Tropospheric NO2 Trends over South Asia during the Last Decade (2004–2014) Using OMI Data

The focus of this study is to assess spatiotemporal variability of tropospheric NO 2 over South Asia using data from spaceborne OMI during the past decade (2004–2015).We find an average value of NO 2 1.0 ± 0.05 × 10molec/cm and a significant decadal increase of 14%.The elevating NO 2 pollution over the region is linked to rise in motor vehicles and industrial and agricultural activities and increase in biomass fuel usage.The observed seasonality ofNO 2 is associatedwith change inmeteorological conditions and seasonal cycles of anthropogenic emissions. OMI data reveal a seasonal peak in spring followed by winter largely linked to metrological conditions and anthropogenic emissions from crop residue and biomass burning for heating purpose, and low concentration in summer ismostly attributed tometeorological conditions. Significant increase, up to 42%, inNO 2 concentrations over northwestern IGB, is observed connected to large scale postmonsoon crop residue events of 2010 and 2012. It is seen that NO 2 is mounting over all the hotspot locations and most of the cities. Dhaka shows the highest increase of 77% followed by Islamabad (69%), Kabul (68%), Korba (64%), Bardhaman (47%), and Lahore (40%). On the contrary, DG Khan has shown negative trend of −11%.


Introduction
Nitrogen dioxide (NO 2 ) plays an important role in the modification of radiative balance of the Earth's atmosphere by changing its oxidizing capacity and chemistry and by influencing the lifetimes of important greenhouse gases.Its high concentration in the troposphere adversely impacts inhabitants of the planet [1,2].NO 2 is mainly emitted during industrial burning, vehicle combustion, biomass fuel and crop residue burning, direct soil emissions, and natural lightning [3,4].The dominant sink of NO 2 is its oxidation process, involving hydroxyl radical (OH) and solar ultraviolet (UV) radiations, by which secondary pollutants such as ozone, nitric acid, methane, and aldehydes are produced [5][6][7][8][9].
Tropospheric NO 2 shows high spatiotemporal variability mainly modulated by local emission changes, seasonal cycles, and meteorological conditions.South Asia is experiencing severe air quality degradation due to high population growth rate, burgeoning urbanization and industrialization, expanding demand of agricultural products, and exponentially increasing energy consumption rate (e.g., [10][11][12][13][14]).Therefore, in order to develop the effective strategies to reduce its emissions, it is necessary to assess spatiotemporal distribution of NO 2 and identify its emission sources over the study region.
As far as tropospheric NO 2 assessments are concerned, no study has so far been carried out over the whole South Asian region.However, a few studies have been conducted only on some parts of South Asia to investigate spatiotemporal variations and to identify NO 2 emission hotspots using satellite remote sensing technique [15][16][17][18][19][20].Recently, Ul-Haq et al. [20] assessed spatial and temporal patterns of NO 2 over Pakistan using Ozone Monitoring Instrument (OMI) retrieved data for the period 2004-2008.Renuka et al. [19] analyzed long-term changes of tropospheric NO 2 over south India using retrievals from GOME/ERS-2 and OMI/Aura during 1996-2014.They reported an increase in NO 2 column over Gadanki (13.48 ∘ N, 79.18 ∘ E).In spite of increase in anthropogenic emissions (e.g., population and vehicular traffic), these authors also found decreasing trend between eastern and western Ghats possibly linked with changes in land-use pattern limiting the soil emissions of NO 2 .Ramachandran et al. [18] studied tropospheric NO 2 over India and identified NO 2 hotspots during 2002-2012 based  on measurements from SCanning Imaging Absorption spec-troMeter for Atmospheric CHartographY (SCIAMACHY, on board Envisat satellite; [21,22]).The study of David and Nair [17] was based on OMI data to detect seasonal changes and trends over Indian region for a period from 2007 to 2008.Ten-year (1996Ten-year ( -2006) ) data obtained from Global Ozone Monitoring Experiment (GOME) and SCIAMACHY were used to study trends and seasonality over India [15] and to detect and analyze global NO 2 hotspots [16].
However, compared to previous studies, the present study offers two advancements, that is, complete coverage of the South Asian region for the first time and usage of higher spatial resolution data from OMI to identify small and localized emission sources of NO 2 .In this study, spatiotemporal distribution of tropospheric NO 2 and identification of hotspots in South Asia are presented by using OMI data for a period from October 2004 to January 2015.

