Detection of Trends and Seasonal Variation in Tropospheric Nitrogen Dioxide over Pakistan

In this study, spatial and temporal distributions of tropospheric NO2 vertical column densities over Pakistan during the time period of 2002–2012 are discussed. Data products from the satellite instrument SCIAMACHY are used. The results show a large NO2 growth over major cities of Pakistan, particularly the areas with rapid urbanization. Different seasonal cycles were observed in different regions of Pakistan. In the provinces of Punjab (north east), Khyber Paktunkhwa (north west) and Sindh (south east), NO2 columns are maximum in winter and minimum in summer months while a reversed seasonality was observed in the province of Baluchistan (south west). We compared the observed spatio-temporal patterns to existing emission inventories and found that for the most populated provinces the NOx emissions are clearly dominated by anthropogenic sources. In these areas also the strongest positive trends were observed. NOx released from soils and produced by lightning both together contribute about 20% for the provinces of Punjab, Sindh, and Khyber Paktunkhwa, while its contribution in Baluchistan is much stronger (~50%). NOx emissions from biomass burning are negligible. This finding can also explain the observed summer maximum in Baluchistan, since the highet lightning activity occurs during the Monsoon season. Our comparison also indicates that the inventories of anthropogenic NOx emissions over Pakistan seem to underestimate the true emissions by about a factor of two.


