai is just 1/7th of that in Pune. The inset gures in Fig. 2 indicate the daily averages (10am to 10am) of PM10 and PM2.5 for the study period at the four locations. It is observed that the concentration shows a signi cant peak on the DD at all locations except Mumbai. The concentration is highest at Delhi, followed by Ahmedabad, Pune and the least at Mumbai.
An anomaly area-averaged map is prepared as explained in Sect. 2.2.4 using ARC-GIS software to have a total picture of PM2.5 surge of each city caused by DD activities and to note probable hotspots under the assumption that all other sources remain unchanged from PD to DD. As illustrated in Fig. 1, central Delhi looks the most polluted. In the case of inter-city comparison, Delhi's south-south-west area exhibited the lowest increase but this is comparable with the most polluted areas of Mumbai in magnitude. Ahmedabad-Gandhinagar stations stands second in particulate density with cleaner Gandhinagar and most polluted south-south-east area. Pune comes third with its south to east quadrant most polluted as it constitutes interstate transport hub and fast developing residential and industrial zone. The only coastal station, Mumbai is least polluted in terms of enhanced PM2.5 during DD. A cohort study was conducted by CPCB in collaboration with medical professionals(CPCB 2019) over the residential areas which coincides with the intensely polluted areas depicted in our anomaly gure (Fig. 1). CPCB had found in their study that cough has increased from 6.7-28.9% ( Daga et al., 2019) and post-Diwali hospital admissions surged by 50% due to cardiac, stroke, respiratory and burns.

Particulate Matter -Chemical Analysis
As detailed in Sect. 2.2.2 low volume samplers (LVS) are used for collecting PM2.5 particulates in pre-baked and weighed quartz lters for gravimetric determination as the difference between pre-weight and post-weight. Flow rate is xed at 1 m 3 hr − 1 while the concentration in µg m − 3 is obtained. In all four cities, LVS was placed at one of the SAFAR stations that represent the residential area for comparison by gravimetric analysis. Figure 4a shows the association between lter and AQMS (online) based PM2.5 quanti cation for the station where lter sampling is done. Technically, the LVS sampler considers the aerodynamic diameter for deciphering PM2.5. It is susceptible to errors due to increased humidity for the continuous measurements (optical detection) and to avoid that sucked air volume is heated at the inlet stage to control humidity. Though there could be a mismatch both the observations compared well except in Mumbai where increased humidity post-Diwali might have played a role. PM2.5 observations reveal that DD value of the station where lter sampling is done exceeded the monthly averaged value of the city by about 150% which implies that choice of station is proper for the study of Diwali impact. On speci c comparison among cities, gravimetric and continuous observations are in tandem in Mumbai on DD, at Delhi gravimetric value is slightly less whereas in Pune and Ahmedabad it exceeds the continuous data.
Vecchi et al., (2008) monitored hourly EC, OC during reworks at Milan, Italy and found that during peak hours both EC and OC rise noticeably. The lter samples of the present campaign are processed to derive organic and elemental carbon content. Our observations show that OC is elevated on DD for all stations though such a pattern is absent for EC. Liu et al., (2004) reports a clear change in OC-EC ratio in connection with spring festival reworks in China but in this study OC-EC ratio that is in the 1-3 range with no systematic variation (Fig. 4b). This may be indicating a non-generalized nature of recrackers and their use, emitting particulates differently (Betha and Balasubramanian, 2013).
Particulates emitted due to rework are rich in elemental fraction since many a times heavy metals are used as colorants. The metal content from the lter samples are derived as detailed in Sect. 2.2.4. We analyzed Ca, K, Mg, Fe, Zn through ame technique and Cu, Cr, Cd, Ni, Mn, Pb by graphite technique; the number of metals is restricted by the present availability of standards present. Figure 5 represents the time series of the total concentration of metals together with individual concentration of each metal in stack format for the Diwali campaign period.
The envelope line of the stack illustrates the temporal variation of total metal contribution in PM2.5. On DD total metals contribution rises and its range is 9-23% while on PD the total metal content is equal to or less than 5% at all stations. These increments are similar to that reported by (Chatterjee et al. 2013)) which is a study of reworks over Kolkata. Delhi witnesses an anomalous increase in other elements post Diwali though other stations generally indicate a decreasing trend; metal content variation at Delhi requires more detailed analysis. While, for Delhi, Ahmedabad and Mumbai, K remains higher than usual on post-Diwali day, Pune has high content of K on PD as well as post Diwali day probably signature of locality and pattern of cracker usage. There are intermittent increases of Ca in this period, at Ahmedabad, Mumbai and small rise at Pune is seen as a possible effect of transport from North West as seen through back trajectories (not shown) and a local wind speed increase is also noticed. Total percentage of analyzed metals in PM2.5 is the highest in Mumbai and lowest in Delhi.