Geography and Meteorology of South Asia
South Asia is the second most populous region in the world having population over 1667.8 million with land surface area of approximately 5,134,613 km 2 .According to the South Asian Association for Regional Cooperation (SAARC), this region consists of eight countries: Afghanistan, Pakistan, India, Nepal, Maldives, Sri Lanka, Bangladesh, and Bhutan [23] (Figure 1).
The study region can be divided into five extensive physical subregions: the high Himalayan and Karakoram mountains in the north; the southern lowlands (Indus-Ganges-Brahmaputra) that expand from Pakistan to the delta lands of Bangladesh making up the core and densely populated areas; Balochistan Plateau, the most dry area of the study region that covers the Suleman and Kirthar mountains in southern boundary of Afghanistan and Pakistan; the peninsular India, dominated by the Deccan Plateau bordered by narrow and fertile coastal plains backed by elongated north-south mountain ranges called western Ghats and eastern Ghats; and the island realm that includes Sri Lanka and Maldives [24].
Monsoon weather systems are the dominant climatic factors for the most of South Asia especially in Pakistan and India.In winter season cold and dry winds flow outward due to a large high-pressure system over the Himalayas and down across South Asia causing small amount of rain, whereas, in spring, these winds diminish, resulting in hot and dry season.In June low pressure over landmass draws in clean, warm, and moist air from the Arabian Sea (ArS), Bay of Bengal (BoB), and Indian Ocean (InO).The uplifting and cooling of these moist monsoon winds result in heavy rain fall, but not all of South Asia receives substantial rainfall from the southwest monsoon.In much of Pakistan and the Indian state of Rajasthan, precipitation is low and variable, resulting in steppe and desert climates [25][26][27][28].
Eleven statistical significant hotspots of NO 2 concentration have been identified which include Delhi, Mumbai, Kolkata, Korba, Singrauli, Bardhaman, Pakur, and Angul located in India, Lahore and Karachi from Pakistan, and Dhaka sited in Bangladesh.The basic climatic parameters of these hotspots are shown in Figure 2.

NO 2
Retrievals by OMI/Aura.NASA's Aura satellite team, celebrating its 10th anniversary of operations, has provided vital data about the chemistry and dynamics of Earth's atmosphere from the surface through the mesosphere.This satellite carries four instruments, that is, High Resolution Dynamics Limb Sounder (HIRDLS), Tropospheric Emission Spectrometer (TES), Microwave Limb Sounder (MLS), and Ozone Monitoring Instrument (OMI).In this study, because of improved algorithms and sensitivity of OMI for NO 2 detection at lower atmosphere, we have used its tropospheric NO 2 daily averaged product (OMNO2d.003,level 3).This sensor is a wide-field-imaging grating spectrometer with horizontal resolution of 13 × 24 km 2 at the nadir point and it uses push-broom mode to measure the backscattered solar radiations (270-500 nm with spectral resolution of about 0.5 nm, [29][30][31]).Tropospheric NO 2 columns are retrieved by using Differential Optical Absorption Spectroscopy (DOAS) analysis in the 405-465 nm spectral range [32,33].Details of data filtering, DOAS analysis and algorithm, and data quality control procedures can be found in NASA's online user's manual for OMI products (https://earthdata.nasa.gov/).
It has been well established that NO 2 measurements retrieved from satellites are in good agreement with in situ measurements and bottom-up emission inventories [34][35][36][37][38][39][40].The studies of Boersma et al. [41] and Celarier et al. [42] reviewed and validated the OMI-NO 2 with ground (MAX-DOAS instruments) and aircraft (DC-8 aircraft) measurements.They showed a good agreement between the tropospheric OMI-NO 2 column and ground-based measurements,       [46,47] is used to identify statistically significant NO 2 concentration hotspots.The Getis-Ord  *  tool can be utilized for spatial clustering and autocorrelation [48].This tool has been applied by using ArcGIS's Spatial Statistics tools and described in the following equations: where   is the attributive value for a point ,  indicates the place of a particular value in a dataset relative to the mean, standardized with respect to the standard deviation.The -score represents the statistical significance (90% significant at  ≥ 1.645, 95% significant at  ≥ 1.960, 99% significant at  ≥ 2.576, and 99.9% significant at  ≥ 3.291) of clustering for a specified distance.For statistically significant positive -scores, the larger the score, the more intense the clustering of high values (hotspot) [49].A high -score for a point indicates its neighbors have high attribute values and -score near zero indicates that neighboring points have a range of values [50].The Inverse Distance Squared Method, appropriate for this type of data where the closer features influence each other, with threshold distance of 1 ∘ has been used in hotspot identification [49].We have obtained -score values ranging from −4.43 to 22.18.Any location with its surrounding areas having sufficiently extreme -score ( ≥ 10) is selected to be a real statistically significant hotspot of NO 2 concentration.