INTRODUCTION
Globally, Nitrogen oxides (NO x = NO + NO 2 ) are released through combustion of fossil fuel, biomass burning, lightning and microbiological processes in the soil (Lee et al., 1997;HSDB, 2007;Seinfeld and Pandis, 2008).On global average about 50% of NO x is released from industry and traffic and 20% from biomass burning (Oliver et al., 1998;Seinfeld and Pandis, 2008).Among natural emissions, approximately 10% are through lightning while 15% are from soil due to microbial activities (Lee et al., 1997;Seinfeld and Pandis, 2008).Here it is interesting to note that over Pakistan the relative contributions of the different NO x sources are different from the global average with a much stronger contribution of anthropogenic sources (see section 'Comparison with NO x emission inventories').
Subject to the different meteorological conditions, seasons, actinic flux and concentrations of the hydroxyl (OH) radical, the lifetime of NO x ranges from a few hours to 1 day (e.g., Crutzen, 1979;Beirle et al., 2003;Platt and Stutz, 2004;Seinfeld and Pandis, 2008).NO 2 is a strong oxidant and has been listed as a criteria pollutant (Seinfeld and Pandis, 2008).It plays a crucial role in tropospheric chemistry mainly involved in the formation of secondary air pollutants (photochemical smog) and acid rain (Logan et al., 1981;Thompson, 1992;Seinfeld and Pandis, 2008).
There has been a 5-fold increase in global NO x emissions since the pre-industrial era (Seinfeld and Pandis, 2008) and in recent years the most rapid increase has been observed in Asia at the rate of 4-6% per year (e.g., Garg et al., 2001;van Aardenne et al., 1999;Gude et al., 2009).Air pollution has emerged as a potential threat for the environment and human health in Pakistan (e.g., Ahmed et. al, 2011).Especially, all major cities of Pakistan are exposed to severe problems of air pollution and are rarely addressed by air quality regulatory authorities (Khattak et al., 2014;Shabbir et al., 2015).Road transport has been increased at a very fast pace in Pakistan during the last decade.According to the National Transport Research Center (NTRC), total number of registered vehicle increased by 52% from 1991-1992 to 2006-2007 (Sindhu, 2008;ESoP, 2013).This continuous increase is leading to severe problems like frequent traffic jams and consequently increased levels of toxic air pollutants.Although there have been some studies presenting the global distribution and temporal trends of atmospheric NO 2 , this is the first study with detailed information about the spatial and temporal variation of NO 2 over different regions of Pakistan.It is worth to mention that different seasonal cycles were identified over different regions as discussed in section (3.5).Recent reports form (WHO, 2014) stated that 16 out of top 20 polluted cities are located in the Indian subcontinent (WHO, 2014) and among them 3 are from Pakistan.
Although air pollution levels in Pakistan (the study area is shown in Fig. 1(a)) is high in many regions, no solid information about NO 2 emissions/sinks and background levels is so far available.However, very few and limited ground observations are conducted mainly in the metropolitan centres of the country (for details see Shabbir et al., 2015).
A study carried out by Ahmed et al. (2011) stated that NO 2 concentrations were above the World Health Organization (WHO) guidelines at all sampling locations within the cities of Islamabad and Rawalpindi.Therefore, the levels of NO 2 should be monitored on a regular basis to design effective policies in order to abate air pollution problems.The measurements offered by satellites have worldwide coverage and their spatial and temporal resolution can meet such needs, especially, a country like Pakistan with no proper/regular air quality monitoring network (ESoP, 2013).Global Ozone Monitoring Experiment-1 (GOME-1 - Burrows et al., 1999;Leue et al., 2001;Richter and Burrows, 2002;Beirle et al., 2003;Wenig et al., 2004) onboard European remote sensing satellite (ERS-1, 1995(ERS-1, -2002)), the Scanning Imaging Absorption spectroMeter for Atmospheric CartograpHY (SCIAMACHY, Bovensmann et al., 1999;Boersma et al., 2004) onboard ENIVISAT-1 (2002ENIVISAT-1 ( -2012) ) the Ozone Monitoring Instrument (OMI - Levelt et al., 2006) onboard AURA (since 2004) and GOME-2 (Callies et al., 2000) onboard MetOP-A (since 2007) have been used since two decades to monitor the atmospheric composition on global scales.Asia has been the major focus of many studies as the air pollutants show greatest variability in this region (e.g., Richter et al., 2005;van der A et al., 2006, 2008;Han et al., 2005;Ghude et al., 2009;Sheel et al., 2010).For instance, Van der A et al. (2008) observed an outflow of NO 2 over the Atlantic and Pacific oceans caused by anthropogenic activities in the eastern coasts of North America and China, respectively.Richter et al. (2005) showed a rapid increase in tropospheric NO 2 columns over China from 1996 to 2005 using GOME and SCIAMACHY observations, caused by extensive industrial and vehicular emissions.Hilboll et al. (2013) and Schneider and van der A (2012) measured NO 2 trends over mega cities around the globe.
In this study, NO 2 tropospheric columns retrieved from SCIAMACHY (Bovensmann et al., 1999) observations by using the Differential Optical Absorption Spectroscopy (DOAS) (Perner and Platt, 1979;Wagner et al., 2002;Beirle et al., 2003;Richter et al., 2005;Platt and Stutz, 2008) technique are used to explore the temporal and spatial variations over Pakistan.Regression analysis was performed to investigate any existing temporal trend and seasonal  The spectral range is between 240 and 2400 nm.The spectral resolution is between 0.2 and 1.5 nm due to which a large variety of trace gases including greenhouse gases can be retrieved (Bovensmann et al., 1999).
Tropospheric slant column densities (SCDs) were determined from a spectral fit (426.3-451.3nm -channel 3, cluster 15) to the back scattered solar irradiance by using the DOAS (Platt and Stutz, 2008) approach, and from assimilating slant columns into a chemistry transport model (CTM) including stratospheric chemistry and wind fields (van der A et al., 2010).The tropospheric SCDs were converted into VCDs by the application of a tropospheric air mass factor (AMF).The calculation of AMF is dependent on various factors such as apriori profile of the trace absorber, cloud fraction, viewing geometry, surface albedo etc.All these parameters affect the accuracy of the retrieval process and thus contribute to the retrieval error.The error estimation during various steps of the VCDs retrieval is presented in Table 1.Further details about the error analysis and retrieval algorithm is available from Boersma et al. (2007).The final product of tropospheric NO 2 VCDs was used in this study for further analysis over Pakistan.
Monthly tropospheric NO 2 VCDs, averaged on a 0.25° × 0.25° grid (in total 1521 grid cells) across Pakistan, were downloaded from the TEMIS website for the years 2002-2012.These data sets were used to create raster images by using ERDAS (Earth Resources Data Analysis System) Imagine software (see, http://geospatial.intergraph.com).The raster images were geo-referenced in ArcGIS (http://www.esri.com/software/arcgis/)software by assigning WGS 1984 coordinate system.A Pakistan shape file (see Fig. 1(a)) was used to extract tropospheric NO 2 columns from the global data.Spatial maps of NO 2 columns over Pakistan were created in order to identify the distribution and variation of NO 2 over major cities during the study period.A database of monthly and annual maps was prepared for tropospheric NO 2 columns over Pakistan.

