Identifying the natural and anthropogenic drivers of absorbing aerosols using OMI data and HYSPLIT model over South Asia

Aerosols absorption contributes significantly to the total radiative effects of aerosols and so an important component of radiative forcing estimates. Therefore, this study explores the spatiotemporal distribution of ultraviolet aerosol index (UVAI), future trends, potential sources of absorbing aerosols and their relationship with temperature, wind speed, precipitation and total ozone column using Ozone Monitoring Instrument retrieved UVAI and HYSPLIT model over South Asia during October 2004 to March 2022. The mean UVAI within the ranges of 0.56–1.62 are observed over Eastern and Southern Pakistan and Northern India associated with dust and biomass burning aerosols. The interannual variations in UVAI show that the values of UVAI increases from 1.73 to 3.11 during 2018–2021 over the Indo-Gangetic Plain. Contrary to this, UVAI < 0 is observed along the Karakorum and Himalaya range during 2005–2021 indicating presence of non-absorbing aerosols. The interaannual variations in UVAI reveal highest UVAI of 0.64 in December followed by 0.51 in July over South Asia. Seasonally, UVAI shows increasing trend at the rate of 0.9064 DJF−1, 0.3810 JJA−1, 0.2707 SON−1 and 0.0774 MAM−1 over South Asia. A positive correlation of 0.56 is observed between UVAI and wind speed followed by over Southern Pakistan followed 0.43 between UVAI and total ozone column Southern Pakistan and India. The UVAI shows increasing trend at the rate of 0.1409, 0.1124, 0.1224, 0.1015, 0.1242 and 0.2054 per year over Lahore, Karachi, Kanpur, New-Delhi, Varanasi, and Dhaka with maximum UVAI of 5.55, 4.47, 4.51, 4.99, 4.61 and 4.65 respectively during the study period. The anthropogenic productivity analysis reveals that primary industry and secondary industry contributes in lowering UVAI values whereas tertiary industry, energy consumption and gross domestic products increase aerosols loading in South Asia. Moreover, HYSPLIT cluster analysis further reveals the localized and trans-boundary sources of absorbing aerosols over the selected cities.


