Impact of Diwali Festival on Aerosol Optical Properties over an Urban City, Ahmedabad (India)

An attempt has been made to investigate changes in the characteristics of aerosol optical properties induced during Diwali/ New Year celebrations of 2012 over an urban city, Ahmedabad (India). Ground based measurements of average Aerosol Optical Depth (AOD) were carried out using a Microtops-II at Physics Department, Gujarat University, Ahmedabad during 10 to 16 Nov, 2012. AOD on the day just after Diwali is found to increase significantly; around 75% compared to pre-Diwali days for visible wavelength, remained at higher level for one more day (next day of New Year). Turbidity factor (β) also changed considerably, to a high value of 0.34 on the next day of Diwali, indicating a significant increase in aerosol loading associated with Diwali/ New Year festivities. Normal anti-correlation between Angstrom exponent (α) and turbidity parameter (β) is not seen on days following Diwali. The AOD and β values in the evening of the next day of Diwali increased to about 164% and 171% respectively compared to pre-Diwali days evenings; such an increase associated with the festivities has not been reported at any part of the country. Spectral variation of AOD also shows a significant change in the pattern during following two days of celebrations compared to pre-Diwali days. It is seen to follow power law more closely. Significant increase in AOD and change in spectral pattern suggest an increase of fine/accumulation mode particles due to Diwali/ New Year festivities.


INTRODRCTION
Aerosols are minute particles in solid or liquid phase suspended in air.They occur over a wide range of sizes extending from 0.001 µm to about 100 µm.Physical and chemical properties of aerosols in the atmosphere depend on its source as well as mechanism of generation.Aerosols in the troposphere are produced directly by various natural processes such as winds, sea-sprays, volcanoes, forest fires etc. and by anthropogenic activities such as burning of fossil and biogenic fuels, forest clearing and activities such as construction, agricultural and transportation, etc.Another contribution to atmospheric aerosols comes from gas-toparticle conversion processes.Aerosols in the atmosphere larger than about 1 µm in size are mainly windblown dust and sea salt from sea-spray.Aerosols smaller than 1 µm are mostly formed by gas-to-particle conversion processes and also by incomplete burning processes.Bursting of fire crackers and burning of sparklers during festive celebrations inject lot of aerosols into the atmosphere locally and temporarily.
Apart from health effects and reducing visibility, aerosols play vital role in controlling radiation balance of earthatmosphere system directly by scattering and absorbing solar radiation (Charlson et al., 1992;Andreae, 1995;Kaufman, 1998).They scatter incoming solar radiation back to space thereby enhance planetary albedo and exert a negative (cooling) radiative forcing (Twomey, 1977).Aerosols also cause negative radiative forcing indirectly through formation of clouds and thereby influencing cloud properties and their lifetime (Penner et al., 2003).On the other hand, certain aerosols such as black carbon and mineral dust cause positive radiative forcing (warming) by absorbing solar radiation (Kaufman et al., 2002;Lohmann and Feichter, 2005;IPCC, 2007).Direct and indirect effects of atmospheric aerosols on radiative forcing and cloud microphysics are strongly dependent on their size characteristics and chemical composition (Hegg et al., 1996;Haywood et al., 1997;Muller et al., 1999).
Atmospheric aerosols exhibit very large spatial as well as temporal variability.Distribution of aerosols in the atmosphere is decided basically by sources and sinks, but is also often influenced by prevailing meteorological conditions.Even though the natural aerosols play a very significant role on a global scale, aerosols of anthropogenic origin are very important on a regional scale (Coakley et al., 1983;Kaufman and Fraser, 1983;Coakley and Cess, 1985;Kiehl and Breigleb, 1993;Andreae, 1995;Ramachandran et al., 2012).Recently Goel et al. (2017) found that incense smoke is a major source of particulate matter in the temple microenvironment of Kanpur city (India) and they observed PM 10 mass concentrations of more than 2100 µg m -3 inside the temples.Impact of aerosols on climate is different from that of greenhouse gases, since lifetime of aerosols is much shorter and also due to the fact that the effects of aerosols are felt more on regional scale.Therefore, studies of aerosols at regional and local levels are very important.