Tropospheric NO 2 Database over Pakistan
A comparison of yearly mean maps of tropospheric NO 2 from SCIAMACHY observations for the years 2003 and 2010 is presented in Fig. 1.In general, tropospheric NO 2 columns have increased significantly over most regions of Pakistan.Especially, maximum NO 2 column amounts are observed over regions with extensive anthropogenic activities and population (Punjab and Sindh provinces) while regions (Baluchistan and Khyber Paktunkhwa) with less population exhibited smaller amounts.Black circles in Fig. 1 are representing the major cities in each district of Pakistan.
It is worth to mention that enhanced NO 2 columns are not confined to the locations of major cities but rather smeared out over larger areas.This is mainly caused by the lifetime of NO x of a few hours, meteorological parameters such as wind direction and speed along with other factors such as aerosol load, surface altitude, surface spectral reflectance Table 1.Features of SCIAMACHY instrument and retrieval error for individual, (mostly) cloud-free pixels (cloud radiance fraction < 50%) Adopted from Boersma et al. (2007) (Blond et al., 2007) σ S = 0.47 × 10 15 molec.cm -2 σ Sst = 0.25 × 10 15 molec.cm -2 σ Mtr = 10-40% σ S is the uncertainty on the slant column, σ Sst the uncertainty on the stratospheric slant column, σ Mtr the uncertainty on the tropospheric AMF. and coarse spatial resolution (30 × 60 km 2 ).These factors lead to such structure (smeared out) of retrieved NO 2 quantities (Hilboll et al., 2013).