Introduction
Aerosols, i.e., liquid or solid particles suspended in atmosphere, have significant impacts on precipitation and hydrological cycle (Gautam et al. 2022), air quality (Zhao et al. 2022), visibility (Li et al. 2021), energy budget (ul-Haq et al. 2017), climate system (Tariq 2020) and human health (Mehmood et al. 2021a;Qayyum et al. 2021;Ren et al. 2021).Aerosols are emitted from natural (e.g., desert dust, volcanoes, sea spray and wildfires) as well as human activities (e.g., fossil fuels, agricultural waste and biomass burning, vehicular and industrial emissions) (Bilal et al. 2021a;Kumar et al. 2018;Tariq et al. 2023).The concentration of atmospheric aerosol varies with geographical area and seasons and depends on the annually varying meteorological and soil moisture conditions, and intensity of fires during fire seasons.Further, the types of aerosols vary significantly in volumetric emissions, seasonality, and location, in their residence time and transport in atmosphere, and in the absorptive properties of aerosols (Samset et al. 2018).According to the degree of absorption of solar radiations, aerosols are categorized as scattering and absorbing.
Absorbing aerosols possess short life span, variable characteristics and their distribution differs in response to spatiotemporal differences (Li et al. 2022;Tariq and Ali 2015).These aerosol particles alter atmospheric stability by absorbing the incoming solar radiations that led to warming of the atmosphere.This heating process also evaporates cloud droplets and reduces cloud cover which ultimately affects the climate.Moreover, the highly variant spatiotemporal distribution and non-uniformity in their sources unable us to understand the aerosols climate interactions (Logan et al. 2013).Therefore, by having depth knowledge of the absorption characteristics and spatiotemporal distribution of aerosols, uncertainties in climate change and weather forecasting can be minimized.
Satellite remote sensing and In-situ measurements are the primary source of acquiring optical, physical, and radiative properties of aerosols.Although, the ground-based measurements are more accurate and reliable, but ground monitoring stations are sparse that make difficult to study large scale variation.Satellite remote sensing enable us to investigate large scale spatiotemporal distribution and characteristics of aerosol particles including ultraviolet aerosol index (UVAI).The UVAI is calculated by comparing spectral variations in two ultraviolet wavelengths (Herman and Celarier 1997).The magnitude of UVAI is dependent upon viewing geometry, aerosol optical depth (AOD), single scattering albedo (SSA) and aerosols layer height (Torres et al. 2007a).Furthermore, UVAI is also used to investigate long-range transport of pollutants, climate forcing, energy budget and to model and forecast air quality (Ahmad et al. 2006).
Several studies analyze aerosols properties, their sources, formation mechanism and their relationship with meteorological parameters in South Asia (Ansari and Ramachandran 2023;Bilal et al. 2023;Mhawish et al. 2022).For example, Tariq et al. (2022a) studied long-term spatiotemporal trends of moderate resolution imaging spectroradiometer (MODIS) retrieved AOD over South Asia and found AOD > 0.7 over Indo Gangetic Plain (IGP) because of increase in economic, agricultural and industrial activities, population density and agriculture residue burning.Ramachandran and Rupakheti (2022) found increase and decrease in AOD over South Asia and East Asia, respectively.They also observed increase in non-absorbing nature of aerosols over IGP and North China Plain.Ramachandran et al.(2022) found decrease in heating rate due to decline in AOD and increase in SSA over East Asia.They also observed a rise in both SSA and AOD over South Asia.Shah et al. (2023) studied temporal and spatial variations of absorbing aerosols over Hindukush-Himalaya-Karakoram regions.They found high concentration of black carbon (0.31 ± 0.04 µg/m 3 ) and organic carbon (1.32 ± 0.32 µg/ m 3 ) during winter and autumn seasons, respectively.Srivastava (2017) observed positive trends in AOD (> 70%) in India during post-monsoon and winter seasons because of lower boundary layer height and increase in biomass burning.They also reveal increase in fine mode aerosol particles over Indian subcontinent and Bay of Bengal.Wang et al. (2021) also found low SSA and high FMF values indicating dominance of absorbing aerosol in Southeast Asian peninsular.Khan et al. (2023a) studied variations of Ozone Monitoring Instrument (OMI) retrieved aerosol index over Pakistan and found increasing trend of UVAI at 0.71% spring −1 , 0.81% summer −1 , 3.24% autumn −1 , and 3 0.97% winter −1 .(Tariq and Ali 2015) observed AI of 1.22 and 1.08 over southern and northern Pakistan respectively during 2004-2008. Tariq et al. (2021a) ) used fine mode fraction (FMF) and SSA to classify different types of aerosols over Lahore and Kanpur during smog days.They found black carbon aerosol as the dominant type during smog days.Tariq et al. (2023) studied nighttime variations of aerosol over Lahore (Pakistan) and found dominance of background conditions, urban/industrial, biomass burning, and mixed type aerosols.
Exploring the trends in aerosol characteristics (e.g., AOD, AE and UVAI), their sources and driving forces using insitu and remotely sensed measurements helps in identifying implications for climate modelling and mitigation.Duan et al. (2021) used pixel-based analysis to examine spatiotemporal distribution of absorbing aerosols over three Northeast provinces in China i.e., Heilongjiang, Jilin, and Liaoning.Li et al. (2019) also employed pixel-based approacsh to estimate variation in UVAI coefficient in Gansu, China.Tariq et al. (2022a) also applied pixel-based approach to study correlation of AOD with enhanced vegetation index, temperature, wind speed and relative humidity over Pakistan and South Asia.Therefore, this study investigates spatiotemporal distribution, interannual and seasonal variations in absorbing aerosols using OMI retrieved UVAI and their relationship with total ozone column, total surface precipitation and temperature over South Asia during October 2004 -March 2022.HYSPLIT cluster analysis is performed to traceback the origin and sources of aerosols over megacities of South Asia.