In the present work, we have used annual mean values of OMI-NO 2 to calculate NO 2 average values, and the percentage increase calculation is based on linear trend line equation; that is, -intercept represents initial concentration value.The correlations of hotspots are based on monthly mean values.

Results and Discussion
4.1.Temporal Distribution of NO 2 .The OMI retrievals show annual average value of tropospheric NO 2 column to be 1.0 ± 0.05 × 10 15 molec/cm 2 over South Asia during the study period.A linear regression on the annual mean data shows a statistically significant (at confidence level of 99.9%) decadal increase of 14% with a slope of 0.013 (±0.002), correlation coefficient ( 2 ) 0.723, and  intercept at 0.930 (±0.017) × 10 15 molec/cm 2 .This positive trend is quite consistent with the trend (1.76 ± 1.1% per year) reported by Ghude et al. [16] for a region consisting of India, Pakistan, Bangladesh, and Nepal and trend reported by Ghude et al. [15] for India (1.67% per year).During the study period the highest annual value of 1.07 × 10 15 molec/cm 2 is found in 2011 and lowest value of 0.93 × 10 15 molec/cm 2 in 2006 (Figure 3).
The observed increase of NO 2 in the region may be attributed to increase in anthropogenic emissions due to expansion in traffic volume, increasing power generation, flourishing industries, rapid urbanization, more demand of agricultural products, and more biomass fuel usage (e.g., [10-16, 19, 20, 51-54]).
The main sources of NO 2 are different for South Asian countries.The level of urbanization in Pakistan is now the highest in South Asia [55].During this period the escalating urban population, coupled with more demand in some sectors such as cement production, industrial expansion, motor vehicles usage, and electricity generation, has resulted in elevated levels of NO 2 emissions.In Pakistan road transport is the dominant mode of passengers that is responsible for carrying 91% of the national passenger traffic and 96% of the freight movements.During 2000 to 2010, the number of vehicles on the roads has grown from 4 to 9.8 million showing an increase of 145% [56].The factors that are responsible for this huge increase are the import of reconditioned vehicles, popularity of vehicle financing schemes, and the President and the Prime Minister's Rozgar schemes (self-employment schemes) with heavy investments at low mark-up to purchase small auto vehicles like rickshaws and taxies [53].The economy of Pakistan is heavily dependent on the agricultural sector.The total agricultural waste burnt is estimated to be 1704.9thousand tons per year in the rice-wheat cropping system in Pakistan [57] contributing significant emissions of NO 2 .
In India, the road transport is the dominant source of NO 2 emissions as compared to industry and power sector.The number of vehicles, registered in India, was 55 million in 2001 which has grown to around 159.5 million by 2012 [58].Another important source of NO 2 emissions is the industrial process, especially the production of nitric acid, used in fertilizer manufacturing.India is an agrarian country and generates a large quantity of agricultural wastes.The annual crop residue generated in India for 2008-2009 is found to be about 620 Mt/year of which 15% is burnt on farms emitting huge amount of NO x for the 2008-2009 [59].Significant amount of NO 2 is emitted from coal fired power plants.India is the biggest energy user, followed by Iran and Pakistan.Coal is India's most abundant source of energy and currently almost 60% of its commercial energy needs are fulfilled by it.Lastly, widespread use of traditional sources of energy such as fuel wood and animal dung has also been contributing to NO 2 emissions.Estimates indicate that nearly 3 in 4 rural households depend on traditional sources of energy for cooking, heating, and so forth [60].