Spatial and Temporal Distribution of NO 2 Columns over Pakistan
The temporal evolution of tropospheric NO 2 VCDs monitored by the SCIAMACHY instrument over Pakistan is presented in Fig. 2. A clear seasonal pattern and temporal increase in tropospheric NO 2 columns can be easily identified.Especially, NO 2 peaks in winter while lowers during the summer months.
Standard regression analysis was also performed in order to identify the relative temporal increase in tropospheric NO 2 columns with respect to the base year 2003.An overall statistically significant temporal increase of 46% was observed during the selected time period.The overall absolute increase in SCIAMACHY data was 0.75 ± 0.25 × 10 15 molecules cm -2 with an annual increase at the rate of 0.83 ± 0.02 × 10 14 molecules cm -2 per year.
The temporal analysis indicated a different yearly growth in the levels of tropospheric NO 2 columns over different regions of Pakistan.Fig. 3 shows the multi-year mean map of SCIAMACHY observations from October 2002 to March 2012.For details, the temporal evolution of NO 2 over four provinces (Brown line as provincial boundary); Punjab, Sindh, Khyber Pakhtunkhwa (KPK) and Baluchistan was studied separately.All of these provinces have different characteristics regarding population density, land area and anthropogenic activities.Therefore, they show different levels of increase in tropospheric NO 2 levels during the selected time period as shown by bar charts in Fig. 3. Total relative increase observed in tropospheric NO 2 columns was in the following order: Punjab (58% at rate of 6.4% per year with absolute increase of 12.74 × 10 13 molecules cm -2 year -1 ), KPK (56% at rate of 6% per year with absolute increase of 6.86 × 10 13 molecule cm -2 year -1 ), Baluchistan (11% at rate of 1.2% per year with absolute increase of 0.32 × 10 13 molecule cm -2 year -1 ) and Sindh (7% at rate of 0.78% per year with absolute increase of 0.64 × 10 13 molecule cm -2 year -1 ).Spatially resolved relative and absolute temporal trends in tropospheric NO 2 columns over major cities of Pakistan derived from SCIAMACHY observations are presented in Fig. 4. White colour shows the grid cells with statistically non-significant trends (below the 95% confidence level).Trends were calculated relative to the base year 2003 by performing Seasonal Mann-Kendal Test (SMKT - Gilbert, 1987).It is a non-parametric test which is robust against seasonality, missing values, non-normality and both interand intra-annual autocorrelations (Hirsch and Slack, 1984).
The results indicate the spatial patterns of i.e. larger trend (up to 3.75% per year) over KPK and Punjab regions while smaller trends (about 1% per year) over southern and western Pakistan regions.
Temporal analysis of tropospheric NO 2 columns over major cities of Pakistan during the time period of 2003-2011 is presented in Fig. 5. NO 2 VCDs for only one grid cell (0.25° × 0.25°) having co-ordinates exactly/close to the geographical co-ordinates of the respective city was selected to represent a city in this study.The values were extracted by using Pakistan's major cities shape file (black circles in Figures) in ArcGIS.The largest increase is observed over the city of Rawalpindi/Islamabad (NO 2 columns are for both cities as satellite pixels are not adequate/small enough to discriminate adjacent twin cities) followed by Lahore, Okara, Peshawar, Karachi and so on.These major cities have high population density and industrial activities thereby exhibited enhanced levels of tropospheric NO 2 columns.Several studies around the world have related enhanced levels of several trace gases with socio-economic activities like industries and fossil fuel burning associated with power and transport sectors (e.g., Garg et al., 2001;Richter and Burrows, 2002;Ricther et al., 2005;Badarinath et al., 2006;Kunhikrishnan et al., 2006;Ghude et al., 2009;Cheng et al., 2012;Schneider and van der A, 2012;Hilboll et al., 2013;Schneider and van der A, 2013).Table 2 enlists  were available for less than 12 months.Seasonal Mann-Kendal test was applied and only the statistically significant pixels (with confidence interval of 95%) were selected.relatively clean air to the city in addition to high humidity.This leads to high OH levels, short lifetime of the NO 2 and therefore resulting in quantitatively lower mean NO 2 VCDs over Karachi.In case of Islamabad/Rawalpindi and Lahore cities, larger mean NO 2 VCDs might be influenced by transboundary NO 2 pollution from neighbouring regions of Afghanistan, Iran and India (Khattak et al., 2014).A recent study Shabbir et al. (2015) by using car MAX-DOAS instrument has indicated that NO 2 levels in Lahore and Rawalpindi cities are exceeding the Pak-NEQS limits of 42.5 ppb.

Possible Influence of Other Quantities on the Observed NO 2 Trends
In the last section it was shown that a strong temporal increase in NO 2 columns over many regions and cities of Pakistan is found during the study period.While these trends are most probably directly related to the increase of anthopogenic actitivities, also other factors influencing the NO 2 levels might have changed during the study period.Global warming caused by climate change is evident worldwide (IPCC, 2013).Haider (2015) investigated the ambient temperature over Pakistan and found an absolute increase of 0.07°C in the mean temperature over Pakistan during the time period of 1951-2010.The increase is especially strong (0.7°C) during the last fourteen years (1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010).The increase in ambient temperature may lead to enhanced levels of OH production through photolysis of water vapours (HO x cycle -Seinfeld and Pandis, 2008), since with increasing temperatures also the amount of water vapour in the atmosphere increases (evaporation and evapotranspiration).In addition, Noreen (2015) indicated a temporal increase of 3 percent in tropospheric ozone columns over Pakistan during the time period of 2004-2014.Therefore, we might expect an overall increase in OH production over Pakistan which might have led to enhanced removal rates of NO 2 over Pakistan.Thus our observed trends in NO 2 might even be underestimated and can be considered as a lower bound of the corresponding increase of the NO x emissions.