Site description
South Asia is located within the latitudes and longitudes of 0-40 ○ N and 60-100 ○ E respectively covering an area of ~ 5,134,613 km 2 .It has population of 1.891 billion which makes it world's most densely populated region (303 peoples per Km 2 ).South Asian region comprises of eight nations including Pakistan, Afghanistan, Nepal, India, Sri-Lanka, Bhutan and Bangladesh (Malatesha Joshi 2015).The study area covers five subregions including Himalayan and Karakoram range in the North, lowlands in the South expanding from Pakistan to densely populated Bangladesh, Baluchistan Plateau along the southern border of Pakistan and Afghanistan, Indian peninsular and the island realm (Maldives and Sri-Lanka) (Ul-Haq et al. 2015).South Asian region is rich with valleys, glaciers, rain forests, grasslands, and deserts.
In South Asia, climate varies from temperate to tropical monsoon in the North and South, respectively.Most of the rainfall is received from monsoon weather systems particularly in Pakistan and India.Winter season is characterized by little rainfall due to outward flow of cold and dry winds over Himalayas.In spring season, these winds diminished making dry and hot season ahead.The variable and low precipitation in Rajasthan state of India and most of Pakistan results in steppe and desert climates (Bandara and Cai 2014;Zhang et al. 2015).
Rapid population growth and increased economic activities in South Asia led to increase in energy consumption which in turn increases high aerosol loading in South Asian countries.Moreover, highly variant topography and complex climatic conditions makes this region vulnerable for studying spatiotemporal patterns of particulate pollution (Dey et al. 2004;Gurjar et al. 2016;Tariq et al. 2022a).The study area map of South Asia portraying spatial distribution of enhanced vegetation index is shown in Fig. 1.

Ozone Monitoring Instrument
Ozone Monitoring Instrument (OMI) is a nadir viewing visible and ultraviolet spectrometer mounted on the NASA polar orbiting sun-synchronous EOS Aura spacecraft.OMI sensor effectively monitors ozone layer as well as differentiates between several types of aerosols and thus offer new perspectives for study of air pollution.It measures backscattered solar radiations within the wavelength ranges of 270 nm-500 nm.The average spectral resolution of OMI is 0.5 nm and it offers daily global coverage with a wide swath of 2600 km (Levelt et al. 2006;Millet et al. 2008;ul-Haq et al. 2014).The altitude of OMI is 705 km with equatorial crossing time of 13:45 ± 15 min.
Reflectance from the top of the atmosphere is utilized to extract several aerosol properties including aerosol index by employing a multi wavelength algorithm.The OMI derived aerosol index is a technique of identifying absorbing aerosols in the atmosphere from remotely sensed measurements in the ultraviolet region of electromagnetic spectrum (Hammer et al. 2016;Torres et al. 2007).This study uses OMI retrieved UVAI daily data product (OMAERUVd) level 3 version 3 acquired from NASA Giovanni online website (https:// giova nni.gsfc.nasa.gov/) during October 2004-March 2022 over South Asia.The OMAERUVd product contains absorption and extinction optical depths at 355 nm, 388 nm and 500 nm of wavelengths.Furthermore, The OMAERUVd data product recognizes clouds by using aerosol index and measured scene reflectivity (Torres et al. 2013).Therefore, all the OMI derived UVAI data, used in this study, is filtered based on high confidence and quality flag 1 (cloud free conditions) and reject those of quality flag of 0 (cloudy conditions).
The Tropical Rainfall Measuring Mission (TRMM) retrieved precipitation, 2-m air temperature from MERRA-2 model, vector wind composite mean from National Oceanic and Atmospheric Administration (NOAA) NCEP/NCAR Re-analysis data (https:// psl.noaa.gov/) and Moderate Resolution Imaging Spectroradiometer (MODIS) retrieved EVI were downloaded from NASA Giovanni (https:// giova nni.gsfc.nasa.gov) while ERA-5 wind speed and wind direction were obtained from Google Earth Engine.The gross domestic products (GDP), primary, secondary, and tertiary gross valueadded industries and energy use (kg of oil equivalent per capita) datasets were downloaded from World Bank indicators (https:// data.world bank.org/).

HYSPLIT model
The Hybrid Single Particle Lagrangian Integrated Trajectory (HYSPLIT) model designed and developed by NOAA is employed to determine backward trajectory of air pollution (Chen et al. 2020;Yan et al. 2021).The re-analysis data from the National Center for Atmospheric Research and National Center for Environmental Prediction (NCAR/NCEP) is used to perform the backward air mass trajectories cluster analysis.In the past, HYSPLT model is extensively used to check forward and backward transport pathways of aerosols from biomass burning, Industrial emissions, and volcanic eruptions (Alam et al. 2010;Chai et al. 2017;Tariq et al. 2022b).The HYSPLIT model suppose that the particle's movement trajectory is (i) floating with the winds and (ii) integration of vector in both time and space.The model compute trajectory using the formula given below: where, P x Initial position.P y 1 st supposed position.v Wind speed.t Time when the position of the particle is P x (t + Δt) = P x and.Δt Step length.
The range of changes should lie between 1 min and ~ 1 h.The air mass trajectory must comply with v max Δt < 0.75 grid distance.