In Afghanistan, usage of electricity generators, biomass burning, and vehicular traffic are important sources of NO 2 .The use of portable generators during power outages is a major source of NO 2 in the country.In Kabul alone, there are about 173,755 diesel and gasoline power generators in which 99.5% are used by households.Animal dung is also used in 85% of rural homes and in about 15% of urban homes for heating and cooking.The transport sector faces challenges of illegal import of used vehicles, continued use of very old and poorly maintained vehicles, poor quality of transport fuel, and limited road capacity leading to high emissions of air pollutants [61].
In Bangladesh, NO 2 is mainly emitted from energy transformation industries, motor vehicles, biomass burning for industrial processing and home cooking, burning of agricultural residues, and iron and steel industries.In Bhutan, the sector-wise emissions estimates of NO 2 indicate that domestic sources and vehicles are responsible for NO 2 emissions.As per Male Declaration-2000 [62], the sources of NO 2 in Sri Lanka include transport (46.8%), domestic use (37.1%), power generation (13.7%), industry (2.3%), and fuel conversion (0.1%).Bhutan is one of the few countries in the world where the environment is still protected largely due to its vast forest cover and widespread use of hydropower and biomass energy.Forest fires are the biggest sources of air pollution in Bhutan [63].In Maldives, road traffic and domestic combustions are mostly responsible for NO x air pollution.In Nepal, major NO x sources are associated with the combustion of fossil fuels in industries, especially in the cement industry.
The estimated lifetime of NO 2 in the Planetary Boundary Layer (PBL) is 18-27 hours with considerable diurnal and seasonal variations [7,64,65].However, in the middle and upper troposphere, NO 2 lifetime varies from several days to a week due to decreased OH and aerosol concentrations [6,66,67].The NO 2 lifetime depends on meteorological conditions, its photolysis rate, surface emissions, length of day and night, aerosols abundance, and OH and H 2 O concentrations [18,68,69].The meteorological conditions such as wind speed, temperature, humidity, and solar radiations flux affect NO 2 concentration via removal, transformation, and transport processes [68].The hot and humid atmosphere enhances the removal of NO 2 through photolysis [18].Also low wind speeds reduce NO 2 transport and its vertical air mixing thus elevating NO 2 concentration near emission sources [69].
The monthly pattern shows NO 2 maximum in spring season with a primary peak in March 1.22 × 10 15 molec/cm 2 and secondary peak in May 1.13 × 10 15 molec/cm 2 .March peak is mainly attributed to low humidity, low wind speeds, and mild temperatures causing reduction in photolysis removal process of NO 2 hence stabilizing its concentration.Because of these factors, the NO 2 stability dominates the NO 2 removal through photolysis process of NO 2 due to the availability of more solar radiations in this month.In March, the lower concentration of OH (if the water vapor concentration is low enough) is considered a limiting factor of NO 2 photolysis to form HNO 3 , the principal sink for NO 2 (e.g., [5]).
Relatively high values in March-May and October-November are also associated with emissions of NO 2 from large scale open field crop residue burning (nearly 7-10 tons of crop waste per hector) in the study area during wheat-rice rotation periods [12,[70][71][72].In the study region, the main area for crop residue burning is IGB consisting of Lahore, Dera Ghazi Khan, Narowal, Hafizabad, and Faisalabad in Pakistan and states of Uttar Pradesh, Punjab, and Haryana in India [13,53,73].
A winter season high with a peak in December 1.07 × 10 15 molec/cm 2 , is due to meteorology (weak winds and dry weather conditions), heavy usage of biomass fuel for wintertime home heating especially in the northern areas, and less UV radiations available for the initialization of photolysis reactions that break down NO 2 [8,20,74,75].In winter, shallower boundary layer results in lower vertical dispersion which reduces the dilution and removal rates of NO 2 .This may also contribute to NO 2 enhancement in wintertime [76].