An increase in 'K', a pointer to biomass burning is also a marker for reworks (Kumar et al., 2016). A similar variation in Delhi on 5th and
10th are also seen which looks similar to the spurt in vehicular tra c with more increase in Pb and Mn and other crustal elements. Evident from elemental variations, it appears that large amount of rework is absent other than DD though in the entire week some scattered ones may be occurring. Potassium (K) is used as an excellent marker of rework as the gun powder, a popular oxidizer has huge contents of potassium nitrates.   In Mumbai the NOx concentration shows an increase from the night of 7th Nov and further peaks on 10th Nov. There is no signi cant change in NOx concentration at Ahemdabad even after Diwali period. A signi cant change in CO with about 1ppm is seen only over Delhi which is followed by Mumbai and Pune with no change in Ahmedabad. When we observe the O3, CO and NOx diurnal pattern inverse correlation is observed between them as the latter two are precursors of O3.
The particulate overload at Delhi should have resulted in considerable radiation cut-off that led to lower ozone generation while in Pune the rains cleared air to some extent not to curb radiation and at Mumbai increase in particulate itself was minimal but availability of precursors rose somewhat helping build up of ozone. Source of NO and CO are oxides produced due to high temperature combustion as in internal combustion engine during an intense rework. In Pune, Mumbai and Delhi a considerable increase in the concentration of O3 is observed after DD, however in Ahmedabad it remains in the same range as before Diwali. NH3 measurements were available only at 3 cities and hence the same is shown in Fig. 6. It is observed that after DD there is a considerable increase in the NH3 concentration at the three locations.  Figure 6 depicts the meteorological parameters during the Diwali period. The daily temperature, Relative Humidity (RH), wind speed (WS) and the Ventilation Coe cient (VC) for the period 3-12 Nov is shown in this gure. The Ventilation coe cient (VC) which is the product of mean wind (in mixing layer) and mixing layer height is considered as a measure of dispersive capacity of atmosphere. VC is consistently higher for Ahmedabad. Over Delhi VC is about 3000 m 2 s − 1 prior to Diwali that drops to less than 1000 m 2 s − 1 post Diwali which relates to the enormous aerosol loading and resulting insolation cut off. All cities except Pune sees VC reduction on the next day of DD. VC is based on the radiosonde measurements at 1700 h every day and Pune being cloudy on DD and it rained too the VC on 7th is small and it could not be determined on 6th. On 8th it was clear to a good extent and the same re ected as rise in VC.