Unary linear regression analysis
The unary linear fit model is applied to analyze spatiotemporal trend in UVAI over South Asia from October 2004 to (1) P x (t + Δt) = P x (t) + 0.5[v(P x , t) + v(P y , t + Δt)]Δt (2) P y (t + Δt) = Px(t) + v(P x , t)Δt March 2022 (Yimir et al. 2019).Mathematically, the model can be expressed as: The positive value of slope signifies increasing trend of UVAI while negative value of slope shows decreasing trend of UVAI.Higher the slope corresponds to significant change trend of UVAI.

Pixel-based spatial analysis
The pixel-based analysis approach is employed to investigate the correlation of the UVAI with total surface precipitation, 2-m air temperature, surface wind speed, enhanced vegetation index and total ozone column (Tariq et al. 2022a(Tariq et al. , 2021b;;Wu et al. 2019).Mathematically, the correlation coefficient can be calculated as: where, r xy is called correlation coefficient whose value lies between -1 and + 1, xi mean UVAI.yi total surface precipitation, 2-m air temperature, enhanced vegetation index, total ozone column and distribution of population.i year.x̄ multiyear average UVAI value.ȳ multi-year average of total surface precipitation, 2-m air temperature, total ozone column and wind speed.n sample size of the data.

UVAI time and spatial distribution
The spatial distribution of UVAI and vector wind composite mean over South Asia during October 2005 to March 2022 has been shown in Fig. 2. High UVAI values in the ranges of Figure 2b shows that winds from the west are approaching India, Bangladesh, Nepal, and Bhutan having wind speed of ~ 1 ms −1 .High speed winds within the range of 3-4.5 ms −1 blew over Pakistan and Afghanistan as shown in Fig. 2b.Two types of wind patterns are found over Pakistan i.e., wind from North and South that causes substantial increase in anthropogenic and dust aerosols over Pakistan, respectively.South Asian region receives natural aerosols from the adjoining ocean and southern dry lands whereas anthropogenic aerosols comes from urban and industrial emissions (Ali et al. 2020;Kumar et al. 2015;Sen et al. 2017).