Air over Ahmedabad is getting polluted by traffic jams, construction works, vehicle emissions, fumes from neighboring industrial areas, careless habits of burning waste etc.National Ambient Air Quality Standards of India prescribes annual average for PM 10 as 60 µg m -3 (The Gazette of India, 2009); which is much less stringent compared to WHO's air quality standards (WHO, 2006).Many parts of Ahmedabad city recorded annual average PM 10 for the year 2012 much higher than the national average (GPCB, 2013); more than 80 µg m -3 over many parts of the city and more than 100 µg m -3 on certain pockets.
Various studies suggest exceptionally high level of pollution, much above the prescribed limit of National Ambient Air Quality standard levels in association with certain festivals and result in short term air pollution episodes.Impact of fireworks during festivals on short term perturbation in air pollution levels, especially in particulate matter, in relation to air quality and health problems have been studied by several researchers (Clark, 1997;Ravindra et al., 2003;Kulshrestha et al., 2004;Barman et al., 2008;Godri et al., 2010;Moreno et al., 2010;Thakur et al., 2010;Singh et al., 2010;Tiwari et al., 2012;Chatterjee et al., 2013).Recent investigation by Sahu and Kota (2017) found that PM 2.5 is the dominant pollutant over New Delhi, and they suggested that mortality over New Delhi can be decreased by 6.2 and 6.5% by meeting the PM 2.5 Indian standards and WHO set limits, respectively.Babu and Moorthy (2001) reported that fireworks are a source of soot in the atmosphere.Singh et al. (2003), Badarinath et al. (2009), Vyas and Saraswat (2012) and Singh et al. (2014) have shown an increase in AOD associated with Diwali celebrations over different parts of country.Increase in BC mass density following Diwali festival has been reported over different cities of the country by Babu and Moorthy (2001), Ramachandran and Rajesh (2007) and Nair et al. (2010).Singh et al. (2014) found a change in surface radiative forcing by -45% and atmospheric radiative forcing by +30% during Diwali days at Varanasi.They also observed an increase in both scattering coefficient as well as absorption coefficient.Such studies related to fireworks in other parts of the world have also been reported.An increase in particle number in accumulation mode (> 100 nm) has been witnessed during the Millennium Fireworks in Leipzig, Germany (Wehner et al., 2000).Wang et al. (2007) reported an increase of about 125% in PM 10 levels over previous day due to fireworks display during Lantern Festival in Beijing.Zhang et al. (2010) found that particle concentrations (PM 10 ) during the peak hour of Chinese New Year's firework event in Shanghai (China) were approximately three times higher than that on the preceding day, with a clear shift in the particles from nuclei mode (10-20 nm) and Aitken mode (20-100 nm) to accumulation mode (0.5-1.0 µm).Do et al. (2012) observed that the particles generated by the fireworks during annual Lantern Festival in Yanshui (Taiwan) possessed higher PM 2.5 -to-PM 10 ratios than those present in the pre-fireworks display period.Investigation by Lin et al. (2016a) showed that Beehive firework display in YanShuei area of southern Taiwan during the Lantern Festival emits significant amounts of PM 1 and PM 2.5 and degrades short-term air quality and Lin et al. (2016b) found that hourly average concentrations of PM 2.5 , BC, total particle number, and ultrafine particle number during YanShui Beehive Firework Festival were 6.9, 2.3, 5.9, and 3.7 times greater than those during the same period on reference days.
Diwali is one of the important festivals celebrated enthusiastically every year during October/November all over India.Since New Year of Gujarat State falls on the next day of Diwali, the festival mood extends and intensifies all over the State.High concentration of anthropogenic aerosols along with gases such as SO 2 , NO 2 etc. are injected into the atmosphere during festival period especially in urban areas by extensive burning of fire crackers.In this context, we investigated changes in aerosol optical properties over an urban city Ahmedabad (India) associated with Diwali/ New Year festivities of the year 2012.