Meteorology During Diwali Period
Mumbai experiences comparatively low VC, below 1500 m 2 s − 1 . The other parameter that assists dispersion of pollution is surface wind. Generally higher winds prevail over Delhi for this period with a rise after Diwali. Ahmedabad falls in the second place with more wind speed but post Diwali there is a little drop, but remained high enough to aid dispersion. Mumbai and Pune experience very low wind speed and these winds could hardly play a part in dispersion. Hence it is clear that Ahmedabad meteorological conditions are conducive for faster dispersal of pollutants in the absence of extreme high levels of pollution. The temperature at the 4 cities does not vary signi cantly throughout the study period and thus it does not play a role in modulating the pollutants. Though at Delhi wind speed does not change much, the pollution level has reached extreme level. Moreover while considering the size distribution of PM it was observed that Ahmedabad had the highest difference between PM10 and PM2.5 range which is bound to help easy gravitational settling.
The swell in pollution is steep and that affected Delhi's pollution distribution by bringing down VC in spite of the strong support from higher winds. Limited rise in PM levels arrested worsening of air quality in Mumbai with increase in pollutant gas levels too after DD. In Pune, the rain washout controlled the impairment of air quality with otherwise non-supportive weather.

Conclusions
The current study is the rst-time analysis of simultaneous networked stations of four metro/mega cities of India on Diwali with online as well as campaign mode observations along with chemical analysis. The festival effect on trace gases, metal and carbon content are also analyzed. An area-averaged map indicated an idea of hotspots within each city and similarities and dissimilarities in pollutant enhancement. Station wise observations indicated speci c nature of city as a whole and differences within.
Trace gas analysis points at enhancement in Ozone, NOx and CO to various levels, and continuing at the elevated levels for many days with much less change at Ahmedabad. The PM collected through lter media was analyzed chemically and the results were: 1. OC showed a systematic increase in all station during Diwali and OC/EC ratio did not show any appreciable change. 2. metal analysis showed that Ca, K, Mg, Mn, Na and Pb shows signi cant increase in all stations from PD to DD. K, an excellent gun powder marker, shoots up by 7 to more than 15 times from PD to DD. Pb increased about 10 times and Cu 77.8 times in Delhi with other cities in lesser quantities. The total metal content which was less than 5% increased to 9-23% on Diwali day. Diwali air pollution raises most concern over health effect which considers the extent in intensity and time duration of the event. An acute health impact due to Diwali event and its endurance hints at a serious possibility revealed by Li et al., (2018) in their study of congestive heart failure (CHF) related hospital admissions at Beijing for 2 years that an increase in PM2.5 concentration by 10µg m − 3 in a day landed 0.35% more patients in hospital with CHF the same day.
In general, the festival is celebrated in the time of the year when weather conditions are not very conducive for dispersing the effect quickly. As expected, the control measures have to be more stringent where the pollution enhancement is maximum (Delhi as per this study) as the generated pollutants are even capable of reversing the otherwise conducive conditions of spread out. At Ahmedabad though the levels go higher on DD the weather helped diminishing the effect quickly. A bit more limiting action could leave the people without getting affected much. In the absence of rain, Pune, with non-conducive weather and very low wind speed, also would have faced longer bad-air days. Here also curbs are needed for certain areas of city where there are intense reworks in general. Over Mumbai, again with very low wind speeds and lower VC, any increase in pollution levels, especially gaseous, are to prevail longer in spite of cleansing land-sea breeze.
The management should look into alternatives like less polluting crackers or crackers released high into the air from an open ground with speci c time for such display so that more can have a view of the same rather than having more ground-based crackers in congested neighbourhoods that is bound to create more noise and air pollution and health concerns. Stricter enforcement based on air quality forecast is needed for regions with adverse conditions for general well-being of society. Sustained cohort studies involving vulnerable group in collaboration with health professionals for many years in different stations, depending upon the prolonged event-induced higher pollution levels, are needed in future to assess the actual health effect.

Declarations
Ethics approval and consent to participate Not Applicable

Consent for publication
Not Applicable Availability of data and materials Not applicable.

Competing interests
We hereby state that we do not have any known con ict of interest with anyone  Metal contribution in PM2.5 as percentage, total percentage of metals shown by line Figure 6 Temporal variation of trace gases, NOx, CO, Ozone and NH3 during Diwali period