Inter-annual UVAI spatial distribution
The interannual variations in spatial distribution of UVAI over South Asia from 2005 to 2021 have been shown in Fig. 3. High values of UVAI are observed over Eastern and   Mhawish et al. (2021) showed that fine mode aerosols contribute largely to the total aerosol burden then coarse mode aerosols during autumn and winter seasons over India, Bangladesh and Nepal whereas coarse particles dominate in the atmosphere over Pakistan.Sijikumar et al. (2016) showed that high dust aerosol loading during summer monsoon over the IGP regions and Arabian sea is transported from the Rajasthan desert, India, and Arabian desert.
The positive values of slope (41-52) with UVAI ≤ 0.18 shows increasing trend during winter season along the Himalaya and Karakoram range as shown in Fig. 5.The slope of mean UVAI ≥ 42 over Afghanistan, Northern and Western Pakistan, Western India, Bhutan, and Sri-Lanka suggest increasing trend of absorbing aerosols with UVAI ≤ 1.16 during summer season.The maximum positive slope of 69 is observed in North and South of Pakistan during spring season having UVAI values of 1.16.In autumn season, maximum slope ( 45) is observed over Northern Pakistan extending towards lowland Terai region of India.Kumar et al. (2018) found statistically insignificant increasing trend of AOD at the rate of 0.002 year −1 over the IGP region.Ramachandran and Rupakheti (2022) also found persistent increase in AOD over the South Asian region.
The temporal variations of the UVAI values over South Asia during winter (DJF), spring (MAM), summer (JJA) and autumn (SON) seasons from 2004 to 2022 has been shown in Fig. 6.The maximum value of UVAI is observed to be 1.92 in winter-2019 followed by 1.86, 1.46 and 0.95 during spring-2021, summer-2021 and autumn-2021, respectively.The coefficient of determination (R 2 ) is found to be 0.4383, 0.153, 0.2417 and 0.4525 during winter, spring, summer, and autumn season, respectively.The values of UVAI show increase at the rate of 0.9064 DJF −1 , 0.0774 MAM −1 , 0.3810 JJA −1 and 0.2707 SON −1 respectively during October 2004-March 2022.Srivastava, (2017) found positive and negative trends in AOD in Southern and Northern South Asia respectively.They also found 2-8% per anum increase in AOD in the Northeastern India during post-monsoon season.Mhawish et al. (2021) observed 30% to 44% increase in aerosol loading over IGP region and Eastern coast of India.Pakistan which includes Gilgit, Astore, Skardu and Sost.Contrary to this, the maximum positive correlation coefficient between UVAI and 2-m air temperature is found to be 0.18 over Southern Pakistan and India, and Sri-Lanka while negative correlation of -0.24 is observed over IGP region, Bhutan, Bangladesh, and Western Nepal.Dave et al. (2020) observed co-variability of temperature maxima and UVAI in the Northwest and central India and this effect was found more persistent on seasonal and heat waves events scales, and in the presence of enhanced UVAI.Negative correlation is observed between UVAI and total ozone column over Western and Southwestern Afghanistan, Northern and Northeastern Pakistan and Northern India within the ranges of -0.09--0.24.Conversely, the maximum positive correlation (0.43) is found over Southern Pakistan and India.This shows that an increase in dust absorbing aerosols will led to increase in total ozone column in the region.Gharibzadeh et al. (2021) found negative correlation of tropospheric ozone with AOD, PM 10 and fine mode fraction of AOD during spring and summer seasons whereas positive correlation of tropospheric ozone with fine mode fraction of AOD during autumn and winter seasons over Zanjan, Iran during 2008-2018.The highest positive correlation coefficient between UVAI and wind speed is to be 0.56 over Southern Pakistan whereas lowest negative correlation is found over Bhutan and Western Bangladesh as shown in Fig. 7d.Low relative humidity, high wind speed and less precipitation causes the dust particles to stay longer in the atmosphere which in turn increases UVAI (Attiya et al. 2019).Figure 10 shows temporal variations of UVAI over megacities of South Asia including Lahore, Karachi, Kanpur, Bangalore, New-Delhi, Varanasi, Dhaka, Kathmandu, Kabul, Sri Jayawardenepura Kotte and Thimphu during October 2004 -March 2022.UVAI shows increasing trend with slope of 0.0049, 0.0038, 0.0029, 0.0028, 0.0035, 0.0030, 0.0033, 0.0036, 0.0030, 0.0027 and 0.0030 and intercept of 0.4324, 0.4166, 0.2966, -0.0461, 0.4318, 0.3057, 0.1745, -0.141, -0.0822, -0.0581 and -0.1731 for Lahore, Karachi, Kanpur, Bangalore, New-Delhi, Varanasi, Dhaka, Kathmandu, Kabul, Sri Jayawardenepura Kotte and Thimphu respectively.The coefficient of determination (R 2 ) is found to be 0.1409, 0.1458, 0.0771, 0.1655, 0.0942, 0.0753, 0.1057, 0.1458, 0.123, 0.1735 and 0.1704 for Lahore, Karachi, Kanpur, Bangalore, New-Delhi, Varanasi, Dhaka, Kathmandu, Kabul, Sri Jayawardenepura Kotte and Thimphu respectively.The UVAI over all megacities show almost similar trend until October 2018, after that remarkable increase is observed with extremely high UVAI values till March 2022.The maximum value of UVAI over Lahore (5.55), Karachi (4.47), Kanpur (4.51), Bangalore  Table 1 shows slope, y-intercept, coefficient of determination (R 2 ) derived from linear regression equation of mean annual UVAI and per year change in UVAI from 2005 to 2021.The UVAI is increasing at the rate of 0.1409, 0.1124, 0.1224, 0.1015, 0.1242 and 0.2054 per year in Lahore, Karachi, Kanpur, New-Delhi, Varanasi, and Dhaka respectively while decreasing at the rate of 0.8173, 0.3197, 0.4307, 0.5969 and 0.2265 per year in Bangalore, Kathmandu, Kabul, Sri Jayawardenepura Kotte, and Thimphu, respectively.In the IGP region and the Himalayan foothills, crop residue burning, forest fires Aerosol Index

Months
and long-range transport of aerosols from the polluted areas causes increase in aerosols loading at high attitude sites including Kathmandu.