Seasonal Variation
NO 2 has a lifetime of a few hours to about one day which depends upon different factors like the photolysis rate, OH concentration and the different metrological conditions (Beirle et al., 2003;Richter et al., 2005;Seinfeld and Pandis, 2008;Sheel et al., 2010).Reaction of NO 2 with OH is by far the largest NO x sink (Stavrakou et al., 2013 and references therein).
Because of less solar flux and low temperature in the winter season, the loss of NO 2 through reaction with the OH radical is smaller leading to an increased NO 2 lifetime and concentrations.Besides temperature, the precipitation rates are also low during winter due to which less amount of water vapours is available for the production of OH (Sheel et al., 2010); leading to relatively small removal of NO 2 molecules.During winter season increased rates of wood, cow dung and coal burning for space heating and cooking results in higher NO 2 emissions.As NO 2 has a longer lifetime and increased emissions in winter season, maximum amount of tropospheric NO 2 is found in winter (Martin et al., 2009) thus introducing a seasonality in tropospheric NO 2 concentrations.Fig. 6 shows the seasonal variation of climatologically averaged NO 2 tropospheric columns over Fig. 6.Exhibited seasonal cycle in tropospheric NO 2 VCDs over major provinces of Pakistan Error bars give the spatial variance of NO 2 VCDs over the region.Grey rectangles are indicating the monsoon period of the year.Also, total amount of precipitation and during the monsoon period is given for each province (province are separated by the brown line in Fig. 4 as province boundary).The precipitation data is taken from Faisal and Gaffar (2012).
the four provinces of Pakistan (Punjab, Sindh, Baluchistan and Khyber Paktunkhwa).NO 2 columns exhibited almost similar seasonal patterns in Punjab, Sindh and KPK provinces; maximum in winter and minimum in summer and monsoon season.Grey rectangles in the figures indicate the monsoon period in Pakistan (it might be different in these regions with an offset of one to two weeks and also over the years).Therefore, the summer minimum exhibited in the observed seasonal cycle is mainly due to lower anthropogenic NO x emissions and higher OH concentrations causing a shorter NO 2 lifetime.Reduced NO 2 concentrations during monsoon season are attributed to advection of moist clean air masses, heavy precipitation (290 mm of rain in KPK), strong actinic fluxes, and enhanced levels of OH radicals (Wang and Jacob, 1998;Jacob, 2003;Gude et al., 2009).
In the province of Baluchistan, a different pattern is observed pointing to non-anthropogenic NO x sources dominating, i.e., lightning and soil emissions.