Low NO 2 during wet summer, with a notable dip in August 0.80 × 10 15 molec/cm 2 , is linked to massive advection of moist clean air mass, increased actinic fluxes enhancing the photodissociation of NO 2 , elevated levels of OH radical helping NO 2 removal from the atmosphere via HNO 3 , and presumably less traffic activity due to reduced social and educational activities during very hot summer [16,20,77,78].During the rainy season, effect of lightning on the increase of NO 2 concentration is not clearly seen due to the opposing influence of rain washout discussed by Yoo et al. [78].The washout of the SO 2 and NO 2 by rainfall has been a global and regional concern, since it plays an important role in producing acidic precipitation [78].In the atmosphere, NO 2 reacts with water to form nitric or nitrous acid [79].A number of previous studies have demonstrated significant negative correlation between NO 2 and rainfall (e.g., [77,80,81]).Martin [80] showed the washout coefficient for NO 2 is about 80% of that for SO 2 .However, the washout effects on the NO 2 and SO 2 concentrations by daily cumulative rainfall are comparable over India, resulting in the reduction (40-45%) of these pollutants [77].Yoo et al. [78] demonstrated the scavenging of air pollutants (CO, NO 2 , SO 2 , and PM 10 ) by summertime precipitation based on the three washout effect indicators such as Absolute Washout Index (AWI), Negative Correlation Fraction (NCF), and Relative Washout Index (RWI).They showed that the washout effect is in the descending magnitude order of PM 10 > SO 2 > NO 2 > CO > O 3 .
We discuss two crop residue burning events of postmonsoon 2010 [82] and postmonsoon 2012 [83].South Asia is the principal niche of rice-wheat system that occupies a total of 13.5 million ha in rice-wheat consortium countries such as India, Pakistan, Bangladesh, and Nepal with rice cultivation areas of 10, 2.2, 0.8, and 0.5 million ha, respectively [57,84].In this regard, South Asian farmers need to manage 5-7 t ha −1 of rice residues and overcome the problems for planting wheat.There are various options for crop waste management out of which the burning of crop residue is one of the most common practices.
Elevated NO 2 concentrations have been found over northwestern parts of IGB including Punjab, Haryana, and western Uttar Pradesh regions as a direct consequence of the crop residue burning emissions (Figures 5(a)-5(c)).In Figure 4, NO 2 enhancements (10-42% on average) can be seen over the rice residue burning areas during the postmonsoon period of 2012, compared with the average value of postmonsoon periods during study years except 2010 and 2012, especially in Pakistani and Indian Punjab and their adjoining territories famous for rice cultivation.For the postmonsoon 2012, we find 15-35% increment in NO 2 levels showing close agreement with the results of Kaskaoutis et al. [83].In a previous study, Kaskaoutis et al. [83] examined the impact of paddy crop residue burning over northern India during the postmonsoon (October-November) season of 2012.They showed about 34-40% increase in NO 2 levels as a direct consequence of these crop residue burning events.

Spatial Distribution of NO 2 .
In Figure 6, an enhanced value of NO 2 observed over IGB is mainly coupled with high human settlement, coal based thermal power plants, industrialization and urbanization, more agricultural activity, large scale crop residue, and biomass mass burning previously reported by many studies (e.g., [3,12,19,71,85,86]).It is evident from the figure that IGB section consisting of Punjab (from India and Pakistan) and eastern region of India show consistent high values due to high pupolation density, crop residue and biomass buning events, power plants, and mining activities.In addition, low population and rural areas scattered all over the region also contribute to NO 2 emissions from domestic cooking, small and medium industries, transport, and open burning of litter and biofuels.NO 2 masking is also observed over marine areas that is associated with NO 2 emissions due to seaport activities and urban pollution.
Our analysis shows significant decadal increasing trend of 17% (slope of 0.02428 ± 0.00657 and  intercept at 1.24751 ± 0.03695 × 10 15 molec/cm 2 ) with average of 1.36 ± 0.01 × 10 15 molec/cm 2 over a region between eastern and western Ghats (12.57-17.58∘ N to 76.05-79.00∘ E).This finding differs with the results by Renuka et al. [19] who reported a decreasing trend over the region between the two Ghats.The disparity in the results may be attributed to the difference in spatial and temporal domains.OMI data reveal that NO 2 columns are mounting over all the hotspots and most of the selected cities.However, the difference in increasing rates is mainly attributed to levels of the industrial activity, traffic volume, urban and rural background, and local meteorological conditions.Dhaka is the fastest growing city among all the hotspots and selected cities [87] linked to the highest increase (77%) in NO 2 column with an average of 3.15 ± 0.51 × 10 15 molec/cm 2 .After Dhaka, we find high growth rates for Islamabad 69%, Kabul 68%, Korba 64%, Bardhaman 47%, and Lahore 40%.On the other hand, DG Khan is found with negative trend at −11%.The average value of NO 2 column over Singrauli is the highest 3.71 ± 0.27 × 10 15 molec/cm 2 followed by Bardhaman 3.23 ± 0.44×10 15 molec/cm 2 and Korba 3.23±0.44×10 15molec/cm 2 , whereas Karachi has exhibited the lowest average value of 2.00 ± 0.10 × 10 15 molec/cm 2 with decadal increase at 6% (Table 1).