Hysplit backward trajectory cluster analysis
The dust aerosols are formed mainly on the less vegetative and dry surfaces and are transported during high speed winds(> 5 ms −1 ) from the source to the receptor region (Prospero et al. 2002).To locate the sources and origin  of atmospheric pollutant, HYSPLIT backward trajectory model is used in this study.HYSLIT backward trajectory cluster analysis and wind rose plot for the year 2021 over all the megacities are shown in Fig. 12.The HYSLIT cluster analysis shows that 36% of air masses are originating from North, 33% from the east of Lahore carrying urban and industrial aerosols while 21% of air masses are coming from the western India carrying biomass burning aerosols over Lahore.Tariq et al. (2015) applied HYSPLIT model backward trajectories analysis and found transport of biomass burning aerosols over Lahore from southeast and northwest directions.For Karachi, about 40% of the backward air mass trajectories brought oceanic aerosols (e.g., sea salt) from the south while 20% and 30% of air masses are originating from Cholistan and Thar deserts, respectively.This reveals that the absorbing aerosols are mostly natural (dust) aerosols over Karachi.Sharif et al. (2015)

Anthropogenic productivity analysis
The results of correlation matrix of the UVAI, primary, secondary and tertiary industry, energy use and GDP from 2004 to 2021 for South Asian countries are given in Table 2.A correlation coefficient (R) is found to be -0.3910,-0.5129, 0.3253, 0.3129 and 0.5601 of UVAI with primary industry, secondary industry, tertiary industry, energy use and GDP.The outcomes signify that tertiary industry, energy consumption and GDP contribute to the UVAI values.The gaseous products emitted during energy consumption are the major sources of absorbing aerosol particles (Duan et al., 2021b).The increase in GDP and energy consumption positively impact gaseous emissions (Mehmood et al. 2021b) which in turn increases waste gases emission and ultimately absorbing aerosols.Contrary to this, negative correlation of the UVAI with primary industry and secondary industry reflect that these two contribute in lowering UVAI values.

Conclusion
The The interaannual variations in UVAI reveal decreasing trend of UVAI with highest UVAI (0.64) during December while lowest UVAI (0.08) during September.The UVAI depicted increasing trend at the rate of 0.1409, 0.1124, 0.1224, 0.1015, 0.1242 and 0.2054 per year in Lahore, Karachi, Kanpur, New-Delhi, Varanasi, and Dhaka respectively while decreasing trend at the rate of 0.8173, 0.3197, 0.4307, 0.5969 and 0.2265 per year in Bangalore, Kathmandu, Kabul, Sri Jayawardenepura Kotte, and Thimphu, respectively.The HYSLIT cluster analysis shows the long-distance transport of the natural dust, urban, industrial and biomass burning aerosols over Lahore, Kanpur, New-Delhi, Varanasi, Dhaka, Kathmandu, and Thimphu.This study enables us to identify the regions with high absorbing aerosols concentrations in South Asia using

Fig. 1
Fig. 1 Study area map of South Asia showing distribution of Aqua-MODIS EVI the year.UVAI i UVAI of the year or a season.
-1.62 are observed over Pakistan, adjoining regions of India and southern Afghanistan.High UVAI values (> 0.69) over Southern and Northeastern Pakistan are associated with long-range transport of dust aerosols from the Cholistan, Thar and Sahara desert (Alam et al. 2011b).(Tariq and Ali 2015) observed mean annual AI of 1.155 ± 0.257 over Pakistan during 2004-2008.In Bangladesh, UVAI varies between 0.46-0.55.The values of UVAI ranges between 0.56-0.85over IGP region and 0.27-0.36over Central, Eastern and Western parts of India associated with the presence of dust, biomass burning aerosols and urban/industrial emissions.Conversely, low UVAI values within the ranges of 0.04-0.17are observed along the Karakorum and Himalaya range.

Fig. 2
Fig. 2 shows (a) spatial distribution of UVAI and (b) wind direction (ms.−1 ) over South Asia during October 2004 to March 2022

Fig
Fig. 3 shows annual variations of the UVAI over South Asia during 2005-2021

Fig. 3
Fig. 3 (continued) The interannual variations in mean UVAI over South Asia during 2005-2021 has been shown in Fig. 8. Highest UVAI of 1.59 is observed in 2021 whereas lowest UVAI of 0.19 is found in 2007.The positive value of UVAI in both years (2007 and 2021) suggest dominance of absorbing aerosols in the atmosphere.The interannual variations in mean UVAI

Fig. 5 Fig. 6 Fig. 7
Fig. 5 The slope of UVAI values over South Asia during each of four seasons from 2004 to 2022