Comparison with NO x Emission Inventories
Fig. 7 presents information on the strength of different NO x emission sources over Pakistan according to emission inventories.In Fig. 7(a) the spatial distribution of anthropogenic NO x emissions is shown.When compared to the spatial pattern of the tropospheric NO 2 VCD derived from SCIAMACHY observations (Figs. 1 and 3) it is obvious that anthropogenic emissions are responsible for the major part of the observed tropospheric NO 2 VCD derived from satellite.The comparison to the patterns shown in Fig. 4 also proves that the highest trends occur in the regions of the strongest anthropogenic NO x emissions.This finding indicates that the increase of anthropogenic emissions is mainly responsible for the observed positive trends.Fig. 7(b) displays the seasonal NO x emissions from lightning (upper panel) and from soil emissions (lower panel) for the investigated provinces.In contrast to the anthropogenic NO x emissions, the highest NO x production by both lightning and soils occurs in the summer months.Fig. 7(c) presents the relative contributions (yearly averages) of the different NO x emission sources for the four provinces.Anthropogenic emissions dominate the total NO x emissions in all provinces.Lightning and soil emissions typically contribute about 10% each for Punjab, Sindh, and KPK, but sum up to about 50% for Baluchistan.The contributions from biomass burning can be neglected.
Thus the summer peak of the NO 2 VCDs observed by satellite over Baluchistan is very probably caused by the seasonality of lightning NO x and soil emissions (see Fig. 7(b)).In the other provinces, where the relative contribution of lightning and soil NO x is much smaller, the observed seasonaliy is dominated by the anthropogenic emissions (with a maximum NO 2 VCD in winter).
We also roughly compared the satellite observations with the existing emission inventories in a quantitative way.In order to convert NO x emisssions into tropospheric NO 2 VCDs several assumptions have to be made: a) The NO 2 VCDs and the NO x emisssions have to be averaged over extended regions and periods (here monthly or seasonal averages over whole provinces are calculated).This assures that the effects of transport across the area boundaries are small compared to the chemical losses.b) An effective NO x lifetime has to be assumed, with which the NO x emissions are multiplied.Since we consider yearly averages, we assume a NO x lifetime of 6 hours for NO x in the boundary layer (from anthropogenic, biomass burning and soil emissions) and of 12 hours in the free troposphere (for NO x produced by lightning).c) Effective ratios of NO 2 to NO x have to be assumed to convert the NO x concentrations to NO 2 , since only NO 2 can be measured by the satellite instrument.Here we assume NO 2 to NO x ratios of 0.7 for the boundary layer and of 0.5 for the free troposphere (see Seinfeld and Pandis, 2008).All of the above assumptions lead to uncertainties in the conversion from NO x emissions to the corresponding corresponding NO 2 VCDs.We estimate the unceratinties related to transport to about 10%, those related to the lifetime and the NO 2 to NO x ratio to 30% each.
In Fig. 8 the yearly averaged tropospheric NO 2 VCDs forecasted from the emission inventories are compared to the corresponding satellite observations.In general the calculated yearly averaged tropospheric NO 2 VCDs are systematically lower than the satellite observations (by about a factor of 1.3 to 2) for all provinces.While at least a part of this underestimation might be caused by the uncertainties of our conversion of the NO x emissions into tropospheric NO 2 VCDs (see above) and also by the uncertainties of the satellite observations (see Table 1), the observed sytematic difference probably indicates that the existing emission inventories of anthropogenic NO x emissions might be too low.For example, a doubling of the anthropogenic emissions would bring the calculated tropospheric NO 2 VCDs into almost perfect agreement with the satellite observations.
Another intersting finding is that the contribution of anthropogenic emissions and NO x production by lightning and soil emissions differ largely between the different provinces (see also Fig. 7(b)).For Punjab, Sindh, and KPK, by far the largest fraction is caused by anthropogenic emissions, while for Baluchistan both sources (lightening and soil emissions) contribute about 50%.Again this indicates that for Baluchistan NO x production by lightning and soil emissions plays an important role, which leads to the maximum of the tropospheric NO 2 VCD occuring in summer (see Fig. 6).

Source Identification
NO 2 has both anthropogenic and natural sources.Among anthropogenic sources, vehicular and industrial emissions are dominant.Electricity generation and other industrial emissions also contribute to a large amount of NO 2 emissions.The boilers of power stations, industrial boilers, and reciprocating internal combustion engines are major contributors in addition to various industrial processes.Several studies around the world have related enhanced levels of various air pollutants with power and transport sectors (e.g., Garg et al., 2001;Ricther et al., 2005;Badarinath et al., 2006;Kunhikrishnan et al., 2006;Ghude et al., 2009;Cheng et al., 2012;Khattak et al., 2014).Punjab with a population of more than 91 million (out of 170 million of total Pakistan) is the most densely populated province of Pakistan followed by Sindh (42 million), KPK (22 million) and Baluchistan (8 million).The population has shown a drastic increase in the recent years (ESoP, 2013).In Pakistan, the numbers of vehicles are also highest in the Punjab region.NO 2 columns showed a strong correlation with the number of vehicles in Punjab over the selected time period 2003-2011 (Fig. 10).The number of vehicles showed a rapid increase in the province of Punjab from around 3 million in 2002 to more than 5 million in 2010.The significant positive co-relation between the number of vehicles and NO 2 tropospheric columns for the province of Punjab has been identified as shown in Fig. 9 with a coefficient of regression i.e., r 2 = 0.94.Wang et al. (2007) reported a decrease of 40% NO 2 columns over Beijing city after reducing 30% of vehicles during 1-6 November, 2006.Similar experiment was repeated by Beijing authorities during the 2008 Olympic and Paralympic Games in Beijing (from 8 August to 17 September) and a decrease in NO 2 concentrations of approximately 40-60% above Beijing area during the Olympic period was observed (Mijling et al., 2009;Wang et al., 2009).
In KPK and Sindh provinces, the NO 2 columns show increase with increase in number of vehicles, evident from the significant correlation shown in Fig. 10.In Baluchistan, no correlation between the changes of population/vehicle number and NO 2 was found.