Advances in Meteorology
Figures 8 and 9 show the seasonality and intercomparison of the hotpots in the region.The megacities of Lahore and Delhi are located in IGB and experience almost the same meteorological conditions (Figure 2), large crop residue burning in surrounding areas, and extremity of monsoon which result in similarity in NO 2 seasonal patterns showing  2 = 0.865 (Figures 7 and 8).Over Lahore and Delhi, higher NO 2 values are found during the winter months followed by spring season (Figure 6).Winter peak is mainly due to biomass burning for domestic heating [20], stable winds, less daily sun hours and sun fraction, and low temperature (Figure 2).In May, a surge in NO 2 over Lahore and Delhi is linked to large scale crop residue burning from wheat fields in the neighboring rural areas of Kasur, Shiekhupura, Narowal, Hafizabad, Mianchannu, and Uttar Pradesh regions.In this month, high temperature also contributes to soil emissions which are further enhanced by the application of fertilizers in the rice fields.The low NO 2 column over these cities during the period of monsoon season is mostly coupled with heavy rains, high humidity, and strong winds (Figure 2).Some important anthropogenic emission sources of NO 2 in Lahore are industrial zones (e.g., Sundar industrial estate, Kot Lakhpat industrial area, Shahdara tractors manufacturing units, steel mills, and refrigerator manufacturing units), power plants (SEPCO, Kohinoor, Japan Power, and Nishat) and large scale crop residue burning events.Also a notable source of high NO 2 , over Lahore, is the usage of diesel and petrol electricity generators during the heavy electric power  load shedding (average 10-16 hours daily in city area) [53].In Delhi, power plants (Badarpur, Rajghat, Pragati, and IPGCL), industrial estates (Badli, Wazirpur, Mangolpuri, Lawrance, Jhilmil Industrial Area, Patparganj, etc.), and biomass and crop residue burning are the notable contributors of NO 2 emissions.We find higher (17%) annual average value of tropospheric NO 2 3.22 ± 0.31 × 10 15 molec/cm 2 than the previously reported value 2.74 ± 0.79 × 10 15 molec/cm 2 by Ul-Haq et al. [20] over Lahore for 2004-2008 using OMI data.The discrepancy between these two reported values is mainly due to the difference in time periods and increasing growth of anthropogenic emissions in the past few years.
The megacities of Karachi and Mumbai are considered as financial hubs of Pakistan and India, respectively.These cities are located in the coastal belt of ArS and show large similarity ( 2 = 0.800) in NO 2 column values, trends, and temporal variability (Table 1, Figure 9) mainly associated to matching meteorological and urban conditions.Karachi showed the lowest average value 2.00 ± 0.10 × 10 15 molec/cm 2 with the lowest increase at 6% followed by Mumbai 2.14 ± 0.08 × 10 15 molec/cm 2 with 7% increase during the study period.Ul-Haq et al. [20] reported higher 3.63 ± 2.19 × 10 15 molec/cm 2 value of NO 2 over Karachi during 2004-2008 than our computed value.This decrease in NO 2 concentration over Karachi may be associated with the political instability and economic downturn in recent years.In Karachi and Mumbai, more than 80% of total NO 2 emissions come from industrial sector and fugitive emissions from oil and gas activities [18,20].It is clear from Figure 8 that high NO 2 occurs in winter over Karachi and Mumbai.These cities have poor conditions for NO 2 removal, transformation, and transport processes (low rainfall, low wind speed, and low ambient temperature) in winter months.Other contributing factor for high NO 2 may be NO 2 carrying wind (NW-NE) from Indian landmass areas amplified by local emissions.On the other hand, low values are observed in summer linked to more rain wash-out and high humidity and strong and clean sea breeze (Figure 2) from ArS.Major sectors of NO 2 emissions in Karachi are power plants (Bin Qasim I&II, Gul Ahmed, and Tapal), textile industry, steel and iron factories, chemical industry, oil refineries, and seaport activities.The core industrial area of major emissions in Mumbai and surrounding areas is the Mumbai-Pune belt that includes textiles, chemicals, engineering, electrical, drugs, transport equipment, plastic and synthetic goods, leather goods, and ship-building.