Fig
Fig. 12 portrayed HYSPLIT backward trajectories cluster analysis(right) and wind rose (left) over megacities of South Asia during 2021

Seasonal trends and spatiotemporal distribution of absorbing aerosols
Choi et al. (2021)8)awish et al. 2021;Tariq et al. 2021b).The westerly winds are responsible for the long distance transport of soil/ dust aerosols from the deserts of Africa and Arabia to the IGP region(David et al. 2018; Prasad and Singh 2007).Pandey et al. (2017)found decrease in dust aerosol loading by 10-20% during the spring season of 2015 as compared to the year 2000 over Indian region because of increase in rainfall.(Nakataetal.2018)observed high aerosol optical thickness during autumn season associated with biomass burning over Southeast Asia.Choi et al. (2021)found dominance of absorbing aerosols over the capital cities of Asia during 2018-2020.
Fadnavis et al. (2019)020) (2020)have found extremely high black carbon and dust aerosol absorption over the IGP region and the Himalayan foothills.Banerjee et al. (2021)studied aerosol climatology over South Asia and South-East Asia using A-Train satellite retrieved products during 2010-2020.They observed increase in UVAI over IGP region during haze dominating days.They also reveal dominance of dust aerosols over Karachi and absorbing smoke particles over IGP region.Fadnavis et al. (2019)found aberrant increase in aerosol in the Asian tropopause aerosols layer that results in weakening of monsoon circulation and thus exacerbates the drought severity.
found that air parcel arriving at Karachi comes from the Arabian sea, Thar Desert of Pakistan and India and Persian Gulf causing presence of present study investigates spatiotemporal distribution of UVAI over South Asia and its relationship with precipitation, air temperature, wind speed and total ozone column during October 2004 -March 2022.We also examined the seasonal distribution of UVAI, seasonal trends of UVAI, interannual and intraannual trends of UVAI over South Asia and selected cities.The HYSPLIT backward trajectories cluster analysis is used to trace back the origin and sources of particulate matter pollution in selected cities.The UVAI > 0.69 is observed over Southern and Northeastern Pakistan associated with long-range transport of dust aerosol particles from Thar, Cholistan and Sahara Deserts.The maximum value of UVAI (3.11) is found over Pakistan during 2021 followed by 2.84 over IGP region, 2.03 over Bangladesh and 1.48 over Sri-Lanka.Seasonally, the UVAI of ~ 1.90 is observed over central and southern Pakistan during summer season while UVAI values within the ranges of 1.16-1.44 are observed over Northeastern Pakistan, Northern India, and Bangladesh during winter season.Spring season is characterized by large scale distribution of absorbing aerosols with UVAI values of 0.61-1.16over Northeastern and southern Pakistan, India, and Bangladesh.The UVAI showed increasing trend at the rate of 0.9064 DJF −1 , 0.0774 MAM −1 , 0.3810 JJA −1 and 0.2707 SON −1 respectively during October 2004-March 2022.

Table 1
shows slope, y-intercept, and coefficient of determination (R 2 ) and per year change in mean annual UVAI values from 2005 to 2021

Table 2
Correlation matrix of UVAI with anthropogenic drivers of UVAI in South Asia during[2004][2005][2006][2007][2008][2009][2010][2011][2012][2013][2014][2015][2016][2017][2018][2019][2020][2021]remote sensing technique.Since, absorbing aerosols causes warming of the atmosphere and affect precipitation therefore, reducing the emissions of absorbing aerosols can help us to mitigate the climate change issue.Moreover, lowering absorbing aerosols during winter times South Asia further helps to control worsen air quality problem.Acknowledgements The authors acknowledged NASA Giovanni for providing OMI retrieved UVAI, TRMM retrieved precipitation, 2-meter air temperature from MERRA-2 model, and MODIS retrieved enhanced vegetation index (https:// giova nni.gsfc.nasa.gov/).We also acknowledged the use of ERA-5 wind speed and wind direction obtained from Google Earth Engine API, and gross domestic products (GDP), primary, secondary, and tertiary gross value-added industries and energy use datasets from World Bank indicators (https:// data.worldbank.org/).Authors contributions HN wrote the manuscript.ST conceptualized the work.ZH supervised the work.UM conducted the analysis.Funding Open Access funding enabled and organized by CAUL and its Member Institutions * = statistically significant at 10%