LOCATION AND CLIMATE OF STUDY AREA
Ahmedabad is a fast developing semi-arid urban city in Gujarat State (India) with very heavy traffic and lot of industries and thermal power plants in suburban areas.Gujarat State is bounded by Arabian Sea in the west and desert of Rajasthan in the northwest (Fig. 1).Climate of Ahmedabad is characterized by hot summer and dry winter.Average annual rainfall of Ahmedabad is 79 cm, in which, 95% occurs during monsoon season.Southwest monsoon brings moisture to the State during mid-June to mid-September and northerly winds bring mild chill during December to February.
Diwali celebrated throughout India on 13th Nov, 2012 and people of Gujarat celebrated the New Year on 14th Nov, 2012.On Diwali day (13th) evening, after lighting diya (lamps usually made from clay with cotton wick dipped in ghee or oil), fireworks started and continued to the next day early morning to welcome the New Year.

INSTRMENTATION AND DATA UTILIZED
Basic measure of aerosol in atmosphere is Aerosol Optical Depth (AOD); which is a measure of depletion of solar radiation while passing through a column of atmosphere.Spectral dependence of depletion by particles in the atmosphere can be approximated by Angstrom power law (Angstrom, 1961) given below where τ(λ) is the AOD at wavelength λ, β is the turbidity coefficient and α is the Angstrom exponent.On taking logarithm, the above equation can be written as Angstrom exponent (α) is a measure of relative size distribution of aerosols.In general, α is found to reduce with increasing aerosol size; large value of α indicates relatively high ratio of smaller particles to larger particles, suggesting dominance of fine/accumulation mode particles over coarse mode particles.Typical values of α range from nearly zero for Sahelian/Saharan desert dust particles, which are dominated by coarse mode particles to > 2.0 for fresh smoke particles, which are dominated by accumulation mode aerosols (Holben et al., 1991;Kaufman et al., 1992;Eck et al., 1999;Reid et al., 1999).Turbidity parameter (β) represents atmospheric aerosol content in the vertical direction, β essentially is the AOD at λ = 1 µm.Generally, β varies between 0.0 and 0.5; β < 0.1 indicates a clean atmosphere, while β > 0.1 depicts a turbid atmosphere (Khoshsima et al., 2014).AOD, Angstrom exponent (α) and turbidity coefficient (β) are generally used to represent optical properties of aerosols.
A better empirical relationship between depletion of solar radiation due to aerosols and wavelength has been obtained by a second order polynomial fit (Eck et al., 1999;O'Neill et al., 2001;Schuster et al., 2006) of the form where, the coefficient a 2 accounts for the curvature of ln τ verses ln λ plot.This curvature indicates aerosol particle size distribution; negative curvature indicates significant contribution of fine mode particle and positive curvature indicates dominance of coarse mode particle.According to Schuster et al. (2006), when a 2 -a 1 is ≥ 2 implies a clear dominance of fine mode aerosol and a 2 -a 1 ≤ 1 implies clear dominance of coarse mode aerosol.Ground based measurements of AOD were carried out using a Microtops-II at Physics Department, Gujarat University, Ahmedabad (23.04°N, 72.54°E and 55 m amsl) during 10th to 16th Nov, 2012 at one hour interval from 0800 hr IST to 1700 hr IST.Observational site has comparatively better greenery and is away from industries.AOD measurements could not be continued beyond afternoon of 16th due to cloudy conditions.Low level winds during the period of study over Ahmedabad were nearly easterly with very low speed, less than 7 km hr -1 .Dew point temperatures were about 15°C for almost all days of observation.Hysplit back trajectories have been used to see any change in source region of aerosol.