We find high correlation ( 2 = 0.864) between the NO 2 seasonality for Dhaka and Kolkata.NO 2 is found high over these cities throughout the dry season (October-March) and low for the rest of the year (April-September).Gurjar et al. [14] have reported that the ambient NO 2 level in Dhaka is 83 g/m 3 which exceeds the World Health Organization (WHO) guideline concentration (40 g/m 3 ).Annual 7-16%  increase of vehicles has been observed for the last 10 years that has worsened the air quality as vehicular emissions are the single largest contributor of NO 2 in Dhaka [10,88,89].The ambient NO 2 in Kolkata is observed to be 37 g/m 3 [14] with the major emission sources of moderate to heavy vehicular traffic, power generation plants, rapid and unplanned urbanization and industrialization, and railway yard [11].In Dhaka, a substantial part of total traffic is nonmotorized vehicles that produce severe congestion and pollution problem especially in road intersections.Other notable sectors which contribute to high NO 2 emissions are brick kilns, urban residential combustion, gas processing and refineries, and energy and manufacturing industries [90].
The highest correlation ( 2 = 0.980) is observed for Pakur and Bardharman and surrounding rural areas.These areas have very similar topography, low population, and meteorological conditions in agreement with Ramachandran et al. [18].These hotspots have major sources of emissions as steel plants, major coal fields (Raniganj and Jharia), and crop residue burning in the adjoining territory.Singurali, Korba, and Angul hotspots are located in the eastern mining region (Singurali coalfield, Talchar coalfield, and Korba coalfields) of India [18].Most of the mining processes and associated industries exist along with many of the major power plants and refineries in this region and cause high emissions and spread of NO 2 in this area [18,85,91,92].We have also found a similar seasonality ( 2 = 0.908) between NO 2 column over Mumbai and Kolkata as well, because both cities lie on almost similar latitude range and have tropical climate.Lower NO 2 values are observed over Mumbai compared to Kolkata primarily linked to strong winds (Table 1).
From Figure 7, it may be noted that there exists some difference between the NO 2 seasonality between the hotspots and South Asia.The NO 2 column over hotspots generally shows peak in December, while over South Asia it exhibits its elevated level in March.This is because of the fact that anthropogenic emissions dominate NO 2 seasonality over hotspots.On the other hand, March peak in South Asia, as a whole, is due to the dominance of natural factors discussed in Section 4.1.

Conclusions
OMI measurements over South Asia have been used to analyze spatial and temporal variability of tropospheric NO 2 column during October 2004 to January 2015.NO 2 emission hotspots and some important cities have been discussed for trends and seasonal cycles.An average value of 1.0 ± 0.05 × 10 15 molec/cm 2 with 14% decadal increase has been reported over the study region.This positive trend is linked to the increasing anthropogenic emissions from power generation, urbanization, and vehicular and industrial sectors.Significant increase has been observed in NO 2 concentrations over some parts of IGB connected to large scale postmonsoon crop residue events of 2010 and 2012.Strong seasonality of NO 2 concentration is observed with the highest value in March and the lowest in August.Statistically significant NO 2 hotspots have been identified and discussed.OMI data reveal that the NO 2 columns are mounting at considerable rates over all the hotspot locations and most of the major cities.Dhaka (Bangladesh) showed the highest decadal increase, whereas Karachi (Pakistan) has exhibited the lowest average value of 2.00 ± 0.10 × 10 15 molec/cm 2 with the lowest decadal increase at 6%.The highest average value is observed over Singrauli (India).

Figure 1 :
Figure 1: Geographical map of South Asia identifying locations of the NO 2 hotspots (image source: Google Earth).

Figure 9 :
Figure 9: Correlation matrix of eleven NO 2 hotspots and study region with best fit curve.Pearson correlations values  2 are also shown.All correlations are significant at the 0.01 level (2-tailed) except correlation between Karachi and the study area indicated by * which is significant at the 0.05 level (2-tailed).