CONCLUSIONS
The level-2 data products of SCIAMACHY instrument were used to prepare a tropospheric NO 2 database from 2002-2012 over Pakistan.Analyses were performed to identify temporal increase, seasonal cycles and source identification of tropospheric NO 2 columns across Pakistan.Enhanced NO 2 levels have been observed over the regions with industrial activities, and regions with higher vehicular and population densities.Punjab province exhibited the highest NO 2 tropospheric column densities due to high population, vehicular density and industrial activities.Baluchistan province with least population, vehicle numbers and industrial activities exhibited smaller tropospheric NO 2 columns, and is significantly affected by NO x emissions from lightning and soil emissions.The regional distribution of NO 2 columns over Pakistan exhibited a gradient of enhanced NO 2 columns on eastern side and lower columns on western parts of Pakistan.Similar East -West gradient exists in population density, vehicular density, agricultural land use pattern, road networks and industrial activities across Pakistan.Results indicated that satellite data can be used to identify not only the regional distribution but also the seasonal cycle of tropospheric NO 2 columns over different regions of Pakistan.In Punjab, Sindh and KPK; the seasonal cycle shows a maximum in winter and minimum in summer because of more anthropogenic emissions in these provinces.While an opposite pattern is observed in Baluchistan province because of a larger relative contribution of NO x produced by lightning and soil emissions.Summer time decrease in NO 2 VCDs over Pakistan is attributed to mainly high OH concentrations, less biomass/biofuel burning activities, larger photolysis rate due to higher actinic flux.We compared the observed spatio-temporal patterns to existing emission inventories.It is found that for most provinces (Punjab, Sindh, and KPK) NO x emissions are dominated by anthropogenic emissions (80-90%), while biomass burning emissions are negligible for all provinces.NO x produced by lightning and soil emissions together contribute about 20% for the provinces of Punjab, Sindh, and KPK, while their contribution in Baluchistan is much higher (~50%).This finding can explain in particular the observed seasonal cycle over Baluchistan, which shows a maximum in summer (in agreement with the highest lightning activity during the Monsoon season).We also converted the reported NO x emissions for the different provinces into tropospheric NO 2 VCDs and compared them to the corresponding satellite observations.We found that the calculated tropospheric NO 2 VCDs are significantly lower than the satellite observations for all provinces indicating that the inventories of anthtropogenic NO x emissions might underestimate the true emissions by about a factor of 1.3 to 2.Here it should, however, be noted that at least a part of the discrepancy might be related to the uncertainties of the satellite observations and the conversion of the NO x emissions into tropospheric NO 2 VCDs.SCIAMACHY observations have enabled to measure the increase of tropospheric NO 2 concentration over Pakistan during the time period of 2002-2012.SCIAMACHY data showed a relative increase of 46% (at rate of about 5% per year).Absolute increase was 0.75 ± 0.25 × 10 15 molecules cm -2 (at rate of 0.83 ± 0.02 × 10 14 per year).Similarly increasing trend in NO 2 columns over major cities of Pakistan was observed.Twin cities of Islamabad/Rawalpindi followed by Lahore, Peshawar, Okara, Faisalabad and Karachi showed the highest increase due to large amount of traffic and industrial activities during the study period.Karachi being the most populated city, exhibited quantitatively less increase in NO 2 columns as compared to cities of Islamabad/Rawalpindi and Lahore, mainly due to frequent sea breeze (that disperse the NO 2 quickly over larger area), high humidity and socio-economic factors (political instability, power outages etc.).However, temporal trend of 6% per year calculated in this study is consistent to trends calculated by other studies (e.g., Schneider and van der A, 2012;Hilboll et al., 2013).The relationship between increasing numbers of vehicles from different provinces showed that the NO 2 columns have also increased especially in Punjab, KPK and Sindh with exception from Baluchistan.Similarly, these provinces have industrial activities (such as thermal power generation, oil refineries etc.), extensive road networks and international airports.The observed trends were statistically significant.