Microtops-II used for the present study is a 5 channel (380, 500, 675, 936 and 1020 nm) hand-held sunphotometer, with field view of 2.5°, designed by Solar Light Company, USA.Microtops calculates AOD values at each wavelength assuming validity of Beer-Lambert's law, using channel's signal, extra-terrestrial constants, atmospheric pressure, time and location.Solar distance correction is applied based on latitude, longitude and solar zenith angle.Optical depth due to Rayleigh scattering has been subtracted from the total optical depth to obtain AOD.Optical depth corrections from other processes such as O 3 and NO 2 absorption are ignored in Microtops.The 936 nm channel data has not been used for the present study, since it is intended to be used for the estimation of water vapour.More detailed description of the instrument and calibration procedures are available in Morys et al. (2001), Porter et al. (2001) and Ichoku et al. (2002).Typical errors in AOD measurements using Microtops-II sunphotometer are quite low (Ichoku et al., 2002).According to Kedia and Ramachandran (2011) cumulative error due to various sources in AOD measurements with Microtops-II is about 2-5% only, which is quite reasonable.

Spectral Dependence of AOD
To see how long the impact of festival on AOD persists, the spectral variations of AOD for forenoon (0900 hr), noon and afternoon (1600 hr) during pre-Diwali days (average of four days before Diwali night, 10th-13th) and three consecutive days after Diwali (14th-16th) are presented in Table 1.Most conspicuous feature of this data is the significant increase of AOD on following day of Diwali (14th) especially at visible wavelengths at all three times; 0.62 to 0.81, 0.55 to 0.82 and 0.58 to 1.17 respectively at forenoon, noon and afternoon.On this day, by 1600 hr, AOD 380 reached more than double of that on the previous day, suggesting an accumulation of aerosols in the atmosphere.On 15th, during forenoon and noon hours AOD values were more than that of 14th and by the evening of 15th, AOD droped to about 0.77; suggesting an increased downsettling of aerosols and the reduced production.The AOD values dropped to pre-Diwali condition on 16th forenoon.Another feature to be noted from this table is that for pre-Diwali days, during both forenoon and afternoon hours, spectral variation follows power law to a good extent, but during noon, AOD 1020 shows higher values, departing from power law.The higher AOD values seen in the longer wavelength are produced mainly by bigger sized particles.But, the spectral variation of AOD for noon plotted in Fig. 2 shows that for two days following Diwali (14th and 15th), spectral variation follows power law more closely; increase in AOD 1020 seen during noon has disappeared.On 16th, atmospheric condition became as that of pre-Diwali days with with comparatively higher value of AOD 1020 .
To ascertain the impact of Diwali festival on wavelength dependence of AOD, hour to hour spectral variations of AOD for pre-Diwali days and following day of Diwali (14th) are shown in Fig. 3.A significant increase in AOD on 14th in comparison with pre-Diwali days especially for shorter wavelength is clearly visible in the figure.During morning and evening period of pre-Diwali days, spectral variation follows power law to a large extent, but during noon hours, AOD 1020 increased compared to AOD 675 .On 14th, AOD increased with time especially in afternoon for all wavelengths and reached more than double compared to   pre-Diwali days by 1600 hrs especailly at visible wavelengths.Difference between AOD 380 and AOD 1020 in morning and evening hours for pre-Diwali days were nearly 0.27, but that has significantly changed on following day of Diwali as 0.5 in morning and 0.8 in evening.At the same time, increase in AOD 1020 seen during noon period for pre-Diwali days are not observed on following day of Diwali.A similar pattern is observed on 15th also.This suggest a change in the size distribution pattern of aerosols in the atmosphere with an increase in smaller sized aerosols due to the fireworks for the following two days of Diwali.Zhao et al. (2014) reported that the firework displays have significant impact on particle number concentrations in accumulation mode (100-1000 nm), especially in 200-500 nm.Vecchia et al.
(2008) observed a significnat increase in particle number concentrations during the firework episode; up to 6.7 times in 1 hour for 0.5 < d < 1 µm size.

Hourly Variation of AOD
Day-time hourly variation of AOD for all four wavelengths during 10th-16th Nov, 2012 is shown in Fig. 4. Hourly values of AOD for averaged 4 days before Diwali (10th-13th) presented in the figure shows that AOD values varied between 0.5 and 0.6 for 380 nm, 0.4 and 0.5 for 500 nm and 0.25 and 0.35 for 675 nm during pre-Diwali days.The AOD 1020 shows a systematic variation with a peak during noon and minimum during morning and evening with a range of about 0.2.
On 14th morning, AOD values were higher compared to pre-Diwali days for all three visible wavelengths, especially for 380 nm, for which AOD increased from about 0.53 to about 0.83.This might be due to continuous firing of crackers and burning of sparklers during Diwali night (13th) and early morning of New Year day (14th).On 14th, AOD remained nearly same up to noon for visible wavelength, but thereafter AOD increased very significantly and systematically and reached to about 160% of pre-Diwali days value at 1700 hr for 380 nm.On the next day (15th) morning, AOD values dropped from previous day evening values, but values were more than that of previous day (14th) morning values.On this day also AOD remained nearly constant up to noon; thereafter it gradually decreased for all three visible wavelengths.On 16th, from morning onwards, AOD values for all wavelengths were very close to pre-Diwali condition.Impact of fireworks on aerosols is clearly seen for two days after the festival with significant increase in AOD values especially for lower wavelengths.Changes in AOD 1020 on the following day of Diwali (14th) with respect to pre-Diwali days were not very significant; evening AOD value has increased only about 0.2.

Hourly Variation of Angstrom Parameters
Hourly values of Angstrom parameters α and β during 10th to 16th Nov, 2012 have been estimated from plot of ln(AOD) versus ln(λ) for visible wavelengths.Such plots for pre-Diwali days (10th-13th) and for next day of Diwali (14th) for few typical hours are shown in Fig. 5.For pre-Diwali days, variation shows almost linear pattern at all times, but on the following day of Diwali (14th), variation indicates a non-linear nature with an obvious bulge.A similar pattern is observed on 15th also.This negative curvature seen on 14th and 15th at all times especially during morning and evening compared to pre-Diwali days indicates domination of fine/accumulation mode particles over coarse mode particles during the festive time (Eck et al., 1999;O'Neil et al., 2001;Kaskaoutis and Kambezidis, 2006).
Hourly variations of Angstrom parameters α and β for the period 10th to 16th Nov, 2012 are shown in Fig. 6.Angstrom exponent α for pre-Diwali days shows a systematic pattern, its value gradually decreased from 1.17 at 0800 hr to about 0.92 at 1300 hr, and then it increased to about 1.12 at 1700 hr.This indicates ratio of fine mode particle to coarse mode particle is high in morning and evening hours and gradually decreased to a minimum during noon period.This dip observed in α during midday hours may be due to local convective activity, leading to a change in aerosol particle size distribution.During the festival period, Angstrom exponent α remained nearly constant about 1.1 throughout the day, suggesting that size distribution remained nearly same during the whole festival period.
Hourly variation of turbidity parameter β during pre-Diwali period shows nearly a constant value of 0.2.The β increased by about 40% in morning of next day of Diwali (14th) and from noon onwards β increased systematically with time and reached to more than 2.7 times at 1700 hr than that in pre-Diwali days.On 15th, β was even higher than that on 14th up to noon time, but in the afternoon hours it gradually decreased.On 16th morning, β reduced to that of pre-Diwali days.Present analysis of Angstrom parameters suggests that aerosol loading in the atmosphere has been significantly affected by the Diwali festival and particle size distribution has changed marginally.Slight increase in Angstrom exponent compared to pre-Diwali days especially in noon hours indicates a slight increase in fine particle concentration in the atmosphere.This suggests the role of gas to particle conversion associated with enhanced release of gases during the fireworks, and possibility of direct generation of more fine mode soot particles.

Variation of Daily Mean AOD and Angstrom Parameters
Average AOD for each day has been calculated using hourly AOD values from 0800 to 1700 hrs.Average AOD for individual days from 10th to 16th Nov, 2012 for all four wavelengths plotted in Fig. 7(a) shows a significant increase in AOD for visible wavelength from 13th to 14th and remained at high on 15th; then dropped to pre-Diwali condition on 16th.Vertical arrow in the figure indicates Diwali day (13th).On 14th, AOD increased by 76%, 79% and 72% respectively for 380 nm, 500 nm and 675 nm, with respect to mean of pre-Diwali days.This nature almost follows the pattern of fireworks.On evening of Diwali day  (13th), after lighting diya, fireworks started, and continued till evening of New Year day (14th).Increase in AOD 1020 is not significant as that of visible wavelength; this suggests a significant increase in smaller sized particle concentration compared to bigger sized particle concentration.Singh et al. (2003) after analyzing aerosol characteristics during Diwali celebrations at Kanpur also observed more increase in AOD for 340 nm and 500 nm compared to 1020 nm.They reported that AOD increased from previous day of Diwali as against our present observation, and they attributed the practice of fireworks one or two days before Diwali, especially over urban areas for this increase.Singh et al. (2014) observed an enhancement of AOD 500 over Varanasi during Diwali period compared to average for the period not affected by Diwali.Babu and Moorthy (2001) reported three times increase in mass concentration of black carbon (BC) associated with Deepavali (Diwali) festival at Trivandrum.Ramachandran and Rajesh (2007) found an increase in BC concentration during late night and early morning hours at Ahmedabad during Diwali time.Surface BC measurements can be compared with vertically integrated AOD values to a large extent during post-monsoon season because the boundary layer during this period is very shallow and winds are comparatively calm so that pollutants are trapped at lower levels.
Spectral variations of day-averaged AOD for the period 12th to 16th Nov, 2012 are shown in Fig. 7(b).Sharp increase in AOD at lower wavelength for the following two days of Diwali (14th and 15th) is clearly visible.For two consecutive days following Diwali, spectral variation of AOD follows the power law to good extent, which was not the case in pre-Diwali and post New Year days.AOD at lower wavelength has increased significantly compared to that at higher wavelength.This indicates that the Diwali/ New Year celebrations produced more fine/ accumulation mode particles in the atmosphere.This may be due to two factors as (i) the gases that are released in the form of SO 2 , NO 2 etc. during fireworks might have converted into minute particles (ii) direct production of more fine mode soot particles during fireworks.Vecchi et al. (2008) and Kumar et al. (2016) also reported increased ambient fine particle concentrations in the atmosphere associated with firework displays.
The ln(day-averaged AOD) versus ln(λ) for a typical day before Diwali (12th) and next day of Diwali (14th) for visible wavelength are shown in Fig. 8.It is clear from the figures that on pre-Diwali day, variation follows almost a linear pattern, but next day of Diwali shows a negative curvature.The second degree polynomial fit (Eq.( 3)) shows a 2 changed from -0.216 on 12th to -0.669 on 14th and a 2 -a 1 increased from 1.144 on 12th to 1.333 on 14th.These results again indicate an increase in the dominance of fine/ accumulation mode particles over coarse mode particles in the atmosphere due to Diwali fireworks.
Temporal variations of Angstrom exponent α and turbidity parameter β calculated based on day-averaged AOD values for the period 10th to 16th Nov, 2012 are shown in Fig. 9. Vertical arrow in figure indicates the Diwali day (13th).Angstrom exponent changed only marginally, but turbidity factor increased significantly from 0.19 on Diwali day to 0.34 on next day (14th) of Diwali.On pre-Diwali days, αand β show a negative correlation as expected, but after Diwali, β also increased along with α.
To see whether wind pattern has any effect on changes Ahmedabad is much more compared to any other parts of the country that is reported.On the following day of Diwali, AOD reached to about 160% of pre-Diwali days at 1700 hr.• Angstrom parameters show that aerosol loading in the atmosphere has been significantly affected by the Diwali/ New Year festival, but the particle size distribution has changed marginally, with a slight dominance of fine mode particle.• Normal anti-correlation between Armstrong parameters α and β are not seen on two days after Diwali.• Negative curvature of ln(AOD) versus ln(λ) plots on following two days of Diwali also suggests a dominance of fine particles over coarse mode particles; this may be due to enhanced gas to particle conversion or production of additional fine mode particles associated with fireworks.

Fig. 2 .
Fig. 2. Spectral variation of AOD during noon for pre-Diwali and three consecutive days after Diwali.

Fig. 3 .
Fig. 3. Hourly spectral variation of AOD for pre-Diwali and following day of Diwali.