Fig. 1 .
(a) exhibiting the study area map of Pakistan and its neighbouring countries.Thumbnail picture is showing the world map.(b) Annual mean maps of tropospheric NO 2 VCDs retrieved from SCIAMACHY observations for the years 2003 and 2010.Black circles are indicating the location of major cities of each district of Pakistan.

Fig. 2 .
Fig. 2. Temporal evolution of tropospheric NO 2 VCDs over Pakistan during the time period of November 2002-November 2011.Linear fit (dashed line) was applied in order to calculate the temporal evolution of tropospheric NO 2 columns from SCIAMACHY observations during the respective time period.Vertical bars are representing the standard deviation.

Fig. 3 .
Fig. 3. Shows a multi-year mean map of tropospheric NO 2 VCDs from SCIAMACHY observations during the time period of 10/2002-03/2012.The temporal evolution of annually averaged tropospheric NO 2 columns over four provinces of Pakistan during 2003-2011 is presented by bar charts.Year 2002 and 2012 were excluded because SCIAMACHY observations were less than 12 months in these years.
Fig. 4. Shows both relative and absolute spatially resolved temporal trends of tropospheric NO 2 columns over Pakistan calculated from SCIAMACHY observations during the period of 2003-2011.Base year 2003 was considered to calculate the relative trend.Years 2002 and 2012 were excluded because SCIAMACHY observationswere available for less than 12 months.Seasonal Mann-Kendal test was applied and only the statistically significant pixels (with confidence interval of 95%) were selected.

Fig. 5 .
Fig. 5. Temporal evolution of annually averaged tropospheric NO 2 columns over major cities (represents only one grid cell (0.25° × 0.25°) having co-ordinates exactly/close to the respective city co-ordinates) of Pakistan during 2003-2011.Year 2002 and 2012 were excluded because SCIAMACHY observations were less than 12 months.

Fig. 7 .
Fig. 7. (a) Map showing the mean anthropogenic emissions of NO x (in kg m -2 s -1 ) over Pakistan for year 2007.Data is taken from MACCity project, for details see Fig. 8.Comparison of yearly mean tropospheric NO 2 VCDs calculated from the emission inventories (for details see text) and the corresponding NO 2 VCDs derived from satellite observations for different provinces.

Fig. 10 .
Fig. 10.Comparison of annually averaged tropospheric NO 2 VCDs (brown line) retrieved from SCIAMACHY observations with population (green line) and vehicle (blue line) densities over four provinces of Pakistan during the time period of 2003-2011.Co-relation between NO 2 column densities and vehicular (blue dots), and population (green dots) densities in four provinces of Pakistan (Sources; Pakistan Bureau of Statistics and National Institute of Population Studies).
cycles of NO 2 over Pakistan.Efforts were made to identify the contribution from different sources (vehicle, population, land use etc.) in different regions/provinces of Pakistan. .

Table 2 .
Temporal Trend of NO 2 VCDs over major cities and Pakistan derived from SCIAMACHY observation during the period of 2003-2011.Schneider and van der A (2012), calculated trend over 1° × 1° grid cell and time period of 2002-2011; ** Hilboll et al. (2013), calculated trend for multi-instrumental data from 1996